sounddiff/.coderabbit.yaml at main · jaimefgdev/sounddiff · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
# yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json
# sounddiff — Python CLI for structured audio comparison
inheritance: true

reviews:
  request_changes_workflow: true
  profile: assertive
  pre_merge_checks:
    docstrings:
      mode: "off"
    title:
      mode: warning
    description:
      mode: warning
  path_filters:
    - "!CLAUDE.md"
    - "!docs/**"
    - "!.coderabbit.yaml"
    - "!scripts/**"
  path_instructions:
    - path: "src/sounddiff/**/*.py"
      instructions: |
        sounddiff is a Python 3.10+ CLI tool for structured audio comparison.
        Uses numpy, scipy, pyloudnorm for DSP. Click for CLI. Rich for terminal output.
        mypy strict mode. Ruff for linting and formatting.

        Review priorities (highest to lowest):
        1. Correctness of DSP algorithms (loudness, spectral, correlation math)
        2. Type safety (all public functions must have complete type annotations)
        3. Edge cases in audio processing (empty signals, mono/stereo, different sample rates)
        4. Performance issues with measurable impact on large audio files
        5. Error handling (clear messages for file not found, unsupported format, corrupt audio)

        Known patterns, DO NOT flag:
        - `from __future__ import annotations` at top of every module (PEP 604 unions on 3.10)
        - `np.ndarray` without generic parameters (numpy typing is limited)
        - `pyloudnorm` and `soundfile` are untyped (ignore_missing_imports in mypy config)
        - Frozen dataclasses for all result types (immutability is intentional)
        - Properties on frozen dataclasses for computed deltas (avoids storing derived values)

        Do NOT flag:
        - Missing docstrings on private functions (underscore-prefixed)
        - Import ordering (ruff handles this)
        - Line length (ruff formatter handles this)
    - path: "src/sounddiff/cli.py"
      instructions: |
        CLI entry point using Click. Review for:
        - Clear error messages for all failure modes
        - Correct exit codes (0 success, 1 for errors)
        - No unhandled exceptions reaching the user
    - path: "src/sounddiff/{loudness,spectral,temporal,detection}.py"
      instructions: |
        Core DSP modules. Review with extra care for:
        - Mathematical correctness (dB conversions, RMS calculations, correlation)
        - Handling of edge cases (silence, single-sample files, DC offset)
        - Numerical stability (division by zero, log of zero)
        - Consistent units (always dB, Hz, seconds — never raw sample counts in output)
    - path: "tests/**/*.py"
      instructions: |
        Test suite using pytest and hypothesis. Review for:
        - Test isolation (no shared mutable state between tests)
        - Meaningful assertions (not just "doesn't crash")
        - Edge case coverage (empty input, mono, extreme values)
        - Deterministic test audio generation (fixed seeds, no randomness without rng fixture)

  tools:
    ruff:
      enabled: true

knowledge_base:
  web_search:
    enabled: false