Merge branch 'dev' into feature-remote-ocr-2

src/documents/management/commands/document_consumer.py

@@ -82,6 +82,13 @@ def _is_ignored(filepath: Path) -> bool:
 
 
 def _consume(filepath: Path) -> None:
+    # Check permissions early
+    try:
+        filepath.stat()
+    except (PermissionError, OSError):
+        logger.warning(f"Not consuming file {filepath}: Permission denied.")
+        return
+
     if filepath.is_dir() or _is_ignored(filepath):
         return
 
@@ -323,7 +330,12 @@ class Command(BaseCommand):
 
                         # Also make sure the file exists still, some scanners might write a
                         # temporary file first
-                        file_still_exists = filepath.exists() and filepath.is_file()
+                        try:
+                            file_still_exists = filepath.exists() and filepath.is_file()
+                        except (PermissionError, OSError):  # pragma: no cover
+                            # If we can't check, let it fail in the _consume function
+                            file_still_exists = True
+                            continue
 
                         if waited_long_enough and file_still_exists:
                             _consume(filepath)

src/documents/tests/test_management_consumer.py

@@ -209,6 +209,26 @@ class TestConsumer(DirectoriesMixin, ConsumerThreadMixin, TransactionTestCase):
         # assert that we have an error logged with this invalid file.
         error_logger.assert_called_once()
 
+    @mock.patch("documents.management.commands.document_consumer.logger.warning")
+    def test_permission_error_on_prechecks(self, warning_logger):
+        filepath = Path(self.dirs.consumption_dir) / "selinux.txt"
+        filepath.touch()
+
+        original_stat = Path.stat
+
+        def raising_stat(self, *args, **kwargs):
+            if self == filepath:
+                raise PermissionError("Permission denied")
+            return original_stat(self, *args, **kwargs)
+
+        with mock.patch("pathlib.Path.stat", new=raising_stat):
+            document_consumer._consume(filepath)
+
+        warning_logger.assert_called_once()
+        (args, _) = warning_logger.call_args
+        self.assertIn("Permission denied", args[0])
+        self.consume_file_mock.assert_not_called()
+
     @override_settings(CONSUMPTION_DIR="does_not_exist")
     def test_consumption_directory_invalid(self):
         self.assertRaises(CommandError, call_command, "document_consumer", "--oneshot")

src/paperless/settings.py

@@ -923,7 +923,7 @@ CELERY_ACCEPT_CONTENT = ["application/json", "application/x-python-serialize"]
 CELERY_BEAT_SCHEDULE = _parse_beat_schedule()
 
 # https://docs.celeryq.dev/en/stable/userguide/configuration.html#beat-schedule-filename
-CELERY_BEAT_SCHEDULE_FILENAME = DATA_DIR / "celerybeat-schedule.db"
+CELERY_BEAT_SCHEDULE_FILENAME = str(DATA_DIR / "celerybeat-schedule.db")
 
 
 # Cachalot: Database read cache.