| from __future__ import annotations |
|
|
| import re |
|
|
| from src.rules import RULES |
| from src.schemas import ChangeSet, Finding |
| from src.domain import ReviewState, Severity |
|
|
| KEYWORD_RULES: list[tuple[str, str, re.Pattern[str]]] = [ |
| ("py-sql-concat", "sql-injection", re.compile(r"\.execute\s*\(")), |
| ("py-cmd-shell", "command-injection", re.compile(r"os\.system|shell\s*=\s*True|os\.popen")), |
| ("py-path-join", "path-traversal", re.compile(r"open\s*\(.*\+|send_file\s*\(")), |
| ("py-pickle-load", "insecure-deserialization", re.compile(r"pickle\.loads|yaml\.load\s*\(")), |
| ("py-ssrf-url", "ssrf", re.compile(r"urlopen\s*\(|requests\.(get|post)\s*\(")), |
| ("py-xss-markup", "xss-template", re.compile(r"Markup\s*\(")), |
| ("py-weak-password-hash", "weak-crypto", re.compile(r"hashlib\.md5|hashlib\.sha1")), |
| ] |
|
|
|
|
| def keyword_findings(change: ChangeSet) -> list[Finding]: |
| findings: list[Finding] = [] |
| index = 1 |
| for item in change.files: |
| source = item.after if item.after.strip() else item.before |
| for line_no, line in enumerate(source.splitlines(), start=1): |
| for rule_id, family, pattern in KEYWORD_RULES: |
| if not pattern.search(line): |
| continue |
| spec = RULES[rule_id] |
| findings.append( |
| Finding( |
| finding_id=f"K{index:03d}-{family}", |
| rule_id=rule_id, |
| family=family, |
| cwe=spec.cwe, |
| severity=spec.severity if isinstance(spec.severity, Severity) else Severity.MEDIUM, |
| path=item.path, |
| start_line=line_no, |
| end_line=line_no, |
| snippet=line.strip(), |
| review_state=ReviewState.HYPOTHETICAL, |
| method="keyword_only", |
| evidence_locator=f"{item.path}:{line_no}", |
| summary="Keyword baseline matched this line without AST or taint evidence.", |
| references=spec.references, |
| inference="Keyword match is not a confirmed weakness.", |
| ) |
| ) |
| index += 1 |
| return findings |
|
|
