Coverage for src/ai_jury/diffprofile.py: 100%

1"""Cheap, pre-review diff profiling for risk-aware auto-depth (issue #120). 

2 

3A multi-vendor fan-out is expensive, so it should scale to the change: a 

4docs-only or few-line diff does not need debate + verification, while a large or 

5security-touching diff warrants the full panel. This module derives a fast, 

6PURE risk profile from the raw diff (size, files, whether it only touches 

7docs/generated files, whether it touches security-sensitive paths) and maps it 

8to a review depth (rounds / verify / early-stop). 

9 

10It never trims the *panel* (vendor diversity is the load-bearing advantage) — 

11only how many rounds run and whether the verification pass runs. Pure and 

12deterministic; the CLI owns applying it (opt-in) and logging it. 

13""" 

14 

15from __future__ import annotations 

16 

17from dataclasses import dataclass, field 

18 

19from .classification import _COMBINED_RX, diff_lines_changed 

20from .largediff import DEFAULT_GENERATED_GLOBS, _matches_any, split_diff 

21 

22# Paths that are low-risk to review at full depth (docs/text/config notes). 

23_DOC_GLOBS: tuple[str, ...] = ("*.md", "*.rst", "*.txt", "docs/**", "*.adoc") 

24 

25RISK_LOW = "low" 

26RISK_MEDIUM = "medium" 

27RISK_HIGH = "high" 

28 

29# Thresholds (changed lines / file count) for the high and low bands. 

30_HIGH_LINES, _HIGH_FILES = 400, 20 

31_LOW_LINES, _LOW_FILES = 15, 2 

32 

33 

34@dataclass 

35class DiffProfile: 

36 changed_lines: int 

37 file_count: int 

38 paths: list[str] = field(default_factory=list) 

39 docs_or_generated_only: bool = False 

40 security_sensitive: bool = False 

41 risk: str = RISK_MEDIUM 

42 

43 

44def _is_doc_or_generated(path: str) -> bool: 

45 return _matches_any(path, _DOC_GLOBS) or _matches_any(path, DEFAULT_GENERATED_GLOBS) 

46 

47 

48def _path_is_security_sensitive(path: str) -> bool: 

49 # bolt: Evaluate multiple regexes at once via the C engine instead of sequential looping. 

50 return bool(_COMBINED_RX.search(path)) 

51 

52 

53def profile_diff(diff: str) -> DiffProfile: 

54 """Profile a unified diff into a deterministic risk band (issue #120).""" 

55 files = split_diff(diff) 

56 paths = [f.path for f in files if f.path] 

57 changed = diff_lines_changed(diff) 

58 file_count = len(files) 

59 

60 # bolt: Consolidate multiple metrics into a single-pass O(N) explicit loop 

61 docs_only = bool(paths) 

62 security = False 

63 for p in paths: 

64 if not security and _path_is_security_sensitive(p): 

65 security = True 

66 if docs_only and not _is_doc_or_generated(p): 

67 docs_only = False 

68 if security and not docs_only: 

69 break 

70 

71 if security or changed > _HIGH_LINES or file_count > _HIGH_FILES: 

72 risk = RISK_HIGH 

73 elif docs_only or (changed <= _LOW_LINES and file_count <= _LOW_FILES): 

74 risk = RISK_LOW 

75 else: 

76 risk = RISK_MEDIUM 

77 

78 return DiffProfile( 

79 changed_lines=changed, 

80 file_count=file_count, 

81 paths=paths, 

82 docs_or_generated_only=docs_only, 

83 security_sensitive=security, 

84 risk=risk, 

85 ) 

86 

87 

88def depth_for(risk: str) -> tuple[int, bool, bool]: 

89 """Map a risk band to ``(rounds, verify, early_stop)``. 

90 

91 - low → 1 round, no verification (trivial change). 

92 - medium → 2 rounds, no verification, early-stop on (skip debate if agreed). 

93 - high → 2 rounds + verification, full (no early-stop). 

94 """ 

95 if risk == RISK_LOW: 

96 return 1, False, False 

97 if risk == RISK_HIGH: 

98 return 2, True, False 

99 return 2, False, True 

100 

101 

102def describe(profile: DiffProfile) -> str: 

103 """One-line, human-readable summary of the profile + chosen depth.""" 

104 rounds, verify, _early = depth_for(profile.risk) 

105 bits = [ 

106 f"risk={profile.risk}", 

107 f"{profile.changed_lines} changed lines", 

108 f"{profile.file_count} file(s)", 

109 ] 

110 if profile.docs_or_generated_only: 

111 bits.append("docs/generated-only") 

112 if profile.security_sensitive: 

113 bits.append("security-sensitive paths") 

114 return f"auto-depth: {', '.join(bits)} → rounds={rounds}, verify={'on' if verify else 'off'}"