diff --git a/.gitignore b/.gitignore
index 331e713c..51464bae 100644
--- a/.gitignore
+++ b/.gitignore
@@ -50,6 +50,7 @@ coverage.xml
*.py,cover
.hypothesis/
.pytest_cache/
+.playwright_cache/
cover/
# Translations
diff --git a/claude_code_log/cache.py b/claude_code_log/cache.py
index d0a3ea00..ad443726 100644
--- a/claude_code_log/cache.py
+++ b/claude_code_log/cache.py
@@ -421,9 +421,9 @@ def _is_cache_version_compatible(self, cache_version: str) -> bool:
# Format: "cache_version": "minimum_library_version_required"
# If cache version is older than the minimum required, it needs invalidation
breaking_changes: dict[str, str] = {
- # Example breaking changes (adjust as needed):
- # "0.3.3": "0.3.4", # 0.3.4 introduced breaking changes to cache format
- # "0.2.x": "0.3.0", # 0.3.0 introduced major cache format changes
+ # 0.9.0 introduced _compact_ide_tags_for_preview() which transforms
+ # first_user_message to use emoji indicators instead of raw IDE tags
+ "0.8.0": "0.9.0",
}
cache_ver = version.parse(cache_version)
diff --git a/claude_code_log/converter.py b/claude_code_log/converter.py
index 960e8b4e..12a2fa83 100644
--- a/claude_code_log/converter.py
+++ b/claude_code_log/converter.py
@@ -318,15 +318,18 @@ def deduplicate_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntr
-> Same timestamp, same message.id or tool_use_id -> SHOULD deduplicate
2. Concurrent tool results: Multiple tool results with same timestamp
-> Same timestamp, different tool_use_ids -> should NOT deduplicate
+ 3. User text messages with same timestamp but different UUIDs (branch switch artifacts)
+ -> Same timestamp, no tool_use_id -> SHOULD deduplicate, keep the one with most content
Args:
messages: List of transcript entries to deduplicate
Returns:
- List of deduplicated messages, preserving order (first occurrence kept)
+ List of deduplicated messages, preserving order (first occurrence kept,
+ but replaced in-place if a better version is found later)
"""
- # Track seen (message_type, timestamp, is_meta, session_id, content_key) tuples
- seen: set[tuple[str, str, bool, str, str]] = set()
+ # Track seen dedup_key -> index in deduplicated list (for in-place replacement)
+ seen: dict[tuple[str, str, bool, str, str], int] = {}
deduplicated: list[TranscriptEntry] = []
for message in messages:
@@ -350,9 +353,10 @@ def deduplicate_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntr
# Get content key for differentiating concurrent messages
# - For assistant messages: use message.id (same for stutters, different for different msgs)
# - For user messages with tool results: use first tool_use_id
+ # - For user text messages: use empty string (deduplicate by timestamp alone)
# - For summary messages: use leafUuid (summaries have no timestamp/uuid)
- # - For other messages: use uuid as fallback
content_key = ""
+ is_user_text = False
if isinstance(message, AssistantTranscriptEntry):
# For assistant messages, use the message id
content_key = message.message.id
@@ -362,20 +366,29 @@ def deduplicate_messages(messages: list[TranscriptEntry]) -> list[TranscriptEntr
if isinstance(item, ToolResultContent):
content_key = item.tool_use_id
break
+ else:
+ # No tool result found - this is a user text message
+ is_user_text = True
+ # content_key stays empty (dedupe by timestamp alone)
elif isinstance(message, SummaryTranscriptEntry):
# Summaries have no timestamp or uuid - use leafUuid to keep them distinct
content_key = message.leafUuid
- # Fallback to uuid if no content key found
- if not content_key:
- content_key = getattr(message, "uuid", "")
- # Create deduplication key - include content_key for proper handling
- # of both version stutters and concurrent tool results
+ # Create deduplication key
dedup_key = (message_type, timestamp, is_meta, session_id, content_key)
- # Keep only first occurrence
- if dedup_key not in seen:
- seen.add(dedup_key)
+ if dedup_key in seen:
+ # For user text messages, replace if new one has more content items
+ if is_user_text and isinstance(message, UserTranscriptEntry):
+ idx = seen[dedup_key]
+ existing = deduplicated[idx]
+ if isinstance(existing, UserTranscriptEntry) and len(
+ message.message.content
+ ) > len(existing.message.content):
+ deduplicated[idx] = message # Replace with better version
+ # Otherwise skip duplicate
+ else:
+ seen[dedup_key] = len(deduplicated)
deduplicated.append(message)
return deduplicated
diff --git a/claude_code_log/html/system_formatters.py b/claude_code_log/html/system_formatters.py
index 14182a25..72eff8b8 100644
--- a/claude_code_log/html/system_formatters.py
+++ b/claude_code_log/html/system_formatters.py
@@ -96,13 +96,21 @@ def format_dedup_notice_content(content: DedupNoticeContent) -> str:
"""Format a deduplication notice as HTML.
Args:
- content: DedupNoticeContent with notice text
+ content: DedupNoticeContent with notice text and optional target link
Returns:
- HTML for the dedup notice display
+ HTML for the dedup notice display with optional anchor link
"""
escaped_notice = html.escape(content.notice_text)
- return f"{escaped_notice}
"
+
+ if content.target_message_id:
+ # Create clickable link to the target message
+ return (
+ f''
+ f"{escaped_notice}
"
+ )
+ else:
+ return f"{escaped_notice}
"
__all__ = [
diff --git a/claude_code_log/html/templates/transcript.html b/claude_code_log/html/templates/transcript.html
index 2f812bdf..b531c688 100644
--- a/claude_code_log/html/templates/transcript.html
+++ b/claude_code_log/html/templates/transcript.html
@@ -103,7 +103,7 @@ 🔍 Search & Filter
{% else %}
{%- set msg_css_class = css_class_from_message(message) %}
{% set markdown = message.type in ['assistant', 'thinking'] or message.modifiers.is_compacted %}
-