Fix: scope ProcessedMail UID deduplication by IMAP UIDVALIDITY (#12839)

This commit is contained in:
Hugo Beauzée-Luyssen
2026-07-15 13:43:16 -07:00
committed by GitHub
parent 735c3b073a
commit f0d934259c
4 changed files with 290 additions and 2 deletions
+39 -2
View File
@@ -18,6 +18,7 @@ from celery import shared_task
from celery.canvas import Signature
from django.conf import settings
from django.db import DatabaseError
from django.db.models import Q
from django.utils import timezone
from django.utils.timezone import is_naive
from django.utils.timezone import make_aware
@@ -245,6 +246,7 @@ def apply_mail_action(
message_uid: str,
message_subject: str,
message_date: datetime.datetime,
uid_validity: str | None = None,
) -> None:
"""
This shared task applies the mail action of a particular mail rule to the
@@ -286,6 +288,7 @@ def apply_mail_action(
rule=rule,
folder=rule.folder,
uid=message_uid,
uid_validity=uid_validity,
subject=message_subject,
received=message_date,
status="SUCCESS",
@@ -297,6 +300,7 @@ def apply_mail_action(
rule=rule,
folder=rule.folder,
uid=message_uid,
uid_validity=uid_validity,
subject=message_subject,
received=message_date,
status="FAILED",
@@ -314,6 +318,7 @@ def error_callback(
message_uid: str,
message_subject: str,
message_date: datetime.datetime,
uid_validity: str | None = None,
) -> None:
"""
A shared task that is called whenever something goes wrong during
@@ -325,6 +330,7 @@ def error_callback(
rule=rule,
folder=rule.folder,
uid=message_uid,
uid_validity=uid_validity,
subject=message_subject,
received=make_aware(message_date) if is_naive(message_date) else message_date,
status="FAILED",
@@ -337,6 +343,7 @@ def queue_consumption_tasks(
consume_tasks: list[Signature],
rule: MailRule,
message: MailMessage,
uid_validity: str | None,
) -> None:
"""
Queue a list of consumption tasks (Signatures for the consume_file shared
@@ -348,6 +355,7 @@ def queue_consumption_tasks(
message_uid=message.uid,
message_subject=message.subject,
message_date=message.date,
uid_validity=uid_validity,
)
chord(header=consume_tasks, body=mail_action_task).on_error(
error_callback.s(
@@ -355,6 +363,7 @@ def queue_consumption_tasks(
message_uid=message.uid,
message_subject=message.subject,
message_date=message.date,
uid_validity=uid_validity,
),
).delay()
@@ -460,6 +469,7 @@ class MailAccountHandler(LoggingMixin):
super().__init__()
self.renew_logging_group()
self._init_preprocessors()
self._current_uid_validity: str | None = None
def _init_preprocessors(self) -> None:
self._message_preprocessors: list[MailMessagePreprocessor] = []
@@ -510,6 +520,21 @@ class MailAccountHandler(LoggingMixin):
"Unknown title selector.",
) # pragma: no cover
def _get_uid_validity(self, M: MailBox, folder: str) -> str | None:
try:
uid_validity = M.folder.status(folder, ["UIDVALIDITY"]).get("UIDVALIDITY")
if uid_validity is not None:
return str(uid_validity)
except errors.MailboxFolderStatusError as e:
self.log.warning(
f"Server does not support retrieving UIDVALIDITY for folder {folder}: {e}",
)
except Exception as e:
self.log.warning(
f"Unable to retrieve UIDVALIDITY for folder {folder}: {e}",
)
return None
def _get_correspondent(
self,
message: MailMessage,
@@ -647,6 +672,8 @@ class MailAccountHandler(LoggingMixin):
f"does not exist in account {rule.account}",
) from err
self._current_uid_validity = self._get_uid_validity(M, rule.folder)
criterias = make_criterias(rule, supports_gmail_labels=supports_gmail_labels)
self.log.debug(
@@ -687,11 +714,17 @@ class MailAccountHandler(LoggingMixin):
)
continue
if ProcessedMail.objects.filter(
already_processed = ProcessedMail.objects.filter(
rule=rule,
uid=message.uid,
folder=rule.folder,
).exists():
)
if self._current_uid_validity is not None:
already_processed = already_processed.filter(
Q(uid_validity=self._current_uid_validity)
| Q(uid_validity__isnull=True),
)
if already_processed.exists():
self.log.debug(
f"Skipping mail '{message.uid}' subject '{message.subject}' from '{message.from_}', already processed.",
)
@@ -921,6 +954,7 @@ class MailAccountHandler(LoggingMixin):
consume_tasks=consume_tasks,
rule=rule,
message=message,
uid_validity=self._current_uid_validity,
)
else:
# No files to consume, just mark as processed if it wasn't by .eml processing
@@ -928,11 +962,13 @@ class MailAccountHandler(LoggingMixin):
rule=rule,
uid=message.uid,
folder=rule.folder,
uid_validity=self._current_uid_validity,
).exists():
ProcessedMail.objects.create(
rule=rule,
folder=rule.folder,
uid=message.uid,
uid_validity=self._current_uid_validity,
subject=message.subject,
received=make_aware(message.date)
if is_naive(message.date)
@@ -1009,6 +1045,7 @@ class MailAccountHandler(LoggingMixin):
consume_tasks=[consume_task],
rule=rule,
message=message,
uid_validity=self._current_uid_validity,
)
processed_elements = 1
@@ -0,0 +1,22 @@
from django.db import migrations
from django.db import models
class Migration(migrations.Migration):
dependencies = [
("paperless_mail", "0003_mailrule_stop_processing"),
]
operations = [
migrations.AddField(
model_name="processedmail",
name="uid_validity",
field=models.CharField(
blank=True,
editable=False,
max_length=64,
null=True,
verbose_name="uid validity",
),
),
]
+8
View File
@@ -338,6 +338,14 @@ class ProcessedMail(document_models.ModelWithOwner):
editable=False,
)
uid_validity = models.CharField(
_("uid validity"),
null=True,
blank=True,
max_length=64,
editable=False,
)
subject = models.CharField(
_("subject"),
null=False,
+221
View File
@@ -54,12 +54,16 @@ class _AttachmentDef:
class BogusFolderManager:
current_folder = "INBOX"
uidvalidity = "1"
def set(self, new_folder) -> None:
if new_folder not in ["INBOX", "spam"]:
raise MailboxFolderSelectError(None, "uhm")
self.current_folder = new_folder
def status(self, folder, options):
return {"UIDVALIDITY": self.uidvalidity}
class BogusClient:
def __init__(self, messages) -> None:
@@ -944,6 +948,223 @@ class TestMail(
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 1)
def test_handle_mail_account_skips_mail_already_processed_in_same_uidvalidity(
self,
) -> None:
"""
GIVEN:
- A ProcessedMail row recorded under the mailbox's current UIDVALIDITY
WHEN:
- A mail with the same UID is fetched from the same UIDVALIDITY epoch
THEN:
- The mail is skipped as a duplicate.
"""
account = MailAccount.objects.create(
name="test",
imap_server="",
username="admin",
password="secret",
)
rule = MailRule.objects.create(
name="testrule",
account=account,
action=MailRule.MailAction.DELETE,
)
message = self.mailMocker.messageBuilder.create_message()
self.mailMocker.bogus_mailbox.messages = [message]
self.mailMocker.bogus_mailbox.updateClient()
self.mailMocker.bogus_mailbox.folder.uidvalidity = "SAME"
ProcessedMail.objects.create(
rule=rule,
folder=rule.folder,
uid=message.uid,
uid_validity="SAME",
subject="Previously processed mail",
status="SUCCESS",
received=timezone.make_aware(timezone.datetime(2023, 1, 1, 12, 0, 0)),
)
self.mail_account_handler.handle_mail_account(account)
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 0)
def test_handle_mail_account_processes_mail_after_uidvalidity_change(
self,
) -> None:
"""
GIVEN:
- A ProcessedMail row recorded under a previous UIDVALIDITY epoch
WHEN:
- A mail with the same UID is fetched after UIDVALIDITY has changed
THEN:
- The mail is processed, not skipped as a duplicate.
"""
account = MailAccount.objects.create(
name="test",
imap_server="",
username="admin",
password="secret",
)
rule = MailRule.objects.create(
name="testrule",
account=account,
action=MailRule.MailAction.DELETE,
)
message = self.mailMocker.messageBuilder.create_message()
self.mailMocker.bogus_mailbox.messages = [message]
self.mailMocker.bogus_mailbox.updateClient()
self.mailMocker.bogus_mailbox.folder.uidvalidity = "NEW"
ProcessedMail.objects.create(
rule=rule,
folder=rule.folder,
uid=message.uid,
uid_validity="OLD",
subject="Previously processed mail",
status="SUCCESS",
received=timezone.make_aware(timezone.datetime(2023, 1, 1, 12, 0, 0)),
)
self.mail_account_handler.handle_mail_account(account)
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 1)
def test_handle_mail_account_skips_mail_processed_before_uidvalidity_tracking(
self,
) -> None:
"""
GIVEN:
- A ProcessedMail row recorded before UIDVALIDITY tracking existed
(uid_validity is NULL)
WHEN:
- A mail with the same UID is fetched
THEN:
- The mail is skipped as a duplicate, to avoid re-ingesting all
previously processed mail after upgrading.
"""
account = MailAccount.objects.create(
name="test",
imap_server="",
username="admin",
password="secret",
)
rule = MailRule.objects.create(
name="testrule",
account=account,
action=MailRule.MailAction.DELETE,
)
message = self.mailMocker.messageBuilder.create_message()
self.mailMocker.bogus_mailbox.messages = [message]
self.mailMocker.bogus_mailbox.updateClient()
ProcessedMail.objects.create(
rule=rule,
folder=rule.folder,
uid=message.uid,
uid_validity=None,
subject="Previously processed mail",
status="SUCCESS",
received=timezone.make_aware(timezone.datetime(2023, 1, 1, 12, 0, 0)),
)
self.mail_account_handler.handle_mail_account(account)
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 0)
def test_handle_mail_account_processes_mail_when_uidvalidity_unavailable(
self,
) -> None:
"""
GIVEN:
- The mail server fails to report a UIDVALIDITY for the folder
WHEN:
- A mail account is processed
THEN:
- The failure is logged and the rule still processes the mail,
instead of the whole rule being disabled.
"""
account = MailAccount.objects.create(
name="test",
imap_server="",
username="admin",
password="secret",
)
_ = MailRule.objects.create(
name="testrule",
account=account,
action=MailRule.MailAction.DELETE,
)
message = self.mailMocker.messageBuilder.create_message()
self.mailMocker.bogus_mailbox.messages = [message]
self.mailMocker.bogus_mailbox.updateClient()
self.mailMocker.bogus_mailbox.folder.status = mock.MagicMock(
side_effect=errors.MailboxFolderStatusError(("NO", [b"unsupported"]), "OK"),
)
with self.assertLogs("paperless_mail", level="WARNING") as cm:
self.mail_account_handler.handle_mail_account(account)
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 1)
self.assertEqual(len(cm.output), 1)
self.assertIn(
"Server does not support retrieving UIDVALIDITY",
cm.output[0],
)
def test_handle_mail_account_skips_mail_when_uidvalidity_unavailable_but_prior_record_exists(
self,
) -> None:
"""
GIVEN:
- A ProcessedMail row recorded with a real uid_validity value
- The mail server fails to report UIDVALIDITY (MailboxFolderStatusError),
so _get_uid_validity returns None
WHEN:
- A mail with the same UID is fetched
THEN:
- The mail is skipped as already-processed rather than re-ingested,
falling back to (rule, uid, folder) matching.
"""
account = MailAccount.objects.create(
name="test",
imap_server="",
username="admin",
password="secret",
)
rule = MailRule.objects.create(
name="testrule",
account=account,
action=MailRule.MailAction.DELETE,
)
message = self.mailMocker.messageBuilder.create_message()
self.mailMocker.bogus_mailbox.messages = [message]
self.mailMocker.bogus_mailbox.updateClient()
ProcessedMail.objects.create(
rule=rule,
folder=rule.folder,
uid=message.uid,
uid_validity="REAL_VALIDITY",
subject="Previously processed mail",
status="SUCCESS",
received=timezone.make_aware(timezone.datetime(2023, 1, 1, 12, 0, 0)),
)
self.mailMocker.bogus_mailbox.folder.status = mock.MagicMock(
side_effect=errors.MailboxFolderStatusError(("NO", [b"unsupported"]), "OK"),
)
self.mail_account_handler.handle_mail_account(account)
self.assertEqual(self.mailMocker._queue_consumption_tasks_mock.call_count, 0)
@pytest.mark.flaky(reruns=4)
def test_handle_mail_account_flag(self) -> None:
account = MailAccount.objects.create(