fix: harden forensic report ingestion

This commit is contained in:
Christian Krakau-Louis
2026-05-23 13:02:01 +02:00
parent 908d4cd2cd
commit 817270b1fb
8 changed files with 89 additions and 23 deletions
+3 -1
View File
@@ -1,7 +1,7 @@
import hashlib
import json
import re
from datetime import datetime
from datetime import datetime, timezone
from email import message_from_bytes
from email.message import Message
from email.parser import Parser
@@ -84,6 +84,8 @@ def _parse_datetime(value: str) -> Optional[datetime]:
return None
if parsed is None:
return None
if parsed.tzinfo is not None:
parsed = parsed.astimezone(timezone.utc)
return parsed.replace(tzinfo=None)
+22 -4
View File
@@ -1,6 +1,7 @@
import json
from typing import Any, Dict, Optional
from sqlalchemy.exc import IntegrityError
from sqlalchemy.orm import Session
from app.models.domain import Domain
@@ -10,7 +11,7 @@ from app.utils.domain_validator import DomainValidationError, validate_domain
def forensic_report_exists(db: Session, report_id: str) -> bool:
"""Return True when a forensic report ID is already persisted."""
if not report_id:
if not str(report_id or "").strip():
return False
return (
db.query(ForensicReport.id).filter(ForensicReport.report_id == report_id).first()
@@ -40,7 +41,10 @@ def save_forensic_report(db: Session, report: Dict[str, Any]) -> tuple[ForensicR
Returns ``(row, created)``. The caller owns the transaction and should
commit after related work has completed.
"""
report_id = str(report.get("report_id") or "")
report_id = str(report.get("report_id") or "").strip()
if not report_id:
raise ValueError("Forensic report_id is required")
existing = db.query(ForensicReport).filter(ForensicReport.report_id == report_id).first()
if existing is not None:
return existing, False
@@ -72,12 +76,26 @@ def save_forensic_report(db: Session, report: Dict[str, Any]) -> tuple[ForensicR
feedback_headers=feedback_headers,
)
db.add(row)
db.flush()
try:
db.flush()
except IntegrityError:
db.rollback()
existing = db.query(ForensicReport).filter(ForensicReport.report_id == report_id).first()
if existing is not None:
return existing, False
raise
return row, True
def forensic_report_to_dict(row: ForensicReport) -> Dict[str, Any]:
"""Convert a forensic report row to an API-safe dictionary."""
feedback_headers = {}
if row.feedback_headers:
try:
feedback_headers = json.loads(row.feedback_headers)
except (json.JSONDecodeError, TypeError):
feedback_headers = {}
return {
"id": row.id,
"report_id": row.report_id,
@@ -98,6 +116,6 @@ def forensic_report_to_dict(row: ForensicReport) -> Dict[str, Any]:
"original_subject": row.original_subject,
"original_message_id": row.original_message_id,
"original_date": row.original_date,
"feedback_headers": json.loads(row.feedback_headers) if row.feedback_headers else {},
"feedback_headers": feedback_headers,
"processed_at": row.processed_at.isoformat() if row.processed_at else None,
}
+14 -13
View File
@@ -59,6 +59,7 @@ DMARC_GMAIL_QUERY = (
# How many message results to fetch per API page
_PAGE_SIZE = 100
RETRYABLE_MESSAGE_FAILURE = -1
class GmailClient:
@@ -247,11 +248,9 @@ class GmailClient:
stats["processed"] += 1
found = self._process_message(service, msg_id, stats)
if found > 0:
stats["new_ingested_ids"].append(msg_id)
self.already_ingested_ids.append(msg_id)
else:
# Track it anyway so we don't re-examine it next run
if found >= 0:
# Track it even when no report is found so we don't re-examine
# unrelated messages on every poll. Retryable failures return -1.
stats["new_ingested_ids"].append(msg_id)
self.already_ingested_ids.append(msg_id)
@@ -334,7 +333,7 @@ class GmailClient:
raw_bytes = base64.urlsafe_b64decode(msg_data.get("raw", ""))
msg = email.message_from_bytes(raw_bytes)
if ForensicParser.is_forensic_report(msg):
return 1 if self._process_forensic_message(raw_bytes, stats, message_id=msg_id) else 0
return self._process_forensic_message(raw_bytes, stats, message_id=msg_id)
return self._process_attachments(msg, stats, message_id=msg_id)
@staticmethod
@@ -383,7 +382,7 @@ class GmailClient:
raw_bytes: bytes,
stats: dict,
message_id: Optional[str] = None,
) -> bool:
) -> int:
"""Parse and persist one DMARC forensic report message."""
try:
report = ForensicParser.parse_bytes(raw_bytes, message_id_hint=message_id)
@@ -399,7 +398,7 @@ class GmailClient:
domain=domain,
report_id=report_id,
)
return False
return RETRYABLE_MESSAGE_FAILURE
if forensic_report_exists(self.db, report_id):
stats["duplicate_forensic_reports"] = stats.get("duplicate_forensic_reports", 0) + 1
@@ -411,7 +410,7 @@ class GmailClient:
domain=domain,
report_id=report_id,
)
return False
return 0
_row, created = save_forensic_report(self.db, report)
if not created:
@@ -424,7 +423,7 @@ class GmailClient:
domain=domain,
report_id=report_id,
)
return False
return 0
stats["forensic_reports_found"] = stats.get("forensic_reports_found", 0) + 1
self._append_detail(
@@ -435,10 +434,12 @@ class GmailClient:
domain=domain,
report_id=report_id,
)
return True
return 1
except Exception as exc: # pylint: disable=broad-exception-caught
logger.error("Failed to parse Gmail forensic report %s: %s", message_id, exc)
stats["errors"].append(f"Failed to parse forensic report {message_id}: {exc}")
stats.setdefault("errors", []).append(
f"Failed to parse forensic report {message_id}: {exc}"
)
self._append_detail(
stats,
status="error",
@@ -446,7 +447,7 @@ class GmailClient:
message_id=message_id,
error=str(exc),
)
return False
return RETRYABLE_MESSAGE_FAILURE
def _process_attachments(
self,