Benjamin_Boenisch/breakpilot-core
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(pipeline): implement golden test suite + fix ontology patterns

Browse Source 
- Add test_golden_controls.py: 37 tests covering all 8 YAML categories
  (container, framework, evidence, negative, title, split, scope, merge_key)
- Fix evidence detection: handle German feminine articles (eine/einer/etc.)
- Fix framework detection: use verb stems for conjugated German verbs
- Add framework patterns: OWASP API6, CCM without CSA prefix, generic category
- Fix negative patterns: use "nicht übertragen/gespeichert/erscheinen" before
  generic "dürfen nicht" to correctly route prevent vs exclude

All 73 tests passing (36 ontology + 37 golden).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Benjamin Admin
2026-04-26 09:48:12 +02:00
parent d1f3b9ffcd
commit d660a45bb5
3 changed files with 259 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
control-pipeline/services/control_ontology.py
View File

@@ -139,10 +139,14 @@ for action_type, info in ACTION_TYPES.items():
_NEGATIVE_PATTERNS: list[tuple[str, str]] = [
# Longer/specific patterns first (checked in order)
("darf nicht wiederverwendet", "prevent"),
("nicht in der URL", "prevent"),
("nicht im Token", "prevent"),
("nicht in Logs", "prevent"),
("nicht wiederverwendet", "prevent"),
("nicht in der url", "prevent"),
("nicht im token", "prevent"),
("nicht in logs", "prevent"),
("nicht in urls", "prevent"),
("nicht gespeichert", "prevent"),
("nicht übertragen", "prevent"),
("nicht erscheinen", "prevent"),
("verhindern", "prevent"),
("unterbinden", "prevent"),
("abweisen", "enforce"),
@@ -199,15 +203,17 @@ EVIDENCE_INDICATORS: set[str] = {
_FRAMEWORK_PATTERNS: list[str] = [
r"OWASP\s+ASVS\s+V\d",
r"OWASP\s+API\d+",
r"OWASP\s+API\s+Top\s+10",
r"NIST\s+SP\s+800-\d+",
r"NIST\s+IA-\d+",
r"NIST\s+AC-\d+",
r"NIST\s+IA[\s-]",
r"NIST\s+AC[\s-]",
r"BSI\s+IT-Grundschutz",
r"BSI\s+200-\d",
r"CSA\s+CCM",
r"(?:CSA\s+)?CCM[\s-]",
r"ISO\s+27001",
r"ISO\s+27002",
r"alle\s+Controls\s+der\s+Kategorie",
]
@@ -258,8 +264,12 @@ def is_evidence(text: str) -> bool:
# Primary check: evidence indicators at the start
for indicator in EVIDENCE_INDICATORS:
if text_lower.startswith(indicator) or f"ein {indicator}" in text_lower:
if text_lower.startswith(indicator):
return True
# German articles: ein/eine/einen/einem/einer + indicator
for article in ("ein ", "eine ", "einen ", "einem ", "einer "):
if f"{article}{indicator}" in text_lower:
return True
# Secondary: "X dokumentieren" where X is another action's result
if text_lower.endswith("dokumentieren") or text_lower.endswith("dokumentiert"):
@@ -276,9 +286,10 @@ def is_framework_reference(text: str) -> bool:
for pattern in _FRAMEWORK_PATTERNS:
if re.search(pattern, text, re.IGNORECASE):
# Only if the text is a generic "implement X framework" statement
implement_words = {"umsetzen", "implementieren", "einhalten", "erfüllen", "anwenden"}
# Use stems to handle German conjugation (umsetzen/umzusetzen/umgesetzt)
implement_stems = ("umsetz", "umzusetz", "implementier", "einhalt", "erfüll", "anwend")
text_lower = text.lower()
if any(w in text_lower for w in implement_words):
if any(s in text_lower for s in implement_stems):
return True
return False