diff --git a/PLAN_PHASE12.md b/PLAN_PHASE12.md deleted file mode 100644 index 1dd2836d..00000000 --- a/PLAN_PHASE12.md +++ /dev/null @@ -1,345 +0,0 @@ -# Phase 12: Format-Neutral Decomposition Plan - -## Overview - -This plan separates format-neutral logic from HTML-specific generation in renderer.py. The goal is to: -1. Create a `TemplateMessage` that stores logical attributes instead of CSS classes -2. Move HTML-specific rendering to a new `html_renderer.py` module -3. Keep format-neutral processing in `renderer.py` (to be renamed later) - -## Key Design Decisions - -### 1. Replace `css_class` with Typed Attributes - -Instead of encoding traits as space-separated CSS classes (e.g., `"user sidechain slash-command"`), we'll use explicit fields: - -```python -# In models.py - add MessageModifiers dataclass -@dataclass -class MessageModifiers: - """Semantic modifiers for message rendering.""" - is_sidechain: bool = False - is_slash_command: bool = False - is_command_output: bool = False - is_compacted: bool = False - is_error: bool = False - is_steering: bool = False - system_level: Optional[str] = None # "info", "warning", "error", "hook" -``` - -The `TemplateMessage` will have: -- `type: MessageType` (already have the enum) -- `modifiers: MessageModifiers` (new) -- Remove `css_class` field - -### 2. HTML Renderer Module (`html_renderer.py`) - -New module containing HTML-specific functions: - -```python -# html_renderer.py - -def css_class_from_message(msg: TemplateMessage) -> str: - """Generate CSS class string from message type and modifiers.""" - parts = [msg.type.value] - if msg.modifiers.is_sidechain: - parts.append("sidechain") - if msg.modifiers.is_slash_command: - parts.append("slash-command") - if msg.modifiers.is_command_output: - parts.append("command-output") - if msg.modifiers.is_compacted: - parts.append("compacted") - if msg.modifiers.is_error: - parts.append("error") - if msg.modifiers.is_steering: - parts.append("steering") - if msg.modifiers.system_level: - parts.append(f"system-{msg.modifiers.system_level}") - return " ".join(parts) - -def get_message_emoji(msg: TemplateMessage) -> str: - """Return emoji for message type.""" - # Move emoji logic from template to here - -def render_content_html(msg: TemplateMessage) -> str: - """Render message content to HTML.""" - # Delegates to format_* functions -``` - -### 3. Keep Format-Neutral Processing in renderer.py - -Functions that stay in renderer.py (format-neutral): -- `_process_messages_loop()` - but sets `modifiers` instead of `css_class` -- `_identify_message_pairs()` - pairing logic -- `_build_message_hierarchy()` - but uses `type` and `modifiers` instead of `css_class` -- `_reorder_paired_messages()` - reordering logic -- Deduplication logic -- Token aggregation - -### 4. Migration Strategy - -The migration will be done in phases to minimize disruption: - -**Phase 12a: Add MessageModifiers** -- Add `MessageModifiers` dataclass to `models.py` -- Add `modifiers` field to `TemplateMessage` -- Keep `css_class` field for backward compatibility - -**Phase 12b: Populate Modifiers** -- Update all TemplateMessage creation sites to set `modifiers` -- Replace `"x" in css_class` checks with `modifiers.is_x` - -**Phase 12c: Create html_renderer.py** -- Move `escape_html()`, `render_markdown()` to html_renderer.py -- Create `css_class_from_message()` function -- Move tool formatters to html_renderer.py - -**Phase 12d: Update Templates** -- Modify template to call `css_class_from_message(message)` -- Update emoji logic to use modifiers - -**Phase 12e: Remove css_class** -- Remove `css_class` parameter from TemplateMessage -- Clean up any remaining references - -## Detailed Implementation - -### Phase 12a: Add MessageModifiers (models.py) - -```python -from dataclasses import dataclass, field -from typing import Optional - -@dataclass -class MessageModifiers: - """Semantic modifiers that affect message display. - - These are format-neutral flags that renderers can use to determine - how to display a message. HTML renderer converts these to CSS classes, - text renderer might use them for indentation or formatting. - """ - is_sidechain: bool = False - is_slash_command: bool = False - is_command_output: bool = False - is_compacted: bool = False - is_error: bool = False - is_steering: bool = False - # System message level (mutually exclusive) - system_level: Optional[str] = None # "info", "warning", "error", "hook" -``` - -Add to TemplateMessage.__init__: -```python -def __init__( - self, - message_type: str, # Will become MessageType - content_html: str, - formatted_timestamp: str, - css_class: str, # Keep for now, will remove in 12e - modifiers: Optional[MessageModifiers] = None, # New - # ... other params -): - self.type = message_type - self.modifiers = modifiers or MessageModifiers() - # ... rest -``` - -### Phase 12b: Populate Modifiers - -Update each TemplateMessage creation site. Example from `_process_system_message`: - -```python -# Before -css_class = f"{message_type}" -if is_sidechain: - css_class = f"{css_class} sidechain" - -# After -modifiers = MessageModifiers(is_sidechain=is_sidechain) -css_class = f"{message_type}" # Keep for backward compat -if is_sidechain: - css_class = f"{css_class} sidechain" -``` - -Update `_get_message_hierarchy_level()`: -```python -# Before -if "sidechain" in css_class: - ... - -# After -def _get_message_hierarchy_level(msg: TemplateMessage) -> int: - is_sidechain = msg.modifiers.is_sidechain - msg_type = msg.type - - if msg_type == MessageType.USER and not is_sidechain: - return 1 - # ... -``` - -### Phase 12c: Create html_renderer.py - -```python -"""HTML-specific rendering utilities. - -This module contains all HTML generation code: -- CSS class computation -- HTML escaping -- Markdown rendering -- Tool-specific formatters -""" - -from html import escape -from typing import Optional, List -import mistune - -from .models import MessageType, MessageModifiers, TemplateMessage - - -def escape_html(text: str) -> str: - """Escape HTML special characters.""" - return escape(text, quote=True) - - -def render_markdown(text: str) -> str: - """Convert markdown to HTML.""" - return mistune.html(text) - - -def css_class_from_message(msg: TemplateMessage) -> str: - """Generate CSS class string from message type and modifiers. - - This reconstructs the original css_class format for backward - compatibility with existing CSS and JavaScript. - """ - parts: List[str] = [msg.type.value if isinstance(msg.type, MessageType) else msg.type] - - mods = msg.modifiers - if mods.is_slash_command: - parts.append("slash-command") - if mods.is_command_output: - parts.append("command-output") - if mods.is_compacted: - parts.append("compacted") - if mods.is_error: - parts.append("error") - if mods.is_steering: - parts.append("steering") - if mods.is_sidechain: - parts.append("sidechain") - if mods.system_level: - parts.append(f"system-{mods.system_level}") - - return " ".join(parts) - - -def get_message_emoji(msg: TemplateMessage) -> str: - """Return appropriate emoji for message type.""" - msg_type = msg.type if isinstance(msg.type, MessageType) else msg.type - - if msg_type == MessageType.SESSION_HEADER: - return "πŸ“‹" - elif msg_type == MessageType.USER: - return "🀷" - elif msg_type == MessageType.ASSISTANT: - return "πŸ€–" - elif msg_type == MessageType.SYSTEM: - return "βš™οΈ" - elif msg_type == MessageType.TOOL_USE: - return "πŸ› οΈ" - elif msg_type == MessageType.TOOL_RESULT: - if msg.modifiers.is_error: - return "🚨" - return "🧰" - elif msg_type == MessageType.THINKING: - return "πŸ’­" - elif msg_type == MessageType.IMAGE: - return "πŸ–ΌοΈ" - return "" - - -# Move format_* tool functions here: -# - format_ask_user_question_tool_content -# - format_todo_write_tool_content -# - format_bash_tool_content -# etc. -``` - -### Phase 12d: Update Templates - -Update transcript.html to use the new functions. Register them as Jinja filters or pass as context: - -```python -# In renderer.py when rendering template -from .html_renderer import css_class_from_message, get_message_emoji - -template = env.get_template("transcript.html") -html = template.render( - messages=messages, - css_class_from_message=css_class_from_message, - get_message_emoji=get_message_emoji, - # ... -) -``` - -Template changes: -```jinja -{# Before #} -
- -{# After #} -
-``` - -### Phase 12e: Remove css_class - -Once all references use modifiers: -1. Remove `css_class` parameter from `TemplateMessage.__init__` -2. Remove `self.css_class = css_class` -3. Clean up all `css_class=...` at creation sites -4. Update tests to use modifiers - -## Files Changed - -| File | Changes | -|------|---------| -| `models.py` | Add `MessageModifiers` dataclass | -| `renderer.py` | Update TemplateMessage, populate modifiers, update hierarchy logic | -| `html_renderer.py` | New file with HTML utilities and css_class_from_message | -| `templates/transcript.html` | Use css_class_from_message filter | -| `test_*.py` | Update tests to use modifiers | - -## Testing Strategy - -1. **Snapshot tests**: Run after each phase to verify HTML output unchanged -2. **Unit tests for css_class_from_message**: Verify it produces same strings -3. **Unit tests for modifiers**: Test each modifier flag -4. **Integration tests**: Full render with real transcripts - -## Commit Plan - -1. `Add MessageModifiers dataclass to models.py` (12a) -2. `Add modifiers field to TemplateMessage` (12a) -3. `Populate modifiers in message processing` (12b part 1) -4. `Update hierarchy logic to use modifiers` (12b part 2) -5. `Create html_renderer.py with css_class_from_message` (12c) -6. `Move escape_html and render_markdown to html_renderer` (12c) -7. `Update template to use css_class_from_message` (12d) -8. `Remove css_class field from TemplateMessage` (12e) - -## Risk Assessment - -- **Low risk**: MessageModifiers is additive, doesn't break existing code -- **Medium risk**: Moving functions to html_renderer.py requires import updates -- **High risk**: Template changes and css_class removal need careful testing - -## Estimated Scope - -- Phase 12a: ~30 lines added to models.py, ~10 lines to renderer.py -- Phase 12b: ~50 modifications across renderer.py -- Phase 12c: ~200 lines new file, ~200 lines moved from renderer.py -- Phase 12d: ~10 lines template changes -- Phase 12e: ~20 lines removed - -Total: Moderate refactoring, ~5-8 commits diff --git a/claude_code_log/cache.py b/claude_code_log/cache.py index ad443726..3f5d43b7 100644 --- a/claude_code_log/cache.py +++ b/claude_code_log/cache.py @@ -172,10 +172,10 @@ def load_cached_entries(self, jsonl_path: Path) -> Optional[list[TranscriptEntry entries_data.extend(cast(list[dict[str, Any]], timestamp_entries)) # Deserialize back to TranscriptEntry objects - from .parser import parse_transcript_entry + from .factories import create_transcript_entry entries = [ - parse_transcript_entry(entry_dict) for entry_dict in entries_data + create_transcript_entry(entry_dict) for entry_dict in entries_data ] return entries except Exception as e: @@ -257,10 +257,10 @@ def load_cached_entries_filtered( ) # Deserialize filtered entries - from .parser import parse_transcript_entry + from .factories import create_transcript_entry entries = [ - parse_transcript_entry(entry_dict) + create_transcript_entry(entry_dict) for entry_dict in filtered_entries_data ] return entries diff --git a/claude_code_log/converter.py b/claude_code_log/converter.py index 12a2fa83..8407f6b3 100644 --- a/claude_code_log/converter.py +++ b/claude_code_log/converter.py @@ -21,7 +21,8 @@ get_warmup_session_ids, ) from .cache import CacheManager, SessionCacheData, get_library_version -from .parser import parse_timestamp, parse_transcript_entry +from .parser import parse_timestamp +from .factories import create_transcript_entry from .models import ( TranscriptEntry, AssistantTranscriptEntry, @@ -183,7 +184,7 @@ def load_transcript( "queue-operation", ]: # Parse using Pydantic models - entry = parse_transcript_entry(entry_dict) + entry = create_transcript_entry(entry_dict) messages.append(entry) elif ( entry_type diff --git a/claude_code_log/factories/__init__.py b/claude_code_log/factories/__init__.py new file mode 100644 index 00000000..108a9c91 --- /dev/null +++ b/claude_code_log/factories/__init__.py @@ -0,0 +1,115 @@ +"""Factory modules for creating typed objects from raw data.""" + +from .meta_factory import ( + # Metadata creation + create_meta, +) +from .system_factory import ( + # System message detection + is_system_message, + # System message creation + create_system_message, +) +from .user_factory import ( + # User message type detection + is_bash_input, + is_bash_output, + is_command_message, + is_local_command_output, + # User message creation + create_bash_input_message, + create_bash_output_message, + create_command_output_message, + create_compacted_summary_message, + create_ide_notification_content, + create_slash_command_message, + create_user_memory_message, + create_user_message, + # Patterns and constants + COMPACTED_SUMMARY_PREFIX, + IDE_DIAGNOSTICS_PATTERN, + IDE_OPENED_FILE_PATTERN, + IDE_SELECTION_PATTERN, +) +from .assistant_factory import ( + # Assistant message creation + create_assistant_message, + create_thinking_message, +) +from .tool_factory import ( + # Tool message creation + create_tool_input, + create_tool_use_message, + create_tool_result_message, + # Tool processing result + ToolItemResult, + # Tool input models mapping + TOOL_INPUT_MODELS, +) +from .transcript_factory import ( + # Content type constants + ASSISTANT_CONTENT_TYPES, + USER_CONTENT_TYPES, + # Conditional casts + as_assistant_entry, + as_user_entry, + # Usage normalization + normalize_usage_info, + # Content item creation + create_content_item, + create_message_content, + # Transcript entry creation + create_transcript_entry, +) + +__all__ = [ + # Metadata creation + "create_meta", + # Content type constants + "USER_CONTENT_TYPES", + "ASSISTANT_CONTENT_TYPES", + # Conditional casts + "as_user_entry", + "as_assistant_entry", + # Usage normalization + "normalize_usage_info", + # Content item creation + "create_content_item", + "create_message_content", + # Transcript entry creation + "create_transcript_entry", + # System message detection + "is_system_message", + # System message creation + "create_system_message", + # User message type detection + "is_bash_input", + "is_bash_output", + "is_command_message", + "is_local_command_output", + # User message creation + "create_bash_input_message", + "create_bash_output_message", + "create_command_output_message", + "create_compacted_summary_message", + "create_ide_notification_content", + "create_slash_command_message", + "create_user_memory_message", + "create_user_message", + # Patterns and constants + "COMPACTED_SUMMARY_PREFIX", + "IDE_DIAGNOSTICS_PATTERN", + "IDE_OPENED_FILE_PATTERN", + "IDE_SELECTION_PATTERN", + # Assistant message creation + "create_assistant_message", + "create_thinking_message", + # Tool message creation + "create_tool_input", + "create_tool_use_message", + "create_tool_result_message", + # Tool processing result + "ToolItemResult", + # Tool input models mapping + "TOOL_INPUT_MODELS", +] diff --git a/claude_code_log/factories/assistant_factory.py b/claude_code_log/factories/assistant_factory.py new file mode 100644 index 00000000..3f8b977d --- /dev/null +++ b/claude_code_log/factories/assistant_factory.py @@ -0,0 +1,114 @@ +"""Factory for assistant transcript entries. + +This module handles creation of AssistantTranscriptEntry content into MessageContent +subclasses: +- AssistantTextMessage: Claude's text responses +- ThinkingMessage: Extended thinking blocks +""" + +from typing import Optional + +from ..models import ( + AssistantTextMessage, + ContentItem, + MessageMeta, + TextContent, + ThinkingContent, + ThinkingMessage, + UsageInfo, +) + + +# ============================================================================= +# Token Usage Formatting +# ============================================================================= + + +def format_token_usage(usage: UsageInfo) -> str: + """Format token usage information as a display string. + + Args: + usage: UsageInfo object with token counts. + + Returns: + Formatted string like "Input: 100 | Output: 50 | Cache Read: 25" + """ + token_parts = [ + f"Input: {usage.input_tokens}", + f"Output: {usage.output_tokens}", + ] + if usage.cache_creation_input_tokens: + token_parts.append(f"Cache Creation: {usage.cache_creation_input_tokens}") + if usage.cache_read_input_tokens: + token_parts.append(f"Cache Read: {usage.cache_read_input_tokens}") + return " | ".join(token_parts) + + +# ============================================================================= +# Message Creation Functions +# ============================================================================= + + +def create_assistant_message( + meta: MessageMeta, + items: list[ContentItem], + usage: Optional[UsageInfo] = None, +) -> Optional[AssistantTextMessage]: + """Create AssistantTextMessage from content items. + + Creates AssistantTextMessage from text/image content items. + + Args: + meta: Message metadata. + items: List of text/image content items (no tool_use, tool_result, thinking). + usage: Optional token usage info to format and attach. + + Returns: + AssistantTextMessage if items is non-empty, None otherwise. + """ + # Create AssistantTextMessage directly from items + # (empty text already filtered by chunk_message_content) + if items: + # Extract text content from items for dedup matching and simple renderers + text_content = "\n".join( + item.text for item in items if isinstance(item, TextContent) + ) + return AssistantTextMessage( + meta, + items=items, # type: ignore[arg-type] + raw_text_content=text_content if text_content else None, + token_usage=format_token_usage(usage) if usage else None, + ) + return None + + +def create_thinking_message( + meta: MessageMeta, + tool_item: ContentItem, + usage: Optional[UsageInfo] = None, +) -> ThinkingMessage: + """Create ThinkingMessage from a thinking content item. + + Args: + meta: Message metadata. + tool_item: ThinkingContent or compatible object with 'thinking' attribute + usage: Optional token usage info to format and attach. + + Returns: + ThinkingMessage containing the thinking text and optional signature. + """ + # Extract thinking text from the content item + if isinstance(tool_item, ThinkingContent): + thinking_text = tool_item.thinking.strip() + signature = getattr(tool_item, "signature", None) + else: + thinking_text = getattr(tool_item, "thinking", str(tool_item)).strip() + signature = None + + # Create the content model (formatting happens in HtmlRenderer) + return ThinkingMessage( + meta, + thinking=thinking_text, + signature=signature, + token_usage=format_token_usage(usage) if usage else None, + ) diff --git a/claude_code_log/factories/meta_factory.py b/claude_code_log/factories/meta_factory.py new file mode 100644 index 00000000..ee6f0d42 --- /dev/null +++ b/claude_code_log/factories/meta_factory.py @@ -0,0 +1,33 @@ +"""Factory for creating MessageMeta from transcript entries. + +This module handles extraction of common metadata from transcript entries +that is shared across all message types. +""" + +from ..models import BaseTranscriptEntry, MessageMeta + + +def create_meta(transcript: BaseTranscriptEntry) -> MessageMeta: + """Create MessageMeta from a transcript entry. + + Extracts all shared fields from BaseTranscriptEntry subclasses. + + Args: + transcript: Any transcript entry inheriting from BaseTranscriptEntry + + Returns: + MessageMeta with identity and context fields + """ + return MessageMeta( + # Identity fields + session_id=transcript.sessionId, + timestamp=transcript.timestamp, + uuid=transcript.uuid, + parent_uuid=transcript.parentUuid, + # Context fields + is_sidechain=transcript.isSidechain, + is_meta=getattr(transcript, "isMeta", False) or False, + agent_id=transcript.agentId, + cwd=transcript.cwd, + git_branch=transcript.gitBranch, + ) diff --git a/claude_code_log/factories/system_factory.py b/claude_code_log/factories/system_factory.py new file mode 100644 index 00000000..3ec9c785 --- /dev/null +++ b/claude_code_log/factories/system_factory.py @@ -0,0 +1,87 @@ +"""Factory for system transcript entries. + +This module handles creation of MessageContent from SystemTranscriptEntry: +- SystemMessage: Regular system messages with level (info, warning, error) +- HookSummaryMessage: Hook execution summaries + +Also provides: +- is_system_message: Check if text content is a system message to filter +""" + +from typing import Optional, Union + +from ..models import ( + HookInfo, + HookSummaryMessage, + SystemMessage, + SystemTranscriptEntry, +) +from .meta_factory import create_meta + + +# ============================================================================= +# System Message Detection +# ============================================================================= + + +def is_system_message(text_content: str) -> bool: + """Check if a message is a system message that should be filtered out.""" + system_message_patterns = [ + "Caveat: The messages below were generated by the user while running local commands. DO NOT respond to these messages or otherwise consider them in your response unless the user explicitly asks you to.", + "[Request interrupted by user for tool use]", + "", + ] + + return any(text_content.startswith(pattern) for pattern in system_message_patterns) + + +# ============================================================================= +# System Message Creation +# ============================================================================= + + +def create_system_message( + transcript: SystemTranscriptEntry, +) -> Optional[Union[SystemMessage, HookSummaryMessage]]: + """Create a MessageContent from a system transcript entry. + + Handles: + - Hook summaries (subtype="stop_hook_summary") + - Regular system messages with level-specific styling (info, warning, error) + + Args: + transcript: The system transcript entry to process + + Returns: + SystemMessage or HookSummaryMessage (with meta attached), + or None if the message should be skipped (e.g., silent hook successes) + + Note: + Slash command messages (, ) are user messages, + not system messages. They are handled separately. + """ + if transcript.subtype == "stop_hook_summary": + # Skip silent hook successes (no output, no errors) + if not transcript.hasOutput and not transcript.hookErrors: + return None + # Create structured hook summary content + meta = create_meta(transcript) + hook_infos = [ + HookInfo(command=info.get("command", "unknown")) + for info in (transcript.hookInfos or []) + ] + return HookSummaryMessage( + has_output=bool(transcript.hasOutput), + hook_errors=transcript.hookErrors or [], + hook_infos=hook_infos, + meta=meta, + ) + + if not transcript.content: + # Skip system messages without content (shouldn't happen normally) + return None + + # Create structured system content + meta = create_meta(transcript) + level = getattr(transcript, "level", "info") + return SystemMessage(level=level, text=transcript.content, meta=meta) diff --git a/claude_code_log/factories/tool_factory.py b/claude_code_log/factories/tool_factory.py new file mode 100644 index 00000000..312472db --- /dev/null +++ b/claude_code_log/factories/tool_factory.py @@ -0,0 +1,769 @@ +"""Factory for tool use and tool result content. + +This module handles creation of tool-related content into MessageContent subclasses: +- ToolUseMessage: Tool invocations with typed inputs (BashInput, ReadInput, etc.) +- ToolResultMessage: Tool results with output and context + +Also provides creation of tool inputs into typed models: +- create_tool_input(): Create typed tool input from raw dict +- create_tool_use_message(): Process ToolUseContent into ToolItemResult +- create_tool_result_message(): Process ToolResultContent into ToolItemResult +""" + +from dataclasses import dataclass +from typing import Any, Callable, Optional, cast + +from pydantic import BaseModel + +import re + +from ..models import ( + # Tool input models + AskUserQuestionInput, + AskUserQuestionItem, + AskUserQuestionOption, + BashInput, + EditInput, + EditItem, + ExitPlanModeInput, + GlobInput, + GrepInput, + MessageContent, + MessageMeta, + MultiEditInput, + ReadInput, + TaskInput, + TodoWriteInput, + TodoWriteItem, + ToolInput, + ToolResultContent, + ToolResultMessage, + ToolUseContent, + ToolUseMessage, + WriteInput, + # Tool output models + AskUserQuestionAnswer, + AskUserQuestionOutput, + BashOutput, + EditOutput, + ExitPlanModeOutput, + ReadOutput, + TaskOutput, + ToolOutput, + WriteOutput, +) + + +# ============================================================================= +# Tool Input Models Mapping +# ============================================================================= + +TOOL_INPUT_MODELS: dict[str, type[BaseModel]] = { + "Bash": BashInput, + "Read": ReadInput, + "Write": WriteInput, + "Edit": EditInput, + "MultiEdit": MultiEditInput, + "Glob": GlobInput, + "Grep": GrepInput, + "Task": TaskInput, + "TodoWrite": TodoWriteInput, + "AskUserQuestion": AskUserQuestionInput, + "ask_user_question": AskUserQuestionInput, # Legacy tool name + "ExitPlanMode": ExitPlanModeInput, +} + + +# ============================================================================= +# Lenient Parsing Helpers +# ============================================================================= +# These functions create typed models even when strict validation fails. +# They use defaults for missing fields and skip invalid nested items. + + +def _parse_todowrite_lenient(data: dict[str, Any]) -> TodoWriteInput: + """Parse TodoWrite input leniently, handling malformed data.""" + todos_raw = data.get("todos", []) + valid_todos: list[TodoWriteItem] = [] + for item in todos_raw: + if isinstance(item, dict): + try: + valid_todos.append(TodoWriteItem.model_validate(item)) + except Exception: + pass + elif isinstance(item, str): + valid_todos.append(TodoWriteItem(content=item)) + return TodoWriteInput(todos=valid_todos) + + +def _parse_bash_lenient(data: dict[str, Any]) -> BashInput: + """Parse Bash input leniently.""" + return BashInput( + command=data.get("command", ""), + description=data.get("description"), + timeout=data.get("timeout"), + run_in_background=data.get("run_in_background"), + ) + + +def _parse_write_lenient(data: dict[str, Any]) -> WriteInput: + """Parse Write input leniently.""" + return WriteInput( + file_path=data.get("file_path", ""), + content=data.get("content", ""), + ) + + +def _parse_edit_lenient(data: dict[str, Any]) -> EditInput: + """Parse Edit input leniently.""" + return EditInput( + file_path=data.get("file_path", ""), + old_string=data.get("old_string", ""), + new_string=data.get("new_string", ""), + replace_all=data.get("replace_all"), + ) + + +def _parse_multiedit_lenient(data: dict[str, Any]) -> MultiEditInput: + """Parse Multiedit input leniently.""" + edits_raw = data.get("edits", []) + valid_edits: list[EditItem] = [] + for edit in edits_raw: + if isinstance(edit, dict): + try: + valid_edits.append(EditItem.model_validate(edit)) + except Exception: + pass + return MultiEditInput(file_path=data.get("file_path", ""), edits=valid_edits) + + +def _parse_task_lenient(data: dict[str, Any]) -> TaskInput: + """Parse Task input leniently.""" + return TaskInput( + prompt=data.get("prompt", ""), + subagent_type=data.get("subagent_type", ""), + description=data.get("description", ""), + model=data.get("model"), + run_in_background=data.get("run_in_background"), + resume=data.get("resume"), + ) + + +def _parse_read_lenient(data: dict[str, Any]) -> ReadInput: + """Parse Read input leniently.""" + return ReadInput( + file_path=data.get("file_path", ""), + offset=data.get("offset"), + limit=data.get("limit"), + ) + + +def _parse_askuserquestion_lenient(data: dict[str, Any]) -> AskUserQuestionInput: + """Parse AskUserQuestion input leniently, handling malformed data.""" + questions_raw = data.get("questions", []) + valid_questions: list[AskUserQuestionItem] = [] + for q in questions_raw: + if isinstance(q, dict): + q_dict = cast(dict[str, Any], q) + try: + # Parse options leniently + options_raw = q_dict.get("options", []) + valid_options: list[AskUserQuestionOption] = [] + for opt in options_raw: + if isinstance(opt, dict): + try: + valid_options.append( + AskUserQuestionOption.model_validate(opt) + ) + except Exception: + pass + valid_questions.append( + AskUserQuestionItem( + question=str(q_dict.get("question", "")), + header=q_dict.get("header"), + options=valid_options, + multiSelect=bool(q_dict.get("multiSelect", False)), + ) + ) + except Exception: + pass + return AskUserQuestionInput( + questions=valid_questions, + question=data.get("question"), + ) + + +def _parse_exitplanmode_lenient(data: dict[str, Any]) -> ExitPlanModeInput: + """Parse ExitPlanMode input leniently.""" + return ExitPlanModeInput( + plan=data.get("plan", ""), + launchSwarm=data.get("launchSwarm"), + teammateCount=data.get("teammateCount"), + ) + + +# Mapping of tool names to their lenient parsers +TOOL_LENIENT_PARSERS: dict[str, Any] = { + "Bash": _parse_bash_lenient, + "Write": _parse_write_lenient, + "Edit": _parse_edit_lenient, + "MultiEdit": _parse_multiedit_lenient, + "Task": _parse_task_lenient, + "TodoWrite": _parse_todowrite_lenient, + "Read": _parse_read_lenient, + "AskUserQuestion": _parse_askuserquestion_lenient, + "ask_user_question": _parse_askuserquestion_lenient, # Legacy tool name + "ExitPlanMode": _parse_exitplanmode_lenient, +} + + +# ============================================================================= +# Tool Input Creation +# ============================================================================= + + +def create_tool_input( + tool_name: str, input_data: dict[str, Any] +) -> Optional[ToolInput]: + """Create typed tool input from raw dictionary. + + Uses strict validation first, then lenient parsing if available. + + Args: + tool_name: The name of the tool (e.g., "Bash", "Read") + input_data: The raw input dictionary from the tool_use content + + Returns: + A typed input model if parsing succeeds, None otherwise. + When None is returned, the caller should use ToolUseContent itself + as the fallback (it's part of the ToolInput union). + """ + model_class = TOOL_INPUT_MODELS.get(tool_name) + if model_class is not None: + try: + return cast(ToolInput, model_class.model_validate(input_data)) + except Exception: + # Try lenient parsing if available + lenient_parser = TOOL_LENIENT_PARSERS.get(tool_name) + if lenient_parser is not None: + return cast(ToolInput, lenient_parser(input_data)) + return None + return None + + +# ============================================================================= +# Tool Output Parsing +# ============================================================================= +# Parse raw tool result content into typed output models (ReadOutput, EditOutput, etc.) +# Symmetric with Tool Input parsing above. + + +def _parse_cat_n_snippet( + lines: list[str], start_idx: int = 0 +) -> Optional[tuple[str, Optional[str], int]]: + """Parse cat-n formatted snippet from lines. + + Args: + lines: List of lines to parse + start_idx: Index to start parsing from (default: 0) + + Returns: + Tuple of (code_content, system_reminder, line_offset) or None if not parseable + """ + code_lines: list[str] = [] + system_reminder: Optional[str] = None + in_system_reminder = False + line_offset = 1 # Default offset + + for line in lines[start_idx:]: + # Check for system-reminder start + if "" in line: + in_system_reminder = True + system_reminder = "" + continue + + # Check for system-reminder end + if "" in line: + in_system_reminder = False + continue + + # If in system reminder, accumulate reminder text + if in_system_reminder: + if system_reminder is not None: + system_reminder += line + "\n" + continue + + # Parse regular code line (format: " 123β†’content") + match = re.match(r"\s+(\d+)β†’(.*)$", line) + if match: + line_num = int(match.group(1)) + # Capture the first line number as offset + if not code_lines: + line_offset = line_num + code_lines.append(match.group(2)) + elif line.strip() == "": # Allow empty lines between cat-n lines + continue + else: # Non-matching non-empty line, stop parsing + break + + if not code_lines: + return None + + # Join code lines and trim trailing reminder text + code_content = "\n".join(code_lines) + if system_reminder: + system_reminder = system_reminder.strip() + + return (code_content, system_reminder, line_offset) + + +def _extract_tool_result_text(tool_result: ToolResultContent) -> str: + """Extract text content from a ToolResultContent. + + Handles both string content and structured content (list of dicts). + + Args: + tool_result: The tool result to extract text from + + Returns: + Extracted text content, or empty string if none found + """ + content = tool_result.content + if isinstance(content, str): + return content + # Structured content - extract text from list of content items + # Format: [{"type": "text", "text": "..."}, ...] + text_parts: list[str] = [] + for item in content: + if item.get("type") == "text": + text_parts.append(str(item.get("text", ""))) + return "\n".join(text_parts) + + +def parse_read_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[ReadOutput]: + """Parse Read tool result into structured content. + + Args: + tool_result: The tool result content + file_path: Path to the file that was read (required for ReadOutput) + + Returns: + ReadOutput if parsing succeeds, None otherwise + """ + if not file_path: + return None + if not (content := _extract_tool_result_text(tool_result)): + return None + + # Check if content matches the cat-n format pattern (line_number β†’ content) + lines = content.split("\n") + if not lines or not re.match(r"\s+\d+β†’", lines[0]): + return None + + result = _parse_cat_n_snippet(lines) + if result is None: + return None + + code_content, system_reminder, line_offset = result + num_lines = len(code_content.split("\n")) + + return ReadOutput( + file_path=file_path, + content=code_content, + start_line=line_offset, + num_lines=num_lines, + total_lines=num_lines, # We don't know total from result + is_truncated=False, # Can't determine from result + system_reminder=system_reminder, + ) + + +def parse_edit_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[EditOutput]: + """Parse Edit tool result into structured content. + + Edit tool results typically have format: + "The file ... has been updated. Here's the result of running `cat -n` on a snippet..." + followed by cat-n formatted lines. + + Args: + tool_result: The tool result content + file_path: Path to the file that was edited (required for EditOutput) + + Returns: + EditOutput if parsing succeeds, None otherwise + """ + if not file_path: + return None + if not (content := _extract_tool_result_text(tool_result)): + return None + + # Look for the cat-n snippet after the preamble + # Pattern: look for first line that matches the cat-n format + lines = content.split("\n") + code_start_idx = None + + for i, line in enumerate(lines): + if re.match(r"\s+\d+β†’", line): + code_start_idx = i + break + + if code_start_idx is None: + return None + + result = _parse_cat_n_snippet(lines, code_start_idx) + if result is None: + return None + + code_content, _system_reminder, line_offset = result + # Edit tool doesn't use system_reminder + + return EditOutput( + file_path=file_path, + success=True, # If we got here, edit succeeded + diffs=[], # We don't have diff info from result + message=code_content, + start_line=line_offset, + ) + + +def parse_write_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[WriteOutput]: + """Parse Write tool result into structured content. + + Write tool results contain an acknowledgment on the first line. + We extract just the first line for display. + + Args: + tool_result: The tool result content + file_path: Path to the file that was written (required for WriteOutput) + + Returns: + WriteOutput if parsing succeeds, None otherwise + """ + if not file_path: + return None + if not (content := _extract_tool_result_text(tool_result)): + return None + + lines = content.split("\n") + if not lines[0]: + return None + + first_line = lines[0] + return WriteOutput( + file_path=file_path, + success=True, # If we got content, write succeeded + message=first_line, + ) + + +def parse_task_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[TaskOutput]: + """Parse Task tool result into structured content. + + Task tool results contain the agent's response as markdown. + + Args: + tool_result: The tool result content (agent's response) + file_path: Unused for Task tool + + Returns: + TaskOutput with the agent's response + """ + del file_path # Unused + if not (content := _extract_tool_result_text(tool_result)): + return None + return TaskOutput(result=content) + + +def _looks_like_bash_output(content: str) -> bool: + """Check if content looks like it's from a Bash tool based on common patterns.""" + if not content: + return False + + # Check for ANSI escape sequences + if "\x1b[" in content: + return True + + # Check for common bash/terminal patterns + bash_indicators = [ + "$ ", # Shell prompt + "❯ ", # Modern shell prompt + "> ", # Shell continuation + "\n+ ", # Bash -x output + "bash: ", # Bash error messages + "/bin/bash", # Bash path + "command not found", # Common bash error + "Permission denied", # Common bash error + "No such file or directory", # Common bash error + ] + + # Check for file path patterns that suggest command output + if re.search(r"/[a-zA-Z0-9_-]+(/[a-zA-Z0-9_.-]+)*", content): # Unix-style paths + return True + + # Check for common command output patterns + if any(indicator in content for indicator in bash_indicators): + return True + + return False + + +def parse_bash_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[BashOutput]: + """Parse Bash tool result into structured content. + + Detects ANSI escape sequences for terminal formatting. + + Args: + tool_result: The tool result content + file_path: Unused for Bash tool + + Returns: + BashOutput with content and ANSI flag + """ + del file_path # Unused + if not (content := _extract_tool_result_text(tool_result)): + return None + has_ansi = _looks_like_bash_output(content) + return BashOutput(content=content, has_ansi=has_ansi) + + +def parse_askuserquestion_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[AskUserQuestionOutput]: + """Parse AskUserQuestion tool result into structured content. + + Parses the result format: + 'User has answered your questions: "Q1"="A1", "Q2"="A2". You can now continue...' + + Args: + tool_result: The tool result content + file_path: Unused for AskUserQuestion tool + + Returns: + AskUserQuestionOutput with Q&A pairs + """ + del file_path # Unused + if not (content := _extract_tool_result_text(tool_result)): + return None + # Check if this is a successful answer + if not content.startswith("User has answered your question"): + return None + + # Extract the Q&A portion between the colon and the final sentence + match = re.match( + r"User has answered your questions?: (.+)\. You can now continue", + content, + re.DOTALL, + ) + if not match: + return None + + qa_portion = match.group(1) + + # Parse "Question"="Answer" pairs + qa_pattern = re.compile(r'"([^"]+)"="([^"]+)"') + pairs = qa_pattern.findall(qa_portion) + + if not pairs: + return None + + answers = [AskUserQuestionAnswer(question=q, answer=a) for q, a in pairs] + return AskUserQuestionOutput(answers=answers, raw_message=content) + + +def parse_exitplanmode_output( + tool_result: ToolResultContent, file_path: Optional[str] +) -> Optional[ExitPlanModeOutput]: + """Parse ExitPlanMode tool result into structured content. + + Truncates redundant plan echo on success. + When a plan is approved, the result contains: + 1. A confirmation message + 2. Path to saved plan file + 3. "## Approved Plan:" followed by full plan text (redundant) + + Args: + tool_result: The tool result content + file_path: Unused for ExitPlanMode tool + + Returns: + ExitPlanModeOutput with truncated message + """ + del file_path # Unused + if not (content := _extract_tool_result_text(tool_result)): + return None + approved = "User has approved your plan" in content + + if approved: + # Truncate at "## Approved Plan:" + marker = "## Approved Plan:" + marker_pos = content.find(marker) + if marker_pos > 0: + message = content[:marker_pos].rstrip() + else: + message = content + else: + message = content + + return ExitPlanModeOutput(message=message, approved=approved) + + +# Type alias for tool output parsers +ToolOutputParser = Callable[[ToolResultContent, Optional[str]], Optional[ToolOutput]] + +# Registry of tool output parsers: tool_name -> parser(tool_result, file_path) -> Optional[ToolOutput] +# Parsers receive the full ToolResultContent and can use _extract_tool_result_text() for text. +TOOL_OUTPUT_PARSERS: dict[str, ToolOutputParser] = { + "Read": parse_read_output, + "Edit": parse_edit_output, + "Write": parse_write_output, + "Bash": parse_bash_output, + "Task": parse_task_output, + "AskUserQuestion": parse_askuserquestion_output, + "ExitPlanMode": parse_exitplanmode_output, +} + + +def create_tool_output( + tool_name: str, + tool_result: ToolResultContent, + file_path: Optional[str] = None, +) -> ToolOutput: + """Create typed tool output from raw ToolResultContent. + + Parses the raw content into specialized output types when possible, + using the TOOL_OUTPUT_PARSERS registry. Each parser receives the full + ToolResultContent and can use _extract_tool_result_text() if it needs text. + + Args: + tool_name: The name of the tool (e.g., "Bash", "Read") + tool_result: The raw tool result content + file_path: Optional file path for file-based tools (Read, Edit, Write) + + Returns: + A typed output model if parsing succeeds, ToolResultContent as fallback. + """ + # Look up parser in registry and parse if available + if (parser := TOOL_OUTPUT_PARSERS.get(tool_name)) and ( + parsed := parser(tool_result, file_path) + ): + return parsed + + # Fallback to raw ToolResultContent + return tool_result + + +# ============================================================================= +# Tool Item Processing +# ============================================================================= + + +@dataclass +class ToolItemResult: + """Result of processing a single tool/thinking/image item. + + Note: Titles are computed at render time by Renderer.title_content() dispatch. + """ + + message_type: str + content: Optional[MessageContent] = None # Structured content for rendering + tool_use_id: Optional[str] = None + is_error: bool = False # For tool_result error state + + +def create_tool_use_message( + meta: MessageMeta, + tool_use: ToolUseContent, + tool_use_context: dict[str, ToolUseContent], +) -> ToolItemResult: + """Create ToolItemResult from a tool_use content item. + + Args: + meta: Message metadata + tool_use: The tool use content item + tool_use_context: Dict to populate with tool_use_id -> ToolUseContent mapping + + Returns: + ToolItemResult with tool_use content model + """ + # Parse tool input into typed model (BashInput, ReadInput, etc.) + parsed = create_tool_input(tool_use.name, tool_use.input) + + # Populate tool_use_context for later use when processing tool results + tool_use_context[tool_use.id] = tool_use + + # Create ToolUseMessage wrapper with parsed input for specialized formatting + # Use ToolUseContent as fallback when no specialized parser exists + tool_use_message = ToolUseMessage( + meta, + input=parsed if parsed is not None else tool_use, + tool_use_id=tool_use.id, + tool_name=tool_use.name, + ) + + return ToolItemResult( + message_type="tool_use", + content=tool_use_message, + tool_use_id=tool_use.id, + ) + + +def create_tool_result_message( + meta: MessageMeta, + tool_result: ToolResultContent, + tool_use_context: dict[str, ToolUseContent], +) -> ToolItemResult: + """Create ToolItemResult from a tool_result content item. + + Args: + meta: Message metadata + tool_result: The tool result content item + tool_use_context: Dict with tool_use_id -> ToolUseContent mapping + + Returns: + ToolItemResult with tool_result content model + """ + # Get file_path and tool_name from tool_use context for specialized rendering + result_file_path: Optional[str] = None + result_tool_name: Optional[str] = None + if tool_result.tool_use_id in tool_use_context: + tool_use_from_ctx = tool_use_context[tool_result.tool_use_id] + result_tool_name = tool_use_from_ctx.name + if ( + result_tool_name in ("Read", "Edit", "Write") + and "file_path" in tool_use_from_ctx.input + ): + result_file_path = tool_use_from_ctx.input["file_path"] + + # Parse into typed output (ReadOutput, EditOutput, etc.) when possible + parsed_output = create_tool_output( + result_tool_name or "", + tool_result, + result_file_path, + ) + + # Create content model with rendering context + content_model = ToolResultMessage( + meta, + tool_use_id=tool_result.tool_use_id, + output=parsed_output, + is_error=tool_result.is_error or False, + tool_name=result_tool_name, + file_path=result_file_path, + ) + + return ToolItemResult( + message_type="tool_result", + content=content_model, + tool_use_id=tool_result.tool_use_id, + is_error=tool_result.is_error or False, + ) diff --git a/claude_code_log/factories/transcript_factory.py b/claude_code_log/factories/transcript_factory.py new file mode 100644 index 00000000..c70821e1 --- /dev/null +++ b/claude_code_log/factories/transcript_factory.py @@ -0,0 +1,265 @@ +"""Factory for creating TranscriptEntry and ContentItem instances from raw data. + +This module creates typed model instances from JSONL transcript data: +- TranscriptEntry subclasses (User, Assistant, Summary, System, QueueOperation) +- ContentItem subclasses (Text, ToolUse, ToolResult, Thinking, Image) + +Also provides: +- Conditional casts for TranscriptEntry discrimination +- Usage info normalization for Anthropic SDK compatibility +""" + +from typing import Any, Callable, Optional, Sequence, cast + +from pydantic import BaseModel + +from ..models import ( + # Content types + ContentItem, + ImageContent, + TextContent, + ThinkingContent, + ToolResultContent, + ToolUseContent, + # Transcript entry types + AssistantTranscriptEntry, + MessageType, + QueueOperationTranscriptEntry, + SummaryTranscriptEntry, + SystemTranscriptEntry, + TranscriptEntry, + UsageInfo, + UserTranscriptEntry, +) + + +# ============================================================================= +# Content Item Registry +# ============================================================================= + +# Maps content type strings to their model classes +CONTENT_ITEM_CREATORS: dict[str, type[BaseModel]] = { + "text": TextContent, + "tool_result": ToolResultContent, + "image": ImageContent, + "tool_use": ToolUseContent, + "thinking": ThinkingContent, +} + +# Content types allowed in each context +USER_CONTENT_TYPES: Sequence[str] = ("text", "tool_result", "image") +ASSISTANT_CONTENT_TYPES: Sequence[str] = ("text", "tool_use", "thinking") + + +# ============================================================================= +# Conditional Casts +# ============================================================================= + + +def as_user_entry(entry: TranscriptEntry) -> UserTranscriptEntry | None: + """Return entry as UserTranscriptEntry if it is one, else None.""" + if entry.type == MessageType.USER: + return cast(UserTranscriptEntry, entry) + return None + + +def as_assistant_entry(entry: TranscriptEntry) -> AssistantTranscriptEntry | None: + """Return entry as AssistantTranscriptEntry if it is one, else None.""" + if entry.type == MessageType.ASSISTANT: + return cast(AssistantTranscriptEntry, entry) + return None + + +# ============================================================================= +# Usage Info Normalization +# ============================================================================= + + +def normalize_usage_info(usage_data: Any) -> Optional[UsageInfo]: + """Normalize usage data from various formats to UsageInfo.""" + if usage_data is None: + return None + + # If it's already a UsageInfo instance, return as-is + if isinstance(usage_data, UsageInfo): + return usage_data + + # If it's a dict, validate and convert + if isinstance(usage_data, dict): + return UsageInfo.model_validate(usage_data) + + # Handle object-like access (e.g., from SDK types) + if hasattr(usage_data, "input_tokens"): + server_tool_use = getattr(usage_data, "server_tool_use", None) + if server_tool_use is not None and hasattr(server_tool_use, "model_dump"): + server_tool_use = server_tool_use.model_dump() + return UsageInfo( + input_tokens=getattr(usage_data, "input_tokens", None), + output_tokens=getattr(usage_data, "output_tokens", None), + cache_creation_input_tokens=getattr( + usage_data, "cache_creation_input_tokens", None + ), + cache_read_input_tokens=getattr( + usage_data, "cache_read_input_tokens", None + ), + service_tier=getattr(usage_data, "service_tier", None), + server_tool_use=server_tool_use, + ) + + return None + + +# ============================================================================= +# Content Item Creation +# ============================================================================= + + +def create_content_item( + item_data: dict[str, Any], + type_filter: Sequence[str] | None = None, +) -> ContentItem: + """Create a ContentItem from raw data using the registry. + + Args: + item_data: The raw dictionary data + type_filter: Sequence of content type strings to allow, or None to allow all + (e.g., USER_CONTENT_TYPES, ASSISTANT_CONTENT_TYPES) + + Returns: + ContentItem instance, with fallback to TextContent for unknown types + """ + try: + content_type = item_data.get("type", "") + + if type_filter is None or content_type in type_filter: + model_class = CONTENT_ITEM_CREATORS.get(content_type) + if model_class is not None: + return cast(ContentItem, model_class.model_validate(item_data)) + + # Fallback to text content for unknown/disallowed types + return TextContent(type="text", text=str(item_data)) + except Exception: + return TextContent(type="text", text=str(item_data)) + + +def create_message_content( + content_data: Any, + type_filter: Sequence[str] | None = None, +) -> list[ContentItem]: + """Create a list of ContentItems from message content data. + + Always returns a list for consistent downstream handling. String content + is wrapped in a TextContent item. + + Args: + content_data: Raw content data (string or list of items) + type_filter: Sequence of content type strings to allow, or None to allow all + """ + if isinstance(content_data, str): + return [TextContent(type="text", text=content_data)] + elif isinstance(content_data, list): + content_list = cast(list[Any], content_data) + result: list[ContentItem] = [] + for item in content_list: + if isinstance(item, dict): + result.append( + create_content_item(cast(dict[str, Any], item), type_filter) + ) + else: + # Non-dict items (e.g., raw strings) become TextContent + result.append(TextContent(type="text", text=str(item))) + return result + else: + return [TextContent(type="text", text=str(content_data))] + + +# ============================================================================= +# Transcript Entry Creation +# ============================================================================= + + +def _create_user_entry(data: dict[str, Any]) -> UserTranscriptEntry: + """Create a UserTranscriptEntry from raw data.""" + data_copy = data.copy() + if "message" in data_copy and "content" in data_copy["message"]: + data_copy["message"] = data_copy["message"].copy() + data_copy["message"]["content"] = create_message_content( + data_copy["message"]["content"], + USER_CONTENT_TYPES, + ) + # Parse toolUseResult if present and it's a list of content items + if "toolUseResult" in data_copy and isinstance(data_copy["toolUseResult"], list): + # Check if it's a list of content items (MCP tool results) + tool_use_result = cast(list[Any], data_copy["toolUseResult"]) + if ( + tool_use_result + and isinstance(tool_use_result[0], dict) + and "type" in tool_use_result[0] + ): + data_copy["toolUseResult"] = [ + create_content_item(cast(dict[str, Any], item)) + for item in tool_use_result + if isinstance(item, dict) + ] + return UserTranscriptEntry.model_validate(data_copy) + + +def _create_assistant_entry(data: dict[str, Any]) -> AssistantTranscriptEntry: + """Create an AssistantTranscriptEntry from raw data.""" + data_copy = data.copy() + + if "message" in data_copy and "content" in data_copy["message"]: + message_copy = data_copy["message"].copy() + message_copy["content"] = create_message_content( + message_copy["content"], + ASSISTANT_CONTENT_TYPES, + ) + + # Normalize usage data to support both Anthropic and custom formats + if "usage" in message_copy: + message_copy["usage"] = normalize_usage_info(message_copy["usage"]) + + data_copy["message"] = message_copy + return AssistantTranscriptEntry.model_validate(data_copy) + + +def _create_queue_operation_entry( + data: dict[str, Any], +) -> QueueOperationTranscriptEntry: + """Create a QueueOperationTranscriptEntry from raw data.""" + data_copy = data.copy() + if "content" in data_copy and isinstance(data_copy["content"], list): + data_copy["content"] = create_message_content(data_copy["content"]) + return QueueOperationTranscriptEntry.model_validate(data_copy) + + +# Registry mapping entry types to their creator functions +ENTRY_CREATORS: dict[str, Callable[[dict[str, Any]], TranscriptEntry]] = { + "user": _create_user_entry, + "assistant": _create_assistant_entry, + "summary": lambda data: SummaryTranscriptEntry.model_validate(data), + "system": lambda data: SystemTranscriptEntry.model_validate(data), + "queue-operation": _create_queue_operation_entry, +} + + +def create_transcript_entry(data: dict[str, Any]) -> TranscriptEntry: + """Create a TranscriptEntry from a JSON dictionary. + + Uses a registry-based dispatch to create the appropriate TranscriptEntry + subclass based on the 'type' field in the data. + + Args: + data: Dictionary parsed from JSON + + Returns: + The appropriate TranscriptEntry subclass + + Raises: + ValueError: If the data doesn't match any known transcript entry type + """ + entry_type = data.get("type") + creator = ENTRY_CREATORS.get(entry_type) # type: ignore[arg-type] + if creator is None: + raise ValueError(f"Unknown transcript entry type: {entry_type}") + return creator(data) diff --git a/claude_code_log/factories/user_factory.py b/claude_code_log/factories/user_factory.py new file mode 100644 index 00000000..22b30f66 --- /dev/null +++ b/claude_code_log/factories/user_factory.py @@ -0,0 +1,477 @@ +"""Factory for user transcript entries. + +This module handles creation of MessageContent from user transcript entries: +- SlashCommandMessage: Slash command invocations +- CommandOutputMessage: Local command output +- BashInputMessage: Bash command input +- BashOutputMessage: Bash command output +- UserTextMessage: Regular user text (with optional IDE notifications) +- UserSlashCommandMessage: Expanded slash command prompts (isMeta) +- CompactedSummaryMessage: Compacted conversation summaries +- UserMemoryMessage: User memory content +- UserSteeringMessage: User steering prompts (queue-operation 'remove') + +Also provides: +- is_command_message: Check if text is a slash command +- is_local_command_output: Check if text is local command output +- is_bash_input: Check if text is bash input +- is_bash_output: Check if text is bash output +""" + +import json +import re +from typing import Any, Optional, Union, cast + +from ..models import ( + BashInputMessage, + BashOutputMessage, + CommandOutputMessage, + CompactedSummaryMessage, + ContentItem, + IdeDiagnostic, + IdeNotificationContent, + IdeOpenedFile, + IdeSelection, + ImageContent, + MessageMeta, + SlashCommandMessage, + TextContent, + UserMemoryMessage, + UserSlashCommandMessage, + UserTextMessage, +) + + +# ============================================================================= +# Message Type Detection +# ============================================================================= + + +def is_command_message(text_content: str) -> bool: + """Check if a message contains command information that should be displayed.""" + return "" in text_content and "" in text_content + + +def is_local_command_output(text_content: str) -> bool: + """Check if a message contains local command output.""" + return "" in text_content + + +def is_bash_input(text_content: str) -> bool: + """Check if a message contains bash input command.""" + return "" in text_content and "" in text_content + + +def is_bash_output(text_content: str) -> bool: + """Check if a message contains bash command output.""" + return "" in text_content or "" in text_content + + +# ============================================================================= +# Slash Command Creation +# ============================================================================= + + +def create_slash_command_message( + meta: MessageMeta, + text: str, +) -> Optional[SlashCommandMessage]: + """Create SlashCommandMessage from text containing command tags. + + Args: + text: Raw text that may contain command-name, command-args, command-contents tags + meta: Message metadata + + Returns: + SlashCommandMessage if tags found, None otherwise + """ + command_name_match = re.search(r"([^<]+)", text) + if not command_name_match: + return None + + command_name = command_name_match.group(1).strip() + + command_args_match = re.search(r"([^<]*)", text) + command_args = command_args_match.group(1).strip() if command_args_match else "" + + # Parse command contents, handling JSON format + command_contents_match = re.search( + r"(.+?)", text, re.DOTALL + ) + command_contents = "" + if command_contents_match: + contents_text = command_contents_match.group(1).strip() + # Try to parse as JSON and extract the text field + try: + contents_json: Any = json.loads(contents_text) + if isinstance(contents_json, dict) and "text" in contents_json: + text_dict = cast(dict[str, Any], contents_json) + text_value = text_dict["text"] + command_contents = str(text_value) + else: + command_contents = contents_text + except json.JSONDecodeError: + command_contents = contents_text + + return SlashCommandMessage( + command_name=command_name, + command_args=command_args, + command_contents=command_contents, + meta=meta, + ) + + +def create_command_output_message( + meta: MessageMeta, + text: str, +) -> Optional[CommandOutputMessage]: + """Create CommandOutputMessage from text containing local-command-stdout tags. + + Args: + text: Raw text that may contain local-command-stdout tags + meta: Message metadata + + Returns: + CommandOutputMessage if tags found, None otherwise + """ + stdout_match = re.search( + r"(.*?)", + text, + re.DOTALL, + ) + if not stdout_match: + return None + + stdout_content = stdout_match.group(1).strip() + # Check if content looks like markdown (starts with markdown headers) + is_markdown = bool(re.match(r"^#+\s+", stdout_content, re.MULTILINE)) + + return CommandOutputMessage( + stdout=stdout_content, is_markdown=is_markdown, meta=meta + ) + + +# ============================================================================= +# Bash Input/Output Creation +# ============================================================================= + + +def create_bash_input_message( + meta: MessageMeta, + text: str, +) -> Optional[BashInputMessage]: + """Create BashInputMessage from text containing bash-input tags. + + Args: + text: Raw text that may contain bash-input tags + meta: Message metadata + + Returns: + BashInputMessage if tags found, None otherwise + """ + bash_match = re.search(r"(.*?)", text, re.DOTALL) + if not bash_match: + return None + + return BashInputMessage(command=bash_match.group(1).strip(), meta=meta) + + +def create_bash_output_message( + meta: MessageMeta, + text: str, +) -> Optional[BashOutputMessage]: + """Create BashOutputMessage from text containing bash-stdout/bash-stderr tags. + + Args: + text: Raw text that may contain bash-stdout/bash-stderr tags + meta: Message metadata + + Returns: + BashOutputMessage if tags found, None otherwise + """ + stdout_match = re.search(r"(.*?)", text, re.DOTALL) + stderr_match = re.search(r"(.*?)", text, re.DOTALL) + + if not stdout_match and not stderr_match: + return None + + stdout = stdout_match.group(1).strip() if stdout_match else None + stderr = stderr_match.group(1).strip() if stderr_match else None + + # Convert empty strings to None for cleaner representation + if stdout == "": + stdout = None + if stderr == "": + stderr = None + + return BashOutputMessage(stdout=stdout, stderr=stderr, meta=meta) + + +# ============================================================================= +# IDE Notification Creation +# ============================================================================= + +# Shared regex patterns for IDE notification tags +IDE_OPENED_FILE_PATTERN = re.compile( + r"(.*?)", re.DOTALL +) +IDE_SELECTION_PATTERN = re.compile(r"(.*?)", re.DOTALL) +IDE_DIAGNOSTICS_PATTERN = re.compile( + r"\s*(.*?)\s*", + re.DOTALL, +) + + +def create_ide_notification_content(text: str) -> Optional[IdeNotificationContent]: + """Create IdeNotificationContent from text containing IDE tags. + + Handles: + - : Simple file open notifications + - : Code selection notifications + - : JSON diagnostic arrays + + Args: + text: Raw text that may contain IDE notification tags + + Returns: + IdeNotificationContent if any tags found, None otherwise + """ + opened_files: list[IdeOpenedFile] = [] + selections: list[IdeSelection] = [] + diagnostics: list[IdeDiagnostic] = [] + remaining_text = text + + # Pattern 1: content + for match in IDE_OPENED_FILE_PATTERN.finditer(remaining_text): + content = match.group(1).strip() + opened_files.append(IdeOpenedFile(content=content)) + + remaining_text = IDE_OPENED_FILE_PATTERN.sub("", remaining_text) + + # Pattern 2: content + for match in IDE_SELECTION_PATTERN.finditer(remaining_text): + content = match.group(1).strip() + selections.append(IdeSelection(content=content)) + + remaining_text = IDE_SELECTION_PATTERN.sub("", remaining_text) + + # Pattern 3: JSON + for match in IDE_DIAGNOSTICS_PATTERN.finditer(remaining_text): + json_content = match.group(1).strip() + try: + parsed_diagnostics: Any = json.loads(json_content) + if isinstance(parsed_diagnostics, list): + diagnostics.append( + IdeDiagnostic( + diagnostics=cast(list[dict[str, Any]], parsed_diagnostics) + ) + ) + else: + # Not a list, store as raw content + diagnostics.append(IdeDiagnostic(raw_content=json_content)) + except (json.JSONDecodeError, ValueError): + # JSON parsing failed, store raw content + diagnostics.append(IdeDiagnostic(raw_content=json_content)) + + remaining_text = IDE_DIAGNOSTICS_PATTERN.sub("", remaining_text) + + # Only return if we found any IDE tags + if not opened_files and not selections and not diagnostics: + return None + + return IdeNotificationContent( + opened_files=opened_files, + selections=selections, + diagnostics=diagnostics, + remaining_text=remaining_text.strip(), + ) + + +# ============================================================================= +# Compacted Summary and User Memory Creation +# ============================================================================= + +# Pattern for compacted session summary detection +COMPACTED_SUMMARY_PREFIX = "This session is being continued from a previous conversation that ran out of context" + + +def create_compacted_summary_message( + meta: MessageMeta, + content_list: list[ContentItem], +) -> Optional[CompactedSummaryMessage]: + """Create CompactedSummaryMessage from content list. + + Compacted summaries are generated when a session runs out of context and + needs to be continued. They contain a summary of the previous conversation. + + If the first text item starts with the compacted summary prefix, all text + items are combined into a single CompactedSummaryMessage. + + Args: + content_list: List of ContentItem from user message + meta: Message metadata + + Returns: + CompactedSummaryMessage if first text is a compacted summary, None otherwise + """ + if not content_list or not isinstance(content_list[0], TextContent): + return None + + first_text = content_list[0].text + if not first_text.startswith(COMPACTED_SUMMARY_PREFIX): + return None + + # Combine all text content for compacted summaries + texts = [item.text for item in content_list if isinstance(item, TextContent)] + all_text = "\n\n".join(texts) + return CompactedSummaryMessage(summary_text=all_text, meta=meta) + + +# Pattern for user memory input tag +USER_MEMORY_PATTERN = re.compile( + r"(.*?)", re.DOTALL +) + + +def create_user_memory_message( + meta: MessageMeta, + text: str, +) -> Optional[UserMemoryMessage]: + """Create UserMemoryMessage from text containing user-memory-input tag. + + User memory input contains context that the user has provided from + their CLAUDE.md or other memory sources. + + Args: + text: Raw text that may contain user memory input tag + meta: Message metadata + + Returns: + UserMemoryMessage if tag found, None otherwise + """ + match = USER_MEMORY_PATTERN.search(text) + if match: + memory_content = match.group(1).strip() + return UserMemoryMessage(memory_text=memory_content, meta=meta) + return None + + +# ============================================================================= +# User Message Content Creation +# ============================================================================= + +# Type alias for content models returned by create_user_message +UserMessageContent = Union[ + SlashCommandMessage, + CommandOutputMessage, + BashInputMessage, + BashOutputMessage, + CompactedSummaryMessage, + UserMemoryMessage, + UserSlashCommandMessage, + UserTextMessage, +] + + +def create_user_message( + meta: MessageMeta, + content_list: list[ContentItem], + text_content: str, + is_slash_command: bool = False, +) -> Optional[UserMessageContent]: + """Create a user message content model from content items. + + This is the main entry point for creating user message content. + It handles all user message types by detecting patterns in the text: + - Slash commands (, ) + - Local command output () + - Bash input () + - Bash output (, ) + - Compacted summaries (special prefix) + - User memory () + - Slash command expanded prompts (isMeta=True) + - Regular user text with IDE notifications + + Args: + content_list: List of ContentItem from user message + text_content: Pre-extracted text content for pattern detection + is_slash_command: True for slash command expanded prompts (isMeta=True) + meta: Message metadata + + Returns: + A content model, or None if content_list is empty. + """ + if not content_list: + return None + + # Check for special message patterns first (before generic parsing) + if is_command_message(text_content): + return create_slash_command_message(meta, text_content) + + if is_local_command_output(text_content): + return create_command_output_message(meta, text_content) + + if is_bash_input(text_content): + return create_bash_input_message(meta, text_content) + + if is_bash_output(text_content): + return create_bash_output_message(meta, text_content) + + # Slash command expanded prompts - combine all text as markdown + if is_slash_command: + all_text = "\n\n".join( + getattr(item, "text", "") for item in content_list if hasattr(item, "text") + ) + return UserSlashCommandMessage(text=all_text, meta=meta) if all_text else None + + # Get first text item for special case detection + first_text_item = next( + (item for item in content_list if hasattr(item, "text")), + None, + ) + first_text = getattr(first_text_item, "text", "") if first_text_item else "" + + # Check for compacted session summary first (handles text combining internally) + if compacted := create_compacted_summary_message(meta, content_list): + return compacted + + # Check for user memory input + if user_memory := create_user_memory_message(meta, first_text): + return user_memory + + # Build items list preserving order, extracting IDE notifications from text + items: list[TextContent | ImageContent | IdeNotificationContent] = [] + + for item in content_list: + # Check for text content + if hasattr(item, "text"): + item_text: str = getattr(item, "text") # type: ignore[assignment] + + if ide_content := create_ide_notification_content(item_text): + # Add IDE notification item first + items.append(ide_content) + remaining_text: str = ide_content.remaining_text + else: + remaining_text = item_text + + # Add remaining text as TextContent if non-empty + if remaining_text.strip(): + items.append(TextContent(type="text", text=remaining_text)) + elif isinstance(item, ImageContent): + # ImageContent model - use as-is + items.append(item) + elif hasattr(item, "source") and getattr(item, "type", None) == "image": + # Duck-typed image content - convert to our Pydantic model + items.append(ImageContent.model_validate(item.model_dump())) # type: ignore[union-attr] + + # Extract text content from items for dedup matching and simple renderers + raw_text_content = "\n".join( + item.text for item in items if isinstance(item, TextContent) + ) + + # Return UserTextMessage with items list and cached raw text + return UserTextMessage( + items=items, + raw_text_content=raw_text_content if raw_text_content else None, + meta=meta, + ) diff --git a/claude_code_log/html/__init__.py b/claude_code_log/html/__init__.py index ef847ef9..168ba9e2 100644 --- a/claude_code_log/html/__init__.py +++ b/claude_code_log/html/__init__.py @@ -8,6 +8,7 @@ escape_html, get_message_emoji, get_template_environment, + is_session_header, render_collapsible_code, render_file_content_collapsible, render_markdown, @@ -15,25 +16,32 @@ starts_with_emoji, ) from .tool_formatters import ( - format_askuserquestion_content, + # Tool input formatters (called by HtmlRenderer.format_{InputClass}) + format_askuserquestion_input, + format_bash_input, + format_edit_input, + format_exitplanmode_input, + format_multiedit_input, + format_read_input, + format_task_input, + format_todowrite_input, + format_write_input, + # Tool output formatters (called by HtmlRenderer.format_{OutputClass}) + format_askuserquestion_output, + format_bash_output, + format_edit_output, + format_exitplanmode_output, + format_read_output, + format_task_output, + format_write_output, + # Fallback formatter + format_tool_result_content_raw, + # Legacy formatters (still used) format_askuserquestion_result, - format_bash_tool_content, - format_edit_tool_content, - format_edit_tool_result, - format_exitplanmode_content, format_exitplanmode_result, - format_multiedit_tool_content, - format_read_tool_content, - format_read_tool_result, - format_task_tool_content, - format_todowrite_content, - format_tool_result_content, - format_tool_use_content, + # Tool summary and title format_tool_use_title, - format_write_tool_content, get_tool_summary, - parse_edit_output, - parse_read_output, render_params_table, ) from .system_formatters import ( @@ -43,28 +51,28 @@ format_system_content, ) from ..models import ( - AssistantTextContent, - BashInputContent, - BashOutputContent, - CommandOutputContent, - CompactedSummaryContent, - DedupNoticeContent, + AssistantTextMessage, + BashInputMessage, + BashOutputMessage, + CommandOutputMessage, + CompactedSummaryMessage, + DedupNoticeMessage, IdeDiagnostic, IdeNotificationContent, IdeOpenedFile, IdeSelection, - SessionHeaderContent, - SlashCommandContent, - ThinkingContentModel, - UserMemoryContent, - UserTextContent, + SessionHeaderMessage, + SlashCommandMessage, + ThinkingMessage, + UserMemoryMessage, + UserTextMessage, ) -from ..parser import ( - parse_bash_input, - parse_bash_output, - parse_command_output, - parse_ide_notifications, - parse_slash_command, +from ..factories import ( + create_bash_input_message, + create_bash_output_message, + create_command_output_message, + create_ide_notification_content, + create_slash_command_message, ) from .user_formatters import ( format_bash_input_content, @@ -89,49 +97,55 @@ "escape_html", "get_message_emoji", "get_template_environment", + "is_session_header", "render_collapsible_code", "render_file_content_collapsible", "render_markdown", "render_markdown_collapsible", "starts_with_emoji", - # tool_formatters (input) - "format_askuserquestion_content", + # tool_formatters (input) - called by HtmlRenderer.format_{InputClass} + "format_askuserquestion_input", + "format_bash_input", + "format_edit_input", + "format_exitplanmode_input", + "format_multiedit_input", + "format_read_input", + "format_task_input", + "format_todowrite_input", + "format_write_input", + # tool_formatters (output) - called by HtmlRenderer.format_{OutputClass} + "format_askuserquestion_output", + "format_bash_output", + "format_edit_output", + "format_exitplanmode_output", + "format_read_output", + "format_task_output", + "format_write_output", + # Fallback formatter + "format_tool_result_content_raw", + # Legacy formatters (still used) "format_askuserquestion_result", - "format_bash_tool_content", - "format_edit_tool_content", - "format_exitplanmode_content", "format_exitplanmode_result", - "format_multiedit_tool_content", - "format_read_tool_content", - "format_task_tool_content", - "format_todowrite_content", - "format_tool_use_content", + # Tool summary and title "format_tool_use_title", - "format_write_tool_content", "get_tool_summary", "render_params_table", - # tool_formatters (output/result) - "parse_read_output", - "format_read_tool_result", - "parse_edit_output", - "format_edit_tool_result", - "format_tool_result_content", # system_formatters "format_dedup_notice_content", "format_hook_summary_content", "format_session_header_content", "format_system_content", # system content models - "DedupNoticeContent", - "SessionHeaderContent", + "DedupNoticeMessage", + "SessionHeaderMessage", # user_formatters (content models) - "SlashCommandContent", - "CommandOutputContent", - "BashInputContent", - "BashOutputContent", - "CompactedSummaryContent", - "UserMemoryContent", - "UserTextContent", + "SlashCommandMessage", + "CommandOutputMessage", + "BashInputMessage", + "BashOutputMessage", + "CompactedSummaryMessage", + "UserMemoryMessage", + "UserTextMessage", "IdeNotificationContent", "IdeOpenedFile", "IdeSelection", @@ -146,15 +160,15 @@ "format_user_text_content", "format_user_text_model_content", "format_ide_notification_content", - # user_formatters (parsing) - "parse_slash_command", - "parse_command_output", - "parse_bash_input", - "parse_bash_output", - "parse_ide_notifications", + # user_factory (message creation) + "create_slash_command_message", + "create_command_output_message", + "create_bash_input_message", + "create_bash_output_message", + "create_ide_notification_content", # assistant_formatters (content models) - "AssistantTextContent", - "ThinkingContentModel", + "AssistantTextMessage", + "ThinkingMessage", # assistant_formatters (formatting) "format_assistant_text_content", "format_thinking_content", diff --git a/claude_code_log/html/assistant_formatters.py b/claude_code_log/html/assistant_formatters.py index cbfe41d0..9bd725f1 100644 --- a/claude_code_log/html/assistant_formatters.py +++ b/claude_code_log/html/assistant_formatters.py @@ -2,19 +2,19 @@ This module formats assistant message content types to HTML. Part of the thematic formatter organization: -- system_formatters.py: SystemContent, HookSummaryContent -- user_formatters.py: SlashCommandContent, CommandOutputContent, BashInputContent -- assistant_formatters.py: AssistantTextContent, ThinkingContentModel, ImageContent +- system_formatters.py: SystemMessage, HookSummaryMessage +- user_formatters.py: SlashCommandMessage, CommandOutputMessage, BashInputMessage +- assistant_formatters.py: AssistantTextMessage, ThinkingMessage, ImageContent - tool_formatters.py: tool use/result content Content models are defined in models.py, this module only handles formatting. """ from ..models import ( - AssistantTextContent, + AssistantTextMessage, ImageContent, - ThinkingContentModel, - UnknownContent, + ThinkingMessage, + UnknownMessage, ) from .utils import escape_html, render_markdown_collapsible @@ -25,7 +25,7 @@ def format_assistant_text_content( - content: AssistantTextContent, + content: AssistantTextMessage, line_threshold: int = 30, preview_line_count: int = 10, ) -> str: @@ -36,7 +36,7 @@ def format_assistant_text_content( - ImageContent: Rendered as inline tag with base64 data URL Args: - content: AssistantTextContent with text/items to render + content: AssistantTextMessage with text/items to render line_threshold: Number of lines before content becomes collapsible preview_line_count: Number of preview lines to show when collapsed @@ -60,14 +60,14 @@ def format_assistant_text_content( def format_thinking_content( - content: ThinkingContentModel, + content: ThinkingMessage, line_threshold: int = 20, preview_line_count: int = 5, ) -> str: """Format thinking content as HTML. Args: - content: ThinkingContentModel with the thinking text + content: ThinkingMessage with the thinking text line_threshold: Number of lines before content becomes collapsible preview_line_count: Number of preview lines to show when collapsed @@ -95,11 +95,11 @@ def format_image_content(image: ImageContent) -> str: return f'Uploaded image' -def format_unknown_content(content: UnknownContent) -> str: +def format_unknown_content(content: UnknownMessage) -> str: """Format unknown content type as HTML. Args: - content: UnknownContent with the type name + content: UnknownMessage with the type name Returns: HTML paragraph with escaped type name diff --git a/claude_code_log/html/renderer.py b/claude_code_log/html/renderer.py index c05bc85d..893768e2 100644 --- a/claude_code_log/html/renderer.py +++ b/claude_code_log/html/renderer.py @@ -1,30 +1,47 @@ """HTML renderer implementation for Claude Code transcripts.""" -from functools import partial from pathlib import Path -from typing import TYPE_CHECKING, Any, Callable, Optional, Tuple +from typing import TYPE_CHECKING, Any, Optional, Tuple, cast from ..cache import get_library_version from ..models import ( - AssistantTextContent, - BashInputContent, - BashOutputContent, - CommandOutputContent, - CompactedSummaryContent, - DedupNoticeContent, - HookSummaryContent, - SessionHeaderContent, - SlashCommandContent, - SystemContent, - ThinkingContentModel, - ToolResultContent, - ToolResultContentModel, - ToolUseContent, + AssistantTextMessage, + BashInputMessage, + BashOutputMessage, + CommandOutputMessage, + CompactedSummaryMessage, + DedupNoticeMessage, + HookSummaryMessage, + SessionHeaderMessage, + SlashCommandMessage, + SystemMessage, + ThinkingMessage, + ToolUseMessage, TranscriptEntry, - UnknownContent, - UserMemoryContent, - UserSlashCommandContent, - UserTextContent, + UnknownMessage, + UserMemoryMessage, + UserSlashCommandMessage, + UserTextMessage, + # Tool input types + AskUserQuestionInput, + BashInput, + EditInput, + ExitPlanModeInput, + MultiEditInput, + ReadInput, + TaskInput, + TodoWriteInput, + ToolUseContent, + WriteInput, + # Tool output types + AskUserQuestionOutput, + BashOutput, + EditOutput, + ExitPlanModeOutput, + ReadOutput, + TaskOutput, + ToolResultContent, + WriteOutput, ) from ..renderer import ( Renderer, @@ -33,7 +50,13 @@ prepare_projects_index, title_for_projects_index, ) -from ..renderer_timings import log_timing +from ..renderer_timings import ( + DEBUG_TIMING, + log_timing, + report_timing_statistics, + set_timing_var, +) +from ..utils import format_timestamp from .system_formatters import ( format_dedup_notice_content, format_hook_summary_content, @@ -55,8 +78,33 @@ format_thinking_content, format_unknown_content, ) -from .tool_formatters import format_tool_result_content, format_tool_use_content -from .utils import css_class_from_message, get_message_emoji, get_template_environment +from .tool_formatters import ( + format_askuserquestion_input, + format_askuserquestion_output, + format_bash_input, + format_bash_output, + format_edit_input, + format_edit_output, + format_exitplanmode_input, + format_exitplanmode_output, + format_multiedit_input, + format_read_input, + format_read_output, + format_task_input, + format_task_output, + format_todowrite_input, + format_tool_result_content_raw, + format_write_input, + format_write_output, + render_params_table, +) +from .utils import ( + css_class_from_message, + escape_html, + get_message_emoji, + get_template_environment, + is_session_header, +) if TYPE_CHECKING: from ..cache import CacheManager @@ -94,77 +142,226 @@ def check_html_version(html_file_path: Path) -> Optional[str]: class HtmlRenderer(Renderer): """HTML renderer for Claude Code transcripts.""" - def _build_dispatcher(self) -> dict[type, Callable[..., str]]: - """Build content type to HTML formatter mapping. + # ------------------------------------------------------------------------- + # System Content Formatters + # ------------------------------------------------------------------------- - Maps MessageContent subclasses to their HTML formatting functions. - Handlers receive the content directly (not the full TemplateMessage). - The cast to the correct type happens in format_content(). - """ - return { - # System content types - SystemContent: format_system_content, - HookSummaryContent: format_hook_summary_content, - SessionHeaderContent: format_session_header_content, - DedupNoticeContent: format_dedup_notice_content, - # User content types - SlashCommandContent: format_slash_command_content, - CommandOutputContent: format_command_output_content, - BashInputContent: format_bash_input_content, - BashOutputContent: format_bash_output_content, - CompactedSummaryContent: format_compacted_summary_content, - UserMemoryContent: format_user_memory_content, - UserSlashCommandContent: format_user_slash_command_content, - UserTextContent: format_user_text_model_content, - # Assistant content types - ThinkingContentModel: partial(format_thinking_content, line_threshold=10), - AssistantTextContent: format_assistant_text_content, - UnknownContent: format_unknown_content, - # Tool content types - ToolUseContent: format_tool_use_content, - ToolResultContentModel: self._format_tool_result_content, - } - - def _format_tool_result_content(self, content: ToolResultContentModel) -> str: - """Format ToolResultContentModel with associated tool context.""" - tool_result = ToolResultContent( - type="tool_result", - tool_use_id=content.tool_use_id, - content=content.content, - is_error=content.is_error, - ) - return format_tool_result_content( - tool_result, - content.file_path, - content.tool_name, + def format_SystemMessage(self, message: SystemMessage) -> str: + return format_system_content(message) + + def format_HookSummaryMessage(self, message: HookSummaryMessage) -> str: + return format_hook_summary_content(message) + + def format_SessionHeaderMessage(self, message: SessionHeaderMessage) -> str: + return format_session_header_content(message) + + def format_DedupNoticeMessage(self, message: DedupNoticeMessage) -> str: + return format_dedup_notice_content(message) + + # ------------------------------------------------------------------------- + # User Content Formatters + # ------------------------------------------------------------------------- + + def format_UserTextMessage(self, message: UserTextMessage) -> str: + return format_user_text_model_content(message) + + def format_UserSlashCommandMessage(self, message: UserSlashCommandMessage) -> str: + return format_user_slash_command_content(message) + + def format_SlashCommandMessage(self, message: SlashCommandMessage) -> str: + return format_slash_command_content(message) + + def format_CommandOutputMessage(self, message: CommandOutputMessage) -> str: + return format_command_output_content(message) + + def format_BashInputMessage(self, message: BashInputMessage) -> str: + return format_bash_input_content(message) + + def format_BashOutputMessage(self, message: BashOutputMessage) -> str: + return format_bash_output_content(message) + + def format_CompactedSummaryMessage(self, message: CompactedSummaryMessage) -> str: + return format_compacted_summary_content(message) + + def format_UserMemoryMessage(self, message: UserMemoryMessage) -> str: + return format_user_memory_content(message) + + # ------------------------------------------------------------------------- + # Assistant Content Formatters + # ------------------------------------------------------------------------- + + def format_AssistantTextMessage(self, message: AssistantTextMessage) -> str: + return format_assistant_text_content(message) + + def format_ThinkingMessage(self, message: ThinkingMessage) -> str: + return format_thinking_content(message, line_threshold=10) + + def format_UnknownMessage(self, message: UnknownMessage) -> str: + return format_unknown_content(message) + + # ------------------------------------------------------------------------- + # Tool Input Formatters + # ------------------------------------------------------------------------- + + def format_BashInput(self, input: BashInput) -> str: + return format_bash_input(input) + + def format_ReadInput(self, input: ReadInput) -> str: + return format_read_input(input) + + def format_WriteInput(self, input: WriteInput) -> str: + return format_write_input(input) + + def format_EditInput(self, input: EditInput) -> str: + return format_edit_input(input) + + def format_MultiEditInput(self, input: MultiEditInput) -> str: + return format_multiedit_input(input) + + def format_TaskInput(self, input: TaskInput) -> str: + return format_task_input(input) + + def format_TodoWriteInput(self, input: TodoWriteInput) -> str: + return format_todowrite_input(input) + + def format_AskUserQuestionInput(self, input: AskUserQuestionInput) -> str: + return format_askuserquestion_input(input) + + def format_ExitPlanModeInput(self, input: ExitPlanModeInput) -> str: + return format_exitplanmode_input(input) + + def format_ToolUseContent(self, content: ToolUseContent) -> str: + return render_params_table(content.input) + + # ------------------------------------------------------------------------- + # Tool Output Formatters + # ------------------------------------------------------------------------- + + def format_ReadOutput(self, output: ReadOutput) -> str: + return format_read_output(output) + + def format_WriteOutput(self, output: WriteOutput) -> str: + return format_write_output(output) + + def format_EditOutput(self, output: EditOutput) -> str: + return format_edit_output(output) + + def format_BashOutput(self, output: BashOutput) -> str: + return format_bash_output(output) + + def format_TaskOutput(self, output: TaskOutput) -> str: + return format_task_output(output) + + def format_AskUserQuestionOutput(self, output: AskUserQuestionOutput) -> str: + return format_askuserquestion_output(output) + + def format_ExitPlanModeOutput(self, output: ExitPlanModeOutput) -> str: + return format_exitplanmode_output(output) + + def format_ToolResultContent(self, output: ToolResultContent) -> str: + return format_tool_result_content_raw(output) + + # ------------------------------------------------------------------------- + # Tool Input Title Methods (for Renderer.title_ToolUseMessage dispatch) + # ------------------------------------------------------------------------- + + def _tool_title( + self, message: TemplateMessage, icon: str, summary: Optional[str] = None + ) -> str: + """Format tool title with icon and optional summary.""" + content = cast(ToolUseMessage, message.content) + escaped_name = escape_html(content.tool_name) + prefix = f"{icon} " if icon else "" + if summary: + escaped_summary = escape_html(summary) + return f"{prefix}{escaped_name} {escaped_summary}" + return f"{prefix}{escaped_name}" + + def title_TodoWriteInput(self, message: TemplateMessage) -> str: # noqa: ARG002 + return "πŸ“ Todo List" + + def title_TaskInput(self, message: TemplateMessage) -> str: + content = cast(ToolUseMessage, message.content) + input = cast(TaskInput, content.input) + escaped_name = escape_html(content.tool_name) + escaped_subagent = ( + escape_html(input.subagent_type) if input.subagent_type else "" ) + if input.description and input.subagent_type: + escaped_desc = escape_html(input.description) + return f"πŸ”§ {escaped_name} {escaped_desc} ({escaped_subagent})" + elif input.description: + return self._tool_title(message, "πŸ”§", input.description) + elif input.subagent_type: + return f"πŸ”§ {escaped_name} ({escaped_subagent})" + return f"πŸ”§ {escaped_name}" + + def title_EditInput(self, message: TemplateMessage) -> str: + input = cast(EditInput, cast(ToolUseMessage, message.content).input) + return self._tool_title(message, "πŸ“", input.file_path) + + def title_WriteInput(self, message: TemplateMessage) -> str: + input = cast(WriteInput, cast(ToolUseMessage, message.content).input) + return self._tool_title(message, "πŸ“", input.file_path) + + def title_ReadInput(self, message: TemplateMessage) -> str: + input = cast(ReadInput, cast(ToolUseMessage, message.content).input) + return self._tool_title(message, "πŸ“„", input.file_path) + + def title_BashInput(self, message: TemplateMessage) -> str: + input = cast(BashInput, cast(ToolUseMessage, message.content).input) + return self._tool_title(message, "πŸ’»", input.description) def _flatten_preorder( self, roots: list[TemplateMessage] - ) -> list[Tuple[TemplateMessage, str]]: + ) -> Tuple[ + list[Tuple[TemplateMessage, str, str, str]], + list[Tuple[str, list[Tuple[float, str]]]], + ]: """Flatten message tree via pre-order traversal, formatting each message. - Traverses the tree depth-first (pre-order), formats each message's - content to HTML, and builds a flat list of (message, html) pairs. + Traverses the tree depth-first (pre-order), computes title and formats + content to HTML, building a flat list of (message, title, html, timestamp) tuples. + + Also tracks timing statistics for Markdown and Pygments operations when + DEBUG_TIMING is enabled. Args: roots: Root messages (typically session headers) with children populated Returns: - Flat list of (message, html_content) tuples in pre-order + Tuple of: + - Flat list of (message, title, html_content, formatted_timestamp) tuples + - Operation timing data for reporting: [("Markdown", timings), ("Pygments", timings)] """ - flat: list[Tuple[TemplateMessage, str]] = [] + flat: list[Tuple[TemplateMessage, str, str, str]] = [] + + # Initialize timing tracking for expensive operations + markdown_timings: list[Tuple[float, str]] = [] + pygments_timings: list[Tuple[float, str]] = [] + set_timing_var("_markdown_timings", markdown_timings) + set_timing_var("_pygments_timings", pygments_timings) def visit(msg: TemplateMessage) -> None: + # Update current message ID for timing tracking + set_timing_var("_current_msg_id", msg.message_id) + title = self.title_content(msg) html = self.format_content(msg) - flat.append((msg, html)) + formatted_ts = format_timestamp(msg.meta.timestamp if msg.meta else None) + flat.append((msg, title, html, formatted_ts)) for child in msg.children: visit(child) for root in roots: visit(root) - return flat + # Return timing data for reporting + operation_timings: list[Tuple[str, list[Tuple[float, str]]]] = [ + ("Markdown", markdown_timings), + ("Pygments", pygments_timings), + ] + + return flat, operation_timings def generate( self, @@ -185,7 +382,11 @@ def generate( # Flatten tree via pre-order traversal, formatting content along the way with log_timing("Content formatting (pre-order)", t_start): - template_messages = self._flatten_preorder(root_messages) + template_messages, operation_timings = self._flatten_preorder(root_messages) + + # Report timing statistics for Markdown/Pygments operations + if DEBUG_TIMING: + report_timing_statistics([], operation_timings) # Render template with log_timing("Template environment setup", t_start): @@ -204,6 +405,7 @@ def generate( library_version=get_library_version(), css_class_from_message=css_class_from_message, get_message_emoji=get_message_emoji, + is_session_header=is_session_header, ) ) diff --git a/claude_code_log/html/system_formatters.py b/claude_code_log/html/system_formatters.py index 72eff8b8..2d71876d 100644 --- a/claude_code_log/html/system_formatters.py +++ b/claude_code_log/html/system_formatters.py @@ -2,28 +2,28 @@ This module formats SystemTranscriptEntry-derived content types to HTML. Part of the thematic formatter organization: -- system_formatters.py: SystemContent, HookSummaryContent -- user_formatters.py: (future) user message variants -- assistant_formatters.py: (future) assistant message variants -- tool_renderers.py: tool use/result content +- system_formatters.py: SystemMessage, HookSummaryMessage +- user_formatters.py: SlashCommandMessage, CommandOutputMessage, etc. +- assistant_formatters.py: AssistantTextMessage, ThinkingMessage, ImageContent +- tool_formatters.py: tool use/result content """ import html from .ansi_colors import convert_ansi_to_html from ..models import ( - DedupNoticeContent, - HookSummaryContent, - SessionHeaderContent, - SystemContent, + DedupNoticeMessage, + HookSummaryMessage, + SessionHeaderMessage, + SystemMessage, ) -def format_system_content(content: SystemContent) -> str: +def format_system_content(content: SystemMessage) -> str: """Format a system message with level-specific icon. Args: - content: SystemContent with level and text + content: SystemMessage with level and text Returns: HTML with icon and ANSI-converted text @@ -33,13 +33,13 @@ def format_system_content(content: SystemContent) -> str: return f"{level_icon} {html_content}" -def format_hook_summary_content(content: HookSummaryContent) -> str: +def format_hook_summary_content(content: HookSummaryMessage) -> str: """Format a hook summary as collapsible details. Shows a compact summary with expandable hook commands and error output. Args: - content: HookSummaryContent with execution details + content: HookSummaryMessage with execution details Returns: HTML with collapsible details section @@ -79,11 +79,11 @@ def format_hook_summary_content(content: HookSummaryContent) -> str: """ -def format_session_header_content(content: SessionHeaderContent) -> str: +def format_session_header_content(content: SessionHeaderMessage) -> str: """Format a session header as HTML. Args: - content: SessionHeaderContent with title, session_id, and optional summary + content: SessionHeaderMessage with title, session_id, and optional summary Returns: HTML for the session header display @@ -92,11 +92,11 @@ def format_session_header_content(content: SessionHeaderContent) -> str: return escaped_title -def format_dedup_notice_content(content: DedupNoticeContent) -> str: +def format_dedup_notice_content(content: DedupNoticeMessage) -> str: """Format a deduplication notice as HTML. Args: - content: DedupNoticeContent with notice text and optional target link + content: DedupNoticeMessage with notice text and optional target link Returns: HTML for the dedup notice display with optional anchor link diff --git a/claude_code_log/html/templates/components/search.html b/claude_code_log/html/templates/components/search.html index cc36d353..634e2006 100644 --- a/claude_code_log/html/templates/components/search.html +++ b/claude_code_log/html/templates/components/search.html @@ -158,7 +158,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } diff --git a/claude_code_log/html/templates/components/session_nav.html b/claude_code_log/html/templates/components/session_nav.html index 99568db2..143d4a89 100644 --- a/claude_code_log/html/templates/components/session_nav.html +++ b/claude_code_log/html/templates/components/session_nav.html @@ -12,7 +12,7 @@

Session Navigation

{% for session in sessions %} -
{% if session.summary %} diff --git a/claude_code_log/html/templates/transcript.html b/claude_code_log/html/templates/transcript.html index 3957cfe7..ef27d002 100644 --- a/claude_code_log/html/templates/transcript.html +++ b/claude_code_log/html/templates/transcript.html @@ -69,15 +69,11 @@

πŸ” Search & Filter

{{ render_session_nav(sessions, "toc") }} {% endif %} - {% for message, html_content in messages %} - {% if message.is_session_header %} + {% for message, message_title, html_content, formatted_timestamp in messages %} + {% if is_session_header(message) %}
-
+
Session: {{ html_content }}
- {% if message.session_subtitle %} -
{{ - message.session_subtitle }} ({{message.session_id}})
- {% endif %} {% if message.has_children %}
{% if message.immediate_children_count == message.total_descendants_count %} @@ -102,16 +98,16 @@

πŸ” Search & Filter

{% else %} {%- set msg_css_class = css_class_from_message(message) %} - {% set markdown = message.has_markdown %} -
+ {% set markdown = message.content.has_markdown if message.content else false %} +
{% set msg_emoji = get_message_emoji(message) -%} - {% if message.message_title %}{% - if message.message_title == 'Memory' %}πŸ’­ {% - elif msg_emoji and (message.type != 'tool_use' or not starts_with_emoji(message.message_title)) %}{{ msg_emoji }} {% endif %}{{ message.message_title | safe }}{% endif %} + {% if message_title %}{% + if message_title == 'Memory' %}πŸ’­ {% + elif msg_emoji and (message.type != 'tool_use' or not starts_with_emoji(message_title)) %}{{ msg_emoji }} {% endif %}{{ message_title | safe }}{% endif %}
- {{ message.formatted_timestamp }} + {{ formatted_timestamp }}
{% if message.token_usage %} {{ message.token_usage }} @@ -531,8 +527,8 @@

πŸ” Search & Filter

if (!classList.includes(targetId)) return false; // Check if it's an immediate child by finding all ancestor IDs - // Ancestor IDs can be either d-XXX (regular messages) or session-XXX (sessions) - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + // Ancestor IDs use unified d-XXX format + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; return lastAncestor === targetId; }); @@ -630,7 +626,7 @@

πŸ” Search & Filter

// Fold: show first level only, hide deeper descendants (State C β†’ State B) descendants.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; // Show immediate children, hide non-immediate children @@ -684,7 +680,7 @@

πŸ” Search & Filter

allMessages.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); if (ancestorIds.length > 0) { const lastAncestor = ancestorIds[ancestorIds.length - 1]; diff --git a/claude_code_log/html/tool_formatters.py b/claude_code_log/html/tool_formatters.py index 962a9f2f..9b0f6ac0 100644 --- a/claude_code_log/html/tool_formatters.py +++ b/claude_code_log/html/tool_formatters.py @@ -28,18 +28,23 @@ from ..models import ( AskUserQuestionInput, AskUserQuestionItem, + AskUserQuestionOutput, BashInput, + BashOutput, EditInput, EditOutput, ExitPlanModeInput, + ExitPlanModeOutput, MultiEditInput, ReadInput, ReadOutput, TaskInput, + TaskOutput, TodoWriteInput, + ToolInput, ToolResultContent, - ToolUseContent, WriteInput, + WriteOutput, ) from .ansi_colors import convert_ansi_to_html from .renderer_code import render_single_diff @@ -81,7 +86,7 @@ def _render_question_item(q: AskUserQuestionItem) -> str: return "".join(html_parts) -def format_askuserquestion_content(ask_input: AskUserQuestionInput) -> str: +def format_askuserquestion_input(ask_input: AskUserQuestionInput) -> str: """Format AskUserQuestion tool use content with prominent question display. Args: @@ -161,7 +166,7 @@ def format_askuserquestion_result(content: str) -> str: # -- ExitPlanMode Tool -------------------------------------------------------- -def format_exitplanmode_content(exit_input: ExitPlanModeInput) -> str: +def format_exitplanmode_input(exit_input: ExitPlanModeInput) -> str: """Format ExitPlanMode tool use content with collapsible plan markdown. Args: @@ -202,7 +207,7 @@ def format_exitplanmode_result(content: str) -> str: # -- TodoWrite Tool ----------------------------------------------------------- -def format_todowrite_content(todo_input: TodoWriteInput) -> str: +def format_todowrite_input(todo_input: TodoWriteInput) -> str: """Format TodoWrite tool use content as a todo list. Args: @@ -251,7 +256,7 @@ def format_todowrite_content(todo_input: TodoWriteInput) -> str: # -- File Tools (Read/Write) -------------------------------------------------- -def format_read_tool_content(read_input: ReadInput) -> str: # noqa: ARG001 +def format_read_input(read_input: ReadInput) -> str: # noqa: ARG001 """Format Read tool use content showing file path. Args: @@ -264,102 +269,11 @@ def format_read_tool_content(read_input: ReadInput) -> str: # noqa: ARG001 return "" -# -- Tool Result Parsing (cat-n format) --------------------------------------- +# -- Tool Result Formatting --------------------------------------------------- +# Parsing (parse_read_output, parse_edit_output) is now in factories/tool_factory.py -def _parse_cat_n_snippet( - lines: list[str], start_idx: int = 0 -) -> Optional[tuple[str, Optional[str], int]]: - """Parse cat-n formatted snippet from lines. - - Args: - lines: List of lines to parse - start_idx: Index to start parsing from (default: 0) - - Returns: - Tuple of (code_content, system_reminder, line_offset) or None if not parseable - """ - code_lines: list[str] = [] - system_reminder: Optional[str] = None - in_system_reminder = False - line_offset = 1 # Default offset - - for line in lines[start_idx:]: - # Check for system-reminder start - if "" in line: - in_system_reminder = True - system_reminder = "" - continue - - # Check for system-reminder end - if "" in line: - in_system_reminder = False - continue - - # If in system reminder, accumulate reminder text - if in_system_reminder: - if system_reminder is not None: - system_reminder += line + "\n" - continue - - # Parse regular code line (format: " 123β†’content") - match = re.match(r"\s+(\d+)β†’(.*)$", line) - if match: - line_num = int(match.group(1)) - # Capture the first line number as offset - if not code_lines: - line_offset = line_num - code_lines.append(match.group(2)) - elif line.strip() == "": # Allow empty lines between cat-n lines - continue - else: # Non-matching non-empty line, stop parsing - break - - if not code_lines: - return None - - # Join code lines and trim trailing reminder text - code_content = "\n".join(code_lines) - if system_reminder: - system_reminder = system_reminder.strip() - - return (code_content, system_reminder, line_offset) - - -def parse_read_output(content: str, file_path: str) -> Optional[ReadOutput]: - """Parse Read tool result into structured content. - - Args: - content: Raw tool result string - file_path: Path to the file that was read - - Returns: - ReadOutput if parsing succeeds, None otherwise - """ - # Check if content matches the cat-n format pattern (line_number β†’ content) - lines = content.split("\n") - if not lines or not re.match(r"\s+\d+β†’", lines[0]): - return None - - result = _parse_cat_n_snippet(lines) - if result is None: - return None - - code_content, system_reminder, line_offset = result - num_lines = len(code_content.split("\n")) - - return ReadOutput( - file_path=file_path, - content=code_content, - start_line=line_offset, - num_lines=num_lines, - total_lines=num_lines, # We don't know total from result - is_truncated=False, # Can't determine from result - system_reminder=system_reminder, - ) - - -def format_read_tool_result(output: ReadOutput) -> str: +def format_read_output(output: ReadOutput) -> str: """Format Read tool result as HTML with syntax highlighting. Args: @@ -385,50 +299,7 @@ def format_read_tool_result(output: ReadOutput) -> str: ) -def parse_edit_output(content: str, file_path: str) -> Optional[EditOutput]: - """Parse Edit tool result into structured content. - - Edit tool results typically have format: - "The file ... has been updated. Here's the result of running `cat -n` on a snippet..." - followed by cat-n formatted lines. - - Args: - content: Raw tool result string - file_path: Path to the file that was edited - - Returns: - EditOutput if parsing succeeds, None otherwise - """ - # Look for the cat-n snippet after the preamble - # Pattern: look for first line that matches the cat-n format - lines = content.split("\n") - code_start_idx = None - - for i, line in enumerate(lines): - if re.match(r"\s+\d+β†’", line): - code_start_idx = i - break - - if code_start_idx is None: - return None - - result = _parse_cat_n_snippet(lines, code_start_idx) - if result is None: - return None - - code_content, _system_reminder, line_offset = result - # Edit tool doesn't use system_reminder - - return EditOutput( - file_path=file_path, - success=True, # If we got here, edit succeeded - diffs=[], # We don't have diff info from result - message=code_content, - start_line=line_offset, - ) - - -def format_edit_tool_result(output: EditOutput) -> str: +def format_edit_output(output: EditOutput) -> str: """Format Edit tool result as HTML with syntax highlighting. Args: @@ -445,7 +316,103 @@ def format_edit_tool_result(output: EditOutput) -> str: ) -def format_write_tool_content(write_input: WriteInput) -> str: +def format_write_output(output: WriteOutput) -> str: + """Format Write tool result as HTML. + + Args: + output: Parsed WriteOutput with first line acknowledgment + + Returns: + HTML string with the acknowledgment message + """ + escaped_message = escape_html(output.message) + return f"
{escaped_message} ...
" + + +def format_bash_output(output: BashOutput) -> str: + """Format Bash tool result as HTML with ANSI color support. + + Args: + output: Parsed BashOutput with content and ANSI flag + + Returns: + HTML string with ANSI colors converted or plain text + """ + content = output.content + if output.has_ansi: + full_html = convert_ansi_to_html(content) + else: + full_html = escape_html(content) + + # For short content, show directly + if len(content) <= 200: + return f"
{full_html}
" + + # For longer content, use collapsible details + preview_html = escape_html(content[:200]) + "..." + return f""" +
+ + 
{preview_html}
+ +
+ 
{full_html}
+
+
+ """ + + +def format_task_output(output: TaskOutput) -> str: + """Format Task tool result as HTML with markdown rendering. + + Args: + output: Parsed TaskOutput with agent's response + + Returns: + HTML string with markdown rendered in collapsible section + """ + return render_markdown_collapsible(output.result, "task-result") + + +def format_askuserquestion_output(output: AskUserQuestionOutput) -> str: + """Format AskUserQuestion tool result with styled Q&A pairs. + + Args: + output: Parsed AskUserQuestionOutput with Q&A pairs + + Returns: + HTML string with styled question/answer blocks + """ + html_parts: list[str] = [ + '
' + ] + + for qa in output.answers: + escaped_q = escape_html(qa.question) + escaped_a = escape_html(qa.answer) + html_parts.append('
') + html_parts.append(f'
❓ {escaped_q}
') + html_parts.append(f'
βœ… {escaped_a}
') + html_parts.append("
") + + html_parts.append("
") + return "".join(html_parts) + + +def format_exitplanmode_output(output: ExitPlanModeOutput) -> str: + """Format ExitPlanMode tool result as HTML. + + Args: + output: Parsed ExitPlanModeOutput with truncated message + + Returns: + HTML string with the (truncated) result message + """ + escaped_content = escape_html(output.message) + return f"
{escaped_content}
" + + +def format_write_input(write_input: WriteInput) -> str: """Format Write tool use content with Pygments syntax highlighting. Args: @@ -460,7 +427,7 @@ def format_write_tool_content(write_input: WriteInput) -> str: # -- Edit Tools (Edit/Multiedit) ---------------------------------------------- -def format_edit_tool_content(edit_input: EditInput) -> str: +def format_edit_input(edit_input: EditInput) -> str: """Format Edit tool use content as a diff view with intra-line highlighting. Args: @@ -481,7 +448,7 @@ def format_edit_tool_content(edit_input: EditInput) -> str: return "".join(html_parts) -def format_multiedit_tool_content(multiedit_input: MultiEditInput) -> str: +def format_multiedit_input(multiedit_input: MultiEditInput) -> str: """Format Multiedit tool use content showing multiple diffs. Args: @@ -512,7 +479,7 @@ def format_multiedit_tool_content(multiedit_input: MultiEditInput) -> str: # -- Bash Tool ---------------------------------------------------------------- -def format_bash_tool_content(bash_input: BashInput) -> str: +def format_bash_input(bash_input: BashInput) -> str: """Format Bash tool use content in VS Code extension style. Args: @@ -531,7 +498,7 @@ def format_bash_tool_content(bash_input: BashInput) -> str: # -- Task Tool ---------------------------------------------------------------- -def format_task_tool_content(task_input: TaskInput) -> str: +def format_task_input(task_input: TaskInput) -> str: """Format Task tool content with markdown-rendered prompt. Args: @@ -549,14 +516,15 @@ def format_task_tool_content(task_input: TaskInput) -> str: # -- Tool Summary and Title --------------------------------------------------- -def get_tool_summary(tool_use: ToolUseContent) -> Optional[str]: - """Extract a one-line summary from tool parameters for display in header. +def get_tool_summary(parsed: Optional[ToolInput]) -> Optional[str]: + """Extract a one-line summary from parsed tool input for display in header. Returns a brief description or filename that can be shown in the message header - to save vertical space. Uses parsed_input for type-safe access. - """ - parsed = tool_use.parsed_input + to save vertical space. + Args: + parsed: Parsed tool input, or None if parsing failed/not available + """ if isinstance(parsed, BashInput): return parsed.description @@ -566,22 +534,25 @@ def get_tool_summary(tool_use: ToolUseContent) -> Optional[str]: if isinstance(parsed, TaskInput): return parsed.description if parsed.description else None - # No summary for other tools + # No summary for other tools or unparsed input return None -def format_tool_use_title(tool_use: ToolUseContent) -> str: +def format_tool_use_title(tool_name: str, parsed: Optional[ToolInput]) -> str: """Generate the title HTML for a tool use message. Returns HTML string for the message header, with tool name, icon, - and optional summary/metadata. Uses parsed_input for type-safe access. + and optional summary/metadata. + + Args: + tool_name: The tool name (e.g., "Bash", "Read", "Edit") + parsed: Parsed tool input, or None if parsing failed/not available """ - escaped_name = escape_html(tool_use.name) - parsed = tool_use.parsed_input - summary = get_tool_summary(tool_use) + escaped_name = escape_html(tool_name) + summary = get_tool_summary(parsed) # TodoWrite: fixed title - if tool_use.name == "TodoWrite": + if tool_name == "TodoWrite": return "πŸ“ Todo List" # Task: show subagent_type and description @@ -695,96 +666,17 @@ def render_params_table(params: dict[str, Any]) -> str: return "".join(html_parts) -# -- Tool Use Dispatcher ------------------------------------------------------ - - -def format_tool_use_content(tool_use: ToolUseContent) -> str: - """Format tool use content as HTML. - - Uses parsed_input which handles lenient parsing at the model layer, - then dispatches to specialized formatters based on type. - """ - parsed = tool_use.parsed_input - - # Dispatch based on parsed type (lenient parsing happens in parsed_input) - if isinstance(parsed, TodoWriteInput): - return format_todowrite_content(parsed) - - if isinstance(parsed, BashInput): - return format_bash_tool_content(parsed) - - if isinstance(parsed, EditInput): - return format_edit_tool_content(parsed) - - if isinstance(parsed, MultiEditInput): - return format_multiedit_tool_content(parsed) - - if isinstance(parsed, WriteInput): - return format_write_tool_content(parsed) - - if isinstance(parsed, TaskInput): - return format_task_tool_content(parsed) - - if isinstance(parsed, ReadInput): - return format_read_tool_content(parsed) - - if isinstance(parsed, AskUserQuestionInput): - return format_askuserquestion_content(parsed) - - if isinstance(parsed, ExitPlanModeInput): - return format_exitplanmode_content(parsed) - - # Default: render as key/value table using shared renderer - return render_params_table(tool_use.input) - - -# -- Tool Result Content Formatter ------------------------------------------- - +# -- Tool Result Content Fallback Formatter ----------------------------------- -def _looks_like_bash_output(content: str) -> bool: - """Check if content looks like it's from a Bash tool based on common patterns.""" - if not content: - return False - # Check for ANSI escape sequences - if "\x1b[" in content: - return True +def format_tool_result_content_raw(tool_result: ToolResultContent) -> str: + """Format raw ToolResultContent as HTML (fallback formatter). - # Check for common bash/terminal patterns - bash_indicators = [ - "$ ", # Shell prompt - "❯ ", # Modern shell prompt - "> ", # Shell continuation - "\n+ ", # Bash -x output - "bash: ", # Bash error messages - "/bin/bash", # Bash path - "command not found", # Common bash error - "Permission denied", # Common bash error - "No such file or directory", # Common bash error - ] - - # Check for file path patterns that suggest command output - if re.search(r"/[a-zA-Z0-9_-]+(/[a-zA-Z0-9_.-]+)*", content): # Unix-style paths - return True - - # Check for common command output patterns - if any(indicator in content for indicator in bash_indicators): - return True - - return False - - -def format_tool_result_content( - tool_result: ToolResultContent, - file_path: Optional[str] = None, - tool_name: Optional[str] = None, -) -> str: - """Format tool result content as HTML, including images. + This handles tool results that don't have specialized output types, + including structured content with embedded images. Args: - tool_result: The tool result content - file_path: Optional file path for context (used for Read/Edit/Write tool rendering) - tool_name: Optional tool name for specialized rendering (e.g., "Write", "Read", "Edit", "Task") + tool_result: The raw tool result content """ # Handle both string and structured content if isinstance(tool_result.content, str): @@ -840,55 +732,11 @@ def format_tool_result_content( raw_content, flags=re.DOTALL, ) - # Remove "String: ..." portions that echo the input (everything after "String:" to end) + # Remove "String: ..." portions that echo the input raw_content = re.sub(r"\nString:.*$", "", raw_content, flags=re.DOTALL) - # Special handling for Write tool: only show first line (acknowledgment) on success - if tool_name == "Write" and not tool_result.is_error and not has_images: - lines = raw_content.split("\n") - if lines: - # Keep only the first acknowledgment line and add ellipsis - first_line = lines[0] - escaped_html = escape_html(first_line) - return f"
{escaped_html} ...
" - - # Try to parse as Read tool result if file_path is provided - if file_path and tool_name == "Read" and not has_images: - read_output = parse_read_output(raw_content, file_path) - if read_output: - return format_read_tool_result(read_output) - - # Try to parse as Edit tool result if file_path is provided - if file_path and tool_name == "Edit" and not has_images: - edit_output = parse_edit_output(raw_content, file_path) - if edit_output: - return format_edit_tool_result(edit_output) - - # Special handling for Task tool: render result as markdown with Pygments (agent's final message) - # Deduplication is now handled retroactively by replacing the sub-assistant content - if tool_name == "Task" and not has_images: - return render_markdown_collapsible(raw_content, "task-result") - - # Special handling for ExitPlanMode tool: truncate redundant plan echo on success - if tool_name == "ExitPlanMode" and not has_images: - processed_content = format_exitplanmode_result(raw_content) - escaped_content = escape_html(processed_content) - return f"
{escaped_content}
" - - # Special handling for AskUserQuestion tool: render Q&A pairs with styling - if tool_name == "AskUserQuestion" and not has_images: - styled_result = format_askuserquestion_result(raw_content) - if styled_result: - return styled_result - # Fall through to default handling if parsing fails - - # Check if this looks like Bash tool output and process ANSI codes - # Bash tool results often contain ANSI escape sequences and terminal output - is_ansi = _looks_like_bash_output(raw_content) - full_html = ( - convert_ansi_to_html(raw_content) if is_ansi else escape_html(raw_content) - ) - # For preview, always use plain escaped text (don't truncate HTML with tags) + # Format the content + full_html = escape_html(raw_content) preview_html = ( escape_html(raw_content[:200]) + "..." if len(raw_content) > 200 @@ -915,12 +763,12 @@ def format_tool_result_content( """ else: - # Text-only content (existing behavior) + # Text-only content # For simple content, show directly without collapsible wrapper if len(raw_content) <= 200: return f"
{full_html}
" - # For longer content, use collapsible details but no extra wrapper + # For longer content, use collapsible details return f"""
@@ -936,36 +784,32 @@ def format_tool_result_content( # -- Public Exports ----------------------------------------------------------- __all__ = [ - # AskUserQuestion - "format_askuserquestion_content", + # Tool input formatters (called by HtmlRenderer.format_{InputClass}) + "format_askuserquestion_input", + "format_exitplanmode_input", + "format_todowrite_input", + "format_read_input", + "format_write_input", + "format_edit_input", + "format_multiedit_input", + "format_bash_input", + "format_task_input", + # Tool output formatters (called by HtmlRenderer.format_{OutputClass}) + "format_read_output", + "format_write_output", + "format_edit_output", + "format_bash_output", + "format_task_output", + "format_askuserquestion_output", + "format_exitplanmode_output", + # Fallback for ToolResultContent + "format_tool_result_content_raw", + # Legacy formatters (still used) "format_askuserquestion_result", - # ExitPlanMode - "format_exitplanmode_content", "format_exitplanmode_result", - # TodoWrite - "format_todowrite_content", - # File tools (input) - "format_read_tool_content", - "format_write_tool_content", - # File tools (output/result) - "parse_read_output", - "format_read_tool_result", - "parse_edit_output", - "format_edit_tool_result", - # Edit tools - "format_edit_tool_content", - "format_multiedit_tool_content", - # Bash - "format_bash_tool_content", - # Task - "format_task_tool_content", # Tool summary and title "get_tool_summary", "format_tool_use_title", # Generic "render_params_table", - # Dispatcher - "format_tool_use_content", - # Tool result - "format_tool_result_content", ] diff --git a/claude_code_log/html/user_formatters.py b/claude_code_log/html/user_formatters.py index 5c6235b4..098a467f 100644 --- a/claude_code_log/html/user_formatters.py +++ b/claude_code_log/html/user_formatters.py @@ -2,27 +2,27 @@ This module formats non-tool user message content types to HTML. Part of the thematic formatter organization: -- system_formatters.py: SystemContent, HookSummaryContent -- user_formatters.py: SlashCommandContent, CommandOutputContent, etc. -- assistant_formatters.py: (future) assistant message variants +- system_formatters.py: SystemMessage, HookSummaryMessage +- user_formatters.py: SlashCommandMessage, CommandOutputMessage, etc. +- assistant_formatters.py: AssistantTextMessage, ThinkingMessage, ImageContent - tool_formatters.py: tool use/result content """ from .ansi_colors import convert_ansi_to_html from ..models import ( - BashInputContent, - BashOutputContent, - CommandOutputContent, - CompactedSummaryContent, + BashInputMessage, + BashOutputMessage, + CommandOutputMessage, + CompactedSummaryMessage, IdeDiagnostic, IdeNotificationContent, IdeOpenedFile, IdeSelection, ImageContent, - SlashCommandContent, - UserMemoryContent, - UserSlashCommandContent, - UserTextContent, + SlashCommandMessage, + UserMemoryMessage, + UserSlashCommandMessage, + UserTextMessage, ) from .tool_formatters import render_params_table from .utils import escape_html, render_collapsible_code, render_markdown_collapsible @@ -33,11 +33,11 @@ # ============================================================================= -def format_slash_command_content(content: SlashCommandContent) -> str: +def format_slash_command_content(content: SlashCommandMessage) -> str: """Format slash command content as HTML. Args: - content: SlashCommandContent with command name, args, and contents + content: SlashCommandMessage with command name, args, and contents Returns: HTML string for the slash command display @@ -75,11 +75,11 @@ def format_slash_command_content(content: SlashCommandContent) -> str: return "
".join(content_parts) -def format_command_output_content(content: CommandOutputContent) -> str: +def format_command_output_content(content: CommandOutputMessage) -> str: """Format command output content as HTML. Args: - content: CommandOutputContent with stdout and is_markdown flag + content: CommandOutputMessage with stdout and is_markdown flag Returns: HTML string for the command output display @@ -96,11 +96,11 @@ def format_command_output_content(content: CommandOutputContent) -> str: return f"
{html_content}
" -def format_bash_input_content(content: BashInputContent) -> str: +def format_bash_input_content(content: BashInputMessage) -> str: """Format bash input content as HTML. Args: - content: BashInputContent with the bash command + content: BashInputMessage with the bash command Returns: HTML string for the bash input display @@ -113,14 +113,14 @@ def format_bash_input_content(content: BashInputContent) -> str: def format_bash_output_content( - content: BashOutputContent, + content: BashOutputMessage, collapse_threshold: int = 10, preview_lines: int = 3, ) -> str: """Format bash output content as HTML. Args: - content: BashOutputContent with stdout and/or stderr + content: BashOutputMessage with stdout and/or stderr collapse_threshold: Number of lines before output becomes collapsible preview_lines: Number of preview lines to show when collapsed @@ -191,8 +191,8 @@ def format_user_text_content(text: str) -> str: return f"
{escaped_text}
" -def format_user_text_model_content(content: UserTextContent) -> str: - """Format UserTextContent model as HTML. +def format_user_text_model_content(content: UserTextMessage) -> str: + """Format UserTextMessage model as HTML. Handles user text with optional IDE notifications, compacted summaries, memory input markers, and inline images. @@ -202,10 +202,8 @@ def format_user_text_model_content(content: UserTextContent) -> str: - ImageContent: Rendered as inline tag with base64 data URL - IdeNotificationContent: Rendered as IDE notification blocks - Falls back to legacy text-only behavior when `items` is None. - Args: - content: UserTextContent with text/items and optional flags/notifications + content: UserTextMessage with text/items and optional flags/notifications Returns: HTML string combining all content items @@ -229,14 +227,14 @@ def format_user_text_model_content(content: UserTextContent) -> str: return "\n".join(parts) -def format_compacted_summary_content(content: CompactedSummaryContent) -> str: +def format_compacted_summary_content(content: CompactedSummaryMessage) -> str: """Format compacted session summary content as HTML. Compacted summaries are rendered as collapsible markdown since they contain structured summary text generated by Claude. Args: - content: CompactedSummaryContent with summary text + content: CompactedSummaryMessage with summary text Returns: HTML string with collapsible markdown rendering @@ -249,14 +247,14 @@ def format_compacted_summary_content(content: CompactedSummaryContent) -> str: ) -def format_user_memory_content(content: UserMemoryContent) -> str: +def format_user_memory_content(content: UserMemoryMessage) -> str: """Format user memory input content as HTML. User memory content (from CLAUDE.md etc.) is rendered as preformatted text to preserve the original formatting. Args: - content: UserMemoryContent with memory text + content: UserMemoryMessage with memory text Returns: HTML string with escaped text in a pre tag @@ -265,14 +263,14 @@ def format_user_memory_content(content: UserMemoryContent) -> str: return f"
{escaped_text}
" -def format_user_slash_command_content(content: UserSlashCommandContent) -> str: +def format_user_slash_command_content(content: UserSlashCommandMessage) -> str: """Format slash command expanded prompt (isMeta) as HTML. These are LLM-generated instruction text from slash commands, rendered as collapsible markdown. Args: - content: UserSlashCommandContent with markdown text + content: UserSlashCommandMessage with markdown text Returns: HTML string with collapsible markdown rendering diff --git a/claude_code_log/html/utils.py b/claude_code_log/html/utils.py index c62657a8..44d5b753 100644 --- a/claude_code_log/html/utils.py +++ b/claude_code_log/html/utils.py @@ -23,24 +23,25 @@ from .renderer_code import highlight_code_with_pygments, truncate_highlighted_preview from ..models import ( - AssistantTextContent, - BashInputContent, - BashOutputContent, - CommandOutputContent, - CompactedSummaryContent, - HookSummaryContent, + AssistantTextMessage, + BashInputMessage, + BashOutputMessage, + CommandOutputMessage, + CompactedSummaryMessage, + DedupNoticeMessage, + HookSummaryMessage, MessageContent, - SessionHeaderContent, - SlashCommandContent, - SystemContent, - ThinkingContentModel, - ToolResultContentModel, - ToolUseContent, - UnknownContent, - UserMemoryContent, - UserSlashCommandContent, - UserSteeringContent, - UserTextContent, + SessionHeaderMessage, + SlashCommandMessage, + SystemMessage, + ThinkingMessage, + ToolResultMessage, + ToolUseMessage, + UnknownMessage, + UserMemoryMessage, + UserSlashCommandMessage, + UserSteeringMessage, + UserTextMessage, ) from ..renderer_timings import timing_stat @@ -55,27 +56,28 @@ CSS_CLASS_REGISTRY: dict[type[MessageContent], list[str]] = { # System message types - SystemContent: ["system"], # level added dynamically - HookSummaryContent: ["system", "system-hook"], + SystemMessage: ["system"], # level added dynamically + HookSummaryMessage: ["system", "system-hook"], # User message types - UserTextContent: ["user"], - UserSteeringContent: ["user", "steering"], - SlashCommandContent: ["user", "slash-command"], - UserSlashCommandContent: ["user", "slash-command"], - UserMemoryContent: ["user"], - CompactedSummaryContent: ["user", "compacted"], - CommandOutputContent: ["user", "command-output"], + UserTextMessage: ["user"], + UserSteeringMessage: ["user", "steering"], + SlashCommandMessage: ["user", "slash-command"], + UserSlashCommandMessage: ["user", "slash-command"], + UserMemoryMessage: ["user"], + CompactedSummaryMessage: ["user", "compacted"], + CommandOutputMessage: ["user", "command-output"], # Assistant message types - AssistantTextContent: ["assistant"], + AssistantTextMessage: ["assistant"], + DedupNoticeMessage: ["assistant", "dedup-notice"], # Styled as assistant # Tool message types - ToolUseContent: ["tool_use"], - ToolResultContentModel: ["tool_result"], # error added dynamically + ToolUseMessage: ["tool_use"], + ToolResultMessage: ["tool_result"], # error added dynamically # Other message types - ThinkingContentModel: ["thinking"], - SessionHeaderContent: ["session_header"], - BashInputContent: ["bash-input"], - BashOutputContent: ["bash-output"], - UnknownContent: ["unknown"], + ThinkingMessage: ["thinking"], + SessionHeaderMessage: ["session_header"], + BashInputMessage: ["bash-input"], + BashOutputMessage: ["bash-output"], + UnknownMessage: ["unknown"], } @@ -91,9 +93,9 @@ def _get_css_classes_from_content(content: MessageContent) -> list[str]: if classes := CSS_CLASS_REGISTRY.get(cls): result = list(classes) # Dynamic modifiers based on content attributes - if isinstance(content, SystemContent): + if isinstance(content, SystemMessage): result.append(f"system-{content.level}") - elif isinstance(content, ToolResultContentModel) and content.is_error: + elif isinstance(content, ToolResultMessage) and content.is_error: result.append("error") return result return [] @@ -134,6 +136,18 @@ def css_class_from_message(msg: "TemplateMessage") -> str: return " ".join(parts) +def is_session_header(msg: "TemplateMessage") -> bool: + """Check if message is a session header. + + Args: + msg: The template message to check + + Returns: + True if message content is a SessionHeaderMessage + """ + return isinstance(msg.content, SessionHeaderMessage) + + def get_message_emoji(msg: "TemplateMessage") -> str: """Return appropriate emoji for message type. @@ -149,7 +163,7 @@ def get_message_emoji(msg: "TemplateMessage") -> str: return "πŸ“‹" elif msg_type == "user": # Command output has no emoji (neutral - can be from built-in or user command) - if isinstance(msg.content, CommandOutputContent): + if isinstance(msg.content, CommandOutputMessage): return "" return "🀷" elif msg_type == "bash-input": @@ -163,7 +177,7 @@ def get_message_emoji(msg: "TemplateMessage") -> str: elif msg_type == "tool_use": return "πŸ› οΈ" elif msg_type == "tool_result": - if isinstance(msg.content, ToolResultContentModel) and msg.content.is_error: + if isinstance(msg.content, ToolResultMessage) and msg.content.is_error: return "🚨" return "🧰" elif msg_type == "thinking": diff --git a/claude_code_log/models.py b/claude_code_log/models.py index 620d3133..f898e353 100644 --- a/claude_code_log/models.py +++ b/claude_code_log/models.py @@ -4,7 +4,7 @@ from enum import Enum from typing import Any, Union, Optional, Literal -from pydantic import BaseModel, PrivateAttr +from pydantic import BaseModel, Field class MessageType(str, Enum): @@ -45,6 +45,229 @@ class MessageType(str, Enum): SYSTEM_ERROR = "system-error" +# ============================================================================= +# JSONL Content Models (Pydantic) +# ============================================================================= +# Low-level content types parsed from JSONL transcript entries. +# These are defined first as they're the "input" types from transcript files. + + +class TextContent(BaseModel): + """Text content block within a message content array.""" + + type: Literal["text"] + text: str + + +class ImageSource(BaseModel): + """Base64-encoded image source data.""" + + type: Literal["base64"] + media_type: str + data: str + + +class ImageContent(BaseModel): + """Image content. + + This represents an image within a content array, not a standalone message. + Images are always part of UserTextMessage.items or AssistantTextMessage.items. + """ + + type: Literal["image"] + source: ImageSource + + +class UsageInfo(BaseModel): + """Token usage information for tracking API consumption.""" + + input_tokens: Optional[int] = None + cache_creation_input_tokens: Optional[int] = None + cache_read_input_tokens: Optional[int] = None + output_tokens: Optional[int] = None + service_tier: Optional[str] = None + server_tool_use: Optional[dict[str, Any]] = None + + +class ToolUseContent(BaseModel): + type: Literal["tool_use"] + id: str + name: str + input: dict[str, Any] + + +class ToolResultContent(BaseModel): + type: Literal["tool_result"] + tool_use_id: str + content: Union[str, list[dict[str, Any]]] + is_error: Optional[bool] = None + agentId: Optional[str] = None # Reference to agent file for sub-agent messages + + +class ThinkingContent(BaseModel): + type: Literal["thinking"] + thinking: str + signature: Optional[str] = None + + +# Content item types that appear in message content arrays +ContentItem = Union[ + TextContent, + ToolUseContent, + ToolResultContent, + ThinkingContent, + ImageContent, +] + + +class UserMessageModel(BaseModel): + role: Literal["user"] + content: list[ContentItem] + usage: Optional["UsageInfo"] = ( + None # For type compatibility with AssistantMessageModel + ) + + +class AssistantMessageModel(BaseModel): + """Assistant message model.""" + + id: str + type: Literal["message"] + role: Literal["assistant"] + model: str + content: list[ContentItem] + stop_reason: Optional[str] = None + stop_sequence: Optional[str] = None + usage: Optional[UsageInfo] = None + + +# Tool result type - flexible to accept various result formats from JSONL +# The specific parsing/formatting happens in tool_formatters.py using +# ReadOutput, EditOutput, etc. (see Tool Output Content Models section) +ToolUseResult = Union[ + str, + list[Any], # Covers list[TodoWriteItem], list[ContentItem], etc. + dict[str, Any], # Covers structured results +] + + +class BaseTranscriptEntry(BaseModel): + parentUuid: Optional[str] + isSidechain: bool + userType: str + cwd: str + sessionId: str + version: str + uuid: str + timestamp: str + isMeta: Optional[bool] = None + agentId: Optional[str] = None # Agent ID for sidechain messages + gitBranch: Optional[str] = None # Git branch name when available + + +class UserTranscriptEntry(BaseTranscriptEntry): + type: Literal["user"] + message: UserMessageModel + toolUseResult: Optional[ToolUseResult] = None + agentId: Optional[str] = None # From toolUseResult when present + + +class AssistantTranscriptEntry(BaseTranscriptEntry): + type: Literal["assistant"] + message: AssistantMessageModel + requestId: Optional[str] = None + + +class SummaryTranscriptEntry(BaseModel): + type: Literal["summary"] + summary: str + leafUuid: str + cwd: Optional[str] = None + sessionId: None = None # Summaries don't have a sessionId + + +class SystemTranscriptEntry(BaseTranscriptEntry): + """System messages like warnings, notifications, hook summaries, etc.""" + + type: Literal["system"] + content: Optional[str] = None + subtype: Optional[str] = None # e.g., "stop_hook_summary" + level: Optional[str] = None # e.g., "warning", "info", "error" + # Hook summary fields (for subtype="stop_hook_summary") + hasOutput: Optional[bool] = None + hookErrors: Optional[list[str]] = None + hookInfos: Optional[list[dict[str, Any]]] = None + preventedContinuation: Optional[bool] = None + + +class QueueOperationTranscriptEntry(BaseModel): + """Queue operations (enqueue/dequeue/remove) for message queueing tracking. + + enqueue/dequeue are internal operations that track when messages are queued and dequeued. + They are parsed but not rendered, as the content duplicates actual user messages. + + 'remove' operations are out-of-band user inputs made visible to the agent while working + for "steering" purposes. These should be rendered as user messages with a 'steering' CSS class. + Content can be a list of ContentItems or a simple string (for 'remove' operations). + """ + + type: Literal["queue-operation"] + operation: Literal["enqueue", "dequeue", "remove", "popAll"] + timestamp: str + sessionId: str + content: Optional[Union[list[ContentItem], str]] = ( + None # List for enqueue, str for remove/popAll + ) + + +TranscriptEntry = Union[ + UserTranscriptEntry, + AssistantTranscriptEntry, + SummaryTranscriptEntry, + SystemTranscriptEntry, + QueueOperationTranscriptEntry, +] + + +# ============================================================================= +# Message Metadata +# ============================================================================= +# Common metadata fields extracted from transcript entries. + + +@dataclass +class MessageMeta: + """Common metadata extracted from transcript entries. + + These fields are shared across all message types and are used to create + the TemplateMessage wrapper for rendering. + + Note: formatted_timestamp is computed at render time, not stored here. + """ + + # Identity fields + session_id: str + timestamp: str # Raw ISO timestamp + uuid: str + parent_uuid: Optional[str] = None + + # Context fields + is_sidechain: bool = False + is_meta: bool = False # User slash command (isMeta=True in transcript) + agent_id: Optional[str] = None + cwd: str = "" + git_branch: Optional[str] = None + + @classmethod + def empty(cls, uuid: str = "") -> "MessageMeta": + """Create a placeholder MessageMeta with empty/default values. + + Useful for cases where full metadata isn't available at creation time + (e.g., SummaryTranscriptEntry where session_id is matched later). + """ + return cls(session_id="", timestamp="", uuid=uuid) + + # ============================================================================= # Message Content Models # ============================================================================= @@ -53,21 +276,43 @@ class MessageType(str, Enum): # renderers (HTML, text, etc.) to format the content appropriately. +@dataclass class MessageContent: """Base class for structured message content. Subclasses represent specific content types that renderers can format appropriately for their output format. - Note: This is a plain class (not dataclass) to allow Pydantic BaseModel - subclasses like ToolUseContent and ImageContent to inherit from it. + The `meta` field is required and first positional, ensuring all message + content always has associated metadata. Use MessageMeta.empty() when + full metadata isn't available at creation time. + + Note: Render-time relationship data (pairing, hierarchy, children) is stored + on TemplateMessage, not here. MessageContent is pure transcript data. """ - pass + meta: MessageMeta + + @property + def message_type(self) -> str: + """Return the message type identifier for this content. + + Subclasses MUST override this to return their specific type. + This is used for CSS classes, filtering, and type-based rendering. + """ + raise NotImplementedError("Subclasses must implement message_type property") + + @property + def has_markdown(self) -> bool: + """Whether this content should be rendered as markdown. + + Subclasses that contain markdown content should override to return True. + """ + return False @dataclass -class SystemContent(MessageContent): +class SystemMessage(MessageContent): """System message with level indicator. Used for info, warning, and error system messages. @@ -76,6 +321,10 @@ class SystemContent(MessageContent): level: str # "info", "warning", "error" text: str # Raw text content (may contain ANSI codes) + @property + def message_type(self) -> str: + return "system" + @dataclass class HookInfo: @@ -86,7 +335,7 @@ class HookInfo: @dataclass -class HookSummaryContent(MessageContent): +class HookSummaryMessage(MessageContent): """Hook execution summary. Used for subtype="stop_hook_summary" system messages. @@ -96,6 +345,10 @@ class HookSummaryContent(MessageContent): hook_errors: list[str] # Error messages from hooks hook_infos: list[HookInfo] # Info about each hook executed + @property + def message_type(self) -> str: + return "system" + # ============================================================================= # User Message Content Models @@ -105,7 +358,7 @@ class HookSummaryContent(MessageContent): @dataclass -class SlashCommandContent(MessageContent): +class SlashCommandMessage(MessageContent): """Content for slash command invocations (e.g., /context, /model). These are user messages containing command-name, command-args, and @@ -116,9 +369,13 @@ class SlashCommandContent(MessageContent): command_args: str command_contents: str + @property + def message_type(self) -> str: + return "user" + @dataclass -class CommandOutputContent(MessageContent): +class CommandOutputMessage(MessageContent): """Content for local command output (e.g., output from /context). These are user messages containing local-command-stdout tags. @@ -127,9 +384,13 @@ class CommandOutputContent(MessageContent): stdout: str is_markdown: bool # True if content appears to be markdown + @property + def message_type(self) -> str: + return "user" + @dataclass -class BashInputContent(MessageContent): +class BashInputMessage(MessageContent): """Content for inline bash commands in user messages. These are user messages containing bash-input tags. @@ -137,9 +398,13 @@ class BashInputContent(MessageContent): command: str + @property + def message_type(self) -> str: + return "bash-input" + @dataclass -class BashOutputContent(MessageContent): +class BashOutputMessage(MessageContent): """Content for bash command output. These are user messages containing bash-stdout and/or bash-stderr tags. @@ -148,24 +413,17 @@ class BashOutputContent(MessageContent): stdout: Optional[str] = None # Raw stdout content (may contain ANSI codes) stderr: Optional[str] = None # Raw stderr content (may contain ANSI codes) + @property + def message_type(self) -> str: + return "bash-output" -@dataclass -class ToolResultContentModel(MessageContent): - """Content model for tool results with rendering context. - Wraps ToolResultContent with additional context needed for rendering, - such as the associated tool name and file path. - """ - - tool_use_id: str - content: Any # Union[str, list[dict[str, Any]]] - is_error: bool = False - tool_name: Optional[str] = None # Name of the tool that produced this result - file_path: Optional[str] = None # File path for Read/Edit/Write tools +# Note: ToolResultMessage and ToolUseMessage are defined in the +# "Tool Message Models" section (before Tool Input Models). @dataclass -class CompactedSummaryContent(MessageContent): +class CompactedSummaryMessage(MessageContent): """Content for compacted session summaries. These are user messages that contain previous conversation context @@ -176,9 +434,17 @@ class CompactedSummaryContent(MessageContent): summary_text: str + @property + def message_type(self) -> str: + return "user" + + @property + def has_markdown(self) -> bool: + return True + @dataclass -class UserMemoryContent(MessageContent): +class UserMemoryMessage(MessageContent): """Content for user memory input. These are user messages containing user-memory-input tags. @@ -188,9 +454,13 @@ class UserMemoryContent(MessageContent): memory_text: str + @property + def message_type(self) -> str: + return "user" + @dataclass -class UserSlashCommandContent(MessageContent): +class UserSlashCommandMessage(MessageContent): """Content for slash command expanded prompts (isMeta=True). These are LLM-generated instruction text from slash commands. @@ -200,6 +470,10 @@ class UserSlashCommandContent(MessageContent): text: str + @property + def message_type(self) -> str: + return "user" + @dataclass class IdeOpenedFile: @@ -227,10 +501,13 @@ class IdeDiagnostic: @dataclass -class IdeNotificationContent(MessageContent): - """Content for IDE notification tags. +class IdeNotificationContent: + """Content for IDE notification tags (embedded within user messages). + + This is NOT a MessageContent subclass - it's used as an item within + UserTextMessage.items alongside TextContent and ImageContent. - These are user messages containing IDE notification tags like: + Represents IDE notification tags like: - : File open notifications - : Code selection notifications - : Diagnostic JSON arrays @@ -244,41 +521,8 @@ class IdeNotificationContent(MessageContent): remaining_text: str # Text after notifications extracted -# ============================================================================= -# Content Item Models (Pydantic) -# ============================================================================= -# These are content items that appear within message content arrays. -# Defined here before the dataclass models that reference them. - - -class TextContent(BaseModel): - """Text content block within a message content array.""" - - type: Literal["text"] - text: str - - -class ImageSource(BaseModel): - """Base64-encoded image source data.""" - - type: Literal["base64"] - media_type: str - data: str - - -class ImageContent(BaseModel): - """Image content from the Anthropic API. - - This represents an image within a content array, not a standalone message. - Images are always part of UserTextContent.items or AssistantTextContent.items. - """ - - type: Literal["image"] - source: ImageSource - - @dataclass -class UserTextContent(MessageContent): +class UserTextMessage(MessageContent): """Content for user text with interleaved images and IDE notifications. The `items` field preserves the original order of content: @@ -294,13 +538,20 @@ class UserTextContent(MessageContent): TextContent | ImageContent | IdeNotificationContent ] = field(default_factory=list) + # Cached raw text extracted from items (for dedup matching, simple renderers) + raw_text_content: Optional[str] = None + + @property + def message_type(self) -> str: + return "user" + @dataclass -class UserSteeringContent(UserTextContent): +class UserSteeringMessage(UserTextMessage): """Content for user steering prompts (queue-operation "remove"). These are user messages that steer the conversation by removing - items from the queue. Inherits from UserTextContent. + items from the queue. Inherits from UserTextMessage. """ pass @@ -314,7 +565,7 @@ class UserSteeringContent(UserTextContent): @dataclass -class AssistantTextContent(MessageContent): +class AssistantTextMessage(MessageContent): """Content for assistant text messages with interleaved images. These are the text portions of assistant messages that get @@ -327,157 +578,69 @@ class AssistantTextContent(MessageContent): Empty items list indicates no content. """ - # Interleaved content items preserving original order - items: list[ # pyright: ignore[reportUnknownVariableType] - TextContent | ImageContent - ] = field(default_factory=list) - - -@dataclass -class ThinkingContentModel(MessageContent): - """Content for assistant thinking/reasoning blocks. - - These are the blocks that show the assistant's - internal reasoning process. - - Note: This is distinct from ThinkingContent (the Pydantic model - for parsing JSONL). This dataclass is for rendering purposes. - """ - - thinking: str - signature: Optional[str] = None - - -@dataclass -class UnknownContent(MessageContent): - """Content for unknown/unrecognized content types. - - Used as a fallback when encountering content types that don't have - specific handlers. Stores the type name for display purposes. - """ - - type_name: str # The name/description of the unknown type - - -# ============================================================================= -# Tool Output Content Models -# ============================================================================= -# Structured content models for tool results (symmetric with Tool Input Models). -# These provide format-neutral representation of tool outputs that renderers -# can format appropriately. - - -@dataclass -class ReadOutput(MessageContent): - """Parsed Read tool output. - - Represents the result of reading a file with optional line range. - Symmetric with ReadInput for tool_use β†’ tool_result pairing. - """ - - file_path: str - content: str # File content (may be truncated) - start_line: int # 1-based starting line number - num_lines: int # Number of lines in content - total_lines: int # Total lines in file - is_truncated: bool # Whether content was truncated - system_reminder: Optional[str] = None # Embedded system reminder text - - -@dataclass -class WriteOutput(MessageContent): - """Parsed Write tool output. - - Symmetric with WriteInput for tool_use β†’ tool_result pairing. - - TODO: Not currently used - tool results handled as raw strings. - """ - - file_path: str - success: bool - message: str # Success or error message - - -@dataclass -class EditDiff: - """Single diff hunk for edit operations.""" - - old_text: str - new_text: str - - -@dataclass -class EditOutput(MessageContent): - """Parsed Edit tool output. - - Contains diff information for file edits. - Symmetric with EditInput for tool_use β†’ tool_result pairing. - """ - - file_path: str - success: bool - diffs: list[EditDiff] # Changes made - message: str # Result message or code snippet - start_line: int = 1 # Starting line number for code display - + # Interleaved content items preserving original order + items: list[ # pyright: ignore[reportUnknownVariableType] + TextContent | ImageContent + ] = field(default_factory=list) -@dataclass -class BashOutput(MessageContent): - """Parsed Bash tool output. + # Cached raw text extracted from items (for dedup matching, simple renderers) + raw_text_content: Optional[str] = None - Symmetric with BashInput for tool_use β†’ tool_result pairing. + # Token usage string (formatted from UsageInfo when available) + token_usage: Optional[str] = None - TODO: Not currently used - tool results handled as raw strings. - """ + @property + def message_type(self) -> str: + return "assistant" - stdout: str - stderr: str - exit_code: Optional[int] - interrupted: bool - is_image: bool # True if output contains image data + @property + def has_markdown(self) -> bool: + return True @dataclass -class TaskOutput(MessageContent): - """Parsed Task (sub-agent) tool output. +class ThinkingMessage(MessageContent): + """Message for assistant thinking/reasoning blocks. - Symmetric with TaskInput for tool_use β†’ tool_result pairing. + These are the blocks that show the assistant's + internal reasoning process. - TODO: Not currently used - tool results handled as raw strings. + Note: This is distinct from ThinkingContent (the Pydantic model + for parsing JSONL). This dataclass is for rendering purposes. """ - agent_id: Optional[str] - result: str # Agent's response - is_background: bool + thinking: str + signature: Optional[str] = None + # Token usage string (formatted from UsageInfo when available) + token_usage: Optional[str] = None -@dataclass -class GlobOutput(MessageContent): - """Parsed Glob tool output. + @property + def message_type(self) -> str: + return "thinking" - Symmetric with GlobInput for tool_use β†’ tool_result pairing. + @property + def has_markdown(self) -> bool: + return True - TODO: Not currently used - tool results handled as raw strings. - """ - pattern: str - files: list[str] # Matching file paths - truncated: bool # Whether list was truncated +# Note: ToolUseMessage is also an assistant content type, defined in +# "Tool Message Models" section (before Tool Input Models). @dataclass -class GrepOutput(MessageContent): - """Parsed Grep tool output. - - Symmetric with GrepInput for tool_use β†’ tool_result pairing. +class UnknownMessage(MessageContent): + """Content for unknown/unrecognized content types. - TODO: Not currently used - tool results handled as raw strings. + Used as a fallback when encountering content types that don't have + specific handlers. Stores the type name for display purposes. """ - pattern: str - matches: list[str] # Matching lines/files - output_mode: str # "content", "files_with_matches", or "count" - truncated: bool + type_name: str # The name/description of the unknown type + + @property + def message_type(self) -> str: + return "unknown" # ============================================================================= @@ -488,7 +651,7 @@ class GrepOutput(MessageContent): @dataclass -class SessionHeaderContent(MessageContent): +class SessionHeaderMessage(MessageContent): """Content for session headers in transcript rendering. Represents the header displayed at the start of each session @@ -499,19 +662,81 @@ class SessionHeaderContent(MessageContent): session_id: str summary: Optional[str] = None + @property + def message_type(self) -> str: + return "session_header" + @dataclass -class DedupNoticeContent(MessageContent): +class DedupNoticeMessage(MessageContent): """Content for deduplication notices. Displayed when content is deduplicated (e.g., sidechain assistant - text that duplicates the Task tool result). + text that duplicates the Task tool result). Styled as assistant message + since it replaces sidechain assistant content. + + The `original` field preserves the original AssistantTextMessage so renderers + can optionally show the full content instead of the link. """ notice_text: str - target_uuid: Optional[str] = None # UUID of target message (for resolving link) - target_message_id: Optional[str] = None # Resolved message ID for anchor link + target_message_id: Optional[str] = None # Message ID for anchor link original_text: Optional[str] = None # Original duplicated content (for debugging) + original: Optional["AssistantTextMessage"] = None # Original message for renderers + + @property + def message_type(self) -> str: + return "assistant" # Styled as assistant (replaces sidechain assistant) + + +# ============================================================================= +# Tool Message Models +# ============================================================================= +# High-level message wrappers for tool invocations and results. +# These wrap the specialized Tool Input/Output models for rendering. + + +@dataclass +class ToolResultMessage(MessageContent): + """Message for tool results with rendering context. + + Wraps ToolResultContent or specialized output with additional context + needed for rendering, such as the associated tool name and file path. + """ + + tool_use_id: str + output: ( + "ToolOutput" # Specialized (ReadOutput, etc.) or generic (ToolResultContent) + ) + is_error: bool = False + tool_name: Optional[str] = None # Name of the tool that produced this result + file_path: Optional[str] = None # File path for Read/Edit/Write tools + + @property + def message_type(self) -> str: + return "tool_result" + + @property + def has_markdown(self) -> bool: + """TaskOutput results contain markdown (agent responses).""" + return isinstance(self.output, TaskOutput) + + +@dataclass +class ToolUseMessage(MessageContent): + """Message for tool invocations. + + Wraps ToolUseContent with the parsed input for specialized formatting. + Falls back to the original ToolUseContent when no specialized parser exists. + """ + + input: "ToolInput" # Specialized (BashInput, etc.) or ToolUseContent fallback + tool_use_id: str # From ToolUseContent.id + tool_name: str # From ToolUseContent.name + + @property + def message_type(self) -> str: + return "tool_use" # ============================================================================= @@ -642,7 +867,9 @@ class AskUserQuestionItem(BaseModel): question: str = "" header: Optional[str] = None - options: list[AskUserQuestionOption] = [] + options: list[AskUserQuestionOption] = Field( + default_factory=lambda: list[AskUserQuestionOption]() + ) multiSelect: bool = False @@ -652,7 +879,9 @@ class AskUserQuestionInput(BaseModel): Supports both modern format (questions list) and legacy format (single question). """ - questions: list[AskUserQuestionItem] = [] + questions: list[AskUserQuestionItem] = Field( + default_factory=lambda: list[AskUserQuestionItem]() + ) question: Optional[str] = None # Legacy single question format @@ -677,172 +906,164 @@ class ExitPlanModeInput(BaseModel): TodoWriteInput, AskUserQuestionInput, ExitPlanModeInput, - dict[str, Any], # Fallback for unknown tools + ToolUseContent, # Generic fallback when no specialized parser ] -class UsageInfo(BaseModel): - """Token usage information for tracking API consumption.""" +# ============================================================================= +# Tool Output Models +# ============================================================================= +# Typed models for tool outputs (symmetric with Tool Input Models). +# These are data containers stored inside ToolResultMessage.output, +# NOT standalone message types (so they don't inherit from MessageContent). - input_tokens: Optional[int] = None - cache_creation_input_tokens: Optional[int] = None - cache_read_input_tokens: Optional[int] = None - output_tokens: Optional[int] = None - service_tier: Optional[str] = None - server_tool_use: Optional[dict[str, Any]] = None +@dataclass +class ReadOutput: + """Parsed Read tool output. -class ToolUseContent(BaseModel, MessageContent): - type: Literal["tool_use"] - id: str - name: str - input: dict[str, Any] - _parsed_input: Optional["ToolInput"] = PrivateAttr( - default=None - ) # Cached parsed input + Represents the result of reading a file with optional line range. + Symmetric with ReadInput for tool_use β†’ tool_result pairing. + """ - @property - def parsed_input(self) -> "ToolInput": - """Get typed input model if available, otherwise return raw dict. + file_path: str + content: str # File content (may be truncated) + start_line: int # 1-based starting line number + num_lines: int # Number of lines in content + total_lines: int # Total lines in file + is_truncated: bool # Whether content was truncated + system_reminder: Optional[str] = None # Embedded system reminder text - Lazily parses the input dict into a typed model. - Uses strict validation first, then lenient parsing if available. - Result is cached for subsequent accesses. - """ - if self._parsed_input is None: - from .parser import parse_tool_input - object.__setattr__( - self, "_parsed_input", parse_tool_input(self.name, self.input) - ) - return self._parsed_input # type: ignore[return-value] +@dataclass +class WriteOutput: + """Parsed Write tool output. + Symmetric with WriteInput for tool_use β†’ tool_result pairing. + """ -class ToolResultContent(BaseModel): - type: Literal["tool_result"] - tool_use_id: str - content: Union[str, list[dict[str, Any]]] - is_error: Optional[bool] = None - agentId: Optional[str] = None # Reference to agent file for sub-agent messages + file_path: str + success: bool + message: str # First line acknowledgment (truncated from full output) -class ThinkingContent(BaseModel): - type: Literal["thinking"] - thinking: str - signature: Optional[str] = None +@dataclass +class EditDiff: + """Single diff hunk for edit operations.""" + + old_text: str + new_text: str -# Content item types that appear in message content arrays -ContentItem = Union[ - TextContent, - ToolUseContent, - ToolResultContent, - ThinkingContent, - ImageContent, -] +@dataclass +class EditOutput: + """Parsed Edit tool output. + Contains diff information for file edits. + Symmetric with EditInput for tool_use β†’ tool_result pairing. + """ -class UserMessage(BaseModel): - role: Literal["user"] - content: list[ContentItem] - usage: Optional["UsageInfo"] = None # For type compatibility with AssistantMessage + file_path: str + success: bool + diffs: list[EditDiff] # Changes made + message: str # Result message or code snippet + start_line: int = 1 # Starting line number for code display -class AssistantMessage(BaseModel): - """Assistant message model.""" +@dataclass +class BashOutput: + """Parsed Bash tool output. - id: str - type: Literal["message"] - role: Literal["assistant"] - model: str - content: list[ContentItem] - stop_reason: Optional[str] = None - stop_sequence: Optional[str] = None - usage: Optional[UsageInfo] = None + Symmetric with BashInput for tool_use β†’ tool_result pairing. + Contains the output with ANSI flag for terminal formatting. + """ + content: str # Output content (stdout/stderr combined) + has_ansi: bool # True if content contains ANSI escape sequences -# Tool result type - flexible to accept various result formats from JSONL -# The specific parsing/formatting happens in tool_formatters.py using -# ReadOutput, EditOutput, etc. (see Tool Output Content Models section) -ToolUseResult = Union[ - str, - list[Any], # Covers list[TodoWriteItem], list[ContentItem], etc. - dict[str, Any], # Covers structured results -] +@dataclass +class TaskOutput: + """Parsed Task (sub-agent) tool output. -class BaseTranscriptEntry(BaseModel): - parentUuid: Optional[str] - isSidechain: bool - userType: str - cwd: str - sessionId: str - version: str - uuid: str - timestamp: str - isMeta: Optional[bool] = None - agentId: Optional[str] = None # Agent ID for sidechain messages + Symmetric with TaskInput for tool_use β†’ tool_result pairing. + Contains the agent's final response as markdown. + """ + result: str # Agent's response (markdown) -class UserTranscriptEntry(BaseTranscriptEntry): - type: Literal["user"] - message: UserMessage - toolUseResult: Optional[ToolUseResult] = None - agentId: Optional[str] = None # From toolUseResult when present +@dataclass +class GlobOutput: + """Parsed Glob tool output. -class AssistantTranscriptEntry(BaseTranscriptEntry): - type: Literal["assistant"] - message: AssistantMessage - requestId: Optional[str] = None + Symmetric with GlobInput for tool_use β†’ tool_result pairing. + TODO: Not currently used - tool results handled as raw strings. + """ -class SummaryTranscriptEntry(BaseModel): - type: Literal["summary"] - summary: str - leafUuid: str - cwd: Optional[str] = None - sessionId: None = None # Summaries don't have a sessionId + pattern: str + files: list[str] # Matching file paths + truncated: bool # Whether list was truncated -class SystemTranscriptEntry(BaseTranscriptEntry): - """System messages like warnings, notifications, hook summaries, etc.""" +@dataclass +class GrepOutput: + """Parsed Grep tool output. - type: Literal["system"] - content: Optional[str] = None - subtype: Optional[str] = None # e.g., "stop_hook_summary" - level: Optional[str] = None # e.g., "warning", "info", "error" - # Hook summary fields (for subtype="stop_hook_summary") - hasOutput: Optional[bool] = None - hookErrors: Optional[list[str]] = None - hookInfos: Optional[list[dict[str, Any]]] = None - preventedContinuation: Optional[bool] = None + Symmetric with GrepInput for tool_use β†’ tool_result pairing. + TODO: Not currently used - tool results handled as raw strings. + """ -class QueueOperationTranscriptEntry(BaseModel): - """Queue operations (enqueue/dequeue/remove) for message queueing tracking. + pattern: str + matches: list[str] # Matching lines/files + output_mode: str # "content", "files_with_matches", or "count" + truncated: bool - enqueue/dequeue are internal operations that track when messages are queued and dequeued. - They are parsed but not rendered, as the content duplicates actual user messages. - 'remove' operations are out-of-band user inputs made visible to the agent while working - for "steering" purposes. These should be rendered as user messages with a 'steering' CSS class. - Content can be a list of ContentItems or a simple string (for 'remove' operations). +@dataclass +class AskUserQuestionAnswer: + """Single Q&A pair from AskUserQuestion result.""" + + question: str + answer: str + + +@dataclass +class AskUserQuestionOutput: + """Parsed AskUserQuestion tool output. + + Symmetric with AskUserQuestionInput for tool_use β†’ tool_result pairing. + Contains the Q&A pairs extracted from the result message. """ - type: Literal["queue-operation"] - operation: Literal["enqueue", "dequeue", "remove", "popAll"] - timestamp: str - sessionId: str - content: Optional[Union[list[ContentItem], str]] = ( - None # List for enqueue, str for remove/popAll - ) + answers: list[AskUserQuestionAnswer] # Q&A pairs + raw_message: str # Original message for fallback -TranscriptEntry = Union[ - UserTranscriptEntry, - AssistantTranscriptEntry, - SummaryTranscriptEntry, - SystemTranscriptEntry, - QueueOperationTranscriptEntry, +@dataclass +class ExitPlanModeOutput: + """Parsed ExitPlanMode tool output. + + Symmetric with ExitPlanModeInput for tool_use β†’ tool_result pairing. + Truncates redundant plan echo on success. + """ + + message: str # Truncated message (without redundant plan) + approved: bool # Whether the plan was approved + + +# Union of all specialized output types + ToolResultContent as generic fallback +ToolOutput = Union[ + ReadOutput, + WriteOutput, + EditOutput, + BashOutput, + TaskOutput, + AskUserQuestionOutput, + ExitPlanModeOutput, + # TODO: Add as parsers are implemented: + # GlobOutput, GrepOutput + ToolResultContent, # Generic fallback for unparsed results ] diff --git a/claude_code_log/parser.py b/claude_code_log/parser.py index a0528849..fc48cd1d 100644 --- a/claude_code_log/parser.py +++ b/claude_code_log/parser.py @@ -1,83 +1,24 @@ #!/usr/bin/env python3 -"""Parse and extract data from Claude transcript JSONL files.""" +"""Parse and extract data from Claude transcript JSONL files. -import json -import re -from typing import Any, Callable, Optional, Union, cast -from datetime import datetime +This module provides utility functions for parsing transcript data: +- extract_text_content: Extract text from content items +- parse_timestamp: Parse ISO timestamps + +For transcript entry and content item creation, see factories/. +""" -from pydantic import BaseModel +from datetime import datetime +from typing import Optional -from .models import ( - # Content types - ContentItem, - TextContent, - ThinkingContent, - ToolUseContent, - ToolResultContent, - ImageContent, - # User message content models - SlashCommandContent, - CommandOutputContent, - BashInputContent, - BashOutputContent, - CompactedSummaryContent, - UserMemoryContent, - UserSlashCommandContent, - UserTextContent, - IdeNotificationContent, - IdeOpenedFile, - IdeSelection, - IdeDiagnostic, - # Assistant message content models - BashInput, - ReadInput, - WriteInput, - EditInput, - EditItem, - MultiEditInput, - GlobInput, - GrepInput, - TaskInput, - TodoWriteInput, - TodoWriteItem, - AskUserQuestionInput, - AskUserQuestionItem, - AskUserQuestionOption, - ExitPlanModeInput, - ToolInput, - # Usage and transcript entry types - UsageInfo, - MessageType, - TranscriptEntry, - UserTranscriptEntry, - AssistantTranscriptEntry, - SummaryTranscriptEntry, - SystemTranscriptEntry, - QueueOperationTranscriptEntry, -) +from .models import ContentItem, TextContent def extract_text_content(content: Optional[list[ContentItem]]) -> str: - """Extract text content from Claude message content structure. - - Supports both custom models (TextContent, ThinkingContent) and official - Anthropic SDK types (TextBlock, ThinkingBlock). - """ + """Extract text content from Claude message content structure.""" if not content: return "" - text_parts: list[str] = [] - for item in content: - # Skip thinking content - if ( - isinstance(item, ThinkingContent) - or getattr(item, "type", None) == "thinking" - ): - continue - # Handle text content - if hasattr(item, "text"): - text_parts.append(getattr(item, "text")) # type: ignore[arg-type] - return "\n".join(text_parts) + return "\n".join(item.text for item in content if isinstance(item, TextContent)) def parse_timestamp(timestamp_str: str) -> Optional[datetime]: @@ -86,863 +27,3 @@ def parse_timestamp(timestamp_str: str) -> Optional[datetime]: return datetime.fromisoformat(timestamp_str.replace("Z", "+00:00")) except (ValueError, AttributeError): return None - - -# ============================================================================= -# User Message Content Parsing -# ============================================================================= - - -def parse_slash_command(text: str) -> Optional[SlashCommandContent]: - """Parse slash command tags from text. - - Args: - text: Raw text that may contain command-name, command-args, command-contents tags - - Returns: - SlashCommandContent if tags found, None otherwise - """ - command_name_match = re.search(r"([^<]+)", text) - if not command_name_match: - return None - - command_name = command_name_match.group(1).strip() - - command_args_match = re.search(r"([^<]*)", text) - command_args = command_args_match.group(1).strip() if command_args_match else "" - - # Parse command contents, handling JSON format - command_contents_match = re.search( - r"(.+?)", text, re.DOTALL - ) - command_contents = "" - if command_contents_match: - contents_text = command_contents_match.group(1).strip() - # Try to parse as JSON and extract the text field - try: - contents_json: Any = json.loads(contents_text) - if isinstance(contents_json, dict) and "text" in contents_json: - text_dict = cast(dict[str, Any], contents_json) - text_value = text_dict["text"] - command_contents = str(text_value) - else: - command_contents = contents_text - except json.JSONDecodeError: - command_contents = contents_text - - return SlashCommandContent( - command_name=command_name, - command_args=command_args, - command_contents=command_contents, - ) - - -def parse_command_output(text: str) -> Optional[CommandOutputContent]: - """Parse command output tags from text. - - Args: - text: Raw text that may contain local-command-stdout tags - - Returns: - CommandOutputContent if tags found, None otherwise - """ - stdout_match = re.search( - r"(.*?)", - text, - re.DOTALL, - ) - if not stdout_match: - return None - - stdout_content = stdout_match.group(1).strip() - # Check if content looks like markdown (starts with markdown headers) - is_markdown = bool(re.match(r"^#+\s+", stdout_content, re.MULTILINE)) - - return CommandOutputContent(stdout=stdout_content, is_markdown=is_markdown) - - -def parse_bash_input(text: str) -> Optional[BashInputContent]: - """Parse bash input tags from text. - - Args: - text: Raw text that may contain bash-input tags - - Returns: - BashInputContent if tags found, None otherwise - """ - bash_match = re.search(r"(.*?)", text, re.DOTALL) - if not bash_match: - return None - - return BashInputContent(command=bash_match.group(1).strip()) - - -def parse_bash_output(text: str) -> Optional[BashOutputContent]: - """Parse bash output tags from text. - - Args: - text: Raw text that may contain bash-stdout/bash-stderr tags - - Returns: - BashOutputContent if tags found, None otherwise - """ - stdout_match = re.search(r"(.*?)", text, re.DOTALL) - stderr_match = re.search(r"(.*?)", text, re.DOTALL) - - if not stdout_match and not stderr_match: - return None - - stdout = stdout_match.group(1).strip() if stdout_match else None - stderr = stderr_match.group(1).strip() if stderr_match else None - - # Convert empty strings to None for cleaner representation - if stdout == "": - stdout = None - if stderr == "": - stderr = None - - return BashOutputContent(stdout=stdout, stderr=stderr) - - -# Shared regex patterns for IDE notification tags -IDE_OPENED_FILE_PATTERN = re.compile( - r"(.*?)", re.DOTALL -) -IDE_SELECTION_PATTERN = re.compile(r"(.*?)", re.DOTALL) -IDE_DIAGNOSTICS_PATTERN = re.compile( - r"\s*(.*?)\s*", - re.DOTALL, -) - - -def parse_ide_notifications(text: str) -> Optional[IdeNotificationContent]: - """Parse IDE notification tags from text. - - Handles: - - : Simple file open notifications - - : Code selection notifications - - : JSON diagnostic arrays - - Args: - text: Raw text that may contain IDE notification tags - - Returns: - IdeNotificationContent if any tags found, None otherwise - """ - opened_files: list[IdeOpenedFile] = [] - selections: list[IdeSelection] = [] - diagnostics: list[IdeDiagnostic] = [] - remaining_text = text - - # Pattern 1: content - for match in IDE_OPENED_FILE_PATTERN.finditer(remaining_text): - content = match.group(1).strip() - opened_files.append(IdeOpenedFile(content=content)) - - remaining_text = IDE_OPENED_FILE_PATTERN.sub("", remaining_text) - - # Pattern 2: content - for match in IDE_SELECTION_PATTERN.finditer(remaining_text): - content = match.group(1).strip() - selections.append(IdeSelection(content=content)) - - remaining_text = IDE_SELECTION_PATTERN.sub("", remaining_text) - - # Pattern 3: JSON - for match in IDE_DIAGNOSTICS_PATTERN.finditer(remaining_text): - json_content = match.group(1).strip() - try: - parsed_diagnostics: Any = json.loads(json_content) - if isinstance(parsed_diagnostics, list): - diagnostics.append( - IdeDiagnostic( - diagnostics=cast(list[dict[str, Any]], parsed_diagnostics) - ) - ) - else: - # Not a list, store as raw content - diagnostics.append(IdeDiagnostic(raw_content=json_content)) - except (json.JSONDecodeError, ValueError): - # JSON parsing failed, store raw content - diagnostics.append(IdeDiagnostic(raw_content=json_content)) - - remaining_text = IDE_DIAGNOSTICS_PATTERN.sub("", remaining_text) - - # Only return if we found any IDE tags - if not opened_files and not selections and not diagnostics: - return None - - return IdeNotificationContent( - opened_files=opened_files, - selections=selections, - diagnostics=diagnostics, - remaining_text=remaining_text.strip(), - ) - - -# Pattern for compacted session summary detection -COMPACTED_SUMMARY_PREFIX = "This session is being continued from a previous conversation that ran out of context" - - -def parse_compacted_summary( - content_list: list[ContentItem], -) -> Optional[CompactedSummaryContent]: - """Parse compacted session summary from content list. - - Compacted summaries are generated when a session runs out of context and - needs to be continued. They contain a summary of the previous conversation. - - If the first text item starts with the compacted summary prefix, all text - items are combined into a single CompactedSummaryContent. - - Args: - content_list: List of ContentItem from user message - - Returns: - CompactedSummaryContent if first text is a compacted summary, None otherwise - """ - if not content_list or not hasattr(content_list[0], "text"): - return None - - first_text = getattr(content_list[0], "text", "") - if not first_text.startswith(COMPACTED_SUMMARY_PREFIX): - return None - - # Combine all text content for compacted summaries - # Use hasattr check to handle both TextContent models and SDK TextBlock objects - texts = cast( - list[str], - [item.text for item in content_list if hasattr(item, "text")], # type: ignore[union-attr] - ) - all_text = "\n\n".join(texts) - return CompactedSummaryContent(summary_text=all_text) - - -# Pattern for user memory input tag -USER_MEMORY_PATTERN = re.compile( - r"(.*?)", re.DOTALL -) - - -def parse_user_memory(text: str) -> Optional[UserMemoryContent]: - """Parse user memory input tag from text. - - User memory input contains context that the user has provided from - their CLAUDE.md or other memory sources. - - Args: - text: Raw text that may contain user memory input tag - - Returns: - UserMemoryContent if tag found, None otherwise - """ - match = USER_MEMORY_PATTERN.search(text) - if match: - memory_content = match.group(1).strip() - return UserMemoryContent(memory_text=memory_content) - return None - - -# Type alias for content models returned by parse_user_message_content -UserMessageContent = Union[ - CompactedSummaryContent, UserMemoryContent, UserSlashCommandContent, UserTextContent -] - - -def parse_user_message_content( - content_list: list[ContentItem], - is_slash_command: bool = False, -) -> Optional[UserMessageContent]: - """Parse user message content into a structured content model. - - Returns a content model for HtmlRenderer to format. The caller can use - isinstance() checks to determine the content type: - - UserSlashCommandContent: Slash command expanded prompts (isMeta=True) - - CompactedSummaryContent: Session continuation summaries - - UserMemoryContent: User memory input from CLAUDE.md - - UserTextContent: Normal user text with optional IDE notifications and images - - This function processes content items preserving their original order: - - TextContent items have IDE notifications extracted, producing - [IdeNotificationContent, TextContent] pairs - - ImageContent items are preserved as-is - - Args: - content_list: List of ContentItem from user message - is_slash_command: True for slash command expanded prompts (isMeta=True) - - Returns: - A content model, or None if content_list is empty. - """ - if not content_list: - return None - - # Slash command expanded prompts - combine all text as markdown - if is_slash_command: - all_text = "\n\n".join( - getattr(item, "text", "") for item in content_list if hasattr(item, "text") - ) - return UserSlashCommandContent(text=all_text) if all_text else None - - # Get first text item for special case detection - first_text_item = next( - (item for item in content_list if hasattr(item, "text")), - None, - ) - first_text = getattr(first_text_item, "text", "") if first_text_item else "" - - # Check for compacted session summary first (handles text combining internally) - compacted = parse_compacted_summary(content_list) - if compacted: - return compacted - - # Check for user memory input - user_memory = parse_user_memory(first_text) - if user_memory: - return user_memory - - # Build items list preserving order, extracting IDE notifications from text - items: list[TextContent | ImageContent | IdeNotificationContent] = [] - - for item in content_list: - # Check for text content - if hasattr(item, "text"): - item_text: str = getattr(item, "text") # type: ignore[assignment] - ide_content = parse_ide_notifications(item_text) - - if ide_content: - # Add IDE notification item first - items.append(ide_content) - remaining_text: str = ide_content.remaining_text - else: - remaining_text = item_text - - # Add remaining text as TextContent if non-empty - if remaining_text.strip(): - items.append(TextContent(type="text", text=remaining_text)) - elif isinstance(item, ImageContent): - # ImageContent model - use as-is - items.append(item) - elif hasattr(item, "source") and getattr(item, "type", None) == "image": - # Anthropic ImageContent - convert to our model - items.append(ImageContent.model_validate(item.model_dump())) # type: ignore[union-attr] - - # Return UserTextContent with items list - return UserTextContent(items=items) - - -# ============================================================================= -# Message Type Detection -# ============================================================================= - - -def is_system_message(text_content: str) -> bool: - """Check if a message is a system message that should be filtered out.""" - system_message_patterns = [ - "Caveat: The messages below were generated by the user while running local commands. DO NOT respond to these messages or otherwise consider them in your response unless the user explicitly asks you to.", - "[Request interrupted by user for tool use]", - "", - ] - - return any(text_content.startswith(pattern) for pattern in system_message_patterns) - - -def is_command_message(text_content: str) -> bool: - """Check if a message contains command information that should be displayed.""" - return "" in text_content and "" in text_content - - -def is_local_command_output(text_content: str) -> bool: - """Check if a message contains local command output.""" - return "" in text_content - - -def is_bash_input(text_content: str) -> bool: - """Check if a message contains bash input command.""" - return "" in text_content and "" in text_content - - -def is_bash_output(text_content: str) -> bool: - """Check if a message contains bash command output.""" - return "" in text_content or "" in text_content - - -def is_warmup_only_session(messages: list[TranscriptEntry], session_id: str) -> bool: - """Check if a session contains only warmup user messages. - - A warmup session is one where ALL user messages are literally just "Warmup". - Sessions with no user messages return False (not considered warmup). - - Args: - messages: List of all transcript entries - session_id: The session ID to check - - Returns: - True if ALL user messages in the session are "Warmup", False otherwise - """ - user_messages_in_session: list[str] = [] - - for message in messages: - if ( - isinstance(message, UserTranscriptEntry) - and getattr(message, "sessionId", "") == session_id - and hasattr(message, "message") - ): - text_content = extract_text_content(message.message.content).strip() - user_messages_in_session.append(text_content) - - # No user messages = not a warmup session - if not user_messages_in_session: - return False - - # All user messages must be exactly "Warmup" - return all(msg == "Warmup" for msg in user_messages_in_session) - - -# ============================================================================= -# Type Guards for TranscriptEntry -# ============================================================================= - - -def as_user_entry(entry: TranscriptEntry) -> UserTranscriptEntry | None: - """Return entry as UserTranscriptEntry if it is one, else None.""" - if entry.type == MessageType.USER: - return cast(UserTranscriptEntry, entry) - return None - - -def as_assistant_entry(entry: TranscriptEntry) -> AssistantTranscriptEntry | None: - """Return entry as AssistantTranscriptEntry if it is one, else None.""" - if entry.type == MessageType.ASSISTANT: - return cast(AssistantTranscriptEntry, entry) - return None - - -# ============================================================================= -# Tool Input Parsing -# ============================================================================= - -TOOL_INPUT_MODELS: dict[str, type[BaseModel]] = { - "Bash": BashInput, - "Read": ReadInput, - "Write": WriteInput, - "Edit": EditInput, - "MultiEdit": MultiEditInput, - "Glob": GlobInput, - "Grep": GrepInput, - "Task": TaskInput, - "TodoWrite": TodoWriteInput, - "AskUserQuestion": AskUserQuestionInput, - "ask_user_question": AskUserQuestionInput, # Legacy tool name - "ExitPlanMode": ExitPlanModeInput, -} - - -# -- Lenient Parsing Helpers -------------------------------------------------- -# These functions create typed models even when strict validation fails. -# They use defaults for missing fields and skip invalid nested items. - - -def _parse_todowrite_lenient(data: dict[str, Any]) -> TodoWriteInput: - """Parse TodoWrite input leniently, handling malformed data.""" - todos_raw = data.get("todos", []) - valid_todos: list[TodoWriteItem] = [] - for item in todos_raw: - if isinstance(item, dict): - try: - valid_todos.append(TodoWriteItem.model_validate(item)) - except Exception: - pass - elif isinstance(item, str): - valid_todos.append(TodoWriteItem(content=item)) - return TodoWriteInput(todos=valid_todos) - - -def _parse_bash_lenient(data: dict[str, Any]) -> BashInput: - """Parse Bash input leniently.""" - return BashInput( - command=data.get("command", ""), - description=data.get("description"), - timeout=data.get("timeout"), - run_in_background=data.get("run_in_background"), - ) - - -def _parse_write_lenient(data: dict[str, Any]) -> WriteInput: - """Parse Write input leniently.""" - return WriteInput( - file_path=data.get("file_path", ""), - content=data.get("content", ""), - ) - - -def _parse_edit_lenient(data: dict[str, Any]) -> EditInput: - """Parse Edit input leniently.""" - return EditInput( - file_path=data.get("file_path", ""), - old_string=data.get("old_string", ""), - new_string=data.get("new_string", ""), - replace_all=data.get("replace_all"), - ) - - -def _parse_multiedit_lenient(data: dict[str, Any]) -> MultiEditInput: - """Parse Multiedit input leniently.""" - edits_raw = data.get("edits", []) - valid_edits: list[EditItem] = [] - for edit in edits_raw: - if isinstance(edit, dict): - try: - valid_edits.append(EditItem.model_validate(edit)) - except Exception: - pass - return MultiEditInput(file_path=data.get("file_path", ""), edits=valid_edits) - - -def _parse_task_lenient(data: dict[str, Any]) -> TaskInput: - """Parse Task input leniently.""" - return TaskInput( - prompt=data.get("prompt", ""), - subagent_type=data.get("subagent_type", ""), - description=data.get("description", ""), - model=data.get("model"), - run_in_background=data.get("run_in_background"), - resume=data.get("resume"), - ) - - -def _parse_read_lenient(data: dict[str, Any]) -> ReadInput: - """Parse Read input leniently.""" - return ReadInput( - file_path=data.get("file_path", ""), - offset=data.get("offset"), - limit=data.get("limit"), - ) - - -def _parse_askuserquestion_lenient(data: dict[str, Any]) -> AskUserQuestionInput: - """Parse AskUserQuestion input leniently, handling malformed data.""" - questions_raw = data.get("questions", []) - valid_questions: list[AskUserQuestionItem] = [] - for q in questions_raw: - if isinstance(q, dict): - q_dict = cast(dict[str, Any], q) - try: - # Parse options leniently - options_raw = q_dict.get("options", []) - valid_options: list[AskUserQuestionOption] = [] - for opt in options_raw: - if isinstance(opt, dict): - try: - valid_options.append( - AskUserQuestionOption.model_validate(opt) - ) - except Exception: - pass - valid_questions.append( - AskUserQuestionItem( - question=str(q_dict.get("question", "")), - header=q_dict.get("header"), - options=valid_options, - multiSelect=bool(q_dict.get("multiSelect", False)), - ) - ) - except Exception: - pass - return AskUserQuestionInput( - questions=valid_questions, - question=data.get("question"), - ) - - -def _parse_exitplanmode_lenient(data: dict[str, Any]) -> ExitPlanModeInput: - """Parse ExitPlanMode input leniently.""" - return ExitPlanModeInput( - plan=data.get("plan", ""), - launchSwarm=data.get("launchSwarm"), - teammateCount=data.get("teammateCount"), - ) - - -# Mapping of tool names to their lenient parsers -TOOL_LENIENT_PARSERS: dict[str, Any] = { - "Bash": _parse_bash_lenient, - "Write": _parse_write_lenient, - "Edit": _parse_edit_lenient, - "MultiEdit": _parse_multiedit_lenient, - "Task": _parse_task_lenient, - "TodoWrite": _parse_todowrite_lenient, - "Read": _parse_read_lenient, - "AskUserQuestion": _parse_askuserquestion_lenient, - "ask_user_question": _parse_askuserquestion_lenient, # Legacy tool name - "ExitPlanMode": _parse_exitplanmode_lenient, -} - - -def parse_tool_input(tool_name: str, input_data: dict[str, Any]) -> ToolInput: - """Parse tool input dictionary into a typed model. - - Uses strict validation first, then lenient parsing if available. - - Args: - tool_name: The name of the tool (e.g., "Bash", "Read") - input_data: The raw input dictionary from the tool_use content - - Returns: - A typed input model if available, otherwise the original dictionary - """ - model_class = TOOL_INPUT_MODELS.get(tool_name) - if model_class is not None: - try: - return cast(ToolInput, model_class.model_validate(input_data)) - except Exception: - # Try lenient parsing if available - lenient_parser = TOOL_LENIENT_PARSERS.get(tool_name) - if lenient_parser is not None: - return cast(ToolInput, lenient_parser(input_data)) - return input_data - return input_data - - -# ============================================================================= -# Usage Info Normalization -# ============================================================================= - - -def normalize_usage_info(usage_data: Any) -> Optional[UsageInfo]: - """Normalize usage data from various formats to UsageInfo.""" - if usage_data is None: - return None - - # If it's already a UsageInfo instance, return as-is - if isinstance(usage_data, UsageInfo): - return usage_data - - # If it's a dict, validate and convert - if isinstance(usage_data, dict): - return UsageInfo.model_validate(usage_data) - - # Handle object-like access (e.g., from SDK types) - if hasattr(usage_data, "input_tokens"): - server_tool_use = getattr(usage_data, "server_tool_use", None) - if server_tool_use is not None and hasattr(server_tool_use, "model_dump"): - server_tool_use = server_tool_use.model_dump() - return UsageInfo( - input_tokens=getattr(usage_data, "input_tokens", None), - output_tokens=getattr(usage_data, "output_tokens", None), - cache_creation_input_tokens=getattr( - usage_data, "cache_creation_input_tokens", None - ), - cache_read_input_tokens=getattr( - usage_data, "cache_read_input_tokens", None - ), - service_tier=getattr(usage_data, "service_tier", None), - server_tool_use=server_tool_use, - ) - - return None - - -# ============================================================================= -# Content Item Parsing -# ============================================================================= -# Functions to parse content items from JSONL data. Organized by entry type -# to clarify which content types can appear in which context. - - -def _parse_text_content(item_data: dict[str, Any]) -> ContentItem: - """Parse text content. - - Common to both user and assistant messages. - """ - return TextContent.model_validate(item_data) - - -def parse_user_content_item(item_data: dict[str, Any]) -> ContentItem: - """Parse a content item from a UserTranscriptEntry. - - User messages can contain: - - text: User-typed text - - tool_result: Results from tool execution - - image: User-attached images - """ - try: - content_type = item_data.get("type", "") - - if content_type == "text": - return _parse_text_content(item_data) - elif content_type == "tool_result": - return ToolResultContent.model_validate(item_data) - elif content_type == "image": - return ImageContent.model_validate(item_data) - else: - # Fallback to text content for unknown types - return TextContent(type="text", text=str(item_data)) - except Exception: - return TextContent(type="text", text=str(item_data)) - - -def parse_assistant_content_item(item_data: dict[str, Any]) -> ContentItem: - """Parse a content item from an AssistantTranscriptEntry. - - Assistant messages can contain: - - text: Assistant's response text - - tool_use: Tool invocations - - thinking: Extended thinking blocks - """ - try: - content_type = item_data.get("type", "") - - if content_type == "text": - return _parse_text_content(item_data) - elif content_type == "tool_use": - return ToolUseContent.model_validate(item_data) - elif content_type == "thinking": - return ThinkingContent.model_validate(item_data) - else: - # Fallback to text content for unknown types - return TextContent(type="text", text=str(item_data)) - except Exception: - return TextContent(type="text", text=str(item_data)) - - -def parse_content_item(item_data: dict[str, Any]) -> ContentItem: - """Parse a content item (generic fallback). - - For cases where the entry type is unknown. Handles all content types. - Prefer parse_user_content_item or parse_assistant_content_item when - the entry type is known. - """ - try: - content_type = item_data.get("type", "") - - if content_type == "tool_result": - return ToolResultContent.model_validate(item_data) - elif content_type == "image": - return ImageContent.model_validate(item_data) - elif content_type == "tool_use": - return ToolUseContent.model_validate(item_data) - elif content_type == "thinking": - return ThinkingContent.model_validate(item_data) - elif content_type == "text": - return _parse_text_content(item_data) - else: - # Fallback to text content for unknown types - return TextContent(type="text", text=str(item_data)) - except Exception: - return TextContent(type="text", text=str(item_data)) - - -def parse_message_content( - content_data: Any, - item_parser: Callable[[dict[str, Any]], ContentItem] = parse_content_item, -) -> list[ContentItem]: - """Parse message content, normalizing to a list of ContentItems. - - Always returns a list for consistent downstream handling. String content - is wrapped in a TextContent item. - - Args: - content_data: Raw content data (string or list of items) - item_parser: Function to parse individual content items. Defaults to - generic parse_content_item, but can be parse_user_content_item or - parse_assistant_content_item for type-specific parsing. - """ - if isinstance(content_data, str): - return [TextContent(type="text", text=content_data)] - elif isinstance(content_data, list): - content_list = cast(list[Any], content_data) - result: list[ContentItem] = [] - for item in content_list: - if isinstance(item, dict): - result.append(item_parser(cast(dict[str, Any], item))) - else: - # Non-dict items (e.g., raw strings) become TextContent - result.append(TextContent(type="text", text=str(item))) - return result - else: - return [TextContent(type="text", text=str(content_data))] - - -# ============================================================================= -# Transcript Entry Parsing -# ============================================================================= - - -def parse_transcript_entry(data: dict[str, Any]) -> TranscriptEntry: - """ - Parse a JSON dictionary into the appropriate TranscriptEntry type. - - Enhanced to optionally use official Anthropic types for assistant messages. - - Args: - data: Dictionary parsed from JSON - - Returns: - The appropriate TranscriptEntry subclass - - Raises: - ValueError: If the data doesn't match any known transcript entry type - """ - entry_type = data.get("type") - - if entry_type == "user": - # Parse message content if present, using user-specific parser - data_copy = data.copy() - if "message" in data_copy and "content" in data_copy["message"]: - data_copy["message"] = data_copy["message"].copy() - data_copy["message"]["content"] = parse_message_content( - data_copy["message"]["content"], - item_parser=parse_user_content_item, - ) - # Parse toolUseResult if present and it's a list of content items - if "toolUseResult" in data_copy and isinstance( - data_copy["toolUseResult"], list - ): - # Check if it's a list of content items (MCP tool results) - tool_use_result = cast(list[Any], data_copy["toolUseResult"]) - if ( - tool_use_result - and isinstance(tool_use_result[0], dict) - and "type" in tool_use_result[0] - ): - data_copy["toolUseResult"] = [ - parse_content_item(cast(dict[str, Any], item)) - for item in tool_use_result - if isinstance(item, dict) - ] - return UserTranscriptEntry.model_validate(data_copy) - - elif entry_type == "assistant": - data_copy = data.copy() - - # Parse assistant message content - if "message" in data_copy and "content" in data_copy["message"]: - message_copy = data_copy["message"].copy() - message_copy["content"] = parse_message_content( - message_copy["content"], - item_parser=parse_assistant_content_item, - ) - - # Normalize usage data to support both Anthropic and custom formats - if "usage" in message_copy: - message_copy["usage"] = normalize_usage_info(message_copy["usage"]) - - data_copy["message"] = message_copy - return AssistantTranscriptEntry.model_validate(data_copy) - - elif entry_type == "summary": - return SummaryTranscriptEntry.model_validate(data) - - elif entry_type == "system": - return SystemTranscriptEntry.model_validate(data) - - elif entry_type == "queue-operation": - # Parse content if present (in enqueue and remove operations) - data_copy = data.copy() - if "content" in data_copy and isinstance(data_copy["content"], list): - data_copy["content"] = parse_message_content(data_copy["content"]) - return QueueOperationTranscriptEntry.model_validate(data_copy) - - else: - raise ValueError(f"Unknown transcript entry type: {entry_type}") diff --git a/claude_code_log/renderer.py b/claude_code_log/renderer.py index 3e5209d9..f8de733e 100644 --- a/claude_code_log/renderer.py +++ b/claude_code_log/renderer.py @@ -1,17 +1,21 @@ #!/usr/bin/env python3 """Render Claude transcript data to HTML format.""" +from __future__ import annotations + +import re import time -from dataclasses import dataclass +from dataclasses import dataclass, field, replace from pathlib import Path -from typing import Any, Callable, Optional, Tuple, TYPE_CHECKING +from typing import TYPE_CHECKING, Any, Optional, Tuple, cast +from datetime import datetime if TYPE_CHECKING: from .cache import CacheManager - from .models import MessageContent -from datetime import datetime from .models import ( + MessageContent, + MessageMeta, MessageType, TranscriptEntry, AssistantTranscriptEntry, @@ -21,34 +25,36 @@ ContentItem, TextContent, ToolResultContent, - ToolResultContentModel, ToolUseContent, ThinkingContent, - ThinkingContentModel, + UsageInfo, # Structured content types - AssistantTextContent, - CommandOutputContent, - CompactedSummaryContent, - DedupNoticeContent, - HookInfo, - HookSummaryContent, - SessionHeaderContent, - SlashCommandContent, - SystemContent, - UnknownContent, - UserMemoryContent, - UserSlashCommandContent, - UserSteeringContent, - UserTextContent, + AssistantTextMessage, + CommandOutputMessage, + DedupNoticeMessage, + SessionHeaderMessage, + SlashCommandMessage, + SystemMessage, + TaskOutput, + ToolResultMessage, + ToolUseMessage, + UnknownMessage, + UserSlashCommandMessage, + UserSteeringMessage, + UserTextMessage, ) -from .parser import ( +from .parser import extract_text_content +from .factories import ( as_assistant_entry, as_user_entry, - extract_text_content, - is_bash_input, - is_bash_output, - is_command_message, - is_local_command_output, + create_assistant_message, + create_meta, + create_system_message, + create_thinking_message, + create_tool_result_message, + create_tool_use_message, + create_user_message, + ToolItemResult, ) from .utils import ( format_timestamp, @@ -59,32 +65,226 @@ create_session_preview, ) from .renderer_timings import ( - DEBUG_TIMING, - report_timing_statistics, - set_timing_var, log_timing, ) -from .html import ( - escape_html, - format_tool_use_title, - parse_bash_input, - parse_bash_output, - parse_command_output, - parse_slash_command, -) -from .parser import parse_user_message_content +# -- Rendering Context -------------------------------------------------------- + + +@dataclass +class RenderingContext: + """Context for a single rendering operation. + + Holds render-time state that should not pollute MessageContent. + This enables parallel-safe rendering where each render gets its own context. + + Attributes: + messages: Registry of all TemplateMessage objects (message_index = index). + tool_use_context: Maps tool_use_id -> ToolUseContent for result rendering. + session_first_message: Maps session_id -> index of first message in session. + """ + + messages: list[TemplateMessage] = field( + default_factory=lambda: [] # type: list[TemplateMessage] + ) + tool_use_context: dict[str, ToolUseContent] = field( + default_factory=lambda: {} # type: dict[str, ToolUseContent] + ) + session_first_message: dict[str, int] = field( + default_factory=lambda: {} # type: dict[str, int] + ) + + def register(self, message: "TemplateMessage") -> int: + """Register a TemplateMessage and assign its message_index. + + Args: + message: The TemplateMessage to register. + + Returns: + The assigned message_index (= index in messages list). + """ + msg_index = len(self.messages) + message.message_index = msg_index + self.messages.append(message) + return msg_index + + def get(self, message_index: int) -> Optional["TemplateMessage"]: + """Get a TemplateMessage by its message_index. + + Args: + message_index: The message_index (index) to look up. + + Returns: + The TemplateMessage if found, None if out of range. + """ + if 0 <= message_index < len(self.messages): + return self.messages[message_index] + return None + + +# -- Template Classes --------------------------------------------------------- + + +class TemplateMessage: + """Structured message data for template rendering. + + This is the primary render-time object that wraps MessageContent. Each + MessageContent has exactly one TemplateMessage wrapper. + + TemplateMessage holds all render-time state: + - message_index: Index in RenderingContext.messages (unique identifier) + - Pairing metadata: pair_first, pair_last, pair_duration + - Hierarchy metadata: ancestry + - Tree structure: children, fold/unfold counts + + All identity/context fields come from meta (timestamp, session_id, etc.) + and content (tool_use_id, has_markdown, token_usage, etc.). + """ + + def __init__( + self, + content: "MessageContent", + *, # Force keyword arguments after this + ancestry: Optional[list[int]] = None, + ): + # Content carries its own meta + self.content = content + self.meta = content.meta + + # Unique index in RenderingContext.messages (assigned by ctx.register()) + self.message_index: Optional[int] = None + + # Pairing metadata (assigned by _mark_pair()) + self.pair_first: Optional[int] = None # Index of first message in pair + self.pair_last: Optional[int] = None # Index of last message in pair + self.pair_duration: Optional[str] = None # Duration string for pair_last + + # Rendering metadata + self.ancestry = ancestry or [] + + # Fold/unfold counts + self.immediate_children_count = 0 # Direct children only + self.total_descendants_count = 0 # All descendants recursively + # Type-aware counting for smarter labels + self.immediate_children_by_type: dict[ + str, int + ] = {} # {"assistant": 2, "tool_use": 3} + self.total_descendants_by_type: dict[str, int] = {} # All descendants by type + + # Children for tree-based rendering + self.children: list["TemplateMessage"] = [] -# -- Content Formatters ------------------------------------------------------- -# NOTE: Content formatters have been moved to html/ submodules: -# - format_thinking_content -> html/assistant_formatters.py -# - format_assistant_text_content -> html/assistant_formatters.py -# - format_tool_result_content -> html/tool_formatters.py -# - format_tool_use_content -> html/tool_formatters.py -# - format_image_content -> html/assistant_formatters.py -# - format_user_text_model_content -> html/user_formatters.py -# - parse_user_message_content -> parser.py + # -- Properties derived from content/meta -- + + @property + def type(self) -> str: + """Get message type from content.""" + return self.content.message_type + + @property + def is_session_header(self) -> bool: + """Check if this message is a session header.""" + return isinstance(self.content, SessionHeaderMessage) + + @property + def has_markdown(self) -> bool: + """Check if this message has markdown content.""" + return self.content.has_markdown + + @property + def has_children(self) -> bool: + """Check if this message has any children.""" + return bool(self.children) + + @property + def is_paired(self) -> bool: + """Check if this message is part of a pair.""" + return self.pair_first is not None or self.pair_last is not None + + @property + def is_first_in_pair(self) -> bool: + """Check if this is the first message in a pair (has pair_last set).""" + return self.pair_last is not None + + @property + def is_last_in_pair(self) -> bool: + """Check if this is the last message in a pair (has pair_first set).""" + return self.pair_first is not None + + @property + def pair_role(self) -> Optional[str]: + """Get the pairing role for CSS class. + + Returns: + "pair_first" if this is the first message in a pair, + "pair_last" if this is the last message in a pair, + None if not paired. + """ + if self.is_first_in_pair: + return "pair_first" + if self.is_last_in_pair: + return "pair_last" + return None + + @property + def message_id(self) -> Optional[str]: + """Get formatted message ID for HTML element IDs. + + Returns "d-{message_index}" for all messages, or None if not registered. + All messages use a unified format based on their index. + """ + if self.message_index is None: + return None + return f"d-{self.message_index}" + + @property + def session_id(self) -> str: + """Get session_id from meta.""" + return self.meta.session_id + + @property + def parent_uuid(self) -> Optional[str]: + """Get parent_uuid from meta.""" + return self.meta.parent_uuid + + @property + def agent_id(self) -> Optional[str]: + """Get agent_id from meta.""" + return self.meta.agent_id + + @property + def token_usage(self) -> Optional[str]: + """Get token_usage from content (if available).""" + return getattr(self.content, "token_usage", None) + + @property + def is_sidechain(self) -> bool: + """Check if this is a sidechain message.""" + return self.meta.is_sidechain + + @property + def tool_use_id(self) -> Optional[str]: + """Get tool_use_id from content (if ToolUseMessage or ToolResultMessage).""" + return getattr(self.content, "tool_use_id", None) + + @property + def title_hint(self) -> Optional[str]: + """Generate title hint from tool_use_id.""" + tool_id = self.tool_use_id + if tool_id: + # Escape for HTML attribute + escaped = tool_id.replace("&", "&").replace('"', """) + return f"ID: {escaped}" + return None + + def get_immediate_children_label(self) -> str: + """Generate human-readable label for immediate children.""" + return _format_type_counts(self.immediate_children_by_type) + + def get_total_descendants_label(self) -> str: + """Generate human-readable label for all descendants.""" + return _format_type_counts(self.total_descendants_by_type) def _format_type_counts(type_counts: dict[str, int]) -> str: @@ -163,108 +363,6 @@ def _format_type_counts(type_counts: dict[str, int]) -> str: return f"{parts[0]}, {parts[1]}, {remaining} more" -# -- Template Classes --------------------------------------------------------- - - -class TemplateMessage: - """Structured message data for template rendering.""" - - def __init__( - self, - message_type: str, - formatted_timestamp: str, - raw_timestamp: Optional[str] = None, - session_summary: Optional[str] = None, - session_id: Optional[str] = None, - is_session_header: bool = False, - token_usage: Optional[str] = None, - tool_use_id: Optional[str] = None, - title_hint: Optional[str] = None, - has_markdown: bool = False, - message_title: Optional[str] = None, - message_id: Optional[str] = None, - ancestry: Optional[list[str]] = None, - has_children: bool = False, - uuid: Optional[str] = None, - parent_uuid: Optional[str] = None, - agent_id: Optional[str] = None, - is_sidechain: bool = False, - content: Optional["MessageContent"] = None, - ): - self.type = message_type - # Structured content for rendering - self.content = content - self.formatted_timestamp = formatted_timestamp - self.is_sidechain = is_sidechain - self.raw_timestamp = raw_timestamp - # Display title for message header (capitalized, with decorations) - self.message_title = ( - message_title if message_title is not None else message_type.title() - ) - self.session_summary = session_summary - self.session_id = session_id - self.is_session_header = is_session_header - self.session_subtitle: Optional[str] = None - self.token_usage = token_usage - self.tool_use_id = tool_use_id - self.title_hint = title_hint - self.message_id = message_id - self.ancestry = ancestry or [] - self.has_children = has_children - self.has_markdown = has_markdown - self.uuid = uuid - self.parent_uuid = parent_uuid - self.agent_id = agent_id # Agent ID for sidechain messages and Task results - # Raw text content for deduplication (sidechain assistants vs Task results) - self.raw_text_content: Optional[str] = None - # Fold/unfold counts - self.immediate_children_count = 0 # Direct children only - self.total_descendants_count = 0 # All descendants recursively - # Type-aware counting for smarter labels - self.immediate_children_by_type: dict[ - str, int - ] = {} # {"assistant": 2, "tool_use": 3} - self.total_descendants_by_type: dict[str, int] = {} # All descendants by type - # Pairing metadata - self.is_paired = False - self.pair_role: Optional[str] = None # "pair_first", "pair_last", "pair_middle" - self.pair_duration: Optional[str] = None # Duration for pair_last messages - # Children for tree-based rendering (future use) - self.children: list["TemplateMessage"] = [] - - def get_immediate_children_label(self) -> str: - """Generate human-readable label for immediate children.""" - return _format_type_counts(self.immediate_children_by_type) - - def get_total_descendants_label(self) -> str: - """Generate human-readable label for all descendants.""" - return _format_type_counts(self.total_descendants_by_type) - - def flatten(self) -> list["TemplateMessage"]: - """Recursively flatten this message and all children into a list. - - Returns a list with this message followed by all descendants in - depth-first order. This provides backward compatibility with the - flat-list template rendering approach. - """ - result: list["TemplateMessage"] = [self] - for child in self.children: - result.extend(child.flatten()) - return result - - @staticmethod - def flatten_all(messages: list["TemplateMessage"]) -> list["TemplateMessage"]: - """Flatten a list of root messages into a single flat list. - - Useful for converting a tree structure back to a flat list for - templates that expect the traditional flat message list. - """ - result: list["TemplateMessage"] = [] - for message in messages: - result.extend(message.flatten()) - return result - - class TemplateProject: """Structured project data for template rendering.""" @@ -450,9 +548,9 @@ def generate_template_messages( if getattr(msg, "sessionId", None) not in warmup_session_ids ] - # Pre-process to find and attach session summaries + # Pre-process to find session summaries with log_timing("Session summary processing", t_start): - prepare_session_summaries(messages) + session_summaries = prepare_session_summaries(messages) # Filter messages (removes summaries, warmup, empty, etc.) with log_timing("Filter messages", t_start): @@ -461,27 +559,23 @@ def generate_template_messages( # Pass 1: Collect session metadata and token tracking with log_timing("Collect session info", t_start): sessions, session_order, show_tokens_for_message = _collect_session_info( - filtered_messages + filtered_messages, session_summaries ) # Pass 2: Render messages to TemplateMessage objects - with log_timing( - lambda: f"Render messages ({len(template_messages)} messages)", t_start - ): - template_messages = _render_messages( - filtered_messages, sessions, show_tokens_for_message - ) + with log_timing(lambda: f"Render messages ({len(ctx.messages)} messages)", t_start): + ctx = _render_messages(filtered_messages, sessions, show_tokens_for_message) - # Prepare session navigation data + # Prepare session navigation data (uses ctx for session header indices) with log_timing( lambda: f"Session navigation building ({len(session_nav)} sessions)", t_start ): - session_nav = prepare_session_navigation(sessions, session_order) + session_nav = prepare_session_navigation(sessions, session_order, ctx) # Reorder messages so each session's messages follow their session header # This fixes interleaving that occurs when sessions are resumed with log_timing("Reorder session messages", t_start): - template_messages = _reorder_session_template_messages(template_messages) + template_messages = _reorder_session_template_messages(ctx.messages) # Identify and mark paired messages (command+output, tool_use+tool_result, etc.) with log_timing("Identify message pairs", t_start): @@ -501,10 +595,6 @@ def generate_template_messages( with log_timing("Build message hierarchy", t_start): _build_message_hierarchy(template_messages) - # Resolve dedup notice targets (needs message_id from hierarchy) - with log_timing("Resolve dedup targets", t_start): - _resolve_dedup_targets(template_messages) - # Mark messages that have children for fold/unfold controls with log_timing("Mark messages with children", t_start): _mark_messages_with_children(template_messages) @@ -515,16 +605,23 @@ def generate_template_messages( with log_timing("Build message tree", t_start): root_messages = _build_message_tree(template_messages) + # Clean up sidechain duplicates on the tree structure + # - Remove first UserTextMessage (duplicate of Task input prompt) + # - Replace last AssistantTextMessage (duplicate of Task output) with DedupNotice + with log_timing("Cleanup sidechain duplicates", t_start): + _cleanup_sidechain_duplicates(root_messages) + return root_messages, session_nav # -- Session Utilities -------------------------------------------------------- -def prepare_session_summaries(messages: list[TranscriptEntry]) -> None: - """Pre-process messages to find and attach session summaries. +def prepare_session_summaries(messages: list[TranscriptEntry]) -> dict[str, str]: + """Extract session summaries from messages. - Modifies messages in place by attaching _session_summary attribute. + Returns: + Dict mapping session_id to summary text. """ session_summaries: dict[str, str] = {} uuid_to_session: dict[str, str] = {} @@ -555,23 +652,20 @@ def prepare_session_summaries(messages: list[TranscriptEntry]) -> None: ): session_summaries[uuid_to_session_backup[leaf_uuid]] = message.summary - # Attach summaries to messages - for message in messages: - if hasattr(message, "sessionId"): - session_id = getattr(message, "sessionId", "") - if session_id in session_summaries: - setattr(message, "_session_summary", session_summaries[session_id]) + return session_summaries def prepare_session_navigation( sessions: dict[str, dict[str, Any]], session_order: list[str], + ctx: RenderingContext, ) -> list[dict[str, Any]]: """Prepare session navigation data for template rendering. Args: sessions: Dictionary mapping session_id to session info dict session_order: List of session IDs in display order + ctx: RenderingContext with session_first_message indices Returns: List of session navigation dicts for template rendering @@ -609,9 +703,13 @@ def prepare_session_navigation( token_parts.append(f"Cache Read: {total_cache_read}") token_summary = "Token usage – " + " | ".join(token_parts) + # Get message_index for session header (for unified d-{index} links) + message_index = ctx.session_first_message.get(session_id) + session_nav.append( { "id": session_id, + "message_index": message_index, "summary": session_info["summary"], "timestamp_range": timestamp_range, "first_timestamp": first_ts, @@ -627,183 +725,6 @@ def prepare_session_navigation( return session_nav -# -- Message Processing Functions --------------------------------------------- -# Note: HTML formatting logic has been moved to html/content_formatters.py -# as part of the refactoring to support format-neutral content models. - - -# def _process_summary_message(message: SummaryTranscriptEntry) -> tuple[str, str, str]: -# """Process a summary message and return (css_class, content_html, message_type).""" -# css_class = "summary" -# content_html = f"Summary: {escape_html(str(message.summary))}" -# message_type = "summary" -# return css_class, content_html, message_type - - -def _process_command_message( - text_content: str, -) -> tuple[Optional["MessageContent"], str, str]: - """Process a slash command message and return (content, message_type, message_title). - - These are user messages containing slash command invocations (e.g., /context, /model). - The JSONL type is "user", not "system". - """ - # Parse to content model (formatting happens in HtmlRenderer) - content = parse_slash_command(text_content) - # If parsing fails, content will be None and caller will handle fallback - - return content, "user", "Slash Command" - - -def _process_local_command_output( - text_content: str, -) -> tuple[Optional["MessageContent"], str, str]: - """Process slash command output and return (content, message_type, message_title). - - These are user messages containing the output from slash commands (e.g., /context, /model). - The JSONL type is "user", not "system". - """ - # Parse to content model (formatting happens in HtmlRenderer) - content = parse_command_output(text_content) - # If parsing fails, content will be None and caller will handle fallback - - return content, "user", "" - - -def _process_bash_input( - text_content: str, -) -> tuple[Optional["MessageContent"], str, str]: - """Process bash input command and return (content, message_type, message_title).""" - # Parse to content model (formatting happens in HtmlRenderer) - content = parse_bash_input(text_content) - # If parsing fails, content will be None and caller will handle fallback - - return content, "bash-input", "Bash command" - - -def _process_bash_output( - text_content: str, -) -> tuple[Optional["MessageContent"], str, str]: - """Process bash output and return (content, message_type, message_title).""" - # Parse to content model (formatting happens in HtmlRenderer) - content = parse_bash_output(text_content) - # If parsing fails, content will be None - caller/renderer handles empty output - - return content, "bash-output", "" - - -def _process_regular_message( - items: list[ContentItem], - message_type: str, - is_sidechain: bool, - is_meta: bool = False, -) -> tuple[bool, Optional["MessageContent"], str, str]: - """Process regular message and return (is_sidechain, content_model, message_type, message_title). - - Returns content_model for user messages, None for non-user messages. - Non-user messages (assistant) are handled by the legacy render_message_content path. - - Note: Sidechain user messages (Sub-assistant prompts) are now skipped entirely - in the main processing loop since they duplicate the Task tool input prompt. - - Args: - items: List of text/image content items (no tool_use, tool_result, thinking). - is_meta: True for slash command expanded prompts (isMeta=True in JSONL) - """ - message_title = message_type.title() # Default title - content_model: Optional["MessageContent"] = None - - # Handle user-specific preprocessing - if message_type == MessageType.USER: - # Note: sidechain user messages are skipped before reaching this function - # Parse user content (is_meta triggers UserSlashCommandContent creation) - content_model = parse_user_message_content(items, is_slash_command=is_meta) - - # Determine message_title from content type - if isinstance(content_model, UserSlashCommandContent): - message_title = "User (slash command)" - elif isinstance(content_model, CompactedSummaryContent): - message_title = "User (compacted conversation)" - elif isinstance(content_model, UserMemoryContent): - message_title = "Memory" - - elif message_type == MessageType.ASSISTANT: - # Create AssistantTextContent directly from items - # (empty text already filtered by chunk_message_content) - if items: - content_model = AssistantTextContent( - items=items # type: ignore[arg-type] - ) - - if is_sidechain: - # Update message title for display (only non-user types reach here) - if not isinstance(content_model, CompactedSummaryContent): - message_title = "Sub-assistant" - - return is_sidechain, content_model, message_type, message_title - - -def _process_system_message( - message: SystemTranscriptEntry, -) -> Optional[TemplateMessage]: - """Process a system message and return a TemplateMessage, or None if it should be skipped. - - Handles: - - Hook summaries (subtype="stop_hook_summary") - - Other system messages with level-specific styling (info, warning, error) - - Note: Slash command messages (, ) are user messages, - not system messages. They are handled by _process_command_message and - _process_local_command_output in the main processing loop. - """ - from .models import MessageContent # Local import to avoid circular dependency - - session_id = getattr(message, "sessionId", "unknown") - timestamp = getattr(message, "timestamp", "") - formatted_timestamp = format_timestamp(timestamp) if timestamp else "" - - # Build structured content based on message subtype - content: MessageContent - if message.subtype == "stop_hook_summary": - # Skip silent hook successes (no output, no errors) - if not message.hasOutput and not message.hookErrors: - return None - # Create structured hook summary content - hook_infos = [ - HookInfo(command=info.get("command", "unknown")) - for info in (message.hookInfos or []) - ] - content = HookSummaryContent( - has_output=bool(message.hasOutput), - hook_errors=message.hookErrors or [], - hook_infos=hook_infos, - ) - level = "hook" - elif not message.content: - # Skip system messages without content (shouldn't happen normally) - return None - else: - # Create structured system content - level = getattr(message, "level", "info") - content = SystemContent(level=level, text=message.content) - - # Store parent UUID for hierarchy rebuild (handled by _build_message_hierarchy) - parent_uuid = getattr(message, "parentUuid", None) - - return TemplateMessage( - message_type="system", - formatted_timestamp=formatted_timestamp, - raw_timestamp=timestamp, - session_id=session_id, - message_title=f"System {level.title()}", - message_id=None, # Will be assigned by _build_message_hierarchy - ancestry=[], # Will be assigned by _build_message_hierarchy - uuid=message.uuid, - parent_uuid=parent_uuid, - content=content, # Level info is in SystemContent - ) - - # Type alias for chunk output: either a list of regular items or a single special item ContentChunk = list[ContentItem] | ContentItem @@ -870,162 +791,6 @@ def chunk_message_content(content: list[ContentItem]) -> list[ContentChunk]: return chunks -@dataclass -class ToolItemResult: - """Result of processing a single tool/thinking/image item.""" - - message_type: str - message_title: str - content: Optional["MessageContent"] = None # Structured content for rendering - tool_use_id: Optional[str] = None - title_hint: Optional[str] = None - pending_dedup: Optional[str] = None # For Task result deduplication - is_error: bool = False # For tool_result error state - - -def _process_tool_use_item( - tool_item: ContentItem, - tool_use_context: dict[str, ToolUseContent], -) -> Optional[ToolItemResult]: - """Process a tool_use content item. - - Args: - tool_item: The tool use content item - tool_use_context: Dict to populate with tool_use_id -> ToolUseContent mapping - - Returns: - ToolItemResult with tool_use content model, or None if item should be skipped - """ - # Convert Anthropic type to our format if necessary - if not isinstance(tool_item, ToolUseContent): - tool_use = ToolUseContent( - type="tool_use", - id=getattr(tool_item, "id", ""), - name=getattr(tool_item, "name", ""), - input=getattr(tool_item, "input", {}), - ) - else: - tool_use = tool_item - - # Title is computed here but content formatting happens in HtmlRenderer - tool_message_title = format_tool_use_title(tool_use) - escaped_id = escape_html(tool_use.id) - item_tool_use_id = tool_use.id - tool_title_hint = f"ID: {escaped_id}" - - # Populate tool_use_context for later use when processing tool results - tool_use_context[item_tool_use_id] = tool_use - - return ToolItemResult( - message_type="tool_use", - message_title=tool_message_title, - content=tool_use, # ToolUseContent is the model - tool_use_id=item_tool_use_id, - title_hint=tool_title_hint, - ) - - -def _process_tool_result_item( - tool_item: ContentItem, - tool_use_context: dict[str, ToolUseContent], -) -> Optional[ToolItemResult]: - """Process a tool_result content item. - - Args: - tool_item: The tool result content item - tool_use_context: Dict with tool_use_id -> ToolUseContent mapping - - Returns: - ToolItemResult with tool_result content model, or None if item should be skipped - """ - # Convert Anthropic type to our format if necessary - if not isinstance(tool_item, ToolResultContent): - tool_result = ToolResultContent( - type="tool_result", - tool_use_id=getattr(tool_item, "tool_use_id", ""), - content=getattr(tool_item, "content", ""), - is_error=getattr(tool_item, "is_error", False), - ) - else: - tool_result = tool_item - - # Get file_path and tool_name from tool_use context for specialized rendering - result_file_path: Optional[str] = None - result_tool_name: Optional[str] = None - if tool_result.tool_use_id in tool_use_context: - tool_use_from_ctx = tool_use_context[tool_result.tool_use_id] - result_tool_name = tool_use_from_ctx.name - if ( - result_tool_name in ("Read", "Edit", "Write") - and "file_path" in tool_use_from_ctx.input - ): - result_file_path = tool_use_from_ctx.input["file_path"] - - # Create content model with rendering context - content_model = ToolResultContentModel( - tool_use_id=tool_result.tool_use_id, - content=tool_result.content, - is_error=tool_result.is_error or False, - tool_name=result_tool_name, - file_path=result_file_path, - ) - - # Retroactive deduplication: if Task result, extract content for later matching - pending_dedup: Optional[str] = None - if result_tool_name == "Task": - # Extract text content from tool result - # Note: tool_result.content can be str or list[dict[str, Any]] - if isinstance(tool_result.content, str): - task_result_content = tool_result.content.strip() - else: - # Handle list of dicts (tool result format) - content_parts: list[str] = [] - for item in tool_result.content: - text_val = item.get("text", "") - if isinstance(text_val, str): - content_parts.append(text_val) - task_result_content = "\n".join(content_parts).strip() - pending_dedup = task_result_content if task_result_content else None - - escaped_id = escape_html(tool_result.tool_use_id) - tool_title_hint = f"ID: {escaped_id}" - tool_message_title = "Error" if tool_result.is_error else "" - - return ToolItemResult( - message_type="tool_result", - message_title=tool_message_title, - content=content_model, - tool_use_id=tool_result.tool_use_id, - title_hint=tool_title_hint, - pending_dedup=pending_dedup, - is_error=tool_result.is_error or False, - ) - - -def _process_thinking_item(tool_item: ContentItem) -> Optional[ToolItemResult]: - """Process a thinking content item. - - Returns: - ToolItemResult with thinking content model - """ - # Extract thinking text from the content item - if isinstance(tool_item, ThinkingContent): - thinking_text = tool_item.thinking.strip() - signature = getattr(tool_item, "signature", None) - else: - thinking_text = getattr(tool_item, "thinking", str(tool_item)).strip() - signature = None - - # Create the content model (formatting happens in HtmlRenderer) - thinking_model = ThinkingContentModel(thinking=thinking_text, signature=signature) - - return ToolItemResult( - message_type="thinking", - message_title="Thinking", - content=thinking_model, - ) - - # -- Message Pairing ---------------------------------------------------------- @@ -1034,46 +799,48 @@ class PairingIndices: """Indices for efficient message pairing lookups. All indices are built in a single pass for efficiency. + Stores message references directly (not list positions). """ - # (session_id, tool_use_id) -> message index for tool_use messages - tool_use: dict[tuple[str, str], int] - # (session_id, tool_use_id) -> message index for tool_result messages - tool_result: dict[tuple[str, str], int] - # uuid -> message index for system messages (parent-child pairing) - uuid: dict[str, int] - # parent_uuid -> message index for slash-command messages - slash_command_by_parent: dict[str, int] + # (session_id, tool_use_id) -> TemplateMessage for tool_use messages + tool_use: dict[tuple[str, str], TemplateMessage] + # (session_id, tool_use_id) -> TemplateMessage for tool_result messages + tool_result: dict[tuple[str, str], TemplateMessage] + # uuid -> TemplateMessage for system messages (parent-child pairing) + uuid: dict[str, TemplateMessage] + # parent_uuid -> TemplateMessage for slash-command messages + slash_command_by_parent: dict[str, TemplateMessage] def _build_pairing_indices(messages: list[TemplateMessage]) -> PairingIndices: """Build indices for efficient message pairing lookups. Single pass through messages to build all indices needed for pairing. + Stores message references directly for robust lookup after reordering. """ - tool_use_index: dict[tuple[str, str], int] = {} - tool_result_index: dict[tuple[str, str], int] = {} - uuid_index: dict[str, int] = {} - slash_command_by_parent: dict[str, int] = {} + tool_use_index: dict[tuple[str, str], TemplateMessage] = {} + tool_result_index: dict[tuple[str, str], TemplateMessage] = {} + uuid_index: dict[str, TemplateMessage] = {} + slash_command_by_parent: dict[str, TemplateMessage] = {} - for i, msg in enumerate(messages): + for msg in messages: # Index tool_use and tool_result by (session_id, tool_use_id) if msg.tool_use_id and msg.session_id: key = (msg.session_id, msg.tool_use_id) if msg.type == "tool_use": - tool_use_index[key] = i + tool_use_index[key] = msg elif msg.type == "tool_result": - tool_result_index[key] = i + tool_result_index[key] = msg # Index system messages by UUID for parent-child pairing - if msg.uuid and msg.type == "system": - uuid_index[msg.uuid] = i + if msg.meta.uuid and msg.type == "system": + uuid_index[msg.meta.uuid] = msg # Index slash-command user messages by parent_uuid if msg.parent_uuid and isinstance( - msg.content, (SlashCommandContent, UserSlashCommandContent) + msg.content, (SlashCommandMessage, UserSlashCommandMessage) ): - slash_command_by_parent[msg.parent_uuid] = i + slash_command_by_parent[msg.parent_uuid] = msg return PairingIndices( tool_use=tool_use_index, @@ -1084,11 +851,12 @@ def _build_pairing_indices(messages: list[TemplateMessage]) -> PairingIndices: def _mark_pair(first: TemplateMessage, last: TemplateMessage) -> None: - """Mark two messages as a pair.""" - first.is_paired = True - first.pair_role = "pair_first" - last.is_paired = True - last.pair_role = "pair_last" + """Mark two messages as a pair by setting their pair indices.""" + first_index = first.message_index + last_index = last.message_index + if first_index is not None and last_index is not None: + first.pair_last = last_index + last.pair_first = first_index def _try_pair_adjacent( @@ -1106,8 +874,8 @@ def _try_pair_adjacent( """ # Slash command + command output (both are user messages) if isinstance( - current.content, (SlashCommandContent, UserSlashCommandContent) - ) and isinstance(next_msg.content, CommandOutputContent): + current.content, (SlashCommandMessage, UserSlashCommandMessage) + ) and isinstance(next_msg.content, CommandOutputMessage): _mark_pair(current, next_msg) return True @@ -1126,7 +894,6 @@ def _try_pair_adjacent( def _try_pair_by_index( current: TemplateMessage, - messages: list[TemplateMessage], indices: PairingIndices, ) -> None: """Try to pair current message with another using index lookups. @@ -1140,20 +907,19 @@ def _try_pair_by_index( if current.type == "tool_use" and current.tool_use_id and current.session_id: key = (current.session_id, current.tool_use_id) if key in indices.tool_result: - result_msg = messages[indices.tool_result[key]] - _mark_pair(current, result_msg) + _mark_pair(current, indices.tool_result[key]) # System child message finding its parent (by parent_uuid) if current.type == "system" and current.parent_uuid: if current.parent_uuid in indices.uuid: - parent_msg = messages[indices.uuid[current.parent_uuid]] - _mark_pair(parent_msg, current) + _mark_pair(indices.uuid[current.parent_uuid], current) # System command finding its slash-command child (by uuid -> parent_uuid) - if current.type == "system" and current.uuid: - if current.uuid in indices.slash_command_by_parent: - slash_msg = messages[indices.slash_command_by_parent[current.uuid]] - _mark_pair(current, slash_msg) + if ( + current.type == "system" + and current.meta.uuid in indices.slash_command_by_parent + ): + _mark_pair(current, indices.slash_command_by_parent[current.meta.uuid]) def _identify_message_pairs(messages: list[TemplateMessage]) -> None: @@ -1190,7 +956,7 @@ def _identify_message_pairs(messages: list[TemplateMessage]) -> None: continue # Try index-based pairing (doesn't skip, continues to next message) - _try_pair_by_index(current, messages, indices) + _try_pair_by_index(current, indices) i += 1 @@ -1211,77 +977,70 @@ def _reorder_paired_messages(messages: list[TemplateMessage]) -> list[TemplateMe # Build index of pair_last messages by (session_id, tool_use_id) # Session ID is included to prevent cross-session pairing when sessions are resumed - pair_last_index: dict[ - tuple[str, str], int - ] = {} # (session_id, tool_use_id) -> message index + # Stores message references directly (not list positions) + pair_last_index: dict[tuple[str, str], TemplateMessage] = {} # Index slash-command pair_last messages by parent_uuid - slash_command_pair_index: dict[str, int] = {} # parent_uuid -> message index + slash_command_pair_index: dict[str, TemplateMessage] = {} - for i, msg in enumerate(messages): - if ( - msg.is_paired - and msg.pair_role == "pair_last" - and msg.tool_use_id - and msg.session_id - ): + for msg in messages: + if msg.is_last_in_pair and msg.tool_use_id and msg.session_id: key = (msg.session_id, msg.tool_use_id) - pair_last_index[key] = i + pair_last_index[key] = msg # Index slash-command messages by parent_uuid if ( - msg.is_paired - and msg.pair_role == "pair_last" + msg.is_last_in_pair and msg.parent_uuid - and isinstance(msg.content, (SlashCommandContent, UserSlashCommandContent)) + and isinstance(msg.content, (SlashCommandMessage, UserSlashCommandMessage)) ): - slash_command_pair_index[msg.parent_uuid] = i + slash_command_pair_index[msg.parent_uuid] = msg # Create reordered list reordered: list[TemplateMessage] = [] - skip_indices: set[int] = set() + already_added: set[int] = set() # Track by message_index (unique per message) - for i, msg in enumerate(messages): - if i in skip_indices: + for msg in messages: + msg_index = msg.message_index + if msg_index in already_added: continue reordered.append(msg) + if msg_index is not None: + already_added.add(msg_index) # If this is the first message in a pair, immediately add its pair_last # Key includes session_id to prevent cross-session pairing on resume - if msg.is_paired and msg.pair_role == "pair_first": - pair_last = None - last_idx = None + if msg.is_first_in_pair: + pair_last: Optional[TemplateMessage] = None # Check for tool_use_id based pairs if msg.tool_use_id and msg.session_id: key = (msg.session_id, msg.tool_use_id) if key in pair_last_index: - last_idx = pair_last_index[key] - pair_last = messages[last_idx] + pair_last = pair_last_index[key] # Check for system + slash-command pairs (via uuid -> parent_uuid) - if pair_last is None and msg.uuid and msg.uuid in slash_command_pair_index: - last_idx = slash_command_pair_index[msg.uuid] - pair_last = messages[last_idx] + if pair_last is None and msg.meta.uuid in slash_command_pair_index: + pair_last = slash_command_pair_index[msg.meta.uuid] # Only append if we haven't already added this pair_last # (handles case where multiple pair_firsts match the same pair_last) - if ( - pair_last is not None - and last_idx is not None - and last_idx not in skip_indices - ): - reordered.append(pair_last) - skip_indices.add(last_idx) + if pair_last is not None: + last_msg_index = pair_last.message_index + if last_msg_index is not None and last_msg_index not in already_added: + reordered.append(pair_last) + already_added.add(last_msg_index) # Calculate duration between pair messages try: - if msg.raw_timestamp and pair_last.raw_timestamp: + first_ts = msg.meta.timestamp if msg.meta else None + last_ts = pair_last.meta.timestamp if pair_last.meta else None + if first_ts and last_ts: # Parse ISO timestamps first_time = datetime.fromisoformat( - msg.raw_timestamp.replace("Z", "+00:00") + first_ts.replace("Z", "+00:00") ) last_time = datetime.fromisoformat( - pair_last.raw_timestamp.replace("Z", "+00:00") + last_ts.replace("Z", "+00:00") ) duration = last_time - first_time @@ -1315,11 +1074,12 @@ def _get_message_hierarchy_level(msg: TemplateMessage) -> int: - Level 1: User messages - Level 2: System commands/errors, Assistant, Thinking - Level 3: Tool use/result, System info/warning (nested under assistant) - - Level 4: Sidechain assistant/thinking (nested under Task tool result) + - Level 4: Sidechain user/assistant/thinking (nested under Task tool result) - Level 5: Sidechain tools (nested under sidechain assistant) - Note: Sidechain user messages (Sub-assistant prompts) are now skipped entirely - since they duplicate the Task tool input prompt. + Note: Sidechain user messages (duplicate of Task input prompt) and the last + sidechain assistant (duplicate of Task output) are cleaned up from the tree + by _cleanup_sidechain_duplicates after tree building. Returns: Integer hierarchy level (1-5, session headers are 0) @@ -1327,14 +1087,13 @@ def _get_message_hierarchy_level(msg: TemplateMessage) -> int: msg_type = msg.type is_sidechain = msg.is_sidechain - # User messages at level 1 (under session) - # Note: sidechain user messages are skipped before reaching this function - if msg_type == "user" and not is_sidechain: - return 1 + # User messages at level 1 (under session), level 4 for sidechain + if msg_type == "user": + return 4 if is_sidechain else 1 # System info/warning at level 3 (tool-related, e.g., hook notifications) - # Get level from SystemContent if available - system_level = msg.content.level if isinstance(msg.content, SystemContent) else None + # Get level from SystemMessage if available + system_level = msg.content.level if isinstance(msg.content, SystemMessage) else None if ( msg_type == "system" and system_level in ("info", "warning") @@ -1367,7 +1126,7 @@ def _get_message_hierarchy_level(msg: TemplateMessage) -> int: def _build_message_hierarchy(messages: list[TemplateMessage]) -> None: - """Build message_id and ancestry for all messages based on their current order. + """Build ancestry for all messages based on their current order. This should be called after all reordering operations (pair reordering, sidechain reordering) to ensure the hierarchy reflects the final display order. @@ -1375,11 +1134,13 @@ def _build_message_hierarchy(messages: list[TemplateMessage]) -> None: The hierarchy is determined by message type using _get_message_hierarchy_level(), and a stack-based approach builds proper parent-child relationships. + Ancestry stores message_index integers. Templates prefix with "d-" for CSS classes. + Args: messages: List of template messages in their final order (modified in place) """ - hierarchy_stack: list[tuple[int, str]] = [] - message_id_counter = 0 + # Stack of (level, message_index) tuples + hierarchy_stack: list[tuple[int, int]] = [] for message in messages: # Session headers are level 0 @@ -1393,30 +1154,21 @@ def _build_message_hierarchy(messages: list[TemplateMessage]) -> None: while hierarchy_stack and hierarchy_stack[-1][0] >= current_level: hierarchy_stack.pop() - # Build ancestry from remaining stack - ancestry = [msg_id for _, msg_id in hierarchy_stack] - - # Generate new message ID - # Session headers use session-{session_id} format for navigation links - if message.is_session_header and message.session_id: - message_id = f"session-{message.session_id}" - else: - message_id = f"d-{message_id_counter}" - message_id_counter += 1 + # Build ancestry from remaining stack (list of message_index integers) + ancestry = [msg_index for _, msg_index in hierarchy_stack] # Push current message onto stack - hierarchy_stack.append((current_level, message_id)) + if message.message_index is not None: + hierarchy_stack.append((current_level, message.message_index)) - # Update the message - message.message_id = message_id + # Update the message ancestry message.ancestry = ancestry def _mark_messages_with_children(messages: list[TemplateMessage]) -> None: - """Mark messages that have children and calculate descendant counts. + """Calculate child and descendant counts for messages. Efficiently calculates: - - has_children: Whether message has any children - immediate_children_count: Count of direct children only - total_descendants_count: Count of all descendants recursively @@ -1425,11 +1177,11 @@ def _mark_messages_with_children(messages: list[TemplateMessage]) -> None: Args: messages: List of template messages to process """ - # Build index of messages by ID for O(1) lookup - message_by_id: dict[str, TemplateMessage] = {} + # Build index of messages by message_index for O(1) lookup + message_by_index: dict[int, TemplateMessage] = {} for message in messages: - if message.message_id: - message_by_id[message.message_id] = message + if message.message_index is not None: + message_by_index[message.message_index] = message # Process each message and update counts for ancestors for message in messages: @@ -1438,29 +1190,28 @@ def _mark_messages_with_children(messages: list[TemplateMessage]) -> None: # Skip counting pair_last messages (second in a pair) # Pairs are visually presented as a single unit, so we only count the first - if message.is_paired and message.pair_role == "pair_last": + if message.is_last_in_pair: continue # Get immediate parent (last in ancestry list) - immediate_parent_id = message.ancestry[-1] + immediate_parent_index = message.ancestry[-1] # Get message type for categorization msg_type = message.type # Increment immediate parent's child count - if immediate_parent_id in message_by_id: - parent = message_by_id[immediate_parent_id] + if immediate_parent_index in message_by_index: + parent = message_by_index[immediate_parent_index] parent.immediate_children_count += 1 - parent.has_children = True # Track by type parent.immediate_children_by_type[msg_type] = ( parent.immediate_children_by_type.get(msg_type, 0) + 1 ) # Increment descendant count for ALL ancestors - for ancestor_id in message.ancestry: - if ancestor_id in message_by_id: - ancestor = message_by_id[ancestor_id] + for ancestor_index in message.ancestry: + if ancestor_index in message_by_index: + ancestor = message_by_index[ancestor_index] ancestor.total_descendants_count += 1 # Track by type ancestor.total_descendants_by_type[msg_type] = ( @@ -1471,7 +1222,7 @@ def _mark_messages_with_children(messages: list[TemplateMessage]) -> None: def _build_message_tree(messages: list[TemplateMessage]) -> list[TemplateMessage]: """Build tree structure by populating children fields based on ancestry. - This function takes a flat list of messages (with message_id and ancestry + This function takes a flat list of messages (with message_index and ancestry already set by _build_message_hierarchy) and populates the children field of each message to form an explicit tree structure. @@ -1481,17 +1232,17 @@ def _build_message_tree(messages: list[TemplateMessage]) -> list[TemplateMessage - More natural parent-child traversal Args: - messages: List of template messages with message_id and ancestry set + messages: List of template messages with message_index and ancestry set Returns: List of root messages (those with empty ancestry). Each message's children field is populated with its direct children. """ - # Build index of messages by ID for O(1) lookup - message_by_id: dict[str, TemplateMessage] = {} + # Build index of messages by message_index for O(1) lookup + message_by_index: dict[int, TemplateMessage] = {} for message in messages: - if message.message_id: - message_by_id[message.message_id] = message + if message.message_index is not None: + message_by_index[message.message_index] = message # Clear any existing children (in case of re-processing) for message in messages: @@ -1507,14 +1258,152 @@ def _build_message_tree(messages: list[TemplateMessage]) -> list[TemplateMessage root_messages.append(message) else: # Has a parent - add to parent's children - immediate_parent_id = message.ancestry[-1] - if immediate_parent_id in message_by_id: - parent = message_by_id[immediate_parent_id] + immediate_parent_index = message.ancestry[-1] + if immediate_parent_index in message_by_index: + parent = message_by_index[immediate_parent_index] parent.children.append(message) return root_messages +# Pattern to match agentId lines added to Task results for resume functionality +# e.g., "agentId: a7c9965 (for resuming to continue this agent's work if needed)" +_AGENT_ID_LINE_PATTERN = re.compile(r"\n*agentId:\s*\w+\s*\([^)]*\)\s*$", re.IGNORECASE) + + +def _normalize_for_dedup(text: str) -> str: + """Normalize text for deduplication matching. + + Strips trailing agentId lines that may be added to Task results + but not present in the sidechain assistant's final message. + """ + return _AGENT_ID_LINE_PATTERN.sub("", text).strip() + + +def _extract_task_result_text(tool_result_message: ToolResultMessage) -> Optional[str]: + """Extract text content from a Task tool result for deduplication matching. + + Args: + tool_result_message: The ToolResultMessage containing Task output + + Returns: + The extracted text content (normalized), or None if extraction fails + """ + output = tool_result_message.output + + # Handle parsed TaskOutput (preferred - has structured result field) + if isinstance(output, TaskOutput): + text = output.result.strip() if output.result else None + return _normalize_for_dedup(text) if text else None + + # Handle raw ToolResultContent (fallback for unparsed results) + if not isinstance(output, ToolResultContent): + return None + + content = output.content + if isinstance(content, str): + text = content.strip() if content else None + return _normalize_for_dedup(text) if text else None + + # Handle list of dicts (tool result format) + content_parts: list[str] = [] + for item in content: + text_val = item.get("text", "") + if isinstance(text_val, str): + content_parts.append(text_val) + result = "\n".join(content_parts).strip() + return _normalize_for_dedup(result) if result else None + + +def _cleanup_sidechain_duplicates(root_messages: list[TemplateMessage]) -> None: + """Clean up duplicate content in sidechains after tree is built. + + For each Task tool_use or tool_result with sidechain children: + - Remove the first UserTextMessage (duplicate of Task input prompt) + - For tool_result: Replace last AssistantTextMessage matching result with DedupNotice + + Sidechain messages can be children of either tool_use or tool_result depending + on timestamp order - tool_use during execution, tool_result after completion. + + Args: + root_messages: List of root messages with children populated + """ + + def process_message(message: TemplateMessage) -> None: + """Recursively process a message and its children.""" + # Recursively process children first (depth-first) + for child in message.children: + process_message(child) + + # Check if this is a Task tool_use or tool_result with sidechain children + is_task_tool_use = ( + message.type == "tool_use" + and isinstance(message.content, ToolUseMessage) + and message.content.tool_name == "Task" + ) + is_task_tool_result = ( + message.type == "tool_result" + and isinstance(message.content, ToolResultMessage) + and message.content.tool_name == "Task" + ) + + if not ((is_task_tool_use or is_task_tool_result) and message.children): + return + + children = message.children + + # Remove first sidechain UserTextMessage (duplicate of Task input prompt) + # Must be specifically UserTextMessage, not ToolResultMessage or other user types + # When removing, adopt its children to preserve sidechain tool messages + if ( + children + and children[0].is_sidechain + and isinstance(children[0].content, UserTextMessage) + ): + removed = children.pop(0) + # Adopt orphaned children (tool_use/tool_result from sidechain) + if removed.children: + # Insert at beginning to maintain order + children[:0] = removed.children + + # For tool_result only: replace last matching AssistantTextMessage with dedup + if not is_task_tool_result: + return + + task_result_text = _extract_task_result_text( + cast(ToolResultMessage, message.content) + ) + if not task_result_text: + return + + for i in range(len(children) - 1, -1, -1): + child = children[i] + child_content = child.content + # Get raw_text_content from content (UserTextMessage/AssistantTextMessage) + child_raw = getattr(child_content, "raw_text_content", None) + child_text = _normalize_for_dedup(child_raw) if child_raw else None + if ( + child.type == "assistant" + and child.is_sidechain + and isinstance(child_content, AssistantTextMessage) + and child_text + and child_text == task_result_text + ): + # Replace with dedup notice pointing to the Task result + # Preserve original meta (sidechain/session flags) and original message + child.content = DedupNoticeMessage( + child_content.meta, + notice_text="Task summary β€” see result above", + target_message_id=message.message_id, + original_text=child_text, + original=child_content, + ) + break + + for root in root_messages: + process_message(root) + + # -- Message Reordering ------------------------------------------------------- @@ -1589,9 +1478,9 @@ def _reorder_sidechain_template_messages( order based on when each agent finishes. This function reorders messages so that each sidechain's messages appear right after the Task result that references them. - This function also handles deduplication: the last sidechain assistant message - typically contains the same content as the Task result, so we replace it with - a forward link to avoid showing the same content twice. + Note: Deduplication of sidechain content (first user message = Task input, + last assistant message = Task output) is handled later by _cleanup_sidechain_duplicates + after the tree structure is built. This must be called AFTER _reorder_paired_messages, since that function moves tool_results next to their tool_uses, which changes where the agentId-bearing @@ -1624,7 +1513,6 @@ def _reorder_sidechain_template_messages( return messages # Second pass: insert sidechains after their Task result messages - # Also perform deduplication of sidechain assistants vs Task results result: list[TemplateMessage] = [] used_agents: set[str] = set() @@ -1644,39 +1532,9 @@ def _reorder_sidechain_template_messages( and agent_id in sidechain_map and agent_id not in used_agents ): - sidechain_msgs = sidechain_map[agent_id] - - # Deduplicate: find the last sidechain assistant with text content - # that matches the Task result content - task_result_content = ( - message.raw_text_content.strip() if message.raw_text_content else None - ) - if task_result_content and message.type == MessageType.TOOL_RESULT: - # Find the last assistant message in this sidechain - for sidechain_msg in reversed(sidechain_msgs): - sidechain_text = ( - sidechain_msg.raw_text_content.strip() - if sidechain_msg.raw_text_content - else None - ) - if ( - sidechain_msg.type == MessageType.ASSISTANT - and sidechain_text - and sidechain_text == task_result_content - ): - # Replace with note pointing to the Task result - sidechain_msg.content = DedupNoticeContent( - notice_text="Task summary β€” see result above", - target_uuid=message.uuid, - original_text=sidechain_text, - ) - # Mark as deduplicated for potential debugging - sidechain_msg.raw_text_content = None - break - # Insert the sidechain messages for this agent right after this message # Note: ancestry will be rebuilt by _build_message_hierarchy() later - result.extend(sidechain_msgs) + result.extend(sidechain_map[agent_id]) used_agents.add(agent_id) # Append any sidechains that weren't matched (shouldn't happen normally) @@ -1687,23 +1545,6 @@ def _reorder_sidechain_template_messages( return result -def _resolve_dedup_targets(messages: list[TemplateMessage]) -> None: - """Resolve dedup notice target UUIDs to message IDs for anchor links. - - Must be called after _build_message_hierarchy assigns message_id values. - """ - # Build uuid -> message_id mapping - uuid_to_id: dict[str, str] = {} - for msg in messages: - if msg.uuid and msg.message_id: - uuid_to_id[msg.uuid] = msg.message_id - - # Resolve dedup notice targets - for msg in messages: - if isinstance(msg.content, DedupNoticeContent) and msg.content.target_uuid: - msg.content.target_message_id = uuid_to_id.get(msg.content.target_uuid) - - def _filter_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntry]: """Filter messages to those that should be rendered. @@ -1712,10 +1553,12 @@ def _filter_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntry]: - Queue operations except 'remove' (steering messages) - Messages with no meaningful content (no text and no tool items) - Messages matching should_skip_message() (warmup, etc.) - - Sidechain user messages without tool results (prompts duplicate Task result) System messages are included as they need special processing in _render_messages. + Note: Sidechain user prompts (duplicates of Task input) are removed later + by _cleanup_sidechain_duplicates after tree building. + Args: messages: List of transcript entries to filter @@ -1725,8 +1568,6 @@ def _filter_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntry]: filtered: list[TranscriptEntry] = [] for message in messages: - message_type = message.type - # Skip summary messages if isinstance(message, SummaryTranscriptEntry): continue @@ -1767,16 +1608,6 @@ def _filter_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntry]: if should_skip_message(text_content): continue - # Skip sidechain user messages that are just prompts (no tool results) - if message_type == MessageType.USER and getattr(message, "isSidechain", False): - has_tool_results = any( - getattr(item, "type", None) == "tool_result" - or isinstance(item, ToolResultContent) - for item in message_content - ) - if not has_tool_results: - continue - # Message passes all filters filtered.append(message) @@ -1785,6 +1616,7 @@ def _filter_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntry]: def _collect_session_info( messages: list[TranscriptEntry], + session_summaries: dict[str, str], ) -> tuple[ dict[str, dict[str, Any]], # sessions list[str], # session_order @@ -1802,6 +1634,7 @@ def _collect_session_info( Args: messages: Pre-filtered list of transcript entries + session_summaries: Dict mapping session_id to summary text Returns: Tuple containing: @@ -1835,7 +1668,7 @@ def _collect_session_info( # Initialize session if new if session_id not in sessions: - current_session_summary = getattr(message, "_session_summary", None) + current_session_summary = session_summaries.get(session_id) # Get first user message content for preview first_user_message = "" @@ -1906,7 +1739,7 @@ def _render_messages( messages: list[TranscriptEntry], sessions: dict[str, dict[str, Any]], show_tokens_for_message: set[str], -) -> list[TemplateMessage]: +) -> RenderingContext: """Pass 2: Render pre-filtered messages to TemplateMessage objects. This pass creates the actual TemplateMessage objects for rendering: @@ -1924,58 +1757,44 @@ def _render_messages( show_tokens_for_message: Set of message UUIDs that should display tokens Returns: - List of TemplateMessage objects ready for template rendering + RenderingContext with all TemplateMessage objects registered """ + # Create rendering context for this operation + ctx = RenderingContext() + # Track which sessions have had headers added seen_sessions: set[str] = set() - # Build mapping of tool_use_id to ToolUseContent for specialized tool result rendering - tool_use_context: dict[str, ToolUseContent] = {} - - # Process messages into template-friendly format - template_messages: list[TemplateMessage] = [] - - # Per-message timing tracking - message_timings: list[ - tuple[float, str, int, str] - ] = [] # (duration, message_type, index, uuid) - - # Track expensive operations - markdown_timings: list[tuple[float, str]] = [] # (duration, context_uuid) - pygments_timings: list[tuple[float, str]] = [] # (duration, context_uuid) - - # Initialize timing tracking - set_timing_var("_markdown_timings", markdown_timings) - set_timing_var("_pygments_timings", pygments_timings) - set_timing_var("_current_msg_uuid", "") - - for msg_idx, message in enumerate(messages): - msg_start_time = time.time() if DEBUG_TIMING else 0.0 + for message in messages: message_type = message.type - msg_uuid = getattr(message, "uuid", f"no-uuid-{msg_idx}") - - # Update current message UUID for timing tracking - set_timing_var("_current_msg_uuid", msg_uuid) - # Handle system messages separately (already filtered in pass 1) + # Handle system messages (already filtered in pass 1) if isinstance(message, SystemTranscriptEntry): - system_template_message = _process_system_message(message) - if system_template_message: - template_messages.append(system_template_message) + system_content = create_system_message(message) + if system_content: + system_msg = TemplateMessage(system_content) + ctx.register(system_msg) + continue + + # Skip summary messages (should be filtered in pass 1, but be defensive) + if isinstance(message, SummaryTranscriptEntry): continue # Handle queue-operation 'remove' messages as user messages if isinstance(message, QueueOperationTranscriptEntry): message_content = message.content if message.content else [] message_type = MessageType.QUEUE_OPERATION + # QueueOperationTranscriptEntry has limited fields (no uuid, agentId, etc.) + meta = MessageMeta( + session_id=message.sessionId, + timestamp=message.timestamp, + uuid="", + ) + effective_type = "user" else: message_content = message.message.content # type: ignore - - # Track sidechain status for user messages - # (sidechain user text is skipped to avoid duplicate Task prompts) - is_sidechain_user = message_type == MessageType.USER and getattr( - message, "isSidechain", False - ) + meta = create_meta(message) + effective_type = message_type # Chunk content: regular items (text/image) accumulate, special items (tool/thinking) separate if isinstance(message_content, list): @@ -1995,267 +1814,136 @@ def _render_messages( continue # Get session info - session_id = getattr(message, "sessionId", "unknown") - session_summary = getattr(message, "_session_summary", None) + session_id = meta.session_id or "unknown" + session_summary = sessions.get(session_id, {}).get("summary") # Add session header if this is a new session if session_id not in seen_sessions: seen_sessions.add(session_id) - current_session_summary = sessions.get(session_id, {}).get("summary") + current_session_summary = session_summary session_title = ( f"{current_session_summary} β€’ {session_id[:8]}" if current_session_summary else session_id[:8] ) - session_header = TemplateMessage( - message_type="session_header", - formatted_timestamp="", - raw_timestamp=None, - session_summary=current_session_summary, + # Create meta with session_id for the session header + session_header_meta = MessageMeta( session_id=session_id, - is_session_header=True, - message_id=None, - ancestry=[], - content=SessionHeaderContent( - title=session_title, - session_id=session_id, - summary=current_session_summary, - ), + timestamp="", + uuid="", ) - template_messages.append(session_header) - - # Get timestamp (only for non-summary messages) - timestamp = getattr(message, "timestamp", "") - formatted_timestamp = format_timestamp(timestamp) if timestamp else "" + session_header_content = SessionHeaderMessage( + session_header_meta, + title=session_title, + session_id=session_id, + summary=current_session_summary, + ) + # Register and track session's first message + session_header = TemplateMessage(session_header_content) + msg_index = ctx.register(session_header) + ctx.session_first_message[session_id] = msg_index # Extract token usage for assistant messages # Only show token usage for the first message with each requestId to avoid duplicates - token_usage_str: Optional[str] = None + usage_to_show: Optional[UsageInfo] = None if assistant_entry := as_assistant_entry(message): assistant_message = assistant_entry.message message_uuid = assistant_entry.uuid - if assistant_message.usage and message_uuid in show_tokens_for_message: - # Only show token usage for messages marked as first occurrence of requestId - usage = assistant_message.usage - token_parts = [ - f"Input: {usage.input_tokens}", - f"Output: {usage.output_tokens}", - ] - if usage.cache_creation_input_tokens: - token_parts.append( - f"Cache Creation: {usage.cache_creation_input_tokens}" - ) - if usage.cache_read_input_tokens: - token_parts.append(f"Cache Read: {usage.cache_read_input_tokens}") - token_usage_str = " | ".join(token_parts) + usage_to_show = assistant_message.usage - # Track whether we've shown token usage (only show on first content chunk) - token_shown = False + # Track whether we've used the usage (only use on first content chunk) + usage_used = False # Process each chunk - regular chunks (list) become text/image messages, # special chunks (single item) become tool/thinking messages - for chunk_idx, chunk in enumerate(chunks): + for chunk in chunks: + # Each chunk needs its own meta copy to preserve original values + chunk_meta = replace(meta) + # Regular chunk: list of text/image items if isinstance(chunk, list): - # Skip text chunks for sidechain user messages - # (prompts duplicate Task result; filtering already done in pass 1) - if is_sidechain_user: - continue - # Extract text for pattern detection chunk_text = extract_text_content(chunk) - # Check for special message patterns - is_command = is_command_message(chunk_text) - is_local_output = is_local_command_output(chunk_text) - is_bash_cmd = is_bash_input(chunk_text) - is_bash_result = is_bash_output(chunk_text) - - # Determine is_sidechain and content based on message type + # Dispatch to user or assistant parser based on effective_type content_model: Optional[MessageContent] = None - chunk_message_type = message_type - chunk_is_sidechain = getattr(message, "isSidechain", False) - - if is_command: - content_model, chunk_message_type, message_title = ( - _process_command_message(chunk_text) - ) - elif is_local_output: - content_model, chunk_message_type, message_title = ( - _process_local_command_output(chunk_text) - ) - elif is_bash_cmd: - content_model, chunk_message_type, message_title = ( - _process_bash_input(chunk_text) + # (user message parsing handles all type detection internally) + if effective_type == "user": + content_model = create_user_message( + chunk_meta, + chunk, # Pass the chunk items + chunk_text, # Pre-extracted text for pattern detection + is_slash_command=chunk_meta.is_meta, ) - elif is_bash_result: - content_model, chunk_message_type, message_title = ( - _process_bash_output(chunk_text) + elif effective_type == "assistant": + # Pass usage only on first chunk + chunk_usage = usage_to_show if not usage_used else None + usage_used = True + content_model = create_assistant_message( + chunk_meta, chunk, chunk_usage ) - else: - # For queue-operation messages, treat them as user messages - if isinstance(message, QueueOperationTranscriptEntry): - effective_type = "user" - else: - effective_type = message_type - - ( - chunk_is_sidechain, - content_model, - chunk_message_type, - message_title, - ) = _process_regular_message( - chunk, # Pass the chunk items - effective_type, - chunk_is_sidechain, - getattr(message, "isMeta", False), + + # Convert to UserSteeringMessage for queue-operation 'remove' messages + if ( + isinstance(message, QueueOperationTranscriptEntry) + and message.operation == "remove" + and isinstance(content_model, UserTextMessage) + ): + content_model = UserSteeringMessage( + items=content_model.items, meta=chunk_meta ) - # Convert to UserSteeringContent for queue-operation 'remove' messages - if ( - isinstance(message, QueueOperationTranscriptEntry) - and message.operation == "remove" - and isinstance(content_model, UserTextContent) - ): - content_model = UserSteeringContent(items=content_model.items) - message_title = "User (steering)" - - # Skip empty chunks - if not chunk: + # Skip empty chunks or when no content model was created + if not chunk or content_model is None: continue - # Only show token usage on first chunk - chunk_token_usage = token_usage_str if not token_shown else None - token_shown = True - - # Generate UUID for this chunk (append index if multiple chunks) - chunk_uuid = getattr(message, "uuid", None) - if chunk_uuid and len(chunks) > 1: - chunk_uuid = f"{chunk_uuid}-chunk-{chunk_idx}" - - # Markdown rendering for assistant, thinking, and compacted content - has_markdown = isinstance( - content_model, - ( - AssistantTextContent, - ThinkingContentModel, - CompactedSummaryContent, - ), - ) - - template_message = TemplateMessage( - message_type=chunk_message_type, - formatted_timestamp=formatted_timestamp, - raw_timestamp=timestamp, - session_summary=session_summary, - session_id=session_id, - token_usage=chunk_token_usage, - message_title=message_title, - message_id=None, # Will be assigned by _build_message_hierarchy - ancestry=[], # Will be assigned by _build_message_hierarchy - agent_id=getattr(message, "agentId", None), - uuid=chunk_uuid, - parent_uuid=getattr(message, "parentUuid", None), - is_sidechain=chunk_is_sidechain, - content=content_model, - has_markdown=has_markdown, - ) - - # Store raw text content for potential future use - template_message.raw_text_content = chunk_text - - template_messages.append(template_message) + chunk_msg = TemplateMessage(content_model) + ctx.register(chunk_msg) else: # Special chunk: single tool_use/tool_result/thinking item tool_item = chunk - tool_timestamp = getattr(message, "timestamp", "") - tool_formatted_timestamp = ( - format_timestamp(tool_timestamp) if tool_timestamp else "" - ) - - # Handle both custom types and Anthropic types - item_type = getattr(tool_item, "type", None) # Dispatch to appropriate handler based on item type - tool_result: Optional[ToolItemResult] = None - if isinstance(tool_item, ToolUseContent) or item_type == "tool_use": - tool_result = _process_tool_use_item(tool_item, tool_use_context) - elif ( - isinstance(tool_item, ToolResultContent) - or item_type == "tool_result" - ): - tool_result = _process_tool_result_item(tool_item, tool_use_context) - elif isinstance(tool_item, ThinkingContent) or item_type == "thinking": - tool_result = _process_thinking_item(tool_item) + tool_result: ToolItemResult + if isinstance(tool_item, ToolUseContent): + tool_result = create_tool_use_message( + chunk_meta, tool_item, ctx.tool_use_context + ) + elif isinstance(tool_item, ToolResultContent): + tool_result = create_tool_result_message( + chunk_meta, tool_item, ctx.tool_use_context + ) + elif isinstance(tool_item, ThinkingContent): + # Pass usage only if not yet used + chunk_usage = usage_to_show if not usage_used else None + usage_used = True + content = create_thinking_message( + chunk_meta, tool_item, chunk_usage + ) + tool_result = ToolItemResult( + message_type=content.message_type, + content=content, + ) else: # Handle unknown content types tool_result = ToolItemResult( message_type="unknown", - content=UnknownContent(type_name=str(type(tool_item))), - message_title="Unknown Content", + content=UnknownMessage( + chunk_meta, type_name=str(type(tool_item)) + ), ) - # Skip if handler returned None (e.g., unsupported image types) - if tool_result is None: + # Skip if no content (shouldn't happen, but be safe) + if tool_result.content is None: continue - # Preserve sidechain context for tool/thinking content - tool_is_sidechain = getattr(message, "isSidechain", False) - - # Generate unique UUID for this tool message - # Use tool_use_id if available, otherwise fall back to msg UUID + index - tool_uuid = ( - tool_result.tool_use_id - if tool_result.tool_use_id - else f"{msg_uuid}-tool-{len(template_messages)}" - ) - - # Thinking content uses markdown - tool_has_markdown = isinstance( - tool_result.content, ThinkingContentModel - ) - - tool_template_message = TemplateMessage( - message_type=tool_result.message_type, - formatted_timestamp=tool_formatted_timestamp, - raw_timestamp=tool_timestamp, - session_summary=session_summary, - session_id=session_id, - tool_use_id=tool_result.tool_use_id, - title_hint=tool_result.title_hint, - message_title=tool_result.message_title, - message_id=None, # Will be assigned by _build_message_hierarchy - ancestry=[], # Will be assigned by _build_message_hierarchy - agent_id=getattr(message, "agentId", None), - uuid=tool_uuid, - is_sidechain=tool_is_sidechain, - content=tool_result.content, # Structured content model - has_markdown=tool_has_markdown, - ) - - # Store raw text for Task result deduplication - # (handled later in _reorder_sidechain_template_messages) - if tool_result.pending_dedup is not None: - tool_template_message.raw_text_content = tool_result.pending_dedup - - template_messages.append(tool_template_message) - - # Track message timing - if DEBUG_TIMING: - msg_duration = time.time() - msg_start_time - message_timings.append((msg_duration, message_type, msg_idx, msg_uuid)) + tool_msg = TemplateMessage(tool_result.content) + ctx.register(tool_msg) - # Report loop statistics - if DEBUG_TIMING: - report_timing_statistics( - message_timings, - [("Markdown", markdown_timings), ("Pygments", pygments_timings)], - ) - - return template_messages + return ctx # -- Project Index Generation ------------------------------------------------- @@ -2368,49 +2056,201 @@ class Renderer: Subclasses implement format-specific rendering (HTML, Markdown, etc.). - The dispatcher pattern enables automatic content formatting based on type: - - Subclasses override _build_dispatcher() to map content types to formatters - - format_content() walks the MRO to find the most specific handler - - Fallback to parent class handlers if no specific handler exists + The method-based dispatcher pattern: + - Base class defines format_xyz_message() methods for each content type + - Each method documents its fallback chain (which method it delegates to) + - format_content() walks the MRO to find the most specific method + - Subclasses override methods to implement format-specific rendering """ - def __init__(self): - self._dispatcher = self._build_dispatcher() + def _dispatch_format(self, obj: Any) -> str: + """Dispatch to format_{ClassName} method based on object type.""" + for cls in type(obj).__mro__: + if cls is object: + break + if method := getattr(self, f"format_{cls.__name__}", None): + return method(obj) + return "" - def _build_dispatcher( - self, - ) -> dict[type, Callable[..., str]]: - """Build the content type to formatter mapping. + def _dispatch_title(self, obj: Any, message: "TemplateMessage") -> Optional[str]: + """Dispatch to title_{ClassName} method based on object type.""" + for cls in type(obj).__mro__: + if cls is object: + break + if method := getattr(self, f"title_{cls.__name__}", None): + return method(message) + return None + + def format_content(self, message: "TemplateMessage") -> str: + """Format message content by dispatching to type-specific method. - Override in subclasses to register format-specific handlers. - The dict maps MessageContent subclasses to formatter functions. - Each formatter receives the content directly (cast to the matched type). + Looks for a method named format_{ClassName} (e.g., format_SystemMessage). + Walks the content type's MRO to find the most specific format method. + + Args: + message: TemplateMessage with content to format. Returns: - Dict mapping content types to formatter functions. + Formatted string (e.g., HTML), or empty string if no handler found. """ - return {} + return self._dispatch_format(message.content) - def format_content(self, message: "TemplateMessage") -> str: - """Format message content by dispatching to type-specific handler. + def title_content(self, message: "TemplateMessage") -> str: + """Get message title by dispatching to type-specific title method. - Walks the content type's MRO to find the most specific registered - handler. This allows handlers for parent classes to serve as fallbacks. + Looks for a method named title_{ClassName} (e.g., title_ToolUseMessage). + Falls back to type-based title derived from message_type. Args: - message: TemplateMessage with content to format. + message: TemplateMessage to get title for. Returns: - Formatted string (e.g., HTML), or empty string if no handler found. + Title string for the message header. """ - if message.content is None: - return "" + # Try title_{ClassName} dispatch for cls in type(message.content).__mro__: if cls is object: break - if fmt := self._dispatcher.get(cls): - return fmt(message.content) - return "" + if method := getattr(self, f"title_{cls.__name__}", None): + return method(message) + # Fallback: convert message_type to title case + return message.content.message_type.replace("_", " ").replace("-", " ").title() + + # ------------------------------------------------------------------------- + # Title Methods (return title strings for message headers) + # ------------------------------------------------------------------------- + # These methods return title strings for specific content types. + # Override in subclasses for format-specific titles (e.g., HTML with icons). + + def title_SystemMessage(self, message: "TemplateMessage") -> str: + content = cast("SystemMessage", message.content) + return f"System {content.level.title()}" + + def title_HookSummaryMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "System Hook" + + def title_SlashCommandMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "Slash Command" + + def title_CommandOutputMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "" # Empty title for command output + + def title_BashInputMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "Bash command" + + def title_BashOutputMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "" # Empty title for bash output + + def title_CompactedSummaryMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "User (compacted conversation)" + + def title_UserMemoryMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "Memory" + + def title_UserSlashCommandMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "User (slash command)" + + def title_UserTextMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "User" + + def title_UserSteeringMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "User (steering)" + + def title_AssistantTextMessage(self, message: "TemplateMessage") -> str: + # Sidechain assistant messages get special title + if message.meta.is_sidechain: + return "Sub-assistant" + return "Assistant" + + def title_ThinkingMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "Thinking" + + def title_UnknownMessage(self, message: "TemplateMessage") -> str: # noqa: ARG002 + return "Unknown Content" + + # Tool title methods (dispatch to input/output title methods) + def title_ToolUseMessage(self, message: "TemplateMessage") -> str: + content = cast("ToolUseMessage", message.content) + if title := self._dispatch_title(content.input, message): + return title + return content.tool_name # Default to tool name + + def title_ToolResultMessage(self, message: "TemplateMessage") -> str: + content = cast("ToolResultMessage", message.content) + if content.is_error: + return "Error" + if title := self._dispatch_title(content.output, message): + return title + return "" # Tool results typically don't need a title + + # Tool input title stubs (override in subclasses for custom titles) + # def title_BashInput(self, message: "TemplateMessage") -> str: ... + # def title_ReadInput(self, message: "TemplateMessage") -> str: ... + # def title_EditInput(self, message: "TemplateMessage") -> str: ... + # def title_TaskInput(self, message: "TemplateMessage") -> str: ... + # def title_TodoWriteInput(self, message: "TemplateMessage") -> str: ... + + # ------------------------------------------------------------------------- + # Format Method Stubs (override in subclasses) + # ------------------------------------------------------------------------- + # System content formatters + # def format_SystemMessage(self, message: "SystemMessage") -> str: return "" + # def format_HookSummaryMessage(self, message: "HookSummaryMessage") -> str: ... + # def format_SessionHeaderMessage(self, message: "SessionHeaderMessage") -> str: ... + # def format_DedupNoticeMessage(self, message: "DedupNoticeMessage") -> str: ... + + # User content formatters + # def format_UserTextMessage(self, message: "UserTextMessage") -> str: ... + # def format_UserSteeringMessage(self, message: "UserSteeringMessage") -> str: ... + # def format_UserSlashCommandMessage(self, message: "UserSlashCommandMessage") -> str: ... + # def format_SlashCommandMessage(self, message: "SlashCommandMessage") -> str: ... + # def format_CommandOutputMessage(self, message: "CommandOutputMessage") -> str: ... + # def format_BashInputMessage(self, message: "BashInputMessage") -> str: ... + # def format_BashOutputMessage(self, message: "BashOutputMessage") -> str: ... + # def format_CompactedSummaryMessage(self, message: "CompactedSummaryMessage") -> str: ... + # def format_UserMemoryMessage(self, message: "UserMemoryMessage") -> str: ... + + # Assistant content formatters + # def format_AssistantTextMessage(self, message: "AssistantTextMessage") -> str: ... + # def format_ThinkingMessage(self, message: "ThinkingMessage") -> str: ... + # def format_UnknownMessage(self, message: "UnknownMessage") -> str: ... + + # Tool content formatters (dispatch to input/output formatters) + def format_ToolUseMessage(self, message: "ToolUseMessage") -> str: + """Dispatch to format_{InputClass} based on message.input type.""" + return self._dispatch_format(message.input) + + def format_ToolResultMessage(self, message: "ToolResultMessage") -> str: + """Dispatch to format_{OutputClass} based on message.output type.""" + return self._dispatch_format(message.output) + + # Tool input formatters + # def format_BashInput(self, input: "BashInput") -> str: ... + # def format_ReadInput(self, input: "ReadInput") -> str: ... + # def format_WriteInput(self, input: "WriteInput") -> str: ... + # def format_EditInput(self, input: "EditInput") -> str: ... + # def format_MultiEditInput(self, input: "MultiEditInput") -> str: ... + # def format_GlobInput(self, input: "GlobInput") -> str: ... + # def format_GrepInput(self, input: "GrepInput") -> str: ... + # def format_TaskInput(self, input: "TaskInput") -> str: ... + # def format_TodoWriteInput(self, input: "TodoWriteInput") -> str: ... + # def format_AskUserQuestionInput(self, input: "AskUserQuestionInput") -> str: ... + # def format_ExitPlanModeInput(self, input: "ExitPlanModeInput") -> str: ... + # def format_ToolUseContent(self, input: "ToolUseContent") -> str: ... # fallback + + # Tool output formatters + # def format_ReadOutput(self, output: "ReadOutput") -> str: ... + # def format_WriteOutput(self, output: "WriteOutput") -> str: ... + # def format_EditOutput(self, output: "EditOutput") -> str: ... + # def format_BashOutput(self, output: "BashOutput") -> str: ... + # def format_TaskOutput(self, output: "TaskOutput") -> str: ... + # def format_AskUserQuestionOutput(self, output: "AskUserQuestionOutput") -> str: ... + # def format_ExitPlanModeOutput(self, output: "ExitPlanModeOutput") -> str: ... + # def format_ToolResultContent(self, output: "ToolResultContent") -> str: ... # fallback + + # ------------------------------------------------------------------------- + # Rendering Entry Points + # ------------------------------------------------------------------------- def generate( self, diff --git a/claude_code_log/renderer_timings.py b/claude_code_log/renderer_timings.py index 96aff4b7..b6d43900 100644 --- a/claude_code_log/renderer_timings.py +++ b/claude_code_log/renderer_timings.py @@ -25,7 +25,7 @@ def set_timing_var(name: str, value: Any) -> None: """Set a timing variable in the global timing data dict. Args: - name: Variable name (e.g., "_markdown_timings", "_pygments_timings", "_current_msg_uuid") + name: Variable name (e.g., "_markdown_timings", "_pygments_timings", "_current_msg_id") value: Value to set """ if DEBUG_TIMING: @@ -106,8 +106,8 @@ def timing_stat(list_name: str) -> Iterator[None]: finally: duration = time.time() - t_start if list_name in _timing_data: - msg_uuid = _timing_data.get("_current_msg_uuid", "") - _timing_data[list_name].append((duration, msg_uuid)) + msg_id = _timing_data.get("_current_msg_id", "") + _timing_data[list_name].append((duration, msg_id)) def report_timing_statistics( @@ -117,30 +117,32 @@ def report_timing_statistics( """Report timing statistics for message rendering. Args: - message_timings: List of (duration, message_type, index, uuid) tuples - operation_timings: List of (name, timings) tuples where timings is a list of (duration, uuid) + message_timings: List of (duration, message_type, index, msg_id) tuples. + Can be empty if only operation timings are being reported. + operation_timings: List of (name, timings) tuples where timings is a list of (duration, msg_id) e.g., [("Markdown", markdown_timings), ("Pygments", pygments_timings)] """ - if not message_timings: - return - - # Sort by duration descending - sorted_timings = sorted(message_timings, key=lambda x: x[0], reverse=True) - - # Calculate statistics - total_msg_time = sum(t[0] for t in message_timings) - avg_time = total_msg_time / len(message_timings) - - # Report slowest messages - print("\n[TIMING] Loop statistics:", flush=True) - print(f"[TIMING] Total messages: {len(message_timings)}", flush=True) - print(f"[TIMING] Average time per message: {avg_time * 1000:.1f}ms", flush=True) - print("[TIMING] Slowest 10 messages:", flush=True) - for duration, msg_type, idx, uuid in sorted_timings[:10]: + # Report message loop statistics if available + if message_timings: + # Sort by duration descending + sorted_timings = sorted(message_timings, key=lambda x: x[0], reverse=True) + + # Calculate statistics + total_msg_time = sum(t[0] for t in message_timings) + avg_time = total_msg_time / len(message_timings) + + # Report slowest messages + print("\n[TIMING] Loop statistics:", flush=True) + print(f"[TIMING] Total messages: {len(message_timings)}", flush=True) print( - f"[TIMING] Message {uuid} (#{idx}, {msg_type}): {duration * 1000:.1f}ms", - flush=True, + f"[TIMING] Average time per message: {avg_time * 1000:.1f}ms", flush=True ) + print("[TIMING] Slowest 10 messages:", flush=True) + for duration, msg_type, idx, msg_id in sorted_timings[:10]: + print( + f"[TIMING] Message {msg_id} (#{idx}, {msg_type}): {duration * 1000:.1f}ms", + flush=True, + ) # Report operation-specific statistics for operation_name, timings in operation_timings: @@ -151,8 +153,8 @@ def report_timing_statistics( print(f"[TIMING] Total operations: {len(timings)}", flush=True) print(f"[TIMING] Total time: {total_time:.3f}s", flush=True) print("[TIMING] Slowest 10 operations:", flush=True) - for duration, uuid in sorted_ops[:10]: + for duration, msg_id in sorted_ops[:10]: print( - f"[TIMING] {uuid}: {duration * 1000:.1f}ms", + f"[TIMING] {msg_id}: {duration * 1000:.1f}ms", flush=True, ) diff --git a/claude_code_log/utils.py b/claude_code_log/utils.py index e0ad25fd..88910e89 100644 --- a/claude_code_log/utils.py +++ b/claude_code_log/utils.py @@ -8,7 +8,7 @@ from claude_code_log.cache import SessionCacheData from .models import ContentItem, TextContent, TranscriptEntry, UserTranscriptEntry -from .parser import ( +from .factories import ( IDE_DIAGNOSTICS_PATTERN, IDE_OPENED_FILE_PATTERN, IDE_SELECTION_PATTERN, diff --git a/dev-docs/MESSAGE_REFACTORING2.md b/dev-docs/MESSAGE_REFACTORING2.md index 87a29c57..fc84aaec 100644 --- a/dev-docs/MESSAGE_REFACTORING2.md +++ b/dev-docs/MESSAGE_REFACTORING2.md @@ -4,23 +4,91 @@ The goal is to achieve a cleaner, type-driven architecture where: 1. **MessageContent type is the source of truth** - No need for separate `MessageModifiers` or `MessageType` checks -2. **Inverted relationship** - Instead of `TemplateMessage.content: MessageContent`, have `MessageContent.meta: MessageMetadata` +2. **Inverted relationship** - Instead of `TemplateMessage.content: MessageContent`, have `MessageContent.meta: MessageMeta` 3. **Leaner models** - Remove derived/redundant fields like `has_children`, `has_markdown`, `is_session_header`, `raw_text_content` -4. **Modular organization** - Split into `user_models.py`, `assistant_models.py`, `tools_models.py` with corresponding parsers +4. **Modular organization** - Split into `user_models.py`, `assistant_models.py`, `tools_models.py` with corresponding factories ## Current State Analysis -### What we've achieved so far -- Content types now determine behavior (e.g., `UserSlashCommandContent` vs `UserTextContent`) -- Dispatcher pattern routes formatting based on content type -- Removed `ContentBlock` from `ContentItem` union - using our own types -- Simplified `_process_regular_message` - content type detection drives rendering +### What we've achieved βœ“ + +- **Content types now determine behavior** (e.g., `UserSlashCommandMessage` vs `UserTextMessage`) +- **Dispatcher pattern** routes formatting based on content type +- **Removed `ContentBlock`** from `ContentItem` union - using our own types +- **Simplified `_process_regular_message`** - content type detection drives rendering - **CSS_CLASS_REGISTRY** derives CSS classes from content types (in `html/utils.py`) -- **MessageModifiers removed** - only `is_sidechain` remains as a flag on `TemplateMessage` -- **UserSteeringContent** created for queue-operation "remove" messages +- **MessageModifiers removed** - only `is_sidechain` remains as a flag on `MessageMeta` +- **UserSteeringMessage** created for queue-operation "remove" messages +- **IdeNotificationContent** is now a plain dataclass (not a MessageContent subclass) +- **Inverted relationship achieved** - `MessageContent.meta` is the source of truth, `TemplateMessage.meta = content.meta` +- **Leaner TemplateMessage** - `has_markdown` delegates to content, `raw_text_content` moved to content classes +- **Title dispatch pattern** - `Renderer.title_content()` dispatches to `title_{ClassName}` methods + +### Factory Organization βœ“ + +Completed reorganization from parsers to factories: + +``` +factories/ +β”œβ”€β”€ __init__.py # Re-exports all public symbols +β”œβ”€β”€ meta_factory.py # create_meta(transcript) -> MessageMeta +β”œβ”€β”€ system_factory.py # create_system_message() +β”œβ”€β”€ user_factory.py # create_user_message(), create_*_message() +β”œβ”€β”€ assistant_factory.py # create_assistant_message(), create_thinking_message() +β”œβ”€β”€ tool_factory.py # create_tool_use_message(), create_tool_result_message() +└── transcript_factory.py # create_transcript_entry(), create_content_item() +``` + +### MessageMeta as Required First Parameter βœ“ + +All factory functions now require `MessageMeta` as the first positional parameter: + +```python +def create_user_message(meta: MessageMeta, content_list: list[ContentItem], ...) -> ... +def create_assistant_message(meta: MessageMeta, items: list[ContentItem]) -> ... +def create_tool_use_message(meta: MessageMeta, tool_item: ContentItem, ...) -> ... +def create_tool_result_message(meta: MessageMeta, tool_item: ContentItem, ...) -> ... +def create_thinking_message(meta: MessageMeta, tool_item: ContentItem) -> ... +``` -### Remaining goals -- `TemplateMessage` still owns `content` rather than the reverse (inverted relationship) +This ensures every `MessageContent` subclass has valid metadata. + +### Goals Status + +| Goal | Status | Notes | +|------|--------|-------| +| Inverted relationship | βœ… Done | `MessageContent.meta` is source of truth, `TemplateMessage.meta = content.meta` | +| Leaner TemplateMessage | βœ… Done | `has_markdown` delegates to content, `raw_text_content` on content classes | +| Title dispatch | βœ… Done | `Renderer.title_content()` with `title_{ClassName}` methods | +| Pure MessageContent | βœ… Done | MessageContent has no render-time fields (relationship data on TemplateMessage) | +| TemplateMessage as primary | βœ… Done | RenderingContext registers TemplateMessage, holds pairing/hierarchy data | +| Models split | ❌ Optional | Still single `models.py` - could split if needed | + +### TemplateMessage Architecture βœ“ + +TemplateMessage is now the primary render-time object, with clear separation of concerns: + +**MessageContent** (pure transcript data): +- `meta: MessageMeta` - metadata from transcript +- `message_type` property - type identifier +- `has_markdown` property - whether content has markdown + +**TemplateMessage** (render-time wrapper): +- `content: MessageContent` - the wrapped content +- `meta = content.meta` - convenience alias +- `message_index: Optional[int]` - index in RenderingContext.messages +- `message_id` property - formatted ID for HTML ("d-{index}" or "session-{id}") +- Pairing fields: `pair_first`, `pair_last`, `pair_duration` +- Pairing properties: `is_paired`, `is_first_in_pair`, `is_last_in_pair`, `pair_role` +- Hierarchy fields: `ancestry`, `children` +- Fold/unfold counts: `immediate_children_count`, `total_descendants_count`, etc. + +**RenderingContext**: +- `messages: list[TemplateMessage]` - registry of all messages +- `register(message: TemplateMessage) -> int` - assigns `message_index` +- `get(message_index: int) -> TemplateMessage` - lookup by index +- `tool_use_context: dict[str, ToolUseContent]` - for tool result pairing +- `session_first_message: dict[str, int]` - session header indices ## Cache Considerations @@ -41,23 +109,13 @@ This means: 2. TemplateMessage is generated fresh from entries on each render 3. The relationship between MessageContent and its metadata is internal to rendering -## Modular Organization Plan +## Future: Models Split (Optional) + +If we decide to split models.py: -### Models split - `models.py` - Base classes (`MessageContent`, `TranscriptEntry`, etc.) - `user_models.py` - User message content types - `assistant_models.py` - Assistant message content types - `tools_models.py` - Tool use/result models -### Parser split -- `parser.py` - Base parsing, entry point -- `user_parser.py` - User message parsing -- `assistant_parser.py` - Assistant message parsing - -### Renderer reorganization -- `renderer.py` - Main message reorganization (`_render_messages`, tree building) -- Move `_process_*` functions to appropriate parser modules - -## Related Work - -See [REMOVE_ANTHROPIC_TYPES.md](REMOVE_ANTHROPIC_TYPES.md) for simplifying Anthropic SDK dependencies. +This is optional and primarily a code organization improvement. diff --git a/dev-docs/REMOVE_ANTHROPIC_TYPES.md b/dev-docs/REMOVE_ANTHROPIC_TYPES.md deleted file mode 100644 index 1b7f96b1..00000000 --- a/dev-docs/REMOVE_ANTHROPIC_TYPES.md +++ /dev/null @@ -1,76 +0,0 @@ -# Remove Anthropic Types Dependency - -## Current Usage - -### Imports - -| Import | File | Purpose | -|--------|------|---------| -| `AnthropicMessage` | models.py, parser.py | Validate message compatibility | -| `AnthropicUsage` | models.py, parser.py | Convert usage info | -| `StopReason` | models.py | Type alias | - -### Methods - -| Method | Location | Used? | -|--------|----------|-------| -| `to_anthropic_usage()` | UsageInfo (models.py:711) | **Never** | -| `from_anthropic_usage()` | UsageInfo (models.py:725) | Yes, in `normalize_usage_info()` | -| `from_anthropic_message()` | AssistantMessage (models.py:808) | **Never** | - -### Call Sites - -| Call | Location | Purpose | Needed? | -|------|----------|---------|---------| -| `AnthropicMessage.model_validate()` | parser.py:933 | Validate JSONL data is compatible | **No** - result unused | -| `AnthropicUsage.model_validate()` | parser.py:731 | Parse usage dict as Anthropic type | **No** - can use UsageInfo directly | -| `UsageInfo.from_anthropic_usage()` | parser.py:725, 732 | Convert Anthropic Usage to ours | **Partially** - simplify | - -## Simplification Plan - -### Phase 1: Remove dead code -1. Remove `to_anthropic_usage()` - never used -2. Remove `from_anthropic_message()` - never used -3. Remove `AnthropicMessage.model_validate()` validation in parser.py - no-op -4. Remove `StopReason` import - just use `Optional[str]` - -### Phase 2: Simplify usage parsing -Replace the Anthropic-aware `normalize_usage_info()` with direct dict-to-UsageInfo conversion: -```python -def normalize_usage_info(usage_data: Any) -> Optional[UsageInfo]: - if usage_data is None: - return None - if isinstance(usage_data, UsageInfo): - return usage_data - if isinstance(usage_data, dict): - return UsageInfo.model_validate(usage_data) - # Handle object-like access for backwards compatibility - return UsageInfo( - input_tokens=getattr(usage_data, "input_tokens", None), - ... - ) -``` - -### Phase 3: Remove Anthropic imports -After Phase 1-2, remove from models.py and parser.py: -- `from anthropic.types import Message as AnthropicMessage` -- `from anthropic.types import Usage as AnthropicUsage` -- `from anthropic.types import StopReason` - -## Benefits - -1. **Simpler dependency** - Don't need anthropic SDK types for parsing our own JSONL -2. **Clearer ownership** - Our models are the canonical types, not wrappers -3. **Easier maintenance** - No need to track Anthropic SDK type changes -4. **Smaller models.py** - Less code, clearer structure - -## Open Questions - -1. **Was there ever a use case for `from_anthropic_message()`?** - - Possibly for direct SDK integration, but we only parse JSONL files - -2. **Why validate against `AnthropicMessage`?** - - Historical artifact from when we considered using SDK types directly - -3. **Could Anthropic types return in content arrays?** - - We already removed `ContentBlock` from `ContentItem` - no SDK types in content diff --git a/dev-docs/RENAME_CONTENT_TO_MESSAGE.md b/dev-docs/RENAME_CONTENT_TO_MESSAGE.md new file mode 100644 index 00000000..def039fb --- /dev/null +++ b/dev-docs/RENAME_CONTENT_TO_MESSAGE.md @@ -0,0 +1,173 @@ +# Refactoring Plan: Content β†’ Message Naming + +## Goal + +Clarify the naming by using consistent suffixes: +- `*Content` = ContentItem members (JSONL parsing layer) +- `*Input` / `*Output` = Tool-specific parsing +- `*Message` = MessageContent subclasses (rendering layer) +- `*Model` = Pydantic JSONL transcript models + +## Phase 1: Free up "Message" names + +Rename Pydantic transcript message models to add `Model` suffix: + +| Current | New | +|---------|-----| +| `UserMessage` | `UserMessageModel` | +| `AssistantMessage` | `AssistantMessageModel` | + +These are only used in `UserTranscriptEntry.message` and `AssistantTranscriptEntry.message` for Pydantic deserialization. + +## Phase 2: Rename MessageContent subclasses to ...Message + +| Current | New | +|---------|-----| +| `UserTextContent` | `UserTextMessage` | +| `UserSteeringContent` | `UserSteeringMessage` | +| `UserSlashCommandContent` | `UserSlashCommandMessage` | +| `UserMemoryContent` | `UserMemoryMessage` | +| `AssistantTextContent` | `AssistantTextMessage` | +| `SlashCommandContent` | `SlashCommandMessage` | +| `CommandOutputContent` | `CommandOutputMessage` | +| `CompactedSummaryContent` | `CompactedSummaryMessage` | +| `BashInputContent` | `BashInputMessage` | +| `BashOutputContent` | `BashOutputMessage` | +| `SystemContent` | `SystemMessage` | +| `HookSummaryContent` | `HookSummaryMessage` | +| `SessionHeaderContent` | `SessionHeaderMessage` | +| `DedupNoticeContent` | `DedupNoticeMessage` | +| `UnknownContent` | `UnknownMessage` | +| `ThinkingContentModel` | `ThinkingMessage` | +| `ToolResultContentModel` | `ToolResultMessage` | + +Also update: +- `CSS_CLASS_REGISTRY` in `html/utils.py` +- All formatters in `html/*_formatters.py` +- All usages in `renderer.py`, `parser.py`, etc. + +## Phase 3: Tool message wrapper pattern with typed inputs/outputs + +### New type aliases + +```python +# Union of all specialized input types + ToolUseContent as generic fallback +ToolInput = Union[ + BashInput, ReadInput, WriteInput, EditInput, MultiEditInput, + GlobInput, GrepInput, TaskInput, TodoWriteInput, AskUserQuestionInput, + ExitPlanModeInput, NotebookEditInput, WebFetchInput, WebSearchInput, + KillShellInput, + ToolUseContent, # Generic fallback when no specialized parser +] + +# Renamed from ToolUseResult for symmetry +# Union of all specialized output types + ToolResultContent as generic fallback +ToolOutput = Union[ + ReadOutput, EditOutput, # ... more as they're implemented + ToolResultContent, # Generic fallback for unparsed results +] +``` + +### New ToolUseMessage + +```python +@dataclass +class ToolUseMessage(MessageContent): + """Message for tool invocations.""" + input: ToolInput # Specialized (BashInput, etc.) or generic (ToolUseContent) + tool_use_id: str # From ToolUseContent.id + tool_name: str # From ToolUseContent.name +``` + +### New ToolResultMessage + +```python +@dataclass +class ToolResultMessage(MessageContent): + """Message for tool results.""" + output: ToolOutput # Specialized (ReadOutput, etc.) or generic (ToolResultContent) + tool_use_id: str + tool_name: Optional[str] = None + file_path: Optional[str] = None + + @property + def is_error(self) -> bool: + if isinstance(self.output, ToolResultContent): + return self.output.is_error or False + return False +``` + +### Simple ThinkingMessage (no wrapper) + +```python +@dataclass +class ThinkingMessage(MessageContent): + thinking_text: str # The thinking content + signature: Optional[str] = None +``` + +## Phase 4: Update CSS_CLASS_REGISTRY + +Update to use new names: + +```python +CSS_CLASS_REGISTRY: dict[type[MessageContent], list[str]] = { + # System messages + SystemMessage: ["system"], + HookSummaryMessage: ["system", "system-hook"], + # User messages + UserTextMessage: ["user"], + UserSteeringMessage: ["user", "steering"], + SlashCommandMessage: ["user", "slash-command"], + UserSlashCommandMessage: ["user", "slash-command"], + UserMemoryMessage: ["user"], + CompactedSummaryMessage: ["user", "compacted"], + CommandOutputMessage: ["user", "command-output"], + # Assistant messages + AssistantTextMessage: ["assistant"], + # Tool messages + ToolUseMessage: ["tool_use"], + ToolResultMessage: ["tool_result"], + # Other messages + ThinkingMessage: ["thinking"], + SessionHeaderMessage: ["session_header"], + BashInputMessage: ["bash-input"], + BashOutputMessage: ["bash-output"], + UnknownMessage: ["unknown"], +} +``` + +## Naming Pattern Summary + +| Suffix | Layer | Examples | +|--------|-------|----------| +| `*Content` | ContentItem (JSONL parsing) | `TextContent`, `ToolUseContent`, `ToolResultContent`, `ThinkingContent`, `ImageContent` | +| `*Input` | Tool input parsing | `BashInput`, `ReadInput`, `TaskInput`, ... | +| `*Output` | Tool output parsing | `ReadOutput`, `EditOutput`, ... | +| `*Message` | MessageContent (rendering) | `UserTextMessage`, `ToolUseMessage`, `ThinkingMessage` | +| `*Model` | Pydantic JSONL models | `UserMessageModel`, `AssistantMessageModel` | + +## Files to Update + +| File | Changes | +|------|---------| +| `models.py` | All renames, new ToolInput/ToolOutput unions | +| `parser.py` | Update imports and usages | +| `renderer.py` | Update imports and usages | +| `html/utils.py` | Update CSS_CLASS_REGISTRY | +| `html/renderer.py` | Update dispatcher and imports | +| `html/user_formatters.py` | Update function signatures and imports | +| `html/assistant_formatters.py` | Update function signatures and imports | +| `html/tool_formatters.py` | Update to use ToolUseMessage/ToolResultMessage | +| `html/system_formatters.py` | Update function signatures and imports | +| `converter.py` | Update imports | +| `dev-docs/messages.md` | Update documentation | + +## Execution Order + +1. Phase 1: Rename `UserMessage` β†’ `UserMessageModel`, `AssistantMessage` β†’ `AssistantMessageModel` +2. Phase 2: Rename all MessageContent subclasses to `*Message` +3. Phase 3: Create `ToolInput`, `ToolOutput` unions; update `ToolUseMessage`, `ToolResultMessage` +4. Phase 4: Update CSS_CLASS_REGISTRY +5. Run tests, fix any remaining issues +6. Update documentation diff --git a/dev-docs/TEMPLATE_MESSAGE_CHILDREN.md b/dev-docs/TEMPLATE_MESSAGE_CHILDREN.md index b1acf510..6a4c142b 100644 --- a/dev-docs/TEMPLATE_MESSAGE_CHILDREN.md +++ b/dev-docs/TEMPLATE_MESSAGE_CHILDREN.md @@ -25,12 +25,83 @@ Level 0: Tree roots (messages without ancestry - typically session headers) Level 1: User messages Level 2: Assistant, System, Thinking Level 3: Tool use/result -Level 4: Sidechain assistant/thinking +Level 4: Sidechain user/assistant/thinking Level 5: Sidechain tools ``` **Note:** Tree roots are any messages with empty `ancestry`. This is typically session headers, but in degenerate cases (no session headers), user messages or other top-level messages become roots. +### Sidechain Hierarchy Details (2025-12-24) + +Sidechain messages come from **Task tool** invocations (subagent spawning). Key findings from investigating real session data: + +#### Where Sidechain Children Attach + +Due to **pair reordering** (tool_result is moved right after its corresponding tool_use), sidechain messages become children of the **Task tool_result**, not the tool_use. + +#### Agent Type Patterns + +Real session data shows two distinct patterns depending on agent type: + +**Plan-type agents** (e.g., `-Users-dain-workspace-coderabbit-*`): +- Start with a **user prompt** (`UserTextMessage` with `isSidechain=true`, `parentUuid=null`) +- This user prompt duplicates the Task input + +Initial tree structure (before cleanup): +``` +tool_use (Task) ← 0 children (pair reordering moves tool_result here) +tool_result (Task) ← sidechain messages become children here + └─ user(sc): UserTextMessage ← Level 4: duplicate of Task input + └─ assistant(sc) ← Level 4: parented to user(sc) + └─ tool_use(sc) ← Level 5: parented to user(sc) + └─ tool_result(sc) ← Level 5 +``` + +After cleanup (user prompt removed, children adopted): +``` +tool_result (Task) + └─ assistant(sc) ← Now direct child of tool_result + └─ tool_use(sc) ← Adopted from removed user(sc) + └─ tool_result(sc) +``` + +**Explore-type agents** (e.g., `-src-deep-manifest`): +- Start directly with **assistant** (no user prompt) +- No cleanup needed for the first message + +``` +tool_result (Task) + └─ assistant(sc): AssistantTextMessage ← First child, kept as-is + └─ tool_use(sc) + └─ tool_result(sc) +``` + +#### Child Adoption During Cleanup + +When `_cleanup_sidechain_duplicates()` removes a UserTextMessage (the duplicate prompt), it must **adopt the removed message's children** to prevent orphaning Level 5 tool messages: + +```python +# In _cleanup_sidechain_duplicates() +if ( + children + and children[0].is_sidechain + and isinstance(children[0].content, UserTextMessage) +): + removed = children.pop(0) + # Adopt orphaned children (tool_use/tool_result from sidechain) + if removed.children: + children[:0] = removed.children +``` + +Without this adoption, the sidechain tool messages would be lost from the tree. + +#### Key Insight + +The hierarchy level is determined by message **type**, not by `parentUuid`. A sidechain user message (`parentUuid=null`) still appears at Level 4 because: +1. It has `isSidechain=true` +2. Its effective parent is determined by the Task tool_result (found via timestamp/session matching) +3. The tree-building algorithm correctly places it as a child of the Task tool_result + ### Template Rendering (current) - Single `{% for message in messages %}` loop over flattened list - Ancestry rendered as CSS classes for JavaScript DOM queries diff --git a/dev-docs/TEMPLATE_MESSAGE_REFACTORING.md b/dev-docs/TEMPLATE_MESSAGE_REFACTORING.md new file mode 100644 index 00000000..fee49101 --- /dev/null +++ b/dev-docs/TEMPLATE_MESSAGE_REFACTORING.md @@ -0,0 +1,99 @@ +# TemplateMessage Simplification Plan + +## Goal + +Simplify `TemplateMessage` by moving redundant fields to `MessageMeta` (accessible via `content.meta`) and adding properties to `MessageContent` subclasses. This prepares for the eventual replacement of `TemplateMessage` with `MessageContent` directly. + +## Completed Changes βœ“ + +### Phase 1: Added `message_type` property to MessageContent subclasses βœ“ + +Added to these subclasses that were missing it: +- `SystemMessage` β†’ returns "system" +- `HookSummaryMessage` β†’ returns "system" +- `ToolResultMessage` β†’ returns "tool_result" +- `ToolUseMessage` β†’ returns "tool_use" +- `UnknownMessage` β†’ returns "unknown" +- `SessionHeaderMessage` β†’ returns "session_header" +- `DedupNoticeMessage` β†’ returns "dedup_notice" + +### Phase 2: Added `has_markdown` property βœ“ + +- Added to `MessageContent` base class (returns `False` by default) +- Override in `AssistantTextMessage` β†’ returns `True` +- Override in `ThinkingMessage` β†’ returns `True` +- Override in `CompactedSummaryMessage` β†’ returns `True` + +### Phase 3: Skip tool_use_id on base βœ“ + +`tool_use_id` already exists as a field on `ToolUseMessage` and `ToolResultMessage`. +No base class property needed - access via `message.content.tool_use_id` when needed. + +### Phase 4: Added `meta` field to TemplateMessage βœ“ + +Added `self.meta = content.meta if content else None` for easy transition. + +### Phase 5: Updated template to use new accessors βœ“ + +- Changed `message.is_session_header` β†’ `is_session_header(message)` (helper function) +- Changed `message.has_markdown` β†’ `message.content.has_markdown if message.content else false` +- Removed dead `session_subtitle` code from template +- Added `is_session_header` helper to `html/utils.py` and template context + +### Phase 6: Converted parameters to properties βœ“ + +In `TemplateMessage`: +- Removed `is_session_header` parameter, added property that checks `isinstance(self.content, SessionHeaderMessage)` +- Removed `has_markdown` parameter, added property that returns `self.content.has_markdown if self.content else False` +- Removed `session_subtitle` assignment (was never set anyway) +- Removed unused imports (`CompactedSummaryMessage`, `ThinkingMessage`) + +## Current TemplateMessage State + +### Parameters (in `__init__`) + +| Parameter | Status | Notes | +|-----------|--------|-------| +| `message_type` | KEEP | Still used for now | +| `raw_timestamp` | KEEP | Still used | +| `session_summary` | KEEP | Complex async matching | +| `session_id` | KEEP | Still used | +| `token_usage` | KEEP | Formatted display string | +| `tool_use_id` | KEEP | Used for tool messages | +| `title_hint` | KEEP | Used for tooltips | +| `message_title` | KEEP | Display title | +| `message_id` | KEEP | Hierarchy-assigned | +| `ancestry` | KEEP | Parent chain | +| `has_children` | KEEP | Tree structure flag | +| `uuid` | KEEP | Still used | +| `parent_uuid` | KEEP | Still used | +| `agent_id` | KEEP | Still used | +| `is_sidechain` | KEEP | Still used | +| `content` | KEEP | The MessageContent | + +### Properties (derived from content) + +| Property | Derivation | +|----------|------------| +| `meta` | `content.meta if content else None` | +| `is_session_header` | `isinstance(self.content, SessionHeaderMessage)` | +| `has_markdown` | `self.content.has_markdown if self.content else False` | + +### Instance attributes (set after init) + +- `raw_text_content` - For deduplication +- Fold/unfold counts and type maps +- Pairing metadata (`is_paired`, `pair_role`, `pair_duration`) +- `children` - Tree structure + +## Future Work + +The following fields could still be derived from `content.meta` in future refactoring: +- `raw_timestamp` β†’ `content.meta.timestamp` +- `session_id` β†’ `content.meta.session_id` +- `uuid` β†’ `content.meta.uuid` +- `parent_uuid` β†’ `content.meta.parent_uuid` +- `agent_id` β†’ `content.meta.agent_id` +- `is_sidechain` β†’ `content.meta.is_sidechain` +- `message_type` β†’ `content.message_type` +- `message_title` β†’ `content.message_title()` diff --git a/dev-docs/messages.md b/dev-docs/messages.md index 09f9e9af..e10a4f60 100644 --- a/dev-docs/messages.md +++ b/dev-docs/messages.md @@ -21,35 +21,35 @@ JSONL Parsing (parser.py) β”‚ β”œβ”€β”€ UserTranscriptEntry β”‚ β”œβ”€β”€ TextContent β†’ User message variants: -β”‚ β”‚ β”œβ”€β”€ UserSlashCommandContent (isMeta) or SlashCommandContent ( tags) -β”‚ β”‚ β”œβ”€β”€ CommandOutputContent ( tags) -β”‚ β”‚ β”œβ”€β”€ BashInputContent ( tags) -β”‚ β”‚ β”œβ”€β”€ CompactedSummaryContent (compacted conversation) -β”‚ β”‚ β”œβ”€β”€ UserSteeringContent (queue-operation "remove") +β”‚ β”‚ β”œβ”€β”€ UserSlashCommandMessage (isMeta) or SlashCommandMessage ( tags) +β”‚ β”‚ β”œβ”€β”€ CommandOutputMessage ( tags) +β”‚ β”‚ β”œβ”€β”€ BashInputMessage ( tags) +β”‚ β”‚ β”œβ”€β”€ CompactedSummaryMessage (compacted conversation) +β”‚ β”‚ β”œβ”€β”€ UserSteeringMessage (queue-operation "remove") β”‚ β”‚ └── Plain user text -β”‚ β”œβ”€β”€ ToolResultContent β†’ Tool result messages: +β”‚ β”œβ”€β”€ ToolResultContent β†’ ToolResultMessage with output: β”‚ β”‚ β”œβ”€β”€ ReadOutput (cat-n formatted file content) β”‚ β”‚ β”œβ”€β”€ EditOutput (cat-n formatted edit result) -β”‚ β”‚ └── Generic tool result text +β”‚ β”‚ └── ToolResultContent (generic fallback) β”‚ └── ImageContent β†’ Image messages β”‚ β”œβ”€β”€ AssistantTranscriptEntry -β”‚ β”œβ”€β”€ TextContent β†’ AssistantTextContent -β”‚ β”œβ”€β”€ ThinkingContent β†’ ThinkingContentModel -β”‚ └── ToolUseContent β†’ Tool use messages with parsed inputs: +β”‚ β”œβ”€β”€ TextContent β†’ AssistantTextMessage +β”‚ β”œβ”€β”€ ThinkingContent β†’ ThinkingMessage +β”‚ └── ToolUseContent β†’ ToolUseMessage with parsed inputs: β”‚ β”œβ”€β”€ ReadInput, WriteInput, EditInput, MultiEditInput β”‚ β”œβ”€β”€ BashInput, GlobInput, GrepInput β”‚ β”œβ”€β”€ TaskInput, TodoWriteInput, AskUserQuestionInput β”‚ └── ExitPlanModeInput β”‚ β”œβ”€β”€ SystemTranscriptEntry -β”‚ β”œβ”€β”€ SystemContent (level: info/warning/error) -β”‚ └── HookSummaryContent (subtype: stop_hook_summary) +β”‚ β”œβ”€β”€ SystemMessage (level: info/warning/error) +β”‚ └── HookSummaryMessage (subtype: stop_hook_summary) β”‚ β”œβ”€β”€ SummaryTranscriptEntry β†’ Session metadata (not rendered) β”‚ └── QueueOperationTranscriptEntry - └── "remove" operation β†’ UserSteeringContent (rendered as user) + └── "remove" operation β†’ UserSteeringMessage (rendered as user) ``` --- @@ -99,22 +99,22 @@ CSS classes are derived from the content type using `CSS_CLASS_REGISTRY` (in `ht | css_class | Content Type | Dynamic Modifier | |-----------|--------------|------------------| -| `"user"` | `UserTextContent` | β€” | -| `"user compacted"` | `CompactedSummaryContent` | β€” | -| `"user slash-command"` | `SlashCommandContent`, `UserSlashCommandContent` | β€” | -| `"user command-output"` | `CommandOutputContent` | β€” | -| `"user steering"` | `UserSteeringContent` | β€” | -| `"assistant"` | `AssistantTextContent` | β€” | -| `"tool_use"` | `ToolUseContent` | β€” | -| `"tool_result"` | `ToolResultContentModel` | β€” | -| `"tool_result error"` | `ToolResultContentModel` | `is_error=True` | -| `"thinking"` | `ThinkingContentModel` | β€” | -| `"bash-input"` | `BashInputContent` | β€” | -| `"bash-output"` | `BashOutputContent` | β€” | -| `"system system-info"` | `SystemContent` | `level="info"` | -| `"system system-warning"` | `SystemContent` | `level="warning"` | -| `"system system-error"` | `SystemContent` | `level="error"` | -| `"system system-hook"` | `HookSummaryContent` | β€” | +| `"user"` | `UserTextMessage` | β€” | +| `"user compacted"` | `CompactedSummaryMessage` | β€” | +| `"user slash-command"` | `SlashCommandMessage`, `UserSlashCommandMessage` | β€” | +| `"user command-output"` | `CommandOutputMessage` | β€” | +| `"user steering"` | `UserSteeringMessage` | β€” | +| `"assistant"` | `AssistantTextMessage` | β€” | +| `"tool_use"` | `ToolUseMessage` | β€” | +| `"tool_result"` | `ToolResultMessage` | β€” | +| `"tool_result error"` | `ToolResultMessage` | `is_error=True` | +| `"thinking"` | `ThinkingMessage` | β€” | +| `"bash-input"` | `BashInputMessage` | β€” | +| `"bash-output"` | `BashOutputMessage` | β€” | +| `"system system-info"` | `SystemMessage` | `level="info"` | +| `"system system-warning"` | `SystemMessage` | `level="warning"` | +| `"system system-error"` | `SystemMessage` | `level="error"` | +| `"system system-hook"` | `HookSummaryMessage` | β€” | The `sidechain` modifier is added when `msg.is_sidechain=True` (a cross-cutting concern that applies to any message type). @@ -158,7 +158,7 @@ Based on flags and tag patterns in `TextContent`, user text messages are classif ### Slash Command (isMeta) - **Condition**: `isMeta: true` flag -- **Content Model**: `UserSlashCommandContent` (models.py) +- **Content Model**: `UserSlashCommandMessage` (models.py) - **CSS Class**: `user slash-command` - **Files**: [user_slash_command.json](messages/user/user_slash_command.json) @@ -172,7 +172,7 @@ Based on flags and tag patterns in `TextContent`, user text messages are classif ```python @dataclass -class UserSlashCommandContent(MessageContent): +class UserSlashCommandMessage(MessageContent): text: str # LLM-generated markdown instruction text ``` @@ -182,13 +182,13 @@ class UserSlashCommandContent(MessageContent): ### Slash Command (Tags) - **Condition**: Contains `` tags -- **Content Model**: `SlashCommandContent` with parsed name/args/contents +- **Content Model**: `SlashCommandMessage` with parsed name/args/contents - **CSS Class**: `user slash-command` - **Files**: [user_command.json](messages/user/user_command.json) ```python @dataclass -class SlashCommandContent(MessageContent): +class SlashCommandMessage(MessageContent): command_name: str # e.g., "/model", "/context" command_args: str # Arguments after command command_contents: str # Content inside command @@ -202,13 +202,13 @@ class SlashCommandContent(MessageContent): ### Command Output - **Condition**: Contains `` tags -- **Content Model**: `CommandOutputContent` +- **Content Model**: `CommandOutputMessage` - **CSS Class**: `user command-output` - **Files**: [command_output.json](messages/user/command_output.json) ```python @dataclass -class CommandOutputContent(MessageContent): +class CommandOutputMessage(MessageContent): stdout: str # Command output text is_markdown: bool # True if content appears to be markdown ``` @@ -216,13 +216,13 @@ class CommandOutputContent(MessageContent): ### Bash Input - **Condition**: Contains `` tags -- **Content Model**: `BashInputContent` +- **Content Model**: `BashInputMessage` - **CSS Class**: `bash-input` (filtered by User) - **Files**: [bash_input.json](messages/user/bash_input.json) ```python @dataclass -class BashInputContent(MessageContent): +class BashInputMessage(MessageContent): command: str # The bash command that was executed ``` @@ -231,34 +231,34 @@ class BashInputContent(MessageContent): The corresponding output uses `` and optionally `` tags: - **Condition**: Contains `` tags -- **Content Model**: `BashOutputContent` +- **Content Model**: `BashOutputMessage` - **CSS Class**: `bash-output` (filtered by User) - **Files**: [bash_output.json](messages/user/bash_output.json) ### Compacted Conversation - **Condition**: Contains "(compacted conversation)" marker -- **Content Model**: `CompactedSummaryContent` +- **Content Model**: `CompactedSummaryMessage` - **CSS Class**: `user compacted` ```python @dataclass -class CompactedSummaryContent(MessageContent): +class CompactedSummaryMessage(MessageContent): summary_text: str # The compacted conversation summary ``` ### User Steering (Queue Remove) - **Condition**: `QueueOperationTranscriptEntry` with `operation: "remove"` -- **Content Model**: `UserSteeringContent` (extends `UserTextContent`) +- **Content Model**: `UserSteeringMessage` (extends `UserTextMessage`) - **CSS Class**: `user steering` - **Title**: "User (steering)" ```python @dataclass -class UserSteeringContent(UserTextContent): - """Content for user steering prompts (queue-operation 'remove').""" - pass # Inherits items from UserTextContent +class UserSteeringMessage(UserTextMessage): + """Message for user steering prompts (queue-operation 'remove').""" + pass # Inherits items from UserTextMessage ``` Steering messages represent user interrupts that cancel queued operations. @@ -296,7 +296,8 @@ class IdeDiagnostic: raw_content: Optional[str] # Fallback if parsing failed @dataclass -class IdeNotificationContent(MessageContent): +class IdeNotificationContent: # NOT a MessageContent subclass + """Embedded within UserTextMessage.items alongside TextContent/ImageContent.""" opened_files: List[IdeOpenedFile] selections: List[IdeSelection] diagnostics: List[IdeDiagnostic] @@ -313,14 +314,16 @@ Tool results appear as `ToolResultContent` items in user messages, linked to the |------|--------------|------------|-------| | Read | `ReadOutput` | file_path, content, start_line, num_lines, is_truncated | [tool_result](messages/tools/Read-tool_result.json) | | Edit | `EditOutput` | file_path, success, diffs, message, start_line | [tool_result](messages/tools/Edit-tool_result.json) | -| Write | `WriteOutput` *(TODO)* | file_path, success, message | [tool_result](messages/tools/Write-tool_result.json) | -| Bash | `BashOutput` *(TODO)* | stdout, stderr, exit_code, interrupted, is_image | [tool_result](messages/tools/Bash-tool_result.json) | +| Write | `WriteOutput` | file_path, success, message | [tool_result](messages/tools/Write-tool_result.json) | +| Bash | `BashOutput` | content, has_ansi | [tool_result](messages/tools/Bash-tool_result.json) | +| Task | `TaskOutput` | result | [tool_result](messages/tools/Task-tool_result.json) | +| AskUserQuestion | `AskUserQuestionOutput` | answers, raw_message | [tool_result](messages/tools/AskUserQuestion-tool_result.json) | +| ExitPlanMode | `ExitPlanModeOutput` | message, approved | [tool_result](messages/tools/ExitPlanMode-tool_result.json) | | Glob | `GlobOutput` *(TODO)* | pattern, files, truncated | [tool_result](messages/tools/Glob-tool_result.json) | | Grep | `GrepOutput` *(TODO)* | pattern, matches, output_mode, truncated | [tool_result](messages/tools/Grep-tool_result.json) | -| Task | `TaskOutput` *(TODO)* | agent_id, result, is_background | [tool_result](messages/tools/Task-tool_result.json) | | (error) | β€” | is_error: true | [Bash error](messages/tools/Bash-tool_result_error.json) | -**(TODO)**: Output model defined in models.py but not yet used - tool results currently handled as raw strings. +**(TODO)**: Glob and Grep output models defined in models.py but not yet used. ### Generic Tool Result @@ -381,16 +384,23 @@ class EditOutput(MessageContent): ### Tool Result Rendering Wrapper -Tool results are wrapped in `ToolResultContentModel` for rendering, which provides additional context: +Tool results are wrapped in `ToolResultMessage` for rendering, which provides additional context and typed output: ```python @dataclass -class ToolResultContentModel(MessageContent): +class ToolResultMessage(MessageContent): tool_use_id: str - content: Any # Union[str, List[Dict[str, Any]]] + output: ToolOutput # Specialized (ReadOutput, EditOutput) or ToolResultContent is_error: bool = False tool_name: Optional[str] = None # Name of the tool file_path: Optional[str] = None # File path for Read/Edit/Write + +# ToolOutput is a union type for tool results +ToolOutput = Union[ + ReadOutput, + EditOutput, + ToolResultContent, # Generic fallback for unparsed results +] ``` ## 1.4 Images (ImageContent) @@ -440,16 +450,16 @@ Assistant messages contain `ContentItem` instances that are: - **ThinkingContent**: Extended thinking blocks - **ToolUseContent**: Tool invocations -## 2.2 Assistant Text β†’ AssistantTextContent +## 2.2 Assistant Text β†’ AssistantTextMessage -- **Content Model**: `AssistantTextContent` (models.py) +- **Content Model**: `AssistantTextMessage` (models.py) - **CSS Class**: `assistant` (or `assistant sidechain`) - **Files**: [assistant.json](messages/assistant/assistant.json) ```python @dataclass -class AssistantTextContent(MessageContent): - text: str # The assistant's response text +class AssistantTextMessage(MessageContent): + items: list[TextContent | ImageContent] # Interleaved text and images ``` ### Sidechain Assistant @@ -459,15 +469,15 @@ class AssistantTextContent(MessageContent): - **Title**: "Sub-assistant" - **Files**: [assistant_sidechain.json](messages/assistant/assistant_sidechain.json) -## 2.3 Thinking Content β†’ ThinkingContentModel +## 2.3 Thinking Content β†’ ThinkingMessage -- **Content Model**: `ThinkingContentModel` (models.py) +- **Content Model**: `ThinkingMessage` (models.py) - **CSS Class**: `thinking` - **Files**: [thinking.json](messages/assistant/thinking.json) ```python @dataclass -class ThinkingContentModel(MessageContent): +class ThinkingMessage(MessageContent): thinking: str # The thinking text signature: Optional[str] # Thinking block signature ``` @@ -481,14 +491,31 @@ class ThinkingContentModel(MessageContent): } ``` -## 2.4 Tool Use β†’ ToolUseContent with Typed Inputs +## 2.4 Tool Use β†’ ToolUseMessage with Typed Inputs + +Tool invocations are parsed from `ToolUseContent` (JSONL) and wrapped in `ToolUseMessage` for rendering: + +```python +@dataclass +class ToolUseMessage(MessageContent): + input: ToolInput # Specialized (BashInput, etc.) or ToolUseContent fallback + tool_use_id: str # From ToolUseContent.id + tool_name: str # From ToolUseContent.name + +# ToolInput is a union of typed input models +ToolInput = Union[ + BashInput, ReadInput, WriteInput, EditInput, MultiEditInput, + GlobInput, GrepInput, TaskInput, TodoWriteInput, + AskUserQuestionInput, ExitPlanModeInput, + ToolUseContent, # Generic fallback when no specialized parser +] +``` -Tool invocations contain a `ToolUseContent` item with: +The original `ToolUseContent` (Pydantic model) provides: - `name`: The tool name (e.g., "Read", "Bash", "Task") - `id`: Unique ID for pairing with results - `input`: Raw input dictionary - -The `parsed_input` property returns a typed input model via `parse_tool_input()`. +- `parsed_input` property: Returns typed input model via `parse_tool_input()` ### Tool Input Models (models.py) @@ -564,18 +591,18 @@ System transcript entries (`type: "system"`) convey notifications and hook summa ## 3.1 Content Types for System Messages System messages are parsed into structured content models in `models.py`: -- **SystemContent**: For info/warning/error messages -- **HookSummaryContent**: For hook execution summaries +- **SystemMessage**: For info/warning/error messages +- **HookSummaryMessage**: For hook execution summaries -## 3.2 System Info/Warning/Error β†’ SystemContent +## 3.2 System Info/Warning/Error β†’ SystemMessage -- **Content Model**: `SystemContent` (models.py) +- **Content Model**: `SystemMessage` (models.py) - **CSS Class**: `system system-info`, `system system-warning`, `system system-error` - **Files**: [system_info.json](messages/system/system_info.json) ```python @dataclass -class SystemContent(MessageContent): +class SystemMessage(MessageContent): level: str # "info", "warning", "error" text: str # Raw text content (may contain ANSI codes) ``` @@ -588,9 +615,9 @@ class SystemContent(MessageContent): } ``` -## 3.3 Hook Summary β†’ HookSummaryContent +## 3.3 Hook Summary β†’ HookSummaryMessage -- **Content Model**: `HookSummaryContent` (models.py) +- **Content Model**: `HookSummaryMessage` (models.py) - **Condition**: `subtype: "stop_hook_summary"` - **CSS Class**: `system system-hook` @@ -600,7 +627,7 @@ class HookInfo: command: str @dataclass -class HookSummaryContent(MessageContent): +class HookSummaryMessage(MessageContent): has_output: bool hook_errors: List[str] hook_infos: List[HookInfo] @@ -646,26 +673,28 @@ The `leafUuid` links the summary to the last message of the session. These models are created during rendering to represent synthesized content not directly from JSONL entries. -## 5.1 SessionHeaderContent +## 5.1 SessionHeaderMessage Session headers are rendered at the start of each session: ```python @dataclass -class SessionHeaderContent(MessageContent): +class SessionHeaderMessage(MessageContent): title: str # e.g., "Session 2025-12-13 10:30" session_id: str # Session UUID summary: Optional[str] = None # Session summary if available ``` -## 5.2 DedupNoticeContent +## 5.2 DedupNoticeMessage Deduplication notices are shown when content is deduplicated (e.g., sidechain assistant text that duplicates the Task tool result): ```python @dataclass -class DedupNoticeContent(MessageContent): +class DedupNoticeMessage(MessageContent): notice_text: str # e.g., "Content omitted (duplicates Task result)" + target_uuid: Optional[str] = None # UUID of target message + target_message_id: Optional[str] = None # Resolved message ID for anchor link ``` --- @@ -678,16 +707,32 @@ Display styling is derived from content types using `CSS_CLASS_REGISTRY` in `htm ```python CSS_CLASS_REGISTRY: dict[type[MessageContent], list[str]] = { + # System message types + SystemMessage: ["system"], # level added dynamically + HookSummaryMessage: ["system", "system-hook"], # User message types - UserTextContent: ["user"], - UserSteeringContent: ["user", "steering"], - SlashCommandContent: ["user", "slash-command"], - CompactedSummaryContent: ["user", "compacted"], - # ... more content types + UserTextMessage: ["user"], + UserSteeringMessage: ["user", "steering"], + SlashCommandMessage: ["user", "slash-command"], + UserSlashCommandMessage: ["user", "slash-command"], + UserMemoryMessage: ["user"], + CompactedSummaryMessage: ["user", "compacted"], + CommandOutputMessage: ["user", "command-output"], + # Assistant message types + AssistantTextMessage: ["assistant"], + # Tool message types + ToolUseMessage: ["tool_use"], + ToolResultMessage: ["tool_result"], # error added dynamically + # Other message types + ThinkingMessage: ["thinking"], + SessionHeaderMessage: ["session_header"], + BashInputMessage: ["bash-input"], + BashOutputMessage: ["bash-output"], + UnknownMessage: ["unknown"], } ``` -The `_get_css_classes_from_content()` function walks the content type's MRO to find the matching registry entry, then adds dynamic modifiers (e.g., `system-{level}` for `SystemContent`). +The `_get_css_classes_from_content()` function walks the content type's MRO to find the matching registry entry, then adds dynamic modifiers (e.g., `system-{level}` for `SystemMessage`). The only cross-cutting modifier is `is_sidechain`, which is stored directly on `TemplateMessage` and appended to CSS classes when true. @@ -820,10 +865,17 @@ Sub-agent messages (from `Task` tool): - [models.py](../claude_code_log/models.py) - Pydantic models for transcript data - [renderer.py](../claude_code_log/renderer.py) - Main rendering module - [html/](../claude_code_log/html/) - HTML-specific formatters (formatting only, content models in models.py) - - [system_formatters.py](../claude_code_log/html/system_formatters.py) - SystemContent, HookSummaryContent formatting + - [system_formatters.py](../claude_code_log/html/system_formatters.py) - SystemMessage, HookSummaryMessage formatting - [user_formatters.py](../claude_code_log/html/user_formatters.py) - User message formatting - - [assistant_formatters.py](../claude_code_log/html/assistant_formatters.py) - AssistantText, Thinking, Image formatting + - [assistant_formatters.py](../claude_code_log/html/assistant_formatters.py) - AssistantTextMessage, ThinkingMessage, ImageContent formatting - [tool_formatters.py](../claude_code_log/html/tool_formatters.py) - Tool use/result formatting -- [parser.py](../claude_code_log/parser.py) - JSONL parsing module +- [parser.py](../claude_code_log/parser.py) - JSONL parsing and text extraction +- [factories/](../claude_code_log/factories/) - Content creation from parsed data + - [user_factory.py](../claude_code_log/factories/user_factory.py) - `create_user_message()`, `create_*_message()` functions + - [assistant_factory.py](../claude_code_log/factories/assistant_factory.py) - `create_assistant_message()`, `create_thinking_message()` + - [tool_factory.py](../claude_code_log/factories/tool_factory.py) - `create_tool_use_message()`, `create_tool_result_message()` + - [system_factory.py](../claude_code_log/factories/system_factory.py) - `create_system_message()` + - [meta_factory.py](../claude_code_log/factories/meta_factory.py) - `create_meta()` - [TEMPLATE_MESSAGE_CHILDREN.md](TEMPLATE_MESSAGE_CHILDREN.md) - Tree architecture exploration -- [MESSAGE_REFACTORING.md](MESSAGE_REFACTORING.md) - Refactoring plan +- [MESSAGE_REFACTORING.md](MESSAGE_REFACTORING.md) - Refactoring plan (Phase 1) +- [MESSAGE_REFACTORING2.md](MESSAGE_REFACTORING2.md) - Refactoring plan (Phase 2) diff --git a/test/__snapshots__/test_snapshot_html.ambr b/test/__snapshots__/test_snapshot_html.ambr index c2b71c39..6ee0eec7 100644 --- a/test/__snapshots__/test_snapshot_html.ambr +++ b/test/__snapshots__/test_snapshot_html.ambr @@ -1003,7 +1003,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } @@ -4852,18 +4852,17 @@
-
+
Session: test_ses
- -
+
-
+
β–Ό 4 users
-
+
β–Όβ–Ό 4 users, 3 assistants, 2 more total
@@ -4875,7 +4874,7 @@ -
+
🀷 User
@@ -4887,10 +4886,10 @@ 
Hello Claude! Can you help me understand how Python decorators work?
-
+
-
+
β–Ό 1 assistant
@@ -4902,7 +4901,7 @@ -
+
πŸ€– Assistant
@@ -4941,7 +4940,7 @@ -
+
🀷 User
@@ -4953,10 +4952,10 @@ 
Great! Can you also show me how to create a decorator that takes parameters?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -4968,7 +4967,7 @@ -
+
πŸ“ Edit /tmp/decorator_example.py
@@ -4985,7 +4984,7 @@ -
+
@@ -5002,7 +5001,7 @@ -
+
πŸ€– Assistant
@@ -5034,7 +5033,7 @@ -
+
🀷 User
@@ -5046,10 +5045,10 @@ 
Can you run that example to show the output?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -5061,9 +5060,9 @@ -
+
- πŸ› οΈ Bash Run the decorator example to show output + πŸ’» Bash Run the decorator example to show output
2025-07-03 16:04:07 @@ -5078,7 +5077,7 @@ -
+
@@ -5097,7 +5096,7 @@ -
+
πŸ€– Assistant
@@ -5125,7 +5124,7 @@ -
+
🀷 User
@@ -5642,8 +5641,8 @@ if (!classList.includes(targetId)) return false; // Check if it's an immediate child by finding all ancestor IDs - // Ancestor IDs can be either d-XXX (regular messages) or session-XXX (sessions) - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + // Ancestor IDs use unified d-XXX format + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; return lastAncestor === targetId; }); @@ -5741,7 +5740,7 @@ // Fold: show first level only, hide deeper descendants (State C β†’ State B) descendants.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; // Show immediate children, hide non-immediate children @@ -5795,7 +5794,7 @@ allMessages.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); if (ancestorIds.length > 0) { const lastAncestor = ancestorIds[ancestorIds.length - 1]; @@ -6032,7 +6031,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } @@ -9653,18 +9652,17 @@
-
+
Session: Tested various edge cases including markdown formatting, long text, tool errors, system messages, command outputs, special characters and emojis. All message types render correctly in the transcript viewer. β€’ edge_cas
- -
+
-
+
β–Ό 5 users
-
+
β–Όβ–Ό 5 users, 2 assistants, 2 more total
@@ -9676,7 +9674,7 @@ -
+
🀷 User
@@ -9688,10 +9686,10 @@ 
Here's a message with some **markdown** formatting, `inline code`, and even a [link](https://example.com). Let's see how it renders!
-
+
-
+
β–Ό 1 assistant
@@ -9703,7 +9701,7 @@ -
+
πŸ€– Assistant
@@ -9761,7 +9759,7 @@ -
+
🀷 User
@@ -9773,10 +9771,10 @@ 
Let's test a very long message to see how it handles text wrapping and layout. Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. Sed ut perspiciatis unde omnis iste natus error sit voluptatem accusantium doloremque laudantium, totam rem aperiam, eaque ipsa quae ab illo inventore veritatis et quasi architecto beatae vitae dicta sunt explicabo. Nemo enim ipsam voluptatem quia voluptas sit aspernatur aut odit aut fugit, sed quia consequuntur magni dolores eos qui ratione voluptatem sequi nesciunt. Neque porro quisquam est, qui dolorem ipsum quia dolor sit amet, consectetur, adipisci velit, sed quia non numquam eius modi tempora incidunt ut labore et dolore magnam aliquam quaerat voluptatem.
-
+
-
+
β–Ό 1 tool
@@ -9788,7 +9786,7 @@ -
+
πŸ› οΈ FailingTool
@@ -9810,7 +9808,7 @@ -
+
🚨 Error
@@ -9827,7 +9825,7 @@ -
+
🀷 Slash Command
@@ -9846,7 +9844,7 @@ -
+
@@ -9863,14 +9861,14 @@ Status: SUCCESS Timestamp: 2025-06-14T11:02:20Z
-
+
-
+
β–Ό 1 assistant
-
+
β–Όβ–Ό 1 assistant, 1 tool total
@@ -9882,7 +9880,7 @@ -
+
πŸ€– Assistant
@@ -9897,10 +9895,10 @@

I see the long Lorem ipsum text wraps nicely! Long text handling is important for readability. The CSS should handle word wrapping automatically.

-
+
-
+
β–Ό 1 tool
@@ -9912,7 +9910,7 @@ -
+
πŸ› οΈ MultiEdit
@@ -9929,7 +9927,7 @@ -
+
🀷 User
@@ -9946,7 +9944,7 @@ -
+
🀷 User
@@ -9963,14 +9961,13 @@
-
+
Session: todowrit
- -
+
-
+
β–Ό 1 tool
@@ -9982,7 +9979,7 @@ -
+
πŸ“ Todo List
@@ -10533,8 +10530,8 @@ if (!classList.includes(targetId)) return false; // Check if it's an immediate child by finding all ancestor IDs - // Ancestor IDs can be either d-XXX (regular messages) or session-XXX (sessions) - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + // Ancestor IDs use unified d-XXX format + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; return lastAncestor === targetId; }); @@ -10632,7 +10629,7 @@ // Fold: show first level only, hide deeper descendants (State C β†’ State B) descendants.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; // Show immediate children, hide non-immediate children @@ -10686,7 +10683,7 @@ allMessages.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); if (ancestorIds.length > 0) { const lastAncestor = ancestorIds[ancestorIds.length - 1]; @@ -10923,7 +10920,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } @@ -14550,7 +14547,7 @@
-
@@ -14567,7 +14564,7 @@ -
@@ -14595,18 +14592,17 @@
-
+
Session: session_
- -
+
-
+
β–Ό 2 users
-
+
β–Όβ–Ό 2 users, 1 assistant total
@@ -14618,7 +14614,7 @@ -
+
🀷 User
@@ -14630,10 +14626,10 @@ 
This is from a different session file to test multi-session handling.
-
+
-
+
β–Ό 1 assistant
@@ -14645,7 +14641,7 @@ -
+
πŸ€– Assistant
@@ -14665,7 +14661,7 @@ -
+
🀷 User
@@ -14682,18 +14678,17 @@
-
+
Session: User learned about Python decorators, including basic decorators and parameterized decorators. Created and ran examples showing how decorators work with functions. User is now ready to implement their own timing decorator. β€’ test_ses
- -
+
-
+
β–Ό 4 users
-
+
β–Όβ–Ό 4 users, 3 assistants, 2 more total
@@ -14705,7 +14700,7 @@ -
+
🀷 User
@@ -14732,7 +14727,7 @@ -
+
πŸ€– Assistant
@@ -14771,7 +14766,7 @@ -
+
🀷 User
@@ -14783,10 +14778,10 @@ 
Great! Can you also show me how to create a decorator that takes parameters?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -14798,7 +14793,7 @@ -
+
πŸ“ Edit /tmp/decorator_example.py
@@ -14815,7 +14810,7 @@ -
+
@@ -14832,7 +14827,7 @@ -
+
πŸ€– Assistant
@@ -14864,7 +14859,7 @@ -
+
🀷 User
@@ -14876,10 +14871,10 @@ 
Can you run that example to show the output?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -14891,9 +14886,9 @@ -
+
- πŸ› οΈ Bash Run the decorator example to show output + πŸ’» Bash Run the decorator example to show output
2025-07-03 16:04:07 @@ -14908,7 +14903,7 @@ -
+
@@ -14927,7 +14922,7 @@ -
+
πŸ€– Assistant
@@ -14955,7 +14950,7 @@ -
+
🀷 User
@@ -15472,8 +15467,8 @@ if (!classList.includes(targetId)) return false; // Check if it's an immediate child by finding all ancestor IDs - // Ancestor IDs can be either d-XXX (regular messages) or session-XXX (sessions) - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + // Ancestor IDs use unified d-XXX format + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; return lastAncestor === targetId; }); @@ -15571,7 +15566,7 @@ // Fold: show first level only, hide deeper descendants (State C β†’ State B) descendants.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; // Show immediate children, hide non-immediate children @@ -15625,7 +15620,7 @@ allMessages.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); if (ancestorIds.length > 0) { const lastAncestor = ancestorIds[ancestorIds.length - 1]; @@ -15862,7 +15857,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } @@ -19483,18 +19478,17 @@
-
+
Session: User learned about Python decorators, including basic decorators and parameterized decorators. Created and ran examples showing how decorators work with functions. User is now ready to implement their own timing decorator. β€’ test_ses
- -
+
-
+
β–Ό 4 users
-
+
β–Όβ–Ό 4 users, 3 assistants, 2 more total
@@ -19506,7 +19500,7 @@ -
+
🀷 User
@@ -19518,10 +19512,10 @@ 
Hello Claude! Can you help me understand how Python decorators work?
-
+
-
+
β–Ό 1 assistant
@@ -19533,7 +19527,7 @@ -
+
πŸ€– Assistant
@@ -19572,7 +19566,7 @@ -
+
🀷 User
@@ -19584,10 +19578,10 @@ 
Great! Can you also show me how to create a decorator that takes parameters?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -19599,7 +19593,7 @@ -
+
πŸ“ Edit /tmp/decorator_example.py
@@ -19616,7 +19610,7 @@ -
+
@@ -19633,7 +19627,7 @@ -
+
πŸ€– Assistant
@@ -19665,7 +19659,7 @@ -
+
🀷 User
@@ -19677,10 +19671,10 @@ 
Can you run that example to show the output?
-
+
-
+
β–Ό 1 tool, 1 assistant
@@ -19692,9 +19686,9 @@ -
+
- πŸ› οΈ Bash Run the decorator example to show output + πŸ’» Bash Run the decorator example to show output
2025-07-03 16:04:07 @@ -19709,7 +19703,7 @@ -
+
@@ -19728,7 +19722,7 @@ -
+
πŸ€– Assistant
@@ -19756,7 +19750,7 @@ -
+
🀷 User
@@ -20273,8 +20267,8 @@ if (!classList.includes(targetId)) return false; // Check if it's an immediate child by finding all ancestor IDs - // Ancestor IDs can be either d-XXX (regular messages) or session-XXX (sessions) - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + // Ancestor IDs use unified d-XXX format + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; return lastAncestor === targetId; }); @@ -20372,7 +20366,7 @@ // Fold: show first level only, hide deeper descendants (State C β†’ State B) descendants.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); const lastAncestor = ancestorIds[ancestorIds.length - 1]; // Show immediate children, hide non-immediate children @@ -20426,7 +20420,7 @@ allMessages.forEach(msg => { const classList = Array.from(msg.classList); - const ancestorIds = classList.filter(cls => cls.startsWith('d-') || cls.startsWith('session-')); + const ancestorIds = classList.filter(cls => cls.startsWith('d-')); if (ancestorIds.length > 0) { const lastAncestor = ancestorIds[ancestorIds.length - 1]; @@ -20663,7 +20657,7 @@ let prev = messageElement.previousElementSibling; while (prev) { if (prev.classList.contains('session-header')) { - return prev.id.replace('session-', ''); + return prev.dataset.sessionId || null; } prev = prev.previousElementSibling; } diff --git a/test/test_askuserquestion_rendering.py b/test/test_askuserquestion_rendering.py index 3d8a3acc..a77f7f98 100644 --- a/test/test_askuserquestion_rendering.py +++ b/test/test_askuserquestion_rendering.py @@ -2,7 +2,7 @@ """Test cases for AskUserQuestion tool rendering.""" from claude_code_log.html import ( - format_askuserquestion_content, + format_askuserquestion_input, format_askuserquestion_result, ) from claude_code_log.models import ( @@ -52,7 +52,7 @@ def test_format_askuserquestion_multiple_questions(self): ] ) - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Check overall structure assert 'class="askuserquestion-content"' in html @@ -104,7 +104,7 @@ def test_format_askuserquestion_single_question(self): ] ) - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Check structure assert 'class="askuserquestion-content"' in html @@ -132,7 +132,7 @@ def test_format_askuserquestion_multiselect(self): ] ) - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Check multi-select hint assert "(select multiple)" in html @@ -144,7 +144,7 @@ def test_format_askuserquestion_legacy_single_question(self): """Test backwards compatibility with single 'question' key format.""" ask_input = AskUserQuestionInput(question="What is your preference?") - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Should still render the question assert 'class="askuserquestion-content"' in html @@ -162,7 +162,7 @@ def test_format_askuserquestion_no_options(self): ] ) - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Should render without options list assert "Please describe the issue in detail." in html @@ -175,7 +175,7 @@ def test_format_askuserquestion_empty_input(self): """Test AskUserQuestion with empty questions returns 'No question' message.""" ask_input = AskUserQuestionInput() # Empty questions list - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # Should show 'No question' message assert "askuserquestion-content" in html @@ -197,7 +197,7 @@ def test_format_askuserquestion_escapes_html(self): ] ) - html = format_askuserquestion_content(ask_input) + html = format_askuserquestion_input(ask_input) # HTML entities should be escaped assert "<script>" in html diff --git a/test/test_bash_rendering.py b/test/test_bash_rendering.py index 775f0726..318d8199 100644 --- a/test/test_bash_rendering.py +++ b/test/test_bash_rendering.py @@ -337,11 +337,9 @@ def test_bash_ansi_color_rendering(): def test_bash_tool_result_ansi_processing(): """Test that Bash tool results have ANSI codes processed.""" - from claude_code_log.html.tool_formatters import ( - format_tool_result_content, - _looks_like_bash_output, - ) - from claude_code_log.models import ToolResultContent + from claude_code_log.factories.tool_factory import _looks_like_bash_output + from claude_code_log.html.tool_formatters import format_bash_output + from claude_code_log.models import BashOutput # Test the detection function bash_content = "❯ npm run build\n\x1b[32mβœ” Build completed\x1b[0m" @@ -350,12 +348,9 @@ def test_bash_tool_result_ansi_processing(): regular_content = "This is just regular text output" assert not _looks_like_bash_output(regular_content) - # Test tool result processing with ANSI codes - tool_result = ToolResultContent( - type="tool_result", tool_use_id="bash_123", content=bash_content, is_error=False - ) - - html = format_tool_result_content(tool_result) + # Test BashOutput formatting with ANSI codes + bash_output = BashOutput(content=bash_content, has_ansi=True) + html = format_bash_output(bash_output) # Should contain colored output assert 'βœ” Build completed' in html @@ -367,20 +362,14 @@ def test_bash_tool_result_ansi_processing(): def test_bash_tool_result_cursor_stripping(): """Test that cursor movement codes are stripped from Bash tool results.""" - from claude_code_log.html import format_tool_result_content - from claude_code_log.models import ToolResultContent + from claude_code_log.html.tool_formatters import format_bash_output + from claude_code_log.models import BashOutput # Content with cursor movement codes content_with_cursor = "Building...\x1b[1A\x1b[2K\x1b[32mβœ” Done!\x1b[0m" - tool_result = ToolResultContent( - type="tool_result", - tool_use_id="bash_456", - content=content_with_cursor, - is_error=False, - ) - - html = format_tool_result_content(tool_result) + bash_output = BashOutput(content=content_with_cursor, has_ansi=True) + html = format_bash_output(bash_output) # Should have colors but no cursor codes assert 'βœ” Done!' in html diff --git a/test/test_cache.py b/test/test_cache.py index e7b481b8..4cb4a23f 100644 --- a/test/test_cache.py +++ b/test/test_cache.py @@ -19,8 +19,8 @@ UserTranscriptEntry, AssistantTranscriptEntry, SummaryTranscriptEntry, - UserMessage, - AssistantMessage, + UserMessageModel, + AssistantMessageModel, UsageInfo, TextContent, ) @@ -60,7 +60,7 @@ def sample_entries(): uuid="user1", timestamp="2023-01-01T10:00:00Z", type="user", - message=UserMessage( + message=UserMessageModel( role="user", content=[TextContent(type="text", text="Hello")] ), ), @@ -74,7 +74,7 @@ def sample_entries(): uuid="assistant1", timestamp="2023-01-01T10:01:00Z", type="assistant", - message=AssistantMessage( + message=AssistantMessageModel( id="msg1", type="message", role="assistant", @@ -222,7 +222,7 @@ def test_filtered_loading_with_dates(self, cache_manager, temp_project_dir): uuid="user1", timestamp="2023-01-01T10:00:00Z", type="user", - message=UserMessage( + message=UserMessageModel( role="user", content=[TextContent(type="text", text="Early message")], ), @@ -237,7 +237,7 @@ def test_filtered_loading_with_dates(self, cache_manager, temp_project_dir): uuid="user2", timestamp="2023-01-02T10:00:00Z", type="user", - message=UserMessage( + message=UserMessageModel( role="user", content=[TextContent(type="text", text="Later message")], ), diff --git a/test/test_context_command.py b/test/test_context_command.py index 9b893439..6b7b6ab9 100644 --- a/test/test_context_command.py +++ b/test/test_context_command.py @@ -1,7 +1,7 @@ """Tests for /context command output rendering.""" +from claude_code_log.factories import create_transcript_entry from claude_code_log.html.renderer import generate_html -from claude_code_log.parser import parse_transcript_entry def test_context_command_rendering(): @@ -33,7 +33,7 @@ def test_context_command_rendering(): ] # Parse the raw messages into TranscriptEntry objects - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Check that ANSI codes were converted to HTML spans with proper classes @@ -91,7 +91,7 @@ def test_context_command_without_ansi(): ] # Parse the raw messages into TranscriptEntry objects - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Check that text is properly rendered even without ANSI codes @@ -130,7 +130,7 @@ def test_mixed_ansi_and_plain_text(): ] # Parse the raw messages into TranscriptEntry objects - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Check that both plain and colored text are present diff --git a/test/test_date_filtering.py b/test/test_date_filtering.py index 722bd41d..4dc3e448 100644 --- a/test/test_date_filtering.py +++ b/test/test_date_filtering.py @@ -7,7 +7,7 @@ from pathlib import Path from claude_code_log.converter import convert_jsonl_to_html from claude_code_log.converter import filter_messages_by_date -from claude_code_log.parser import parse_transcript_entry +from claude_code_log.factories import create_transcript_entry def create_test_message(timestamp_str: str, text: str) -> dict: @@ -47,7 +47,7 @@ def to_utc_iso(dt: datetime) -> str: ] # Parse dictionaries into TranscriptEntry objects - messages = [parse_transcript_entry(msg_dict) for msg_dict in message_dicts] + messages = [create_transcript_entry(msg_dict) for msg_dict in message_dicts] # Test filtering from yesterday onwards filtered = filter_messages_by_date(messages, "yesterday", None) @@ -78,7 +78,7 @@ def to_utc_iso(dt: datetime) -> str: def test_invalid_date_handling(): """Test handling of invalid date strings.""" messages = [ - parse_transcript_entry( + create_transcript_entry( create_test_message("2025-06-08T12:00:00Z", "Test message") ) ] @@ -163,7 +163,7 @@ def test_natural_language_dates(): """Test various natural language date formats.""" message_dict = create_test_message("2025-06-08T12:00:00Z", "Test message") - messages = [parse_transcript_entry(message_dict)] + messages = [create_transcript_entry(message_dict)] # Test various natural language formats date_formats = ["today", "yesterday", "last week", "3 days ago", "1 week ago"] diff --git a/test/test_exitplanmode_rendering.py b/test/test_exitplanmode_rendering.py index 68d11893..808ac81c 100644 --- a/test/test_exitplanmode_rendering.py +++ b/test/test_exitplanmode_rendering.py @@ -2,7 +2,7 @@ """Test cases for ExitPlanMode tool rendering.""" from claude_code_log.html import ( - format_exitplanmode_content, + format_exitplanmode_input, format_exitplanmode_result, ) from claude_code_log.models import ExitPlanModeInput @@ -17,7 +17,7 @@ def test_format_exitplanmode_with_plan(self): plan="# My Plan\n\n## Overview\n\nThis is the plan.\n\n## Steps\n\n1. First step\n2. Second step" ) - html = format_exitplanmode_content(exit_input) + html = format_exitplanmode_input(exit_input) # Should render as markdown in a plan-content div assert 'class="plan-content' in html @@ -33,7 +33,7 @@ def test_format_exitplanmode_long_plan_collapsible(self): ) exit_input = ExitPlanModeInput(plan=long_plan) - html = format_exitplanmode_content(exit_input) + html = format_exitplanmode_input(exit_input) # Should be collapsible due to length assert "collapsible" in html.lower() or "details" in html.lower() @@ -43,7 +43,7 @@ def test_format_exitplanmode_empty_plan(self): """Test ExitPlanMode with empty plan shows 'No plan' message.""" exit_input = ExitPlanModeInput() # Empty plan - html = format_exitplanmode_content(exit_input) + html = format_exitplanmode_input(exit_input) # Should show 'No plan' message assert "plan-content" in html diff --git a/test/test_hook_summary.py b/test/test_hook_summary.py index a40b265c..2873a740 100644 --- a/test/test_hook_summary.py +++ b/test/test_hook_summary.py @@ -1,7 +1,7 @@ """Tests for hook summary (stop_hook_summary) parsing and rendering.""" +from claude_code_log.factories import create_transcript_entry from claude_code_log.models import SystemTranscriptEntry -from claude_code_log.parser import parse_transcript_entry from claude_code_log.html.renderer import generate_html @@ -30,7 +30,7 @@ def test_parse_hook_summary_without_content(self): "uuid": "test-uuid", } - entry = parse_transcript_entry(data) + entry = create_transcript_entry(data) assert isinstance(entry, SystemTranscriptEntry) assert entry.subtype == "stop_hook_summary" @@ -66,7 +66,7 @@ def test_parse_hook_summary_with_errors(self): "uuid": "test-uuid", } - entry = parse_transcript_entry(data) + entry = create_transcript_entry(data) assert isinstance(entry, SystemTranscriptEntry) assert entry.subtype == "stop_hook_summary" @@ -91,7 +91,7 @@ def test_parse_system_message_with_content_still_works(self): "uuid": "test-uuid", } - entry = parse_transcript_entry(data) + entry = create_transcript_entry(data) assert isinstance(entry, SystemTranscriptEntry) assert entry.content == "init" @@ -124,7 +124,7 @@ def test_silent_hook_success_not_rendered(self): } ] - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Should not contain actual hook content (skipped) @@ -156,7 +156,7 @@ def test_hook_with_errors_rendered(self): } ] - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Should contain hook summary elements @@ -188,7 +188,7 @@ def test_hook_with_output_but_no_errors_rendered(self): } ] - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Should contain hook summary elements @@ -218,7 +218,7 @@ def test_hook_with_ansi_errors_rendered(self): } ] - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # ANSI codes should be converted, not present raw @@ -243,7 +243,7 @@ def test_regular_system_message_still_renders(self): } ] - parsed_messages = [parse_transcript_entry(msg) for msg in messages] + parsed_messages = [create_transcript_entry(msg) for msg in messages] html = generate_html(parsed_messages) # Should render the command name diff --git a/test/test_ide_tags.py b/test/test_ide_tags.py index 49554eef..b542f336 100644 --- a/test/test_ide_tags.py +++ b/test/test_ide_tags.py @@ -1,35 +1,40 @@ """Tests for IDE tag parsing and formatting in user messages. Split into: -- Parsing tests: test parse_ide_notifications() from parser.py +- Parsing tests: test create_ide_notification_content() from factories - Formatting tests: test format_ide_notification_content() from user_formatters.py -- User message tests: test parse_user_message_content() and formatters +- User message tests: test create_user_message() and formatters - Assistant text tests: test format_assistant_text_content() """ -from claude_code_log.parser import parse_ide_notifications, parse_user_message_content +from claude_code_log.factories import ( + create_ide_notification_content, + create_user_message, +) +from claude_code_log.parser import extract_text_content from claude_code_log.html.user_formatters import ( format_ide_notification_content, format_user_text_content, ) from claude_code_log.html.assistant_formatters import format_assistant_text_content from claude_code_log.models import ( - AssistantTextContent, + AssistantTextMessage, IdeNotificationContent, ImageContent, ImageSource, + MessageMeta, TextContent, - UserTextContent, + UserTextMessage, ) # ============================================================================= -# Parsing Tests - parse_ide_notifications() +# Parsing Tests - create_ide_notification_content() # ============================================================================= class TestParseIdeNotifications: - """Tests for parse_ide_notifications() parser function.""" + """Tests for create_ide_notification_content() parser function.""" def test_parse_ide_opened_file_tag(self): """Test that tags are parsed correctly.""" @@ -40,7 +45,7 @@ def test_parse_ide_opened_file_tag(self): "Here is my actual question." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.opened_files) == 1 @@ -55,7 +60,7 @@ def test_parse_multiple_ide_tags(self): "Second file opened." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.opened_files) == 2 @@ -68,7 +73,7 @@ def test_parse_no_ide_tags(self): """Test that messages without IDE tags return None.""" text = "This is a regular user message without any IDE tags." - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is None @@ -81,7 +86,7 @@ def test_parse_multiline_ide_tag(self): "User question follows." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.opened_files) == 1 @@ -100,7 +105,7 @@ def test_parse_ide_diagnostics(self): "Here is my question." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.diagnostics) == 1 @@ -122,7 +127,7 @@ def test_parse_ide_selection_short(self): "Can you explain this?" ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.selections) == 1 @@ -141,7 +146,7 @@ def test_parse_all_ide_tag_types(self): "Please help." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None assert len(result.opened_files) == 1 @@ -170,7 +175,7 @@ def test_format_ide_opened_file_tag(self): "Question here." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -186,7 +191,7 @@ def test_format_multiple_ide_tags(self): "Second file opened." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -197,7 +202,7 @@ def test_format_special_chars_escaped(self): """Test that special HTML characters are escaped in IDE tag content.""" text = 'File with & "characters" in path.' - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -217,7 +222,7 @@ def test_format_ide_diagnostics(self): "Question." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -236,7 +241,7 @@ def test_format_ide_selection_short(self): "Question." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -250,7 +255,7 @@ def test_format_ide_selection_long(self): long_selection = "The user selected lines 1 to 50:\n" + ("line content\n" * 30) text = f"{long_selection}\nQuestion." - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -270,7 +275,7 @@ def test_format_mixed_ide_tags(self): "Please review." ) - result = parse_ide_notifications(text) + result = create_ide_notification_content(text) assert result is not None notifications = format_ide_notification_content(result) @@ -282,12 +287,12 @@ def test_format_mixed_ide_tags(self): # ============================================================================= -# User Message Content Tests - parse_user_message_content() +# User Message Content Tests - create_user_message() # ============================================================================= class TestParseUserMessageContent: - """Tests for parse_user_message_content() function.""" + """Tests for create_user_message() function.""" def test_parse_user_message_with_multi_item_content(self): """Test parsing user message with multiple content items (text + image).""" @@ -309,10 +314,12 @@ def test_parse_user_message_with_multi_item_content(self): image_item, ] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) - # Should return UserTextContent with items - assert isinstance(content_model, UserTextContent) + # Should return UserTextMessage with items + assert isinstance(content_model, UserTextMessage) assert len(content_model.items) == 3 # IDE notification, text, image # First item should be IDE notification @@ -349,8 +356,9 @@ def test_format_user_text_content(self): def test_format_assistant_text_content(self): """Test that assistant text is formatted as markdown.""" - content = AssistantTextContent( - items=[TextContent(type="text", text="**Bold** response")] + content = AssistantTextMessage( + MessageMeta.empty(), + items=[TextContent(type="text", text="**Bold** response")], ) html = format_assistant_text_content(content) diff --git a/test/test_message_filtering.py b/test/test_message_filtering.py index e937fa38..bba56763 100644 --- a/test/test_message_filtering.py +++ b/test/test_message_filtering.py @@ -6,7 +6,7 @@ from pathlib import Path from claude_code_log.converter import load_transcript from claude_code_log.html.renderer import generate_html -from claude_code_log.parser import is_system_message +from claude_code_log.factories import is_system_message def test_caveat_message_filtering(): diff --git a/test/test_phase8_message_variants.py b/test/test_phase8_message_variants.py index 3d4da469..2973364e 100644 --- a/test/test_phase8_message_variants.py +++ b/test/test_phase8_message_variants.py @@ -21,31 +21,14 @@ class TestSlashCommandRendering: def test_slash_command_css_class(self): """Test that isMeta=True user messages get 'slash-command' CSS class.""" - # Parent user message (normal) - parent_message = { - "type": "user", - "timestamp": "2025-06-11T22:45:17.436Z", - "parentUuid": None, - "isSidechain": False, - "isMeta": False, - "userType": "human", - "cwd": "/tmp", - "sessionId": "test_session", - "version": "1.0.0", - "uuid": "parent_001", - "message": { - "role": "user", - "content": [{"type": "text", "text": "/review"}], - }, - } - # Expanded slash command prompt (isMeta=True) + # In real transcripts, slash command expansions are standalone messages slash_command_message = { "type": "user", - "timestamp": "2025-06-11T22:45:17.436Z", # Same timestamp as parent - "parentUuid": "parent_001", + "timestamp": "2025-06-11T22:45:17.436Z", + "parentUuid": None, "isSidechain": False, - "isMeta": True, # This is the key flag + "isMeta": True, # This is the key flag for slash command expanded prompts "userType": "external", "cwd": "/tmp", "sessionId": "test_session", @@ -63,14 +46,13 @@ def test_slash_command_css_class(self): } with tempfile.NamedTemporaryFile(mode="w", suffix=".jsonl", delete=False) as f: - f.write(json.dumps(parent_message) + "\n") f.write(json.dumps(slash_command_message) + "\n") f.flush() test_file_path = Path(f.name) try: messages = load_transcript(test_file_path) - assert len(messages) == 2 + assert len(messages) == 1 html = generate_html(messages, "Test Slash Command") @@ -93,7 +75,12 @@ def test_slash_command_css_class(self): test_file_path.unlink() def test_slash_command_sidechain(self): - """Test slash command in sidechain context.""" + """Test slash command in sidechain context renders correctly. + + Standalone sidechain user messages (not children of Task) are rendered, + since they're not duplicates of Task input prompts. Only the first + UserTextMessage child of a Task tool_result is removed as duplicate. + """ slash_command_sidechain = { "type": "user", "timestamp": "2025-06-11T22:45:17.436Z", @@ -126,12 +113,12 @@ def test_slash_command_sidechain(self): messages = load_transcript(test_file_path) html = generate_html(messages, "Test Sidechain Slash Command") - # Sidechain user messages without tool results are skipped during filtering - # (see _filter_messages in renderer.py). Even with isMeta=True, they don't - # contain tool results so they are not rendered. - assert "Sub-agent Slash Command" not in html, ( - "Sidechain user messages without tool results should be skipped" + # Standalone sidechain user messages (not Task children) are rendered + assert "Sub-agent Slash Command" in html, ( + "Standalone sidechain user messages should be rendered" ) + # Should have sidechain CSS class + assert "sidechain" in html finally: test_file_path.unlink() @@ -344,7 +331,12 @@ def test_tool_result_error_sidechain(self): test_file_path.unlink() def test_user_compacted_sidechain(self): - """Test user message with compacted and sidechain modifiers.""" + """Test user message with compacted and sidechain modifiers renders. + + Standalone sidechain user messages (not children of Task) are rendered, + since they're not duplicates of Task input prompts. Only the first + UserTextMessage child of a Task tool_result is removed as duplicate. + """ # Compacted messages have specific structure compacted_message = { "type": "user", @@ -377,14 +369,13 @@ def test_user_compacted_sidechain(self): messages = load_transcript(test_file_path) html = generate_html(messages, "Test Compacted Sidechain") - # Sidechain user messages are skipped (duplicate of Task prompt input) - # Verify the raw content is not rendered - assert "context-messages" not in html, ( - "Sidechain user messages should be skipped" - ) - assert "[Compacted conversation]" not in html, ( - "Sidechain user message content should not be rendered" + # Standalone sidechain compacted messages are rendered + # They get special formatting as CompactedSummaryMessage + assert "Compacted conversation" in html, ( + "Standalone sidechain compacted messages should be rendered" ) + # Should have sidechain CSS class + assert "sidechain" in html finally: test_file_path.unlink() diff --git a/test/test_sidechain_agents.py b/test/test_sidechain_agents.py index 880cf028..5d15fbd0 100644 --- a/test/test_sidechain_agents.py +++ b/test/test_sidechain_agents.py @@ -197,6 +197,63 @@ def test_sidechain_tool_results_rendered(): ), "Sidechain user prompts should be skipped" +def test_explore_agent_no_user_prompt(): + """Test Explore-type agents that start with assistant (no user prompt). + + Real-world pattern from -src-deep-manifest project: Explore agents don't have + a user prompt message at the start - they begin directly with assistant messages. + This differs from Plan-type agents (like in coderabbit) which start with a user + prompt that duplicates the Task input. + + The cleanup should handle both patterns correctly: + - Plan agents: Remove first UserTextMessage (duplicate prompt) + - Explore agents: No UserTextMessage to remove, just process normally + """ + with tempfile.TemporaryDirectory() as tmpdir: + tmpdir_path = Path(tmpdir) + + # Create main session with Task tool_use (Explore type) + main_file = tmpdir_path / "main.jsonl" + main_file.write_text( + # User request + '{"parentUuid":null,"isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","type":"user","message":{"role":"user","content":[{"type":"text","text":"How does the ZIP processing work?"}]},"uuid":"u-0","timestamp":"2025-01-15T15:24:45.000Z"}\n' + # Assistant invokes Task (Explore subagent) + '{"parentUuid":"u-0","isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","message":{"model":"claude-sonnet-4-5-20250929","id":"msg_main","type":"message","role":"assistant","content":[{"type":"tool_use","id":"task-explore","name":"Task","input":{"prompt":"Explore ZIP processing","subagent_type":"Explore"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":100,"output_tokens":50}},"requestId":"req_main","type":"assistant","uuid":"a-0","timestamp":"2025-01-15T15:24:47.000Z"}\n' + # Task result (triggers agent loading) + '{"parentUuid":"a-0","isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","type":"user","message":{"role":"user","content":[{"type":"tool_result","tool_use_id":"task-explore","content":"ZIP files are processed recursively..."}]},"uuid":"u-1","timestamp":"2025-01-15T15:25:00.000Z","toolUseResult":{"agentId":"explore-agent","content":"ZIP files are processed recursively..."},"agentId":"explore-agent"}\n' + ) + + # Create Explore-type agent file: starts with ASSISTANT (no user prompt) + # This matches real data from -src-deep-manifest/agent-c8d9b115.jsonl + agent_file = tmpdir_path / "agent-explore-agent.jsonl" + agent_file.write_text( + # First message is assistant (NOT user) - Explore agents don't echo the prompt + '{"parentUuid":null,"isSidechain":true,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","agentId":"explore-agent","message":{"model":"claude-sonnet-4-5-20250929","id":"msg_explore_1","type":"message","role":"assistant","content":[{"type":"text","text":"Let me search for ZIP-related code."},{"type":"tool_use","id":"grep-1","name":"Grep","input":{"pattern":"zip"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":50,"output_tokens":30}},"requestId":"req_explore_1","type":"assistant","uuid":"explore-a-0","timestamp":"2025-01-15T15:24:51.000Z"}\n' + # User provides Grep result + '{"parentUuid":"explore-a-0","isSidechain":true,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","agentId":"explore-agent","type":"user","message":{"role":"user","content":[{"type":"tool_result","tool_use_id":"grep-1","content":"deep_manifest.py:42: def process_zip()"}]},"uuid":"explore-u-0","timestamp":"2025-01-15T15:24:52.000Z"}\n' + # Assistant continues with another tool + '{"parentUuid":"explore-u-0","isSidechain":true,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","agentId":"explore-agent","message":{"model":"claude-sonnet-4-5-20250929","id":"msg_explore_2","type":"message","role":"assistant","content":[{"type":"text","text":"Found it. Let me read the file."},{"type":"tool_use","id":"read-1","name":"Read","input":{"file_path":"deep_manifest.py"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":80,"output_tokens":40}},"requestId":"req_explore_2","type":"assistant","uuid":"explore-a-1","timestamp":"2025-01-15T15:24:53.000Z"}\n' + # User provides Read result + '{"parentUuid":"explore-a-1","isSidechain":true,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","agentId":"explore-agent","type":"user","message":{"role":"user","content":[{"type":"tool_result","tool_use_id":"read-1","content":"def process_zip(path):\\n with zipfile.ZipFile(path) as zf:\\n ..."}]},"uuid":"explore-u-1","timestamp":"2025-01-15T15:24:54.000Z"}\n' + # Final assistant summary + '{"parentUuid":"explore-u-1","isSidechain":true,"userType":"external","cwd":"/workspace","sessionId":"test-explore","version":"2.0.46","gitBranch":"main","agentId":"explore-agent","message":{"model":"claude-sonnet-4-5-20250929","id":"msg_explore_final","type":"message","role":"assistant","content":[{"type":"text","text":"ZIP files are processed recursively..."}],"stop_reason":"end_turn","stop_sequence":null,"usage":{"input_tokens":100,"output_tokens":50}},"requestId":"req_explore_final","type":"assistant","uuid":"explore-a-2","timestamp":"2025-01-15T15:24:58.000Z"}\n' + ) + + messages = load_transcript(main_file) + html = generate_html(messages, title="Test Explore Agent") + + # Verify sidechain tool results are rendered (Grep and Read results) + assert "deep_manifest.py:42" in html, "Grep tool result should be rendered" + assert "def process_zip" in html, "Read tool result should be rendered" + + # Verify assistant messages are rendered + assert "Let me search for ZIP-related code" in html + assert "Found it. Let me read the file" in html + + # The final assistant message matches Task result, should become dedup notice + # (but the intermediate messages should still be visible) + + def test_multiple_agent_invocations(): """Test handling of multiple Task invocations in same session.""" with tempfile.TemporaryDirectory() as tmpdir: @@ -231,3 +288,92 @@ def test_multiple_agent_invocations(): html = generate_html(messages, title="Test") assert "First done" in html assert "Second done" in html + + +def test_task_output_structured_content_markdown(): + """Test that Task results with structured content render as markdown. + + Real-world Task tool results have structured content like: + [{"type": "text", "text": "## Summary\n\nMarkdown content..."}] + + This tests that: + 1. Structured content is extracted and parsed into TaskOutput + 2. has_markdown returns True for TaskOutput + 3. HTML output includes task-result class with rendered markdown headers + """ + import json + from claude_code_log.renderer import generate_template_messages + from claude_code_log.models import ToolResultMessage, TaskOutput + + with tempfile.TemporaryDirectory() as tmpdir: + tmpdir_path = Path(tmpdir) + + # Task result with structured content containing markdown + markdown_content = ( + "## Summary\n\n" + "This is the agent's response with **bold** and `code`.\n\n" + "### Section 1\n\n" + "Some content here." + ) + + main_file = tmpdir_path / "main.jsonl" + main_file.write_text( + # User request + '{"parentUuid":null,"isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-md","version":"2.0.55","gitBranch":"main","type":"user","message":{"role":"user","content":[{"type":"text","text":"Explore the codebase"}]},"uuid":"u-0","timestamp":"2025-01-15T15:00:00.000Z"}\n' + # Assistant invokes Task + '{"parentUuid":"u-0","isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-md","version":"2.0.55","gitBranch":"main","message":{"model":"claude-opus-4-5-20251101","id":"msg_main","type":"message","role":"assistant","content":[{"type":"tool_use","id":"task-md","name":"Task","input":{"prompt":"Explore this","subagent_type":"Explore"}}],"stop_reason":"tool_use","stop_sequence":null,"usage":{"input_tokens":100,"output_tokens":50}},"requestId":"req_main","type":"assistant","uuid":"a-0","timestamp":"2025-01-15T15:00:05.000Z"}\n' + # Task result with STRUCTURED content (list of text items) + + '{"parentUuid":"a-0","isSidechain":false,"userType":"external","cwd":"/workspace","sessionId":"test-md","version":"2.0.55","gitBranch":"main","type":"user","message":{"role":"user","content":[{"type":"tool_result","tool_use_id":"task-md","content":' + + json.dumps([{"type": "text", "text": markdown_content}]) + + '}]},"uuid":"u-1","timestamp":"2025-01-15T15:00:30.000Z"}\n' + ) + + messages = load_transcript(main_file) + + # Generate template messages to check TaskOutput parsing + root_messages, _ = generate_template_messages(messages) + + # Flatten to find all messages + def flatten(msg): + yield msg + for child in msg.children: + yield from flatten(child) + + # Find the Task tool result + task_result = None + for root in root_messages: + for msg in flatten(root): + if ( + isinstance(msg.content, ToolResultMessage) + and msg.content.tool_name == "Task" + ): + task_result = msg.content + break + + assert task_result is not None, "Task tool result should be found" + + # Verify it's parsed as TaskOutput (not raw ToolResultContent) + assert isinstance(task_result.output, TaskOutput), ( + f"Expected TaskOutput, got {type(task_result.output).__name__}" + ) + + # Verify has_markdown returns True + assert task_result.has_markdown, "has_markdown should be True for TaskOutput" + + # Verify the result text is preserved + assert "## Summary" in task_result.output.result + assert "### Section 1" in task_result.output.result + + # Generate HTML and verify markdown rendering + html = generate_html(messages, title="Test Markdown") + + # Should have task-result CSS class + assert "task-result" in html, "HTML should include task-result class" + + # Should have rendered markdown headers (h2, h3) + assert "

" in html, "Markdown ## should render as

" + assert "

" in html, "Markdown ### should render as

" + + # Should include the rendered content + assert "bold" in html or "bold" in html + assert "code" in html diff --git a/test/test_template_data.py b/test/test_template_data.py index d509495c..c748fe8c 100644 --- a/test/test_template_data.py +++ b/test/test_template_data.py @@ -7,10 +7,19 @@ from claude_code_log.converter import load_transcript, load_directory_transcripts from claude_code_log.html.renderer import generate_html, generate_projects_index_html from claude_code_log.renderer import ( + Renderer, TemplateMessage, TemplateProject, TemplateSummary, ) +from claude_code_log.models import ( + MessageMeta, + UserTextMessage, + AssistantTextMessage, + SessionHeaderMessage, + ToolUseMessage, + ToolUseContent, +) class TestTemplateMessage: @@ -18,32 +27,40 @@ class TestTemplateMessage: def test_template_message_creation(self): """Test creating a TemplateMessage with all fields.""" - msg = TemplateMessage( - message_type="user", - formatted_timestamp="2025-06-14 10:00:00", - raw_timestamp=None, + meta = MessageMeta( + session_id="test-session", + timestamp="2025-06-14T10:00:00Z", + uuid="test-uuid", ) + content = UserTextMessage(meta=meta) + msg = TemplateMessage(content) + renderer = Renderer() assert msg.type == "user" - assert msg.formatted_timestamp == "2025-06-14 10:00:00" - assert msg.message_title == "User" - - def test_template_message_title_capitalization(self): - """Test that message_title properly capitalizes message types.""" - test_cases = [ - ("user", "User"), - ("assistant", "Assistant"), - ("system", "System"), - ("summary", "Summary"), - ] - - for msg_type, expected_display in test_cases: - msg = TemplateMessage( - message_type=msg_type, - formatted_timestamp="time", - raw_timestamp=None, - ) - assert msg.message_title == expected_display + assert msg.meta.timestamp == "2025-06-14T10:00:00Z" + assert renderer.title_content(msg) == "User" + + def test_template_message_title_generation(self): + """Test that Renderer.title_content generates correct titles.""" + meta = MessageMeta.empty() + renderer = Renderer() + + # Test UserTextMessage + user_content = UserTextMessage(meta=meta) + user_msg = TemplateMessage(user_content) + assert renderer.title_content(user_msg) == "User" + + # Test AssistantTextMessage + assistant_content = AssistantTextMessage(meta=meta) + assistant_msg = TemplateMessage(assistant_content) + assert renderer.title_content(assistant_msg) == "Assistant" + + # Test SessionHeaderMessage - fallback to type-based title + session_content = SessionHeaderMessage( + meta=meta, title="Test Session", session_id="test-id" + ) + session_msg = TemplateMessage(session_content) + assert renderer.title_content(session_msg) == "Session Header" class TestTemplateProject: @@ -369,121 +386,59 @@ def test_malformed_message_handling(self): class TestTemplateMessageTree: - """Test TemplateMessage tree building and flatten functionality.""" + """Test TemplateMessage tree building.""" + + _message_counter = 0 + + def setup_method(self): + """Reset counter before each test to avoid order-dependent tests.""" + TestTemplateMessageTree._message_counter = 0 def _create_message( - self, msg_type: str, msg_id: str | None = None, ancestry: list | None = None + self, + msg_type: str, + msg_id: str | None = None, + ancestry: list[int] | None = None, ) -> TemplateMessage: """Helper to create a minimal TemplateMessage for testing.""" - msg = TemplateMessage( - message_type=msg_type, - formatted_timestamp="2025-06-14 10:00:00", - raw_timestamp=None, + # Parse int message_index from string if provided (e.g., "d-0" -> 0) + if msg_id and msg_id.startswith("d-"): + int_msg_index = int(msg_id[2:]) + else: + int_msg_index = TestTemplateMessageTree._message_counter + TestTemplateMessageTree._message_counter += 1 + + meta = MessageMeta( + session_id="test-session", + timestamp="2025-06-14T10:00:00Z", + uuid=msg_id or "test-uuid", ) - if msg_id: - msg.message_id = msg_id - if ancestry: - msg.ancestry = ancestry - return msg - - def test_flatten_single_message(self): - """Test flattening a single message with no children.""" - msg = self._create_message("user", "m1", []) - - result = msg.flatten() - - assert len(result) == 1 - assert result[0] is msg - - def test_flatten_with_children(self): - """Test flattening a message with children.""" - parent = self._create_message("user", "m1", []) - child1 = self._create_message("assistant", "m2", ["m1"]) - child2 = self._create_message("tool_use", "m3", ["m1"]) - - parent.children = [child1, child2] - - result = parent.flatten() - - assert len(result) == 3 - assert result[0] is parent - assert result[1] is child1 - assert result[2] is child2 - - def test_flatten_nested_children(self): - """Test flattening with nested children (depth-first order).""" - root = self._create_message("user", "m1", []) - child = self._create_message("assistant", "m2", ["m1"]) - grandchild = self._create_message("tool_use", "m3", ["m1", "m2"]) - - child.children = [grandchild] - root.children = [child] - - result = root.flatten() - - assert len(result) == 3 - # Depth-first order: root, child, grandchild - assert result[0] is root - assert result[1] is child - assert result[2] is grandchild - - def test_flatten_multiple_branches(self): - """Test flattening with multiple branches (depth-first order).""" - root = self._create_message("user", "m1", []) - branch1 = self._create_message("assistant", "m2", ["m1"]) - branch2 = self._create_message("assistant", "m3", ["m1"]) - leaf1 = self._create_message("tool_use", "m4", ["m1", "m2"]) - leaf2 = self._create_message("tool_use", "m5", ["m1", "m3"]) - - branch1.children = [leaf1] - branch2.children = [leaf2] - root.children = [branch1, branch2] - - result = root.flatten() - - # Depth-first: root -> branch1 -> leaf1 -> branch2 -> leaf2 - assert len(result) == 5 - assert result[0] is root - assert result[1] is branch1 - assert result[2] is leaf1 - assert result[3] is branch2 - assert result[4] is leaf2 - def test_flatten_all_single_root(self): - """Test flatten_all with a single root message.""" - root = self._create_message("user", "m1", []) - child = self._create_message("assistant", "m2", ["m1"]) - root.children = [child] - - result = TemplateMessage.flatten_all([root]) - - assert len(result) == 2 - assert result[0] is root - assert result[1] is child - - def test_flatten_all_multiple_roots(self): - """Test flatten_all with multiple root messages.""" - root1 = self._create_message("user", "m1", []) - child1 = self._create_message("assistant", "m2", ["m1"]) - root1.children = [child1] - - root2 = self._create_message("user", "m3", []) - child2 = self._create_message("assistant", "m4", ["m3"]) - root2.children = [child2] - - result = TemplateMessage.flatten_all([root1, root2]) - - assert len(result) == 4 - assert result[0] is root1 - assert result[1] is child1 - assert result[2] is root2 - assert result[3] is child2 - - def test_flatten_all_empty_list(self): - """Test flatten_all with an empty list.""" - result = TemplateMessage.flatten_all([]) + # Create appropriate content based on message type + if msg_type == "user": + content = UserTextMessage(meta=meta) + elif msg_type == "assistant": + content = AssistantTextMessage(meta=meta) + elif msg_type == "tool_use": + content = ToolUseMessage( + meta=meta, + input=ToolUseContent( + type="tool_use", id="test-id", name="TestTool", input={} + ), + tool_use_id="test-id", + tool_name="TestTool", + ) + elif msg_type == "session": + content = SessionHeaderMessage( + meta=meta, title="Test Session", session_id="test-session" + ) + else: + # Fallback to UserTextMessage for unknown types + content = UserTextMessage(meta=meta) - assert result == [] + msg = TemplateMessage(content, ancestry=ancestry) + msg.message_index = int_msg_index # Set message_index on TemplateMessage + return msg def test_children_field_default_empty(self): """Test that children field defaults to empty list.""" @@ -491,21 +446,6 @@ def test_children_field_default_empty(self): assert msg.children == [] - def test_flatten_preserves_order(self): - """Test that flatten preserves insertion order of children.""" - root = self._create_message("user", "m1", []) - children = [ - self._create_message("assistant", f"m{i}", ["m1"]) for i in range(2, 7) - ] - root.children = children - - result = root.flatten() - - # First element is root, rest are children in order - assert result[0] is root - for i, child in enumerate(children): - assert result[i + 1] is child - class TestTreeBuildingIntegration: """Integration tests for tree building with real transcript data.""" @@ -524,54 +464,6 @@ def test_tree_built_from_representative_messages(self): # _build_message_tree is private. This test just verifies the # tree building doesn't break normal HTML generation. - def test_flatten_roundtrip_preserves_count(self): - """Test that flatten of built tree gives same count as input.""" - # Create a manual tree and verify flatten returns all messages - root = TemplateMessage( - message_type="session", - formatted_timestamp="2025-06-14 10:00:00", - raw_timestamp=None, - ) - root.message_id = "session-1" - root.ancestry = [] - - user = TemplateMessage( - message_type="user", - formatted_timestamp="2025-06-14 10:00:01", - raw_timestamp=None, - ) - user.message_id = "d-1" - user.ancestry = ["session-1"] - - assistant = TemplateMessage( - message_type="assistant", - formatted_timestamp="2025-06-14 10:00:02", - raw_timestamp=None, - ) - assistant.message_id = "d-2" - assistant.ancestry = ["session-1", "d-1"] - - tool = TemplateMessage( - message_type="tool_use", - formatted_timestamp="2025-06-14 10:00:03", - raw_timestamp=None, - ) - tool.message_id = "d-3" - tool.ancestry = ["session-1", "d-1", "d-2"] - - # Build tree manually - assistant.children = [tool] - user.children = [assistant] - root.children = [user] - - # Flatten and verify - flat = TemplateMessage.flatten_all([root]) - assert len(flat) == 4 - assert flat[0].message_id == "session-1" - assert flat[1].message_id == "d-1" - assert flat[2].message_id == "d-2" - assert flat[3].message_id == "d-3" - if __name__ == "__main__": pytest.main([__file__]) diff --git a/test/test_template_utils.py b/test/test_template_utils.py index 196510d2..9c44137a 100644 --- a/test/test_template_utils.py +++ b/test/test_template_utils.py @@ -4,10 +4,15 @@ import pytest from datetime import datetime from claude_code_log.parser import parse_timestamp, extract_text_content -from claude_code_log.html import parse_slash_command +from claude_code_log.factories import create_slash_command_message from claude_code_log.html import escape_html from claude_code_log.utils import format_timestamp -from claude_code_log.models import TextContent, ToolUseContent, ToolResultContent +from claude_code_log.models import ( + MessageMeta, + TextContent, + ToolUseContent, + ToolResultContent, +) class TestTimestampHandling: @@ -97,41 +102,41 @@ def test_extract_text_content_no_text_items(self): class TestCommandExtraction: """Test command information extraction from system messages.""" - def test_parse_slash_command_complete(self): + def test_create_slash_command_message_complete(self): """Test parsing complete slash command information.""" text = 'Testing...\ntest-cmd\n--verbose\n{"type": "text", "text": "Test content"}' - result = parse_slash_command(text) + result = create_slash_command_message(MessageMeta.empty(), text) assert result is not None assert result.command_name == "test-cmd" assert result.command_args == "--verbose" assert result.command_contents == "Test content" - def test_parse_slash_command_missing_parts(self): + def test_create_slash_command_message_missing_parts(self): """Test parsing slash command with missing parts.""" text = "minimal-cmd" - result = parse_slash_command(text) + result = create_slash_command_message(MessageMeta.empty(), text) assert result is not None assert result.command_name == "minimal-cmd" assert result.command_args == "" assert result.command_contents == "" - def test_parse_slash_command_no_command(self): + def test_create_slash_command_message_no_command(self): """Test parsing text without command tags returns None.""" text = "This is just regular text with no command tags" - result = parse_slash_command(text) + result = create_slash_command_message(MessageMeta.empty(), text) assert result is None # No command-name tag found - def test_parse_slash_command_malformed_json(self): + def test_create_slash_command_message_malformed_json(self): """Test parsing command contents with malformed JSON.""" text = 'bad-json\n{"invalid": json' - result = parse_slash_command(text) + result = create_slash_command_message(MessageMeta.empty(), text) assert result is not None assert result.command_name == "bad-json" @@ -181,9 +186,9 @@ def test_extract_text_content_none(self): result = extract_text_content(None) assert result == "" - def test_parse_slash_command_empty_string(self): + def test_create_slash_command_message_empty_string(self): """Test parsing slash command from empty string returns None.""" - result = parse_slash_command("") + result = create_slash_command_message(MessageMeta.empty(), "") assert result is None # No command-name tag found diff --git a/test/test_todowrite_rendering.py b/test/test_todowrite_rendering.py index 3973269e..e3577e07 100644 --- a/test/test_todowrite_rendering.py +++ b/test/test_todowrite_rendering.py @@ -6,8 +6,15 @@ from pathlib import Path import pytest from claude_code_log.converter import convert_jsonl_to_html -from claude_code_log.html import format_todowrite_content, format_tool_use_content -from claude_code_log.models import TodoWriteInput, TodoWriteItem, ToolUseContent +from claude_code_log.html import format_todowrite_input +from claude_code_log.html.renderer import HtmlRenderer +from claude_code_log.models import ( + EditInput, + MessageMeta, + TodoWriteInput, + TodoWriteItem, + ToolUseMessage, +) class TestTodoWriteRendering: @@ -38,7 +45,7 @@ def test_format_todowrite_basic(self): ] ) - html = format_todowrite_content(todo_input) + html = format_todowrite_input(todo_input) # Check overall structure (TodoWrite now has streamlined format) assert 'class="todo-list"' in html @@ -69,7 +76,7 @@ def test_format_todowrite_empty(self): """Test TodoWrite formatting with no todos.""" todo_input = TodoWriteInput(todos=[]) - html = format_todowrite_content(todo_input) + html = format_todowrite_input(todo_input) assert 'class="todo-content"' in html # Title and ID are now in the message header, not in content @@ -88,7 +95,7 @@ def test_format_todowrite_html_escaping(self): ] ) - html = format_todowrite_content(todo_input) + html = format_todowrite_input(todo_input) # Check that HTML is escaped assert "<script>" in html @@ -110,7 +117,7 @@ def test_format_todowrite_invalid_status_priority(self): ] ) - html = format_todowrite_content(todo_input) + html = format_todowrite_input(todo_input) # Should use default emojis for unknown values assert "⏳" in html # default status emoji @@ -194,33 +201,37 @@ def test_todowrite_vs_regular_tool_use(self): "This is a very long content that should definitely exceed 200 characters so that we can test the collapsible details functionality properly. " * 3 ) - regular_tool = ToolUseContent( - type="tool_use", - id="toolu_regular", - name="Edit", - input={"file_path": "/tmp/test.py", "content": long_content}, + regular_tool = ToolUseMessage( + MessageMeta.empty(), + input=EditInput( + file_path="/tmp/test.py", + old_string="", + new_string=long_content, + ), + tool_use_id="toolu_regular", + tool_name="Edit", ) # Create TodoWrite tool use - todowrite_tool = ToolUseContent( - type="tool_use", - id="toolu_todowrite", - name="TodoWrite", - input={ - "todos": [ - { - "id": "1", - "content": "Test todo", - "status": "pending", - "priority": "medium", - } + todowrite_tool = ToolUseMessage( + MessageMeta.empty(), + input=TodoWriteInput( + todos=[ + TodoWriteItem( + content="Test todo", + status="pending", + activeForm="Testing todo", + ) ] - }, + ), + tool_use_id="toolu_todowrite", + tool_name="TodoWrite", ) - # Test both through the main format function - regular_html = format_tool_use_content(regular_tool) - todowrite_html = format_tool_use_content(todowrite_tool) + # Test both through the HtmlRenderer + renderer = HtmlRenderer() + regular_html = renderer.format_ToolUseMessage(regular_tool) + todowrite_html = renderer.format_ToolUseMessage(todowrite_tool) # Edit tool should use diff formatting (not table) assert "edit-diff" in regular_html diff --git a/test/test_toggle_functionality.py b/test/test_toggle_functionality.py index 9ca59788..7f239e12 100644 --- a/test/test_toggle_functionality.py +++ b/test/test_toggle_functionality.py @@ -1,12 +1,12 @@ """Tests for the collapsible details toggle functionality.""" from typing import Any, Dict, List +from claude_code_log.factories import create_content_item from claude_code_log.models import ( AssistantTranscriptEntry, - AssistantMessage, + AssistantMessageModel, UsageInfo, ) -from claude_code_log.parser import parse_content_item from claude_code_log.html.renderer import generate_html @@ -17,11 +17,11 @@ def _create_assistant_message( self, content_items: List[Dict[str, Any]] ) -> AssistantTranscriptEntry: """Helper to create a properly structured AssistantTranscriptEntry.""" - # Convert raw content items to proper ContentItem objects - parsed_content = [parse_content_item(item) for item in content_items] + # Convert raw content items to proper ContentItem objects (no filter = all types) + parsed_content = [create_content_item(item) for item in content_items] - # Create AssistantMessage with proper types - message = AssistantMessage( + # Create AssistantMessageModel with proper types + message = AssistantMessageModel( id="msg_001", type="message", role="assistant", diff --git a/test/test_tool_result_image_rendering.py b/test/test_tool_result_image_rendering.py index 961de78a..b3b9b999 100644 --- a/test/test_tool_result_image_rendering.py +++ b/test/test_tool_result_image_rendering.py @@ -1,6 +1,6 @@ """Test image rendering within tool results.""" -from claude_code_log.html import format_tool_result_content +from claude_code_log.html.tool_formatters import format_tool_result_content_raw from claude_code_log.models import ToolResultContent @@ -27,7 +27,7 @@ def test_tool_result_with_image(): is_error=False, ) - html = format_tool_result_content(tool_result) + html = format_tool_result_content_raw(tool_result) # Should be collapsible when images are present assert '
' in html @@ -66,7 +66,7 @@ def test_tool_result_with_only_image(): is_error=False, ) - html = format_tool_result_content(tool_result) + html = format_tool_result_content_raw(tool_result) # Should be collapsible assert '
' in html @@ -106,7 +106,7 @@ def test_tool_result_with_multiple_images(): is_error=False, ) - html = format_tool_result_content(tool_result) + html = format_tool_result_content_raw(tool_result) # Should contain both images assert html.count("' not in html @@ -145,7 +145,7 @@ def test_tool_result_structured_text_only(): is_error=False, ) - html = format_tool_result_content(tool_result) + html = format_tool_result_content_raw(tool_result) # Should contain both text lines assert "First line" in html diff --git a/test/test_user_renderer.py b/test/test_user_renderer.py index a301a685..01bd03b7 100644 --- a/test/test_user_renderer.py +++ b/test/test_user_renderer.py @@ -1,8 +1,8 @@ """Tests for user message parsing and rendering. Split into: -- Parsing tests: test parse_compacted_summary(), parse_user_memory() -- Content model tests: test parse_user_message_content() +- Parsing tests: test create_compacted_summary_message(), create_user_memory_message() +- Content model tests: test create_user_message() - HTML rendering tests: test full pipeline from JSONL to HTML """ @@ -19,28 +19,31 @@ format_user_text_model_content, ) from claude_code_log.models import ( - CompactedSummaryContent, + CompactedSummaryMessage, + ContentItem, + MessageMeta, TextContent, - UserMemoryContent, - UserTextContent, + UserMemoryMessage, + UserTextMessage, ) -from claude_code_log.parser import ( +from claude_code_log.factories import ( COMPACTED_SUMMARY_PREFIX, - parse_compacted_summary, - parse_user_memory, - parse_user_message_content, + create_compacted_summary_message, + create_user_memory_message, + create_user_message, ) +from claude_code_log.parser import extract_text_content # ============================================================================= -# Parsing Tests - parse_compacted_summary() +# Parsing Tests - create_compacted_summary_message() # ============================================================================= -class TestParseCompactedSummary: - """Tests for parse_compacted_summary() parser function (takes content list).""" +class TestCreateCompactedSummaryMessage: + """Tests for create_compacted_summary_message() factory function (takes content list).""" - def test_parse_compacted_summary_detected(self): + def test_create_compacted_summary_message_detected(self): """Test that compacted summary is detected and content combined.""" text = ( f"{COMPACTED_SUMMARY_PREFIX}. The conversation is summarized below:\n" @@ -48,27 +51,27 @@ def test_parse_compacted_summary_detected(self): ) content_list = [TextContent(type="text", text=text)] - result = parse_compacted_summary(content_list) + result = create_compacted_summary_message(MessageMeta.empty(), content_list) assert result is not None - assert isinstance(result, CompactedSummaryContent) + assert isinstance(result, CompactedSummaryMessage) assert result.summary_text == text - def test_parse_compacted_summary_not_detected(self): + def test_create_compacted_summary_message_not_detected(self): """Test that regular text is not detected as compacted summary.""" text = "This is a regular user message." content_list = [TextContent(type="text", text=text)] - result = parse_compacted_summary(content_list) + result = create_compacted_summary_message(MessageMeta.empty(), content_list) assert result is None - def test_parse_compacted_summary_empty_list(self): + def test_create_compacted_summary_message_empty_list(self): """Test that empty content list returns None.""" - result = parse_compacted_summary([]) + result = create_compacted_summary_message(MessageMeta.empty(), []) assert result is None - def test_parse_compacted_summary_combines_multiple_texts(self): + def test_create_compacted_summary_message_combines_multiple_texts(self): """Test that multiple text items are combined with double newlines.""" first_text = f"{COMPACTED_SUMMARY_PREFIX}. Part 1." second_text = "Part 2." @@ -79,7 +82,7 @@ def test_parse_compacted_summary_combines_multiple_texts(self): TextContent(type="text", text=third_text), ] - result = parse_compacted_summary(content_list) + result = create_compacted_summary_message(MessageMeta.empty(), content_list) assert result is not None expected = "\n\n".join([first_text, second_text, third_text]) @@ -87,77 +90,79 @@ def test_parse_compacted_summary_combines_multiple_texts(self): # ============================================================================= -# Parsing Tests - parse_user_memory() +# Parsing Tests - create_user_memory_message() # ============================================================================= class TestParseUserMemory: - """Tests for parse_user_memory() parser function.""" + """Tests for create_user_memory_message() parser function.""" - def test_parse_user_memory_detected(self): + def test_create_user_memory_message_detected(self): """Test that user memory input tag is detected correctly.""" text = "Memory content from CLAUDE.md" - result = parse_user_memory(text) + result = create_user_memory_message(MessageMeta.empty(), text) assert result is not None - assert isinstance(result, UserMemoryContent) + assert isinstance(result, UserMemoryMessage) assert result.memory_text == "Memory content from CLAUDE.md" - def test_parse_user_memory_with_surrounding_text(self): + def test_create_user_memory_message_with_surrounding_text(self): """Test memory tag extraction from mixed content.""" text = "Some prefix The actual memory suffix" - result = parse_user_memory(text) + result = create_user_memory_message(MessageMeta.empty(), text) assert result is not None assert result.memory_text == "The actual memory" - def test_parse_user_memory_multiline(self): + def test_create_user_memory_message_multiline(self): """Test multiline memory content.""" memory_content = "Line 1\nLine 2\nLine 3" text = f"{memory_content}" - result = parse_user_memory(text) + result = create_user_memory_message(MessageMeta.empty(), text) assert result is not None assert result.memory_text == memory_content - def test_parse_user_memory_not_detected(self): + def test_create_user_memory_message_not_detected(self): """Test that regular text without tag returns None.""" text = "Regular text without memory tag." - result = parse_user_memory(text) + result = create_user_memory_message(MessageMeta.empty(), text) assert result is None - def test_parse_user_memory_strips_whitespace(self): + def test_create_user_memory_message_strips_whitespace(self): """Test that memory content whitespace is stripped.""" text = " \n Content with spaces \n " - result = parse_user_memory(text) + result = create_user_memory_message(MessageMeta.empty(), text) assert result is not None assert result.memory_text == "Content with spaces" # ============================================================================= -# Content Model Tests - parse_user_message_content() +# Content Model Tests - create_user_message() # ============================================================================= class TestParseUserMessageContentCompacted: - """Tests for parse_user_message_content() handling compacted summaries.""" + """Tests for create_user_message() handling compacted summaries.""" def test_compacted_summary_single_text_item(self): """Test compacted summary with single text content item.""" text = f"{COMPACTED_SUMMARY_PREFIX}. The conversation summary." content_list = [TextContent(type="text", text=text)] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) assert content_model is not None - assert isinstance(content_model, CompactedSummaryContent) + assert isinstance(content_model, CompactedSummaryMessage) assert content_model.summary_text == text def test_compacted_summary_multiple_text_items(self): @@ -171,51 +176,59 @@ def test_compacted_summary_multiple_text_items(self): TextContent(type="text", text=third_text), ] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) assert content_model is not None - assert isinstance(content_model, CompactedSummaryContent) + assert isinstance(content_model, CompactedSummaryMessage) # All text items should be combined with double newlines expected = "\n\n".join([first_text, second_text, third_text]) assert content_model.summary_text == expected class TestParseUserMessageContentMemory: - """Tests for parse_user_message_content() handling user memory.""" + """Tests for create_user_message() handling user memory.""" def test_user_memory_detected(self): """Test user memory content is detected and returned.""" text = "CLAUDE.md content here" content_list = [TextContent(type="text", text=text)] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) assert content_model is not None - assert isinstance(content_model, UserMemoryContent) + assert isinstance(content_model, UserMemoryMessage) assert content_model.memory_text == "CLAUDE.md content here" class TestParseUserMessageContentRegular: - """Tests for parse_user_message_content() handling regular user text.""" + """Tests for create_user_message() handling regular user text.""" def test_regular_text(self): """Test regular user text without special markers.""" text = "Hello, please help me with this code." content_list = [TextContent(type="text", text=text)] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) assert content_model is not None - assert isinstance(content_model, UserTextContent) + assert isinstance(content_model, UserTextMessage) assert len(content_model.items) == 1 assert isinstance(content_model.items[0], TextContent) assert content_model.items[0].text == text def test_empty_content_list(self): """Test empty content list returns None.""" - content_list = [] + content_list: list[ContentItem] = [] - content_model = parse_user_message_content(content_list) + content_model = create_user_message( + MessageMeta.empty(), content_list, extract_text_content(content_list) + ) assert content_model is None @@ -225,12 +238,14 @@ def test_empty_content_list(self): # ============================================================================= -class TestFormatCompactedSummaryContent: +class TestFormatCompactedSummaryMessage: """Tests for format_compacted_summary_content() formatter function.""" def test_format_compacted_summary_basic(self): """Test basic compacted summary formatting.""" - content = CompactedSummaryContent(summary_text="Summary:\n- Point 1\n- Point 2") + content = CompactedSummaryMessage( + MessageMeta.empty(), summary_text="Summary:\n- Point 1\n- Point 2" + ) html = format_compacted_summary_content(content) @@ -243,7 +258,9 @@ def test_format_compacted_summary_collapsible(self): """Test that long compacted summaries are collapsible.""" # Create long content that exceeds threshold long_summary = "Summary:\n" + "\n".join([f"- Point {i}" for i in range(50)]) - content = CompactedSummaryContent(summary_text=long_summary) + content = CompactedSummaryMessage( + MessageMeta.empty(), summary_text=long_summary + ) html = format_compacted_summary_content(content) @@ -257,12 +274,14 @@ def test_format_compacted_summary_collapsible(self): # ============================================================================= -class TestFormatUserMemoryContent: +class TestFormatUserMemoryMessage: """Tests for format_user_memory_content() formatter function.""" def test_format_user_memory_basic(self): """Test basic user memory formatting.""" - content = UserMemoryContent(memory_text="CLAUDE.md content") + content = UserMemoryMessage( + MessageMeta.empty(), memory_text="CLAUDE.md content" + ) html = format_user_memory_content(content) @@ -272,7 +291,9 @@ def test_format_user_memory_basic(self): def test_format_user_memory_escapes_html(self): """Test that HTML characters are escaped.""" - content = UserMemoryContent(memory_text="") + content = UserMemoryMessage( + MessageMeta.empty(), memory_text="" + ) html = format_user_memory_content(content) @@ -290,8 +311,9 @@ class TestFormatUserTextModelContent: def test_format_user_text_basic(self): """Test basic user text formatting.""" - content = UserTextContent( - items=[TextContent(type="text", text="User question here")] + content = UserTextMessage( + MessageMeta.empty(), + items=[TextContent(type="text", text="User question here")], ) html = format_user_text_model_content(content) @@ -301,8 +323,9 @@ def test_format_user_text_basic(self): def test_format_user_text_escapes_html(self): """Test that HTML characters are escaped.""" - content = UserTextContent( - items=[TextContent(type="text", text='Test bold & "quotes"')] + content = UserTextMessage( + MessageMeta.empty(), + items=[TextContent(type="text", text='Test bold & "quotes"')], ) html = format_user_text_model_content(content) diff --git a/test/test_utils.py b/test/test_utils.py index 642390e8..ebfb6d4f 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -2,13 +2,12 @@ """Test cases for the utils module functions.""" import pytest -from claude_code_log.parser import ( +from claude_code_log.factories import ( is_bash_input, is_bash_output, is_command_message, is_local_command_output, is_system_message, - is_warmup_only_session, ) from claude_code_log.utils import ( should_skip_message, @@ -23,9 +22,7 @@ TextContent, ToolUseContent, UserTranscriptEntry, - UserMessage, - AssistantTranscriptEntry, - AssistantMessage, + UserMessageModel, ) @@ -606,139 +603,6 @@ def test_create_session_preview_multiple_ide_tags(self): assert "Please review this code for bugs" in preview -class TestWarmupOnlySessionDetection: - """Test detection of warmup-only sessions.""" - - def _create_user_entry( - self, session_id: str, content: str, uuid: str, timestamp: str - ) -> UserTranscriptEntry: - """Helper to create a UserTranscriptEntry with all required fields.""" - return UserTranscriptEntry( - type="user", - sessionId=session_id, - parentUuid=None, - isSidechain=False, - userType="external", - cwd="/test", - version="1.0.0", - message=UserMessage( - role="user", content=[TextContent(type="text", text=content)] - ), - uuid=uuid, - timestamp=timestamp, - ) - - def _create_assistant_entry( - self, - session_id: str, - content: str, - uuid: str, - timestamp: str, - parent_uuid: str, - ) -> AssistantTranscriptEntry: - """Helper to create an AssistantTranscriptEntry with all required fields.""" - return AssistantTranscriptEntry( - type="assistant", - sessionId=session_id, - parentUuid=parent_uuid, - isSidechain=False, - userType="external", - cwd="/test", - version="1.0.0", - message=AssistantMessage( - id="msg-id", - type="message", - role="assistant", - model="claude-3-5-sonnet", - content=[TextContent(type="text", text=content)], - ), - uuid=uuid, - timestamp=timestamp, - ) - - def test_session_with_only_warmup_messages(self): - """Test that a session with only warmup messages is detected.""" - session_id = "test-session-1" - messages = [ - self._create_user_entry( - session_id, "Warmup", "msg-1", "2025-01-01T10:00:00Z" - ), - self._create_assistant_entry( - session_id, - "I'm ready to help!", - "msg-2", - "2025-01-01T10:00:01Z", - "msg-1", - ), - ] - - assert is_warmup_only_session(messages, session_id) is True - - def test_session_with_real_messages(self): - """Test that a session with real messages is not detected as warmup-only.""" - session_id = "test-session-2" - messages = [ - self._create_user_entry( - session_id, "Hello, can you help me?", "msg-1", "2025-01-01T10:00:00Z" - ), - self._create_assistant_entry( - session_id, "Sure!", "msg-2", "2025-01-01T10:00:01Z", "msg-1" - ), - ] - - assert is_warmup_only_session(messages, session_id) is False - - def test_session_with_warmup_and_real_messages(self): - """Test that a session with both warmup and real messages is not warmup-only.""" - session_id = "test-session-3" - messages = [ - self._create_user_entry( - session_id, "Warmup", "msg-1", "2025-01-01T10:00:00Z" - ), - self._create_assistant_entry( - session_id, "Ready!", "msg-2", "2025-01-01T10:00:01Z", "msg-1" - ), - self._create_user_entry( - session_id, - "Now help me debug this code", - "msg-3", - "2025-01-01T10:00:02Z", - ), - ] - - assert is_warmup_only_session(messages, session_id) is False - - def test_session_with_multiple_warmup_messages(self): - """Test session with multiple warmup messages.""" - session_id = "test-session-4" - messages = [ - self._create_user_entry( - session_id, " Warmup ", "msg-1", "2025-01-01T10:00:00Z" - ), - self._create_user_entry( - session_id, "Warmup", "msg-2", "2025-01-01T10:00:01Z" - ), - ] - - assert is_warmup_only_session(messages, session_id) is True - - def test_nonexistent_session(self): - """Test checking a session ID that doesn't exist.""" - messages = [ - self._create_user_entry( - "different-session", "Hello", "msg-1", "2025-01-01T10:00:00Z" - ), - ] - - # Should return False (no user messages may mean system messages exist) - assert is_warmup_only_session(messages, "nonexistent-session") is False - - def test_empty_messages_list(self): - """Test with empty messages list.""" - # Should return False (no user messages may mean system messages exist) - assert is_warmup_only_session([], "any-session") is False - - class TestGetWarmupSessionIds: """Test bulk warmup session ID detection.""" @@ -754,7 +618,7 @@ def _create_user_entry( userType="external", cwd="/test", version="1.0.0", - message=UserMessage( + message=UserMessageModel( role="user", content=[TextContent(type="text", text=content)] ), uuid=uuid, diff --git a/test/test_version_deduplication.py b/test/test_version_deduplication.py index 0675fa85..6888683e 100644 --- a/test/test_version_deduplication.py +++ b/test/test_version_deduplication.py @@ -4,9 +4,9 @@ from datetime import datetime from claude_code_log.models import ( AssistantTranscriptEntry, - AssistantMessage, + AssistantMessageModel, UserTranscriptEntry, - UserMessage, + UserMessageModel, ToolUseContent, ToolResultContent, ) @@ -32,7 +32,7 @@ def test_assistant_message_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=AssistantMessage( + message=AssistantMessageModel( id="msg_duplicate", type="message", role="assistant", @@ -63,7 +63,7 @@ def test_assistant_message_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=AssistantMessage( + message=AssistantMessageModel( id="msg_duplicate", # SAME message.id type="message", role="assistant", @@ -112,7 +112,7 @@ def test_tool_result_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ ToolResultContent( @@ -134,7 +134,7 @@ def test_tool_result_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ ToolResultContent( @@ -170,7 +170,7 @@ def test_full_stutter_pair(self): userType="external", cwd="/test", sessionId="session-test", - message=AssistantMessage( + message=AssistantMessageModel( id="msg_full_test", type="message", role="assistant", @@ -197,7 +197,7 @@ def test_full_stutter_pair(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ ToolResultContent( @@ -220,7 +220,7 @@ def test_full_stutter_pair(self): userType="external", cwd="/test", sessionId="session-test", - message=AssistantMessage( + message=AssistantMessageModel( id="msg_full_test", # SAME type="message", role="assistant", @@ -247,7 +247,7 @@ def test_full_stutter_pair(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ ToolResultContent( @@ -292,7 +292,7 @@ def test_user_text_message_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ TextContent( @@ -318,7 +318,7 @@ def test_user_text_message_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ TextContent( @@ -340,7 +340,7 @@ def test_user_text_message_deduplication(self): userType="external", cwd="/test", sessionId="session-test", - message=UserMessage( + message=UserMessageModel( role="user", content=[ TextContent(