Files
govoplan-mail/src/govoplan_mail/backend/pop3_imports.py
T
zemion 218fef11f1
Module Package Release / publish-packages (push) Successful in 11s
feat(mail): add governed POP3 legacy import
2026-08-22 04:52:25 +02:00

247 lines
8.0 KiB
Python

from __future__ import annotations
import base64
import hashlib
from dataclasses import dataclass
from datetime import datetime, timezone
from typing import Iterable
from sqlalchemy import or_, select
from sqlalchemy.orm import Session
from govoplan_core.security.secrets import encrypt_secret
from govoplan_mail.backend.db.models import MailPop3Import, MailServerEndpoint
from govoplan_mail.backend.sending.pop3 import Pop3DownloadedMessage
class Pop3ImportError(RuntimeError):
pass
@dataclass(frozen=True, slots=True)
class Pop3ImportResult:
imported: tuple[MailPop3Import, ...]
duplicates: tuple[MailPop3Import, ...]
def create_pop3_imports(
session: Session,
*,
tenant_id: str,
profile_id: str,
pop3_server_id: str,
pop3_credential_id: str | None,
transport_revision: str,
messages: Iterable[Pop3DownloadedMessage],
user_id: str | None,
deletion_requested: bool,
) -> Pop3ImportResult:
downloaded = tuple(messages)
if not downloaded:
raise Pop3ImportError("No POP3 messages were downloaded for import")
uidls = [item.uidl for item in downloaded]
if len(uidls) != len(set(uidls)):
raise Pop3ImportError("The POP3 download contained duplicate UIDL identifiers")
# Serialize imports per source before checking UIDLs. The database unique
# constraint remains the last line of defense, while this lock lets a
# concurrent request observe the first request's committed rows and report
# them as duplicates instead of surfacing an integrity error.
source = session.scalar(
select(MailServerEndpoint)
.where(
MailServerEndpoint.id == pop3_server_id,
MailServerEndpoint.profile_id == profile_id,
or_(
MailServerEndpoint.tenant_id == tenant_id,
MailServerEndpoint.tenant_id.is_(None),
),
MailServerEndpoint.protocol == "pop3",
)
.with_for_update()
)
if source is None:
raise Pop3ImportError("The selected POP3 source is unavailable")
existing = {
row.provider_uidl: row
for row in session.scalars(
select(MailPop3Import).where(
MailPop3Import.tenant_id == tenant_id,
MailPop3Import.profile_id == profile_id,
MailPop3Import.pop3_server_id == pop3_server_id,
MailPop3Import.provider_uidl.in_(uidls),
)
)
}
imported: list[MailPop3Import] = []
duplicates: list[MailPop3Import] = []
now = datetime.now(timezone.utc)
for message in downloaded:
duplicate = existing.get(message.uidl)
if duplicate is not None:
duplicates.append(duplicate)
continue
encrypted = encrypt_secret(base64.b64encode(message.raw).decode("ascii"))
if not encrypted:
raise Pop3ImportError("The downloaded POP3 message could not be encrypted")
summary = message.summary
row = MailPop3Import(
tenant_id=tenant_id,
profile_id=profile_id,
pop3_server_id=pop3_server_id,
pop3_credential_id=pop3_credential_id,
transport_revision=_required_revision(transport_revision),
provider_uidl=message.uidl,
provider_message_number=message.message_number,
fingerprint=_fingerprint(
tenant_id=tenant_id,
profile_id=profile_id,
pop3_server_id=pop3_server_id,
uidl=message.uidl,
raw_sha256=message.raw_sha256,
),
raw_sha256=message.raw_sha256,
raw_message_encrypted=encrypted,
message_id=summary.message_id,
subject=summary.subject,
from_header=summary.from_header,
to_header=summary.to_header,
date=summary.date,
body_preview=summary.body_preview,
size_bytes=len(message.raw),
status="pending_review",
imported_at=now,
imported_by_user_id=user_id,
deletion_requested=bool(deletion_requested),
deletion_status=("pending" if deletion_requested else "not_requested"),
)
session.add(row)
imported.append(row)
session.flush()
return Pop3ImportResult(imported=tuple(imported), duplicates=tuple(duplicates))
def list_pop3_imports(
session: Session,
*,
tenant_id: str,
profile_id: str | None = None,
profile_ids: Iterable[str] | None = None,
limit: int = 100,
) -> tuple[MailPop3Import, ...]:
statement = select(MailPop3Import).where(
MailPop3Import.tenant_id == tenant_id
)
if profile_id:
statement = statement.where(MailPop3Import.profile_id == profile_id)
elif profile_ids is not None:
allowed = tuple(dict.fromkeys(str(value) for value in profile_ids if value))
if not allowed:
return ()
statement = statement.where(MailPop3Import.profile_id.in_(allowed))
rows = session.scalars(
statement.order_by(
MailPop3Import.imported_at.desc(),
MailPop3Import.id.desc(),
).limit(max(1, min(int(limit), 500)))
)
return tuple(rows)
def mark_pop3_deletion_result(
session: Session,
*,
tenant_id: str,
import_ids: Iterable[str],
status: str,
error: str | None = None,
) -> tuple[MailPop3Import, ...]:
clean_status = str(status or "").strip().casefold()
if clean_status not in {"succeeded", "failed", "outcome_unknown"}:
raise Pop3ImportError("Unsupported POP3 deletion result")
ids = tuple(dict.fromkeys(str(value).strip() for value in import_ids if str(value).strip()))
if not ids:
return ()
rows = tuple(
session.scalars(
select(MailPop3Import)
.where(
MailPop3Import.tenant_id == tenant_id,
MailPop3Import.id.in_(ids),
MailPop3Import.deletion_requested.is_(True),
)
.with_for_update()
)
)
if len(rows) != len(ids):
raise Pop3ImportError("One or more POP3 import records are unavailable")
now = datetime.now(timezone.utc)
safe_error = _bounded_error(error)
for row in rows:
row.deletion_status = clean_status
row.deletion_attempted_at = now
row.deletion_error = safe_error
session.flush()
return rows
def pop3_import_payload(row: MailPop3Import) -> dict[str, object]:
return {
"id": row.id,
"profile_id": row.profile_id,
"pop3_server_id": row.pop3_server_id,
"transport_revision": row.transport_revision,
"provider_uidl": row.provider_uidl,
"message_id": row.message_id,
"subject": row.subject,
"from_header": row.from_header,
"to_header": row.to_header,
"date": row.date,
"body_preview": row.body_preview,
"size_bytes": row.size_bytes,
"raw_sha256": row.raw_sha256,
"status": row.status,
"imported_at": row.imported_at,
"deletion_requested": row.deletion_requested,
"deletion_status": row.deletion_status,
"deletion_attempted_at": row.deletion_attempted_at,
"deletion_error": row.deletion_error,
}
def _fingerprint(
*,
tenant_id: str,
profile_id: str,
pop3_server_id: str,
uidl: str,
raw_sha256: str,
) -> str:
material = "\x1f".join(
(tenant_id, profile_id, pop3_server_id, uidl, raw_sha256)
).encode("utf-8")
return hashlib.sha256(material).hexdigest()
def _required_revision(value: object) -> str:
clean = str(value or "").strip()
if not clean or len(clean) > 120:
raise Pop3ImportError("A valid POP3 transport revision is required")
return clean
def _bounded_error(value: str | None) -> str | None:
clean = " ".join(str(value or "").split())
return clean[:500] or None
__all__ = [
"Pop3ImportError",
"Pop3ImportResult",
"create_pop3_imports",
"list_pop3_imports",
"mark_pop3_deletion_result",
"pop3_import_payload",
]