126 lines
4.1 KiB
Python
126 lines
4.1 KiB
Python
"""Strict parser for reviewer reports, verdicts, and workflow tags.
|
|
|
|
Issue #5 Phase 3.
|
|
Implements fence-aware parsing so markdown code blocks (e.g. documentation or
|
|
examples quoted inside reports) cannot trigger or spoof verdicts, escalations,
|
|
or rebuttals.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import os
|
|
import re
|
|
from pathlib import Path
|
|
from typing import List, Optional
|
|
|
|
# Fences start with at least 3 backticks or tildes: ``` or ~~~
|
|
_FENCE_START_RE = re.compile(r"^[ \t]*(```+|~~~+)")
|
|
_VERDICT_RE = re.compile(r"^\[VERDICT:\s*(PASS|NOT PASS)\]\s*$")
|
|
|
|
|
|
def strip_code_fences(text: str) -> str:
|
|
"""Remove lines that are inside markdown code blocks (``` or ~~~)."""
|
|
clean_lines = []
|
|
in_fence = False
|
|
fence_char = ""
|
|
fence_len = 0
|
|
|
|
for line in text.splitlines():
|
|
match = _FENCE_START_RE.match(line)
|
|
if match:
|
|
marker = match.group(1)
|
|
char = marker[0]
|
|
length = len(marker)
|
|
if not in_fence:
|
|
in_fence = True
|
|
fence_char = char
|
|
fence_len = length
|
|
continue
|
|
else:
|
|
# Closing fence must match the fence character and have at least the opening length
|
|
if char == fence_char and length >= fence_len:
|
|
in_fence = False
|
|
fence_char = ""
|
|
fence_len = 0
|
|
continue
|
|
if not in_fence:
|
|
clean_lines.append(line)
|
|
|
|
return "\n".join(clean_lines)
|
|
|
|
|
|
def parse_verdict(content: str) -> Optional[str]:
|
|
"""Extract verdict ('PASS' or 'NOT PASS') from the last non-blank line outside code fences.
|
|
|
|
Returns None if no valid verdict line is found or if content is malformed.
|
|
"""
|
|
stripped = strip_code_fences(content)
|
|
lines = [line.strip() for line in stripped.splitlines() if line.strip()]
|
|
if not lines:
|
|
return None
|
|
|
|
last_line = lines[-1]
|
|
match = _VERDICT_RE.match(last_line)
|
|
if match:
|
|
return match.group(1)
|
|
return None
|
|
|
|
|
|
def has_verdict(file_or_content: str | Path, expected: str) -> bool:
|
|
"""Check whether a report file or content ends with [VERDICT: <expected>]."""
|
|
content = ""
|
|
if isinstance(file_or_content, Path) or (isinstance(file_or_content, str) and os.path.isfile(file_or_content)):
|
|
try:
|
|
content = Path(file_or_content).read_text(encoding="utf-8", errors="replace")
|
|
except OSError:
|
|
return False
|
|
else:
|
|
content = str(file_or_content)
|
|
|
|
verdict = parse_verdict(content)
|
|
return verdict == expected
|
|
|
|
|
|
def has_tag_line(file_or_content: str | Path, tag: str) -> bool:
|
|
"""Detect a standalone '[TAG]' line anywhere outside code fences."""
|
|
content = ""
|
|
if isinstance(file_or_content, Path) or (isinstance(file_or_content, str) and os.path.isfile(file_or_content)):
|
|
try:
|
|
content = Path(file_or_content).read_text(encoding="utf-8", errors="replace")
|
|
except OSError:
|
|
return False
|
|
else:
|
|
content = str(file_or_content)
|
|
|
|
stripped = strip_code_fences(content)
|
|
pattern = re.compile(rf"^\[{re.escape(tag)}\]\s*$", re.MULTILINE)
|
|
return bool(pattern.search(stripped))
|
|
|
|
|
|
def extract_tag_values(file_or_content: str | Path, key: str) -> List[str]:
|
|
"""Emit the trimmed payload of every standalone '[KEY: value]' line outside code fences.
|
|
|
|
Preserves appearance order and eliminates duplicates.
|
|
"""
|
|
content = ""
|
|
if isinstance(file_or_content, Path) or (isinstance(file_or_content, str) and os.path.isfile(file_or_content)):
|
|
try:
|
|
content = Path(file_or_content).read_text(encoding="utf-8", errors="replace")
|
|
except OSError:
|
|
return []
|
|
else:
|
|
content = str(file_or_content)
|
|
|
|
stripped = strip_code_fences(content)
|
|
pattern = re.compile(rf"^\[{re.escape(key)}:\s*([^\]]*)\]\s*$", re.MULTILINE)
|
|
results = []
|
|
seen = set()
|
|
|
|
for match in pattern.finditer(stripped):
|
|
val = match.group(1).strip()
|
|
if val and val not in seen:
|
|
seen.add(val)
|
|
results.append(val)
|
|
|
|
return results
|