Generalize parser tests and allow parsing of the old log files

.github/copilot-instructions.md

@@ -0,0 +1,106 @@
+# Dosview Codebase AI Assistant Instructions
+
+## Project Overview
+
+**Dosview** is a lightweight PyQt5-based log viewer for analyzing radiation detector data from UST (Universal Scientific Technologies) dosimeters. It parses binary/text log files and visualizes spectral and temporal radiation data with matplotlib and pyqtgraph.
+
+### Core Architecture
+- **Parser-GUI separation**: `dosview/parsers.py` handles all file format detection and parsing (no Qt dependencies) to enable unit testing without GUI initialization
+- **Multi-format support**: Two main log formats (AIRDOS04C v2.0 and legacy pre-2.0) with automatic detection
+- **Threading model**: File loading happens in `LoadDataThread` to prevent UI blocking during large file parsing
+- **PyQt5 multi-window support**: Handles multiple log files and app instances (single instance via QLocalSocket/QLocalServer pattern)
+
+## Critical Patterns
+
+### Parser Architecture (`dosview/parsers.py`)
+The parser module is **parser-agnostic by design**. Add new format support by:
+1. Subclass `BaseLogParser` with `detect()` (static) and `parse()` (instance) methods
+2. Return tuple: `[time_axis (ndarray), sums (ndarray), hist (ndarray), metadata (dict)]`
+3. Add to `LOG_PARSERS` sequence for auto-detection
+4. Return data must have consistent ndarray shapes with `time_axis.shape[0] == sums.shape[0]`
+
+**Example metadata structure**:
+```python
+{
+    "log_device_info": {"DOS": {...}, "AIRDOS": {...}},
+    "log_info": {
+        "log_type": "xDOS_SPECTRAL",
+        "detector_type": "AIRDOS04C",  # or legacy detector name
+        "histogram_channels": 1024,
+        "events_total": 12345
+    }
+}
+```
+
+### Data Format Conventions
+- **Time axis**: Seconds (converted to minutes for UI via `/60`)
+- **Histogram**: 1D numpy array, channel indices as x-axis
+- **Log format detection**: First line matching triggers parser (e.g., `$DOS` prefix with/without `AIRDOS04C`)
+- **Legacy format quirk**: Pads histogram to 1024 channels regardless of actual data width
+
+### GUI Patterns (`dosview/__init__.py`)
+- **PlotCanvas**: Uses `pg.GraphicsLayoutWidget` for dual-plot layout (evolution + spectrum)
+- **Spectrum plot**: Log scale on both axes with marker overlays for channel events
+- **Evolution plot**: Linear scale with rolling average overlay (20-point window)
+- **Telemetry support**: Temperature, humidity, pressure, voltage metadata—already defined but not fully integrated
+
+## Developer Workflows
+
+### Testing
+```bash
+pytest  # Automatically discovers tests in tests/test_parser.py
+```
+- Tests require data fixtures in `data/` directory (e.g., `DATALOG_AIRDOS_GEO.TXT`)
+- Parser tests parametrized by fixture files—each format should have at least one sample log
+- No GUI testing (parsers are unit-testable independently)
+
+### Building & Distribution
+```bash
+pip install .              # Development mode
+python setup.py build sdist  # Build source distribution
+pytest && python setup.py build sdist  # Full CI/CD flow (see .github/workflows/tests.yml)
+```
+- Entry point: `dosview:main()` defined in `pyproject.toml`
+- Desktop integration: `dosview.desktop` installed to `/usr/local/share/applications` post-install
+- Icon: `media/icon_ust.png` bundled via `setup.py` custom `PostInstallCommand`
+
+### Running
+```bash
+dosview <filename>        # CLI entry point
+```
+- App auto-detects file format and loads appropriately
+- Supports command-line file opening from system file manager
+
+## Important Constraints & Decisions
+
+### Why Parsers Don't Depend on Qt
+Allows unit testing without initializing X11/display (critical for CI/CD headless environments). Import graph: `parsers.py` → `numpy/pandas` only; `__init__.py` imports `parsers` after Qt setup.
+
+### Multi-Format Detection Strategy
+- Linear first-line scan (fast for large files)
+- `Airdos04CLogParser` detected first (specificity: `$DOS` + `AIRDOS04C`)
+- `OldLogParser` catches everything else (fallback: `$DOS`, `$AIRDOS`, or `$HIST` prefixes)
+- Raises `ValueError("Neznámý typ logu...")` if no match—never silently fails
+
+### Known Limitations
+- Telemetry fields defined but not wired to plots (see `PlotCanvas.telemetry_lines`)
+- Spectrum plot hard-coded to 1024 channels (assumes all detectors standardize here)
+- No support for real-time streaming (file-based only)
+
+## When Adding Features
+
+- **New detector format?** Extend `OldLogParser.detect()` or create `NewDetectorParser` subclass
+- **New plot type?** Add to `PlotCanvas.plot()` before `self.show()` to avoid blocking
+- **New metadata?** Update `metadata` dict in parser; PlotCanvas will have access via `self.data[3]`
+- **New dependencies?** Add to `dependencies` list in `pyproject.toml` AND `requirements.txt` (both are used by CI)
+
+## Testing New Code
+
+Ensure parser tests pass with existing fixtures:
+```bash
+pytest tests/test_parser.py -v
+```
+
+For new parser format, commit sample log file to `data/` and verify:
+- `test_any_parser_detects_fixture` finds your parser
+- `test_parse_fixture_returns_consistent_shapes` validates output shapes

.github/workflows/tests.yml

@@ -0,0 +1,22 @@
+name: Python tests
+
+on:
+  push:
+  pull_request:
+
+jobs:
+  pytest:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.12'
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install . pytest
+      - name: Run pytest
+        run: pytest