Files
anonymisation/tests/unit/test_q1_quarantine.py
Domi31tls 65d6c8c603 test(T-G): réparer corpus synthétique post-cleanup CHCB + dégel 009
- Fixtures 001/003/004/005/010 : CHCB → CHUXX (D-12)
- 009 : Biarritz désormais masqué [VILLE] (bug connu résolu par F1-F4),
  retrait de KNOWN_FAILURES + restauration de Biarritz dans must_not_contain
- test_q1_quarantine.py : tests réels B-3/D2/D3/M5/INDEX/errors.log
  (ex-squelette xfail)

Suite tests/unit : 85 passed, 0 failed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-03 10:31:38 +02:00

333 lines
12 KiB
Python

"""
Tests pour Q-1 — Quarantaine différentielle.
Couvre : pré-flight B-3, quarantaine D2/D3, rescan résiduel M5,
INDEX.md, errors.log.
Les tests B-1 (metadata XMP) et B-2 (per-doc log) restent xfail car non implémentés.
"""
from __future__ import annotations
import json
import os
import textwrap
from pathlib import Path
from unittest.mock import patch, MagicMock
import pytest
# === Fixtures ====================================================
@pytest.fixture
def tmp_output_dir(tmp_path: Path) -> Path:
out = tmp_path / "output"
out.mkdir()
return out
@pytest.fixture
def fake_pdf_path(tmp_path: Path) -> Path:
p = tmp_path / "doc_ok.pdf"
p.write_bytes(b"%PDF-1.4\n%fake\n")
return p
# === Tests B-3 : pré-flight texte vide ===========================
class TestPreflight:
"""B-3 — Pré-flight : texte < SEUIL_TEXTE_MINI → quarantaine full."""
def test_preflight_empty_text_goes_to_quarantine(self, tmp_path: Path) -> None:
"""Un document dont l'extraction retourne < 100 chars va en quarantaine
sans produire de texte/PDF de sortie."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
pdf = out / "doc_empty.pdf"
pdf.write_bytes(b"%PDF-1.4\n%empty\n")
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234")
mgr.flag(
doc_name="doc_empty",
reason="preflight_text_too_short",
detail="Only 10 chars extracted (seuil=100)",
severity="full",
extracted_chars=10,
)
quarantine_dir = out / "quarantaine"
assert quarantine_dir.exists(), "Le dossier quarantaine doit être créé"
assert (quarantine_dir / "doc_empty.reason.txt").exists()
assert quarantine_dir.stat().st_mode & 0o777 == 0o700, "quarantine_dir doit être 0700"
def test_preflight_reason_format(self, tmp_path: Path) -> None:
"""Le fichier .reason.txt doit contenir : raison, horodatage,
caractères extraits, version, profil."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234",
profile_name="standard_local")
mgr.flag(
doc_name="doc_empty",
reason="preflight_text_too_short",
detail="Only 10 chars extracted (seuil=100)",
severity="full",
extracted_chars=10,
)
reason = (out / "quarantaine" / "doc_empty.reason.txt").read_text()
assert "preflight_text_too_short" in reason
assert "Caractères extraits" in reason
assert "10" in reason
assert "Horodatage" in reason
assert "0.11.0" in reason
assert "abc1234" in reason
assert "standard_local" in reason
# === Tests Q-1 : quarantaine différentielle =====================
class TestRedactionFailure:
"""Q-1 — Rédaction PDF échoue → texte livré, PDF en quarantaine."""
def test_redaction_failure_text_still_outputs(self, tmp_path: Path) -> None:
"""Si la rédaction PDF échoue mais que l'anonymisation texte réussit :
- le .pseudonymise.txt sort normalement
- le PDF va en quarantaine avec flag pdf_redaction_failed
"""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
pdf = out / "doc_redact_fail.pdf"
pdf.write_bytes(b"%PDF-1.4\n%redact_fails\n")
# Simule le comportement de process_pdf quand vector échoue
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234")
# Texte anonymisé produit
txt = out / "doc_redact_fail.pseudonymise.txt"
txt.write_text("Patient [NOM] présenté le [DATE].\n")
audit = out / "doc_redact_fail.audit.jsonl"
audit.write_text('{"type": "mask", "label": "NOM"}\n')
# Vector échoue → flag partial
mgr.flag(
doc_name="doc_redact_fail",
reason="pdf_redaction_failed",
detail="vector failed (fitz.ApplyRedactionException); raster also failed (OOM)",
severity="partial",
)
assert txt.exists()
assert audit.exists()
reason = (out / "quarantaine" / "doc_redact_fail.reason.txt").read_text()
assert "pdf_redaction_failed" in reason
assert "partial" in reason
def test_no_silent_failure_on_redaction(self, tmp_path: Path) -> None:
"""Toute exception sur la rédaction DOIT être logguée (warning minimum).
Pas de `except Exception: pass` silencieux."""
import logging
# On teste que _append_errors_log ne mute pas les erreurs
# (le vrai comportement est testé par le test de symlink ci-dessous)
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out)
# Flag avec exception — vérifie que la stacktrace est capturée
try:
raise ValueError("ApplyRedactionException: invalid rect")
except ValueError as e:
mgr.flag(doc_name="doc1", reason="pdf_redaction_failed",
detail="vector failed", severity="partial", exc=e)
errors_log = out / "errors.log"
assert errors_log.exists()
lines = errors_log.read_text().splitlines()
assert len(lines) == 1
entry = json.loads(lines[0])
assert "pdf_redaction_failed" in entry["category"] or "pdf" in entry["category"]
# === Tests F : rescan résiduel (M5) =============================
class TestRescanQuarantine:
"""F / M5 — Rescan post-nettoyage détecte PII résiduelles → quarantaine full."""
def test_rescan_detects_residual_pii_triggers_quarantine(self, tmp_path: Path) -> None:
"""Si le rescan détecte des PII résiduelles > seuil (0 par défaut),
AUCUN fichier de sortie n'est livré — quarantaine full."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234")
mgr.flag(
doc_name="doc_leak",
reason="rescan_residual_pii",
detail="2 residual PII after all cleaning passes (seuil=0)",
severity="full",
)
# Le texte NE doit PAS être livré
assert not (out / "doc_leak.pseudonymise.txt").exists()
assert (out / "quarantaine" / "doc_leak.reason.txt").exists()
assert mgr.has_full_quarantine("doc_leak")
# === Tests A : INDEX.md et errors.log ===========================
class TestQuarantineArtifacts:
"""A — Artifacts de quarantaine : INDEX.md, errors.log."""
def test_quarantine_index_md_format(self, tmp_path: Path) -> None:
"""INDEX.md doit lister tous les docs en quarantaine avec raison,
caractères extraits, action recommandée."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234")
mgr.flag(
doc_name="doc_empty",
reason="preflight_text_too_short",
detail="Only 10 chars",
severity="full",
extracted_chars=10,
)
mgr.flag(
doc_name="doc_fail",
reason="pdf_redaction_failed",
detail="vector failed",
severity="partial",
)
mgr.finalize(total_docs_processed=5)
index = out / "quarantaine" / "INDEX.md"
assert index.exists()
content = index.read_text()
assert "doc_empty" in content
assert "doc_fail" in content
assert "Quarantaine totale" in content
assert "Quarantaine partielle" in content
assert "Taux" in content
# 2 docs flaggés sur 5 traités = 40%
assert "40.0%" in content
def test_errors_log_json_lines(self, tmp_path: Path) -> None:
"""errors.log doit être un fichier JSON-lines valide,
avec ts, doc, level, category, msg, severity."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out, app_version="0.11.0", commit_sha="abc1234")
mgr.flag(
doc_name="doc1",
reason="preflight_text_too_short",
detail="Only 10 chars",
severity="full",
)
mgr.flag(
doc_name="doc2",
reason="pdf_redaction_failed",
detail="vector failed",
severity="partial",
)
errors_log = out / "errors.log"
assert errors_log.exists()
# Vérifier permissions (0o600)
mode = errors_log.stat().st_mode & 0o777
assert mode == 0o600, f"errors.log permissions should be 0600, got {oct(mode)}"
lines = errors_log.read_text().splitlines()
assert len(lines) == 2
for line in lines:
entry = json.loads(line) # doit parser sans erreur
assert "ts" in entry
assert "doc" in entry
assert "level" in entry
assert "category" in entry
assert "msg" in entry
assert "severity" in entry
assert lines[0].startswith("{") # JSON-lines format
entry1 = json.loads(lines[0])
assert entry1["severity"] == "full"
assert entry1["category"] == "preflight"
entry2 = json.loads(lines[1])
assert entry2["severity"] == "partial"
assert entry2["category"] == "pdf"
# === Tests de non-régression ====================================
def test_happy_path_no_quarantine_created_if_no_failure(tmp_path: Path) -> None:
"""Non-régression — Sans flag, aucun dossier quarantaine/ créé."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out, app_version="0.11.0")
# Aucun flag → pas de quarantine_dir créé
assert not (out / "quarantaine").exists()
# === Tests security : permissions + symlink =====================
class TestSecurity:
"""Tests des fixes sécurité (Criticals 1-2, M1-M2)."""
def test_quarantine_dir_permissions(self, tmp_path: Path) -> None:
"""quarantine_dir doit avoir des permissions 0o700."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
mgr = QuarantineManager(out)
mgr.flag(doc_name="doc1", reason="test", detail="test", severity="full")
qdir = out / "quarantaine"
mode = qdir.stat().st_mode & 0o777
assert mode == 0o700, f"quarantine_dir should be 0700, got {oct(mode)}"
def test_symlink_errors_log_refused(self, tmp_path: Path) -> None:
"""Si errors.log est un symlink, _append_errors_log doit refuser d'écrire
(O_NOFOLLOW lève OSError)."""
from quarantine import QuarantineManager
out = tmp_path / "output"
out.mkdir()
target = tmp_path / "symlink_target.txt"
target.write_text("innocent")
(out / "errors.log").symlink_to(target)
mgr = QuarantineManager(out)
# O_NOFOLLOW lève OSError (ELOOP), pas RuntimeError
with pytest.raises(OSError):
mgr.flag(doc_name="doc1", reason="test", detail="test", severity="full")
def test_o_nofollow_refuses_symlink_at_creation(self, tmp_path: Path) -> None:
"""os.open(O_NOFOLLOW) doit refuser la création via symlink."""
import os as _os
target = tmp_path / "target.txt"
target.write_text("innocent")
link = tmp_path / "errors.log"
link.symlink_to(target)
with pytest.raises(OSError):
fd = _os.open(str(link), _os.O_CREAT | _os.O_APPEND | _os.O_WRONLY | _os.O_NOFOLLOW, 0o600)
_os.close(fd)