Merge pull request #102 from christianlouis/codex/import-detail-events

feat: add mail import detail events
This commit is contained in:
Christian Krakau-Louis
2026-05-22 20:15:29 +02:00
committed by GitHub
13 changed files with 520 additions and 67 deletions
@@ -0,0 +1,27 @@
"""add mail source import details
Revision ID: f6a7b8c9d0e1
Revises: e5f6a7b8c9d0
Create Date: 2026-05-22 20:05:00.000000
"""
from typing import Sequence, Union
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision: str = "f6a7b8c9d0e1"
down_revision: Union[str, Sequence[str], None] = "e5f6a7b8c9d0"
branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None
def upgrade() -> None:
"""Add a sanitized details payload to import history rows."""
op.add_column("mail_source_imports", sa.Column("details", sa.Text(), nullable=True))
def downgrade() -> None:
"""Remove import details payloads."""
op.drop_column("mail_source_imports", "details")
@@ -121,6 +121,7 @@ class MailSourceImportResponse(BaseModel):
error_count: int error_count: int
new_domains: List[str] new_domains: List[str]
errors: List[str] errors: List[str]
details: List[Dict[str, str]]
started_at: datetime started_at: datetime
finished_at: datetime finished_at: datetime
created_at: datetime created_at: datetime
@@ -183,6 +184,24 @@ def _decode_json_list(value: Optional[str]) -> List[str]:
return [str(item) for item in decoded] return [str(item) for item in decoded]
def _decode_json_details(value: Optional[str]) -> List[Dict[str, str]]:
"""Decode a sanitized JSON detail list stored on import history rows."""
if not value:
return []
try:
decoded = json.loads(value)
except (json.JSONDecodeError, TypeError):
return []
if not isinstance(decoded, list):
return []
details: List[Dict[str, str]] = []
for item in decoded:
if isinstance(item, dict):
details.append({str(key): str(val) for key, val in item.items()})
return details
def _import_to_response(row: MailSourceImport) -> MailSourceImportResponse: def _import_to_response(row: MailSourceImport) -> MailSourceImportResponse:
"""Convert an import-history ORM row to an API response.""" """Convert an import-history ORM row to an API response."""
return MailSourceImportResponse( return MailSourceImportResponse(
@@ -196,6 +215,7 @@ def _import_to_response(row: MailSourceImport) -> MailSourceImportResponse:
error_count=row.error_count, error_count=row.error_count,
new_domains=_decode_json_list(row.new_domains), new_domains=_decode_json_list(row.new_domains),
errors=_decode_json_list(row.errors), errors=_decode_json_list(row.errors),
details=_decode_json_details(row.details),
started_at=row.started_at, started_at=row.started_at,
finished_at=row.finished_at, finished_at=row.finished_at,
created_at=row.created_at, created_at=row.created_at,
+1
View File
@@ -24,6 +24,7 @@ class MailSourceImport(Base):
new_domains = Column(Text, nullable=True) new_domains = Column(Text, nullable=True)
errors = Column(Text, nullable=True) errors = Column(Text, nullable=True)
details = Column(Text, nullable=True)
started_at = Column(DateTime, nullable=False, default=datetime.utcnow, index=True) started_at = Column(DateTime, nullable=False, default=datetime.utcnow, index=True)
finished_at = Column(DateTime, nullable=False, default=datetime.utcnow, index=True) finished_at = Column(DateTime, nullable=False, default=datetime.utcnow, index=True)
+69 -3
View File
@@ -214,6 +214,7 @@ class GmailClient:
"new_domains": [], "new_domains": [],
"errors": [], "errors": [],
"new_ingested_ids": [], "new_ingested_ids": [],
"details": [],
} }
try: try:
@@ -232,6 +233,12 @@ class GmailClient:
for msg_id in message_ids: for msg_id in message_ids:
if msg_id in self.already_ingested_ids: if msg_id in self.already_ingested_ids:
self._append_detail(
stats,
status="skipped",
reason="already_ingested_message",
message_id=msg_id,
)
continue continue
stats["processed"] += 1 stats["processed"] += 1
@@ -292,6 +299,11 @@ class GmailClient:
return ids return ids
@staticmethod
def _append_detail(stats: dict, **detail: str) -> None:
"""Append a compact attachment/message outcome to the import stats."""
stats.setdefault("details", []).append({key: value for key, value in detail.items() if value})
def _process_message(self, service, msg_id: str, stats: dict) -> int: def _process_message(self, service, msg_id: str, stats: dict) -> int:
""" """
Download a Gmail message and process any DMARC-report attachments. Download a Gmail message and process any DMARC-report attachments.
@@ -305,11 +317,17 @@ class GmailClient:
except HttpError as exc: except HttpError as exc:
logger.error("Gmail API: failed to fetch message %s: %s", msg_id, exc) logger.error("Gmail API: failed to fetch message %s: %s", msg_id, exc)
stats["errors"].append(f"Failed to fetch message {msg_id}") stats["errors"].append(f"Failed to fetch message {msg_id}")
self._append_detail(
stats,
status="error",
reason="message_fetch_failed",
message_id=msg_id,
)
return 0 return 0
raw_bytes = base64.urlsafe_b64decode(msg_data.get("raw", "")) raw_bytes = base64.urlsafe_b64decode(msg_data.get("raw", ""))
msg = email.message_from_bytes(raw_bytes) msg = email.message_from_bytes(raw_bytes)
return self._process_attachments(msg, stats) return self._process_attachments(msg, stats, message_id=msg_id)
@staticmethod @staticmethod
def _decode_part_filename(part: email.message.Message) -> str: def _decode_part_filename(part: email.message.Message) -> str:
@@ -352,33 +370,81 @@ class GmailClient:
self.report_store.add_report(report) self.report_store.add_report(report)
return True return True
def _process_attachments(self, msg: email.message.Message, stats: dict) -> int: def _process_attachments(
self,
msg: email.message.Message,
stats: dict,
message_id: Optional[str] = None,
) -> int:
"""Walk a parsed email message and extract DMARC report attachments.""" """Walk a parsed email message and extract DMARC report attachments."""
reports_found = 0 reports_found = 0
for part in msg.walk(): for part in msg.walk():
filename = self._decode_part_filename(part) filename = self._decode_part_filename(part)
if not filename or not self._is_dmarc_attachment(filename): if not filename:
continue continue
disposition = part.get_content_disposition() disposition = part.get_content_disposition()
if disposition not in ("attachment", None): if disposition not in ("attachment", None):
continue continue
if not self._is_dmarc_attachment(filename):
self._append_detail(
stats,
status="skipped",
reason="unsupported_attachment",
message_id=message_id,
filename=filename,
)
continue
content = part.get_payload(decode=True) content = part.get_payload(decode=True)
if not content: if not content:
self._append_detail(
stats,
status="skipped",
reason="empty_attachment",
message_id=message_id,
filename=filename,
)
continue continue
try: try:
report = DMARCParser.parse_file(content, filename) report = DMARCParser.parse_file(content, filename)
domain = str(report.get("domain", "unknown"))
report_id = str(report.get("report_id", ""))
if self._store_report_if_new(report): if self._store_report_if_new(report):
stats["reports_found"] += 1 stats["reports_found"] += 1
reports_found += 1 reports_found += 1
self._append_detail(
stats,
status="imported",
message_id=message_id,
filename=filename,
domain=domain,
report_id=report_id,
)
else: else:
stats["duplicate_reports"] = stats.get("duplicate_reports", 0) + 1 stats["duplicate_reports"] = stats.get("duplicate_reports", 0) + 1
self._append_detail(
stats,
status="duplicate",
message_id=message_id,
filename=filename,
domain=domain,
report_id=report_id,
)
except Exception as exc: # pylint: disable=broad-exception-caught except Exception as exc: # pylint: disable=broad-exception-caught
logger.error("Failed to parse DMARC attachment %s: %s", filename, exc) logger.error("Failed to parse DMARC attachment %s: %s", filename, exc)
stats["errors"].append(f"Failed to parse {filename}: {exc}") stats["errors"].append(f"Failed to parse {filename}: {exc}")
self._append_detail(
stats,
status="error",
reason="parse_failed",
message_id=message_id,
filename=filename,
error=str(exc),
)
return reports_found return reports_found
+148 -57
View File
@@ -140,17 +140,24 @@ class IMAPClient:
def _process_single_email(self, mail, email_id: bytes, stats: dict) -> None: def _process_single_email(self, mail, email_id: bytes, stats: dict) -> None:
"""Fetch, parse, and store DMARC attachments from one email message.""" """Fetch, parse, and store DMARC attachments from one email message."""
message_id = email_id.decode("utf-8", errors="replace")
try: try:
status, msg_data = mail.fetch(email_id, "(RFC822)") status, msg_data = mail.fetch(email_id, "(RFC822)")
if status != "OK": if status != "OK":
logger.error("Error fetching email ID %s", email_id) logger.error("Error fetching email ID %s", email_id)
self._append_detail(
stats,
status="error",
reason="message_fetch_failed",
message_id=message_id,
)
return return
raw_email = msg_data[0][1] raw_email = msg_data[0][1]
msg = email.message_from_bytes(raw_email) msg = email.message_from_bytes(raw_email)
if self._is_dmarc_report_email(msg): if self._is_dmarc_report_email(msg):
reports_found = self._process_attachments(msg, stats) reports_found = self._process_attachments(msg, stats, message_id=message_id)
stats["reports_found"] += reports_found stats["reports_found"] += reports_found
# Mark email as read (and optionally delete) # Mark email as read (and optionally delete)
@@ -163,6 +170,13 @@ class IMAPClient:
error_msg = f"Error processing email ID {email_id}: {str(e)}" error_msg = f"Error processing email ID {email_id}: {str(e)}"
logger.error(error_msg) logger.error(error_msg)
stats["errors"].append(error_msg) stats["errors"].append(error_msg)
self._append_detail(
stats,
status="error",
reason="message_processing_failed",
message_id=message_id,
error=str(e),
)
def fetch_reports(self, days: int = 7) -> Dict[str, Any]: def fetch_reports(self, days: int = 7) -> Dict[str, Any]:
""" """
@@ -185,6 +199,7 @@ class IMAPClient:
"duplicate_reports": 0, "duplicate_reports": 0,
"new_domains": [], "new_domains": [],
"errors": [], "errors": [],
"details": [],
} }
try: try:
@@ -334,12 +349,7 @@ class IMAPClient:
filename = self._decode_email_header(filename) filename = self._decode_email_header(filename)
# Check file extension # Check file extension
if ( if self._is_dmarc_filename(filename):
filename.lower().endswith(".xml")
or filename.lower().endswith(".zip")
or filename.lower().endswith(".gz")
or filename.lower().endswith(".gzip")
):
return True return True
# Check content type # Check content type
@@ -355,7 +365,115 @@ class IMAPClient:
return False return False
def _process_attachments(self, msg: email.message.Message, stats: dict | None = None) -> int: @staticmethod
def _is_dmarc_filename(filename: str) -> bool:
lower = filename.lower()
return (
lower.endswith(".xml")
or lower.endswith(".zip")
or lower.endswith(".gz")
or lower.endswith(".gzip")
)
@staticmethod
def _append_detail(stats: dict | None, **detail: str) -> None:
"""Append a compact attachment/message outcome to the import stats."""
if stats is None:
return
stats.setdefault("details", []).append(
{key: value for key, value in detail.items() if value}
)
def _store_report_if_new(
self,
report: Dict[str, Any],
*,
filename: str,
stats: dict | None,
message_id: str | None,
) -> bool:
domain = report.get("domain", "unknown")
report_id = report.get("report_id", "")
if report_id and (
self.report_store.has_report(domain, report_id)
or (self.db is not None and report_exists(self.db, domain, report_id))
):
logger.info("Skipping duplicate DMARC report %s for %s", report_id, domain)
if stats is not None:
stats["duplicate_reports"] = stats.get("duplicate_reports", 0) + 1
self._append_detail(
stats,
status="duplicate",
message_id=message_id,
filename=filename,
domain=str(domain),
report_id=str(report_id),
)
return False
if self.db is not None:
save_parsed_report(self.db, report)
self.report_store.add_report(report)
self._append_detail(
stats,
status="imported",
message_id=message_id,
filename=filename,
domain=str(domain),
report_id=str(report_id),
)
return True
def _process_dmarc_attachment(
self,
part: email.message.Message,
*,
filename: str,
stats: dict | None,
message_id: str | None,
) -> bool:
try:
content = part.get_payload(decode=True)
if not content:
self._append_detail(
stats,
status="skipped",
reason="empty_attachment",
message_id=message_id,
filename=filename,
)
return False
report = DMARCParser.parse_file(content, filename)
stored = self._store_report_if_new(
report,
filename=filename,
stats=stats,
message_id=message_id,
)
if stored:
logger.info("Successfully processed DMARC report: %s", filename)
return stored
except Exception as exc: # pylint: disable=broad-exception-caught
logger.error("Error processing attachment %s: %s", filename, str(exc))
if stats is not None:
stats.setdefault("errors", []).append(f"Failed to parse {filename}: {exc}")
self._append_detail(
stats,
status="error",
reason="parse_failed",
message_id=message_id,
filename=filename,
error=str(exc),
)
return False
def _process_attachments(
self,
msg: email.message.Message,
stats: dict | None = None,
message_id: str | None = None,
) -> int:
""" """
Process email attachments that might be DMARC reports Process email attachments that might be DMARC reports
@@ -368,57 +486,30 @@ class IMAPClient:
reports_found = 0 reports_found = 0
for part in msg.walk(): for part in msg.walk():
content_disposition = part.get_content_disposition() if part.get_content_disposition() != "attachment":
continue
if content_disposition == "attachment": filename = part.get_filename()
filename = part.get_filename() if not filename:
if filename: continue
# Decode filename if needed
filename = self._decode_email_header(filename)
# Check if it's a likely DMARC report file filename = self._decode_email_header(filename)
if ( if not self._is_dmarc_filename(filename):
filename.lower().endswith(".xml") self._append_detail(
or filename.lower().endswith(".zip") stats,
or filename.lower().endswith(".gz") status="skipped",
or filename.lower().endswith(".gzip") reason="unsupported_attachment",
): message_id=message_id,
filename=filename,
)
continue
try: if self._process_dmarc_attachment(
# Get attachment content part,
content = part.get_payload(decode=True) filename=filename,
stats=stats,
# Parse the DMARC report message_id=message_id,
report = DMARCParser.parse_file(content, filename) ):
reports_found += 1
domain = report.get("domain", "unknown")
report_id = report.get("report_id", "")
if report_id and (
self.report_store.has_report(domain, report_id)
or (
self.db is not None
and report_exists(self.db, domain, report_id)
)
):
logger.info(
"Skipping duplicate DMARC report %s for %s",
report_id,
domain,
)
if stats is not None:
stats["duplicate_reports"] = (
stats.get("duplicate_reports", 0) + 1
)
continue
# Add the report to the store
if self.db is not None:
save_parsed_report(self.db, report)
self.report_store.add_report(report)
reports_found += 1
logger.info("Successfully processed DMARC report: %s", filename)
except Exception as e: # pylint: disable=broad-exception-caught
logger.error("Error processing attachment %s: %s", filename, str(e))
return reports_found return reports_found
+34 -1
View File
@@ -9,13 +9,24 @@ from app.models.mail_source_import import MailSourceImport
MAX_STORED_ERRORS = 10 MAX_STORED_ERRORS = 10
MAX_ERROR_LENGTH = 500 MAX_ERROR_LENGTH = 500
MAX_STORED_DETAILS = 50
MAX_DETAIL_VALUE_LENGTH = 300
DETAIL_FIELDS = {
"status",
"reason",
"message_id",
"filename",
"domain",
"report_id",
"error",
}
def _sanitize_error(value: object) -> str: def _sanitize_error(value: object) -> str:
"""Return a compact, log-safe error string for storage and UI display.""" """Return a compact, log-safe error string for storage and UI display."""
text = str(value).replace("\r", "").replace("\n", " ").strip() text = str(value).replace("\r", "").replace("\n", " ").strip()
if len(text) > MAX_ERROR_LENGTH: if len(text) > MAX_ERROR_LENGTH:
return text[: MAX_ERROR_LENGTH - 1] + "..." return text[: MAX_ERROR_LENGTH - 3] + "..."
return text return text
@@ -23,6 +34,27 @@ def _json_list(values: Optional[Iterable[Any]]) -> str:
return json.dumps([str(value) for value in values or []]) return json.dumps([str(value) for value in values or []])
def _sanitize_detail_value(value: object) -> str:
text = _sanitize_error(value)
if len(text) > MAX_DETAIL_VALUE_LENGTH:
return text[: MAX_DETAIL_VALUE_LENGTH - 3] + "..."
return text
def _json_details(values: Optional[Iterable[Any]]) -> str:
details = []
for value in list(values or [])[:MAX_STORED_DETAILS]:
if not isinstance(value, dict):
continue
entry = {}
for key in DETAIL_FIELDS:
if key in value and value[key] not in (None, ""):
entry[key] = _sanitize_detail_value(value[key])
if entry:
details.append(entry)
return json.dumps(details)
def record_import_attempt( def record_import_attempt(
db: Session, db: Session,
source: MailSource, source: MailSource,
@@ -47,6 +79,7 @@ def record_import_attempt(
error_count=len(result_errors), error_count=len(result_errors),
new_domains=_json_list(results.get("new_domains", [])), new_domains=_json_list(results.get("new_domains", [])),
errors=json.dumps(errors), errors=json.dumps(errors),
details=_json_details(results.get("details", [])),
started_at=started_at, started_at=started_at,
finished_at=datetime.utcnow(), finished_at=datetime.utcnow(),
) )
+45
View File
@@ -178,6 +178,7 @@
<th>Duplicates</th> <th>Duplicates</th>
<th>New Domains</th> <th>New Domains</th>
<th>Errors</th> <th>Errors</th>
<th>Details</th>
</tr> </tr>
</thead> </thead>
<tbody> <tbody>
@@ -207,6 +208,32 @@
<span class="text-muted-foreground">None</span> <span class="text-muted-foreground">None</span>
</template> </template>
</td> </td>
<td>
<template x-if="entry.details && entry.details.length">
<details class="max-w-sm">
<summary class="cursor-pointer" x-text="formatDetailsSummary(entry.details)"></summary>
<div class="mt-2 space-y-2 text-xs">
<template x-for="detail in entry.details" :key="`${detail.status}-${detail.message_id || ''}-${detail.filename || ''}-${detail.report_id || ''}`">
<div class="rounded border border-base-300 p-2">
<div class="flex flex-wrap items-center gap-2">
<span class="badge badge-sm" :class="detailBadgeClass(detail.status)" x-text="detail.status"></span>
<span class="font-mono" x-show="detail.report_id" x-text="detail.report_id"></span>
</div>
<div class="mt-1 text-muted-foreground">
<span x-show="detail.filename" x-text="detail.filename"></span>
<span x-show="detail.domain" x-text="` • ${detail.domain}`"></span>
<span x-show="detail.reason" x-text="` • ${(detail.reason || '').replaceAll('_', ' ')}`"></span>
<span x-show="detail.error" x-text="` • ${detail.error}`"></span>
</div>
</div>
</template>
</div>
</details>
</template>
<template x-if="!entry.details || entry.details.length === 0">
<span class="text-muted-foreground">None</span>
</template>
</td>
</tr> </tr>
</template> </template>
</tbody> </tbody>
@@ -579,6 +606,17 @@ function mailSourcesApp() {
return value && value.length ? value.join(', ') : '—'; return value && value.length ? value.join(', ') : '—';
}, },
formatDetailsSummary(details) {
const counts = details.reduce((acc, detail) => {
const key = detail.status || 'unknown';
acc[key] = (acc[key] || 0) + 1;
return acc;
}, {});
return Object.entries(counts)
.map(([key, count]) => `${count} ${key}`)
.join(', ');
},
statusBadgeClass(status) { statusBadgeClass(status) {
if (status === 'success') return 'badge-success'; if (status === 'success') return 'badge-success';
if (status === 'warning') return 'badge-warning'; if (status === 'warning') return 'badge-warning';
@@ -586,6 +624,13 @@ function mailSourcesApp() {
return 'badge-outline'; return 'badge-outline';
}, },
detailBadgeClass(status) {
if (status === 'imported') return 'badge-success';
if (status === 'duplicate' || status === 'skipped') return 'badge-warning';
if (status === 'error') return 'badge-error';
return 'badge-outline';
},
openAddForm() { openAddForm() {
this.editingId = null; this.editingId = null;
this.gmailConnected = false; this.gmailConnected = false;
+43 -1
View File
@@ -439,6 +439,8 @@ class TestProcessMessage:
assert count == 0 assert count == 0
assert len(stats["errors"]) == 1 assert len(stats["errors"]) == 1
assert "bad-id" in stats["errors"][0] assert "bad-id" in stats["errors"][0]
assert stats["details"][0]["status"] == "error"
assert stats["details"][0]["message_id"] == "bad-id"
# =========================================================================== # ===========================================================================
@@ -463,9 +465,36 @@ class TestProcessAttachments:
raw = _make_raw_email([{"filename": "photo.png", "content": b"\x89PNG"}]) raw = _make_raw_email([{"filename": "photo.png", "content": b"\x89PNG"}])
msg = email_mod.message_from_bytes(raw) msg = email_mod.message_from_bytes(raw)
stats = {"reports_found": 0, "errors": []} stats = {"reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats) count = client._process_attachments(msg, stats, message_id="msg-1")
assert count == 0 assert count == 0
assert stats["reports_found"] == 0 assert stats["reports_found"] == 0
assert stats["details"] == [
{
"status": "skipped",
"reason": "unsupported_attachment",
"message_id": "msg-1",
"filename": "photo.png",
}
]
def test_inline_dmarc_attachment_is_skipped(self):
client = _make_client()
raw = _make_raw_email(
[
{
"filename": "report.xml",
"content": SAMPLE_XML.encode(),
"disposition": "inline",
}
]
)
msg = email_mod.message_from_bytes(raw)
stats = {"reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats)
assert count == 0
assert stats.get("details") is None
def test_google_style_zip_attachment_is_parsed(self): def test_google_style_zip_attachment_is_parsed(self):
"""A Google DMARC ZIP attachment is parsed and counted.""" """A Google DMARC ZIP attachment is parsed and counted."""
@@ -485,6 +514,9 @@ class TestProcessAttachments:
assert count == 1 assert count == 1
assert stats["reports_found"] == 1 assert stats["reports_found"] == 1
assert stats["details"][0]["status"] == "imported"
assert stats["details"][0]["filename"].endswith(".zip")
assert stats["details"][0]["report_id"] == "123456789"
assert "example.com" in client.report_store.get_domains() assert "example.com" in client.report_store.get_domains()
def test_google_style_zip_attachment_is_persisted(self, db_session): def test_google_style_zip_attachment_is_persisted(self, db_session):
@@ -524,6 +556,8 @@ class TestProcessAttachments:
assert client._process_attachments(msg, first_stats) == 1 assert client._process_attachments(msg, first_stats) == 1
assert client._process_attachments(msg, second_stats) == 0 assert client._process_attachments(msg, second_stats) == 0
assert second_stats["details"][0]["status"] == "duplicate"
assert second_stats["details"][0]["report_id"] == "123456789"
assert client.report_store.get_domain_summary("example.com")["reports_processed"] == 1 assert client.report_store.get_domain_summary("example.com")["reports_processed"] == 1
def test_dmarc_attachment_with_empty_content_skipped(self): def test_dmarc_attachment_with_empty_content_skipped(self):
@@ -537,6 +571,8 @@ class TestProcessAttachments:
# Empty payload → `get_payload(decode=True)` returns b"" which is # Empty payload → `get_payload(decode=True)` returns b"" which is
# falsy, so the attachment is skipped # falsy, so the attachment is skipped
assert count == 0 assert count == 0
assert stats["details"][0]["status"] == "skipped"
assert stats["details"][0]["reason"] == "empty_attachment"
def test_parse_exception_adds_error_and_continues(self): def test_parse_exception_adds_error_and_continues(self):
"""A parse error should be recorded in stats but not raise.""" """A parse error should be recorded in stats but not raise."""
@@ -567,6 +603,9 @@ class TestProcessAttachments:
assert len(stats["errors"]) == 1 assert len(stats["errors"]) == 1
assert "bad.xml" in stats["errors"][0] assert "bad.xml" in stats["errors"][0]
assert stats["details"][0]["status"] == "error"
assert stats["details"][0]["filename"] == "bad.xml"
assert stats["details"][1]["status"] == "imported"
assert count == 1 # second attachment still parsed assert count == 1 # second attachment still parsed
@@ -622,6 +661,9 @@ class TestFetchReports:
call_args = mock_proc.call_args_list[0][0] call_args = mock_proc.call_args_list[0][0]
assert call_args[1] == "id2" assert call_args[1] == "id2"
assert result["processed"] == 1 assert result["processed"] == 1
assert result["details"][0]["status"] == "skipped"
assert result["details"][0]["reason"] == "already_ingested_message"
assert result["details"][0]["message_id"] == "id1"
def test_tracks_new_ingested_ids(self): def test_tracks_new_ingested_ids(self):
client = _make_client() client = _make_client()
+72 -2
View File
@@ -427,8 +427,12 @@ class TestProcessAttachments:
msg = email.message_from_bytes( msg = email.message_from_bytes(
_make_email_with_attachment("report.xml", MINIMAL_DMARC_XML, "application/xml") _make_email_with_attachment("report.xml", MINIMAL_DMARC_XML, "application/xml")
) )
count = client._process_attachments(msg) stats = {"processed": 0, "reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats, message_id="1")
assert count == 1 assert count == 1
assert stats["details"][0]["status"] == "imported"
assert stats["details"][0]["message_id"] == "1"
assert stats["details"][0]["report_id"] == "abc-123"
def test_processes_zip_attachment(self): def test_processes_zip_attachment(self):
client = self._make_client() client = self._make_client()
@@ -450,12 +454,73 @@ class TestProcessAttachments:
assert count == 1 assert count == 1
assert db_session.query(DMARCReport).filter_by(report_id="abc-123").count() == 1 assert db_session.query(DMARCReport).filter_by(report_id="abc-123").count() == 1
def test_duplicate_report_adds_detail(self):
client = self._make_client()
msg = email.message_from_bytes(
_make_email_with_attachment("report.xml", MINIMAL_DMARC_XML, "application/xml")
)
first_stats = {"processed": 0, "reports_found": 0, "errors": []}
second_stats = {"processed": 0, "reports_found": 0, "errors": []}
assert client._process_attachments(msg, first_stats) == 1
assert client._process_attachments(msg, second_stats) == 0
assert second_stats["details"][0]["status"] == "duplicate"
assert second_stats["details"][0]["report_id"] == "abc-123"
def test_bad_attachment_does_not_raise(self): def test_bad_attachment_does_not_raise(self):
client = self._make_client() client = self._make_client()
msg = email.message_from_bytes(_make_email_with_attachment("report.xml", b"not xml at all")) msg = email.message_from_bytes(_make_email_with_attachment("report.xml", b"not xml at all"))
# Should not raise; just returns 0 # Should not raise; just returns 0
count = client._process_attachments(msg) stats = {"processed": 0, "reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats)
assert count == 0 assert count == 0
assert stats["errors"]
assert stats["details"][0]["status"] == "error"
assert stats["details"][0]["filename"] == "report.xml"
def test_empty_attachment_adds_detail(self):
client = self._make_client()
msg = email.message_from_bytes(_make_email_with_attachment("report.xml", b""))
stats = {"processed": 0, "reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats)
assert count == 0
assert stats["details"][0]["status"] == "skipped"
assert stats["details"][0]["reason"] == "empty_attachment"
def test_unsupported_attachment_adds_detail(self):
client = self._make_client()
msg = email.message_from_bytes(
_make_email_with_attachment("notes.txt", b"not xml", "text/plain")
)
stats = {"processed": 0, "reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats, message_id="2")
assert count == 0
assert stats["details"] == [
{
"status": "skipped",
"reason": "unsupported_attachment",
"message_id": "2",
"filename": "notes.txt",
}
]
def test_attachment_without_filename_is_skipped(self):
client = self._make_client()
msg = MIMEMultipart()
msg.attach(MIMEText("body"))
part = MIMEApplication(MINIMAL_DMARC_XML)
part["Content-Disposition"] = "attachment"
msg.attach(part)
stats = {"processed": 0, "reports_found": 0, "errors": []}
count = client._process_attachments(msg, stats)
assert count == 0
assert stats.get("details") is None
def test_no_attachments_returns_zero(self): def test_no_attachments_returns_zero(self):
client = self._make_client() client = self._make_client()
@@ -497,6 +562,7 @@ class TestProcessSingleEmail:
assert stats["processed"] == 1 assert stats["processed"] == 1
assert stats["reports_found"] == 1 assert stats["reports_found"] == 1
assert stats["details"][0]["status"] == "imported"
def test_fetch_error_skips_email(self): def test_fetch_error_skips_email(self):
client = self._make_client() client = self._make_client()
@@ -507,6 +573,8 @@ class TestProcessSingleEmail:
client._process_single_email(mock_mail, b"1", stats) client._process_single_email(mock_mail, b"1", stats)
assert stats["processed"] == 0 assert stats["processed"] == 0
assert stats["details"][0]["status"] == "error"
assert stats["details"][0]["reason"] == "message_fetch_failed"
def test_exception_adds_to_errors(self): def test_exception_adds_to_errors(self):
client = self._make_client() client = self._make_client()
@@ -517,6 +585,7 @@ class TestProcessSingleEmail:
client._process_single_email(mock_mail, b"1", stats) client._process_single_email(mock_mail, b"1", stats)
assert len(stats["errors"]) == 1 assert len(stats["errors"]) == 1
assert stats["details"][0]["reason"] == "message_processing_failed"
def test_marks_deleted_when_flag_set(self): def test_marks_deleted_when_flag_set(self):
client = self._make_client() client = self._make_client()
@@ -599,6 +668,7 @@ class TestFetchReports:
assert result["success"] is True assert result["success"] is True
assert result["reports_found"] >= 1 assert result["reports_found"] >= 1
assert result["details"][0]["status"] == "imported"
def test_connection_error_returns_failure(self): def test_connection_error_returns_failure(self):
client = self._make_client() client = self._make_client()
+57
View File
@@ -3,6 +3,8 @@ Tests for MailSource model and mail-sources API endpoints.
""" """
import asyncio import asyncio
import json
from datetime import datetime
from unittest.mock import MagicMock, patch from unittest.mock import MagicMock, patch
from urllib.parse import parse_qs, urlparse from urllib.parse import parse_qs, urlparse
@@ -12,6 +14,7 @@ from sqlalchemy.orm import Session
from app.models.mail_source import MailSource from app.models.mail_source import MailSource
from app.models.mail_source_import import MailSourceImport from app.models.mail_source_import import MailSourceImport
from app.services.import_history import record_import_attempt
class TestMailSourceModel: class TestMailSourceModel:
@@ -95,6 +98,7 @@ class TestMailSourceImportModel:
error_count=1, error_count=1,
new_domains='["example.com"]', new_domains='["example.com"]',
errors='["bad attachment"]', errors='["bad attachment"]',
details='[{"status": "imported", "filename": "report.xml"}]',
) )
db_session.add(row) db_session.add(row)
db_session.commit() db_session.commit()
@@ -103,8 +107,51 @@ class TestMailSourceImportModel:
assert row.id is not None assert row.id is not None
assert row.mail_source_id == source.id assert row.mail_source_id == source.id
assert row.duplicate_reports == 1 assert row.duplicate_reports == 1
assert '"imported"' in row.details
assert row.mail_source.name == "History Source" assert row.mail_source.name == "History Source"
def test_record_import_attempt_sanitizes_details(self, db_session: Session):
source = MailSource(name="Detail Sanitizer", method="IMAP")
db_session.add(source)
db_session.commit()
db_session.refresh(source)
attempt = record_import_attempt(
db_session,
source,
{
"success": True,
"errors": ["x" * 600],
"details": [
"skip-me",
{"status": "imported", "filename": "a" * 400, "ignored": "secret"},
],
},
started_at=datetime.utcnow(),
trigger="manual",
)
details = json.loads(attempt.details)
errors = json.loads(attempt.errors)
assert len(errors[0]) == 500
assert details[0]["status"] == "imported"
assert len(details[0]["filename"]) == 300
assert "ignored" not in details[0]
class TestImportHistoryDecoding:
"""Unit tests for import-history JSON decoding helpers."""
def test_decode_details_handles_empty_and_malformed_values(self):
from app.api.api_v1.endpoints.mail_sources import _decode_json_details
assert _decode_json_details(None) == []
assert _decode_json_details("not-json") == []
assert _decode_json_details('{"not": "a list"}') == []
assert _decode_json_details('["skip", {"status": "imported", "report_id": 123}]') == [
{"status": "imported", "report_id": "123"}
]
class TestMailSourcesAPI: class TestMailSourcesAPI:
"""Integration tests for /api/v1/mail-sources endpoints (no auth).""" """Integration tests for /api/v1/mail-sources endpoints (no auth)."""
@@ -267,6 +314,7 @@ class TestMailSourcesAPIAuthed:
error_count=1, error_count=1,
new_domains='["example.com"]', new_domains='["example.com"]',
errors='["sanitized error"]', errors='["sanitized error"]',
details='[{"status": "duplicate", "report_id": "abc-123"}]',
) )
) )
db_session.commit() db_session.commit()
@@ -281,6 +329,7 @@ class TestMailSourcesAPIAuthed:
assert data[0]["duplicate_reports"] == 1 assert data[0]["duplicate_reports"] == 1
assert data[0]["new_domains"] == ["example.com"] assert data[0]["new_domains"] == ["example.com"]
assert data[0]["errors"] == ["sanitized error"] assert data[0]["errors"] == ["sanitized error"]
assert data[0]["details"] == [{"status": "duplicate", "report_id": "abc-123"}]
def test_list_import_history_handles_malformed_json( def test_list_import_history_handles_malformed_json(
self, authed_client: TestClient, db_session: Session self, authed_client: TestClient, db_session: Session
@@ -297,6 +346,7 @@ class TestMailSourcesAPIAuthed:
status="warning", status="warning",
new_domains="not-json", new_domains="not-json",
errors='{"not": "a list"}', errors='{"not": "a list"}',
details='{"not": "a list"}',
) )
) )
db_session.commit() db_session.commit()
@@ -306,6 +356,7 @@ class TestMailSourcesAPIAuthed:
assert resp.status_code == 200 assert resp.status_code == 200
assert resp.json()[0]["new_domains"] == [] assert resp.json()[0]["new_domains"] == []
assert resp.json()[0]["errors"] == [] assert resp.json()[0]["errors"] == []
assert resp.json()[0]["details"] == []
def test_list_import_history_unknown_source_returns_404(self, authed_client: TestClient): def test_list_import_history_unknown_source_returns_404(self, authed_client: TestClient):
resp = authed_client.get("/api/v1/mail-sources/99999/imports") resp = authed_client.get("/api/v1/mail-sources/99999/imports")
@@ -764,6 +815,7 @@ class TestManualSourceFetchEndpoint:
"duplicate_reports": 1, "duplicate_reports": 1,
"new_domains": ["example.com"], "new_domains": ["example.com"],
"errors": ["bad attachment\nwith newline"], "errors": ["bad attachment\nwith newline"],
"details": [{"status": "error", "filename": "bad.xml"}],
} }
with ( with (
@@ -780,6 +832,10 @@ class TestManualSourceFetchEndpoint:
assert data["duplicate_reports"] == 1 assert data["duplicate_reports"] == 1
assert data["error_count"] == 1 assert data["error_count"] == 1
assert "bad attachment with newline" in caplog.text assert "bad attachment with newline" in caplog.text
history_resp = authed_client.get(f"/api/v1/mail-sources/{source_id}/imports")
assert history_resp.json()[0]["details"] == [
{"status": "error", "filename": "bad.xml"}
]
mock_imap.fetch_reports.assert_called_once_with(days=30) mock_imap.fetch_reports.assert_called_once_with(days=30)
def test_fetch_gmail_source(self, authed_client: TestClient, db_session: Session): def test_fetch_gmail_source(self, authed_client: TestClient, db_session: Session):
@@ -807,6 +863,7 @@ class TestManualSourceFetchEndpoint:
"new_domains": [], "new_domains": [],
"errors": [], "errors": [],
"new_ingested_ids": ["id1"], "new_ingested_ids": ["id1"],
"details": [{"status": "imported", "report_id": "abc-123"}],
} }
mock_gmail.get_refreshed_tokens.return_value = None mock_gmail.get_refreshed_tokens.return_value = None
+2 -2
View File
@@ -28,6 +28,7 @@ Recently improved:
- Report/domain API reads can hydrate the dashboard projection from persisted data after restart. - Report/domain API reads can hydrate the dashboard projection from persisted data after restart.
- Mail source import history is visible in the Mail Sources UI. - Mail source import history is visible in the Mail Sources UI.
- Individual mail sources can be manually imported from the Mail Sources UI. - Individual mail sources can be manually imported from the Mail Sources UI.
- Import-history rows include sanitized per-attachment outcomes and imported report IDs.
- The current Alpine-based UI is allowed by CSP and renders dynamic tables in real browsers. - The current Alpine-based UI is allowed by CSP and renders dynamic tables in real browsers.
Implementation note: Implementation note:
@@ -38,9 +39,8 @@ Implementation note:
Objective: make mailbox imports auditable and make report totals trustworthy. Objective: make mailbox imports auditable and make report totals trustworthy.
Priority tasks: Priority tasks:
- Expand import attempts with message ID, source, attachment filename, outcome, and sanitized error details. - Add mailbox search controls for date-range backfills.
- Report duplicate skips separately from parse failures. - Report duplicate skips separately from parse failures.
- Add backfill controls for Gmail and IMAP sources.
- Improve source rollups so a source IP tracks pass/fail counts over time. - Improve source rollups so a source IP tracks pass/fail counts over time.
Quality bar: Quality bar:
+1 -1
View File
@@ -63,10 +63,10 @@ Recently delivered:
- Uploaded, Gmail-imported, and IMAP-imported reports are now persisted to report/record tables and can be reloaded into report/domain views. - Uploaded, Gmail-imported, and IMAP-imported reports are now persisted to report/record tables and can be reloaded into report/domain views.
- Mail source import history is now visible from the Mail Sources UI. - Mail source import history is now visible from the Mail Sources UI.
- A single mail source can be manually imported from the Mail Sources UI, with the result recorded in import history. - A single mail source can be manually imported from the Mail Sources UI, with the result recorded in import history.
- Import history now includes per-attachment details for imported reports, duplicates, parse errors, unsupported attachments, and imported report IDs.
- The current Alpine-based UI can run under the configured CSP, so dynamic tables render in real browsers. - The current Alpine-based UI can run under the configured CSP, so dynamic tables render in real browsers.
Next tasks: Next tasks:
- Add per-import result details: skipped duplicates, parse failures, unsupported attachments, and imported report IDs.
- Add mailbox search controls for date range/backfill without requiring code changes. - Add mailbox search controls for date range/backfill without requiring code changes.
- Improve source aggregation so each sender IP keeps pass/fail totals instead of only the latest result. - Improve source aggregation so each sender IP keeps pass/fail totals instead of only the latest result.
+1
View File
@@ -92,6 +92,7 @@ This file tracks the specific implementation tasks for each milestone of the DMA
- [x] Count duplicate skips separately from parse failures - [x] Count duplicate skips separately from parse failures
- [x] Show recent import history in the Mail Sources UI - [x] Show recent import history in the Mail Sources UI
- [x] Add manual import trigger per mail source - [x] Add manual import trigger per mail source
- [x] Add per-import result details for duplicates, parse failures, unsupported attachments, and imported report IDs
- [ ] Add retry/backfill controls per mail source - [ ] Add retry/backfill controls per mail source
## Milestone 3: Database Integration ## Milestone 3: Database Integration