Chore: Reuse the initialized Django ASGI application for HTTP routing

src/documents/barcodes.py

@@ -10,9 +10,6 @@ from typing import TYPE_CHECKING
 import regex as regex_mod
 from django.conf import settings
 from pdf2image import convert_from_path
-from pikepdf import Page
-from pikepdf import PasswordError
-from pikepdf import Pdf
 
 from documents.converters import convert_from_tiff_to_pdf
 from documents.data_models import ConsumableDocument
@@ -34,6 +31,7 @@ from documents.utils import maybe_override_pixel_limit
 from paperless.config import BarcodeConfig
 
 if TYPE_CHECKING:
+    from pikepdf import Page
     from PIL import Image
 
 logger = logging.getLogger("paperless.barcodes")
@@ -282,6 +280,8 @@ class BarcodePlugin(ConsumeTaskPlugin):
         # No op if not a TIFF
         self.convert_from_tiff_to_pdf()
 
+        from pikepdf import PasswordError
+
         try:
             self.barcodes = scan_pdf(
                 self.pdf_file,
@@ -433,6 +433,8 @@ class BarcodePlugin(ConsumeTaskPlugin):
 
         document_paths = []
         fname: str = self.input_doc.original_file.stem
+        from pikepdf import Pdf
+
         with Pdf.open(self.pdf_file) as input_pdf:
             # Start with an empty document
             current_document: list[Page] = []
@@ -516,6 +518,8 @@ def scan_pdf(pdf_path: Path, settings: BarcodeConfig, work_dir: Path) -> list[Ba
     """
     barcodes: list[Barcode] = []
 
+    from pikepdf import Pdf
+
     with Pdf.open(pdf_path) as pdf:
         num_of_pages = len(pdf.pages)
     logger.debug(f"PDF has {num_of_pages} pages")

src/documents/converters.py

@@ -1,6 +1,5 @@
 from pathlib import Path
 
-import img2pdf
 from PIL import Image
 
 from documents.utils import copy_basic_file_stats
@@ -17,6 +16,8 @@ def convert_from_tiff_to_pdf(tiff_path: Path, target_directory: Path) -> Path:
 
     Returns the path of the PDF created.
     """
+    import img2pdf
+
     # override pixel setting if needed
     maybe_override_pixel_limit()
 

src/documents/double_sided.py

@@ -6,7 +6,6 @@ from pathlib import Path
 from typing import Final
 
 from django.conf import settings
-from pikepdf import Pdf
 
 from documents.consumer import ConsumerError
 from documents.converters import convert_from_tiff_to_pdf
@@ -85,6 +84,8 @@ class CollatePlugin(NoCleanupPluginMixin, NoSetupPluginMixin, ConsumeTaskPlugin)
                 valid_staging_exists = True
 
         if valid_staging_exists:
+            from pikepdf import Pdf
+
             try:
                 # Collate pages from second PDF in reverse order
                 with Pdf.open(staging) as pdf1, Pdf.open(pdf_file) as pdf2:

src/documents/filters.py

@@ -11,7 +11,6 @@ from decimal import InvalidOperation
 from typing import TYPE_CHECKING
 from typing import Any
 
-import dateparser
 from django.conf import settings
 from django.contrib.contenttypes.models import ContentType
 from django.db.models import Case
@@ -350,6 +349,8 @@ class DateQueryField(serializers.DateField):
             if not isinstance(value, str):
                 raise
 
+        import dateparser
+
         try:
             parsed = dateparser.parse(
                 value,

src/documents/pdf_ops.py

@@ -14,8 +14,6 @@ from __future__ import annotations
 from typing import TYPE_CHECKING
 from typing import NamedTuple
 
-import pikepdf
-
 if TYPE_CHECKING:
     from collections.abc import Callable
     from collections.abc import Iterable
@@ -43,6 +41,8 @@ def rotate_pdf(src: Path, dst: Path, degrees: int) -> None:
     Rotate every page relatively on the opened document, not a rebuild, so Info,
     XMP and outlines are kept. ``src`` is not modified.
     """
+    import pikepdf
+
     with pikepdf.open(src) as pdf:
         for page in pdf.pages:
             page.rotate(degrees, relative=True)
@@ -58,6 +58,8 @@ def remove_pages(src: Path, dst: Path, pages: Iterable[int]) -> None:
     Duplicates are ignored. Pages are removed highest first so earlier removals
     never shift the index of later ones.
     """
+    import pikepdf
+
     unique = sorted(set(pages))
     _require_positive(unique)
     with pikepdf.open(src) as pdf:
@@ -81,6 +83,8 @@ def build_pdfs(
     (Info, XMP, outlines) is not carried over. Returns the written paths in output
     order.
     """
+    import pikepdf
+
     for specs, _ in outputs:
         _require_positive(spec.page for spec in specs)
 
@@ -140,6 +144,8 @@ def needs_decrypt(src: Path) -> bool:
     True if ``src`` is encrypted. A PDF that needs a password to open at all
     counts as encrypted.
     """
+    import pikepdf
+
     try:
         with pikepdf.open(src) as pdf:
             return bool(pdf.is_encrypted)
@@ -154,6 +160,8 @@ def decrypt_pdf(src: Path, make_dst: Callable[[], Path], password: str) -> Path:
     ``make_dst`` is only called once the password has been accepted, so a wrong
     password never leaves a destination behind.
     """
+    import pikepdf
+
     with pikepdf.open(src, password=password) as pdf:
         pdf.remove_unreferenced_resources()
         dst = make_dst()
@@ -170,6 +178,8 @@ class PdfMerger:
     """
 
     def __init__(self) -> None:
+        import pikepdf
+
         self._pdf = pikepdf.new()
         self._version: str = self._pdf.pdf_version
 
@@ -185,6 +195,8 @@ class PdfMerger:
         self._pdf.close()
 
     def add(self, path: Path) -> None:
+        import pikepdf
+
         with pikepdf.open(str(path)) as pdf:
             self._version = max(self._version, pdf.pdf_version)
             self._pdf.pages.extend(pdf.pages)

src/documents/plugins/date_parsing/base.py

@@ -7,8 +7,6 @@ from dataclasses import dataclass
 from types import TracebackType
 from typing import Self
 
-import dateparser
-
 logger = logging.getLogger(__name__)
 
 
@@ -80,6 +78,8 @@ class DateParserPluginBase(ABC):
 
         Uses configuration from `self.config`.
         """
+        import dateparser
+
         try:
             return dateparser.parse(
                 date_string,

src/documents/tasks.py

@@ -76,9 +76,6 @@ from paperless.logging import consume_task_id
 from paperless.parsers import ParserContext
 from paperless.parsers.registry import get_parser_registry
 from paperless_ai.exceptions import LLMTimeoutError
-from paperless_ai.indexing import llm_index_add_or_update_document
-from paperless_ai.indexing import llm_index_remove_document
-from paperless_ai.indexing import update_llm_index
 
 if settings.AUDIT_LOG_ENABLED:
     from auditlog.models import LogEntry
@@ -339,6 +336,8 @@ def bulk_update_documents(document_ids) -> None:
 
     ai_config = AIConfig()
     if ai_config.llm_index_enabled:
+        from paperless_ai.indexing import update_llm_index
+
         update_llm_index(
             rebuild=False,
             document_ids=document_ids,
@@ -499,6 +498,8 @@ def update_document_content_maybe_archive_file(
 
             ai_config = AIConfig()
             if ai_config.llm_index_enabled:
+                from paperless_ai.indexing import llm_index_add_or_update_document
+
                 llm_index_add_or_update_document(document)
 
             clear_document_caches(document.pk)
@@ -800,11 +801,15 @@ def apply_ai_suggestions(self, action_id: int, document_id: int) -> None:
 
 @shared_task
 def update_document_in_llm_index(document) -> None:
+    from paperless_ai.indexing import llm_index_add_or_update_document
+
     llm_index_add_or_update_document(document)
 
 
 @shared_task
 def remove_document_from_llm_index(document) -> None:
+    from paperless_ai.indexing import llm_index_remove_document
+
     llm_index_remove_document(document)
 
 

src/documents/tests/date_parsing/test_date_parsing.py

@@ -71,7 +71,7 @@ class TestParseString:
         ):
             # We still need to mock dateparser.parse to force the exception
             mocker.patch(
-                "documents.plugins.date_parsing.base.dateparser.parse",
+                "dateparser.parse",
                 side_effect=ValueError(
                     "Parsing error: 01/01/2024",
                 ),
@@ -260,7 +260,7 @@ class TestRegexDateParser:
         parser = RegexDateParserPlugin(base_config)
 
         # Patch the dateparser.parse
-        target = "documents.plugins.date_parsing.base.dateparser.parse"
+        target = "dateparser.parse"
 
         def fake_parse(
             date_string: str,
@@ -335,7 +335,7 @@ class TestRegexDateParser:
         )
         parser = RegexDateParserPlugin(cfg)
 
-        target = "documents.plugins.date_parsing.base.dateparser.parse"
+        target = "dateparser.parse"
 
         def fake_parse(
             date_string: str,
@@ -400,7 +400,7 @@ class TestRegexDateParser:
         parser = RegexDateParserPlugin(cfg)
 
         # Patch the module's dateparser.parse so we can inspect calls
-        target = "documents.plugins.date_parsing.base.dateparser.parse"
+        target = "dateparser.parse"
 
         def fake_parse(
             date_string: str,

src/documents/tests/test_api_app_config.py

@@ -1070,7 +1070,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
 
         with (
             patch("documents.tasks.llmindex_index.apply_async") as mock_update,
-            patch("paperless.views.llm_index_exists") as mock_exists,
+            patch("paperless_ai.indexing.llm_index_exists") as mock_exists,
         ):
             mock_exists.return_value = False
             self.client.patch(
@@ -1095,7 +1095,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
 
         with (
             patch("documents.tasks.llmindex_index.apply_async") as mock_update,
-            patch("paperless.views.llm_index_exists", return_value=False),
+            patch("paperless_ai.indexing.llm_index_exists", return_value=False),
         ):
             self.client.patch(
                 f"{self.ENDPOINT}1/",
@@ -1115,7 +1115,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
 
         with (
             patch("documents.tasks.llmindex_index.apply_async") as mock_update,
-            patch("paperless.views.llm_index_exists") as mock_exists,
+            patch("paperless_ai.indexing.llm_index_exists") as mock_exists,
         ):
             mock_exists.return_value = True
             self.client.patch(
@@ -1136,7 +1136,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
 
         with (
             patch("documents.tasks.llmindex_index.apply_async") as mock_update,
-            patch("paperless.views.llm_index_exists") as mock_exists,
+            patch("paperless_ai.indexing.llm_index_exists") as mock_exists,
         ):
             mock_exists.return_value = True
             self.client.patch(
@@ -1174,7 +1174,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
 
         with (
             patch("documents.tasks.llmindex_index.apply_async") as mock_update,
-            patch("paperless.views.llm_index_exists") as mock_exists,
+            patch("paperless_ai.indexing.llm_index_exists") as mock_exists,
         ):
             mock_exists.return_value = True
             self.client.patch(

src/documents/tests/test_api_chat.py

@@ -61,7 +61,7 @@ class TestChatStreamingViewInputValidation(APITestCase):
                 return_value=self._mock_ai_enabled(),
             ),
             mock.patch(
-                "documents.views.stream_chat_with_documents",
+                "paperless_ai.chat.stream_chat_with_documents",
                 return_value=iter(chunks),
             ),
         ):
@@ -104,7 +104,7 @@ class TestChatStreamingViewUnrestrictedFlag:
         """
         mocker.patch("documents.views.AIConfig").return_value.ai_enabled = True
         return mocker.patch(
-            "documents.views.stream_chat_with_documents",
+            "paperless_ai.chat.stream_chat_with_documents",
             return_value=iter(()),
         )
 

src/documents/tests/test_api_document_versions.py

@@ -1110,7 +1110,7 @@ class TestVersionActionPermissions(DirectoriesMixin, APITestCase):
             self.assertEqual(response.status_code, 403)
             with (
                 mock.patch("documents.views.AIConfig") as ai_config,
-                mock.patch("documents.views.stream_chat_with_documents") as chat,
+                mock.patch("paperless_ai.chat.stream_chat_with_documents") as chat,
             ):
                 ai_config.return_value.ai_enabled = True
                 response = self.client.post(

src/documents/tests/test_api_documents.py

@@ -2686,7 +2686,7 @@ class TestDocumentApi(DirectoriesMixin, ConsumeTaskMixin, APITestCase):
         response = self.client.get("/api/documents/34676/suggestions/")
         self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
 
-    @mock.patch("documents.views.get_ai_document_classification")
+    @mock.patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(AI_ENABLED=True)
     def test_suggestions_still_uses_classifier_when_ai_enabled(
         self,

src/documents/tests/test_permission_filtering_security.py

@@ -572,7 +572,7 @@ class TestAiChatAllDocumentsPermissionBoundary:
     ENDPOINT = "/api/documents/chat/"
 
     @override_settings(AI_ENABLED=True)
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     def test_chat_all_documents_excludes_unshared_document(self, mock_stream_chat):
         mock_stream_chat.return_value = iter([b"data"])
 

src/documents/tests/test_tasks.py

@@ -322,7 +322,7 @@ class TestUpdateContent(DirectoriesMixin, TestCase):
         )
 
     @override_settings(AI_ENABLED=True, LLM_EMBEDDING_BACKEND="huggingface")
-    @mock.patch("documents.tasks.llm_index_add_or_update_document")
+    @mock.patch("paperless_ai.indexing.llm_index_add_or_update_document")
     @mock.patch("documents.search.get_backend")
     def test_update_content_version_updates_llm_index(
         self,
@@ -450,7 +450,7 @@ class TestAIIndex(DirectoriesMixin, TestCase):
             checksum="wow",
         )
         with mock.patch(
-            "documents.tasks.llm_index_add_or_update_document",
+            "paperless_ai.indexing.llm_index_add_or_update_document",
         ) as llm_index_add_or_update_document:
             tasks.update_document_in_llm_index(doc)
             llm_index_add_or_update_document.assert_called_once_with(doc)
@@ -470,7 +470,7 @@ class TestAIIndex(DirectoriesMixin, TestCase):
             checksum="wow",
         )
         with mock.patch(
-            "documents.tasks.llm_index_remove_document",
+            "paperless_ai.indexing.llm_index_remove_document",
         ) as llm_index_remove_document:
             tasks.remove_document_from_llm_index(doc)
             llm_index_remove_document.assert_called_once_with(doc)
@@ -495,7 +495,7 @@ class TestAIIndex(DirectoriesMixin, TestCase):
                 "documents.tasks.update_document_in_llm_index",
             ) as update_document_in_llm_index,
             mock.patch(
-                "documents.tasks.update_llm_index",
+                "paperless_ai.indexing.update_llm_index",
             ) as update_llm_index,
         ):
             doc_ids = [doc.pk for doc in docs]

src/documents/tests/test_views.py

@@ -474,7 +474,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
         self.assertEqual(response.json()["tags"], [])
         self.assertEqual(response.json()["suggested_tags"], [])
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -524,7 +524,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             None,
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -562,7 +562,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             "KI Title",
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -601,7 +601,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             "Titre IA",
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -637,7 +637,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             ),
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -709,7 +709,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             ),
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="openai-like",
@@ -738,7 +738,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             get_llm_suggestion_cache(self.document.pk, backend="openai-like"),
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="openai-like",
@@ -765,7 +765,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             get_llm_suggestion_cache(self.document.pk, backend="openai-like"),
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="openai-like",
@@ -797,7 +797,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             get_llm_suggestion_cache(self.document.pk, backend="openai-like"),
         )
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -835,7 +835,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
         self.assertEqual(response.json()["tags"], [self.tag1.pk])
         self.assertEqual(response.json()["suggested_tags"], ["Follow-up"])
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -874,7 +874,7 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
         self.assertEqual(response.json()["tags"], [self.tag1.pk])
         self.assertEqual(response.json()["suggested_tags"], [])
 
-    @patch("documents.views.get_ai_document_classification")
+    @patch("paperless_ai.ai_classifier.get_ai_document_classification")
     @override_settings(
         AI_ENABLED=True,
         LLM_BACKEND="mock_backend",
@@ -996,7 +996,7 @@ class TestAIChatStreamingView(DirectoriesMixin, TestCase):
         self.assertEqual(response.status_code, 400)
         self.assertIn(b"AI is required for this feature", response.content)
 
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     @patch("documents.views.permitted_document_ids")
     @override_settings(AI_ENABLED=True)
     def test_post_no_document_id(self, mock_permitted_ids, mock_stream_chat) -> None:
@@ -1016,7 +1016,7 @@ class TestAIChatStreamingView(DirectoriesMixin, TestCase):
         self.assertEqual(list(call_kwargs["documents"]), [self.document])
         self.assertIsNone(call_kwargs["output_language"])
 
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     @patch("documents.views.permitted_document_ids")
     @override_settings(AI_ENABLED=True)
     def test_post_uses_user_display_language(
@@ -1042,7 +1042,7 @@ class TestAIChatStreamingView(DirectoriesMixin, TestCase):
         self.assertEqual(list(call_kwargs["documents"]), [self.document])
         self.assertEqual(call_kwargs["output_language"], "de-de")
 
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     @override_settings(AI_ENABLED=True)
     def test_post_with_document_id(self, mock_stream_chat) -> None:
         self.grant_view_document_permission()
@@ -1079,7 +1079,7 @@ class TestAIChatStreamingView(DirectoriesMixin, TestCase):
         self.assertEqual(response.status_code, 403)
         self.assertIn(b"Insufficient permissions", response.content)
 
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     @override_settings(AI_ENABLED=True)
     def test_post_no_document_id_requires_view_document_permission(
         self,
@@ -1093,7 +1093,7 @@ class TestAIChatStreamingView(DirectoriesMixin, TestCase):
         self.assertEqual(response.status_code, 403)
         mock_stream_chat.assert_not_called()
 
-    @patch("documents.views.stream_chat_with_documents")
+    @patch("paperless_ai.chat.stream_chat_with_documents")
     @override_settings(AI_ENABLED=True)
     def test_post_with_document_id_requires_view_document_permission(
         self,

src/documents/views.py

@@ -250,20 +250,8 @@ from paperless.parsers.remote import RemoteEngineConfig
 from paperless.serialisers import GroupSerializer
 from paperless.serialisers import UserSerializer
 from paperless.views import StandardPagination
-from paperless_ai.ai_classifier import get_ai_document_classification
-from paperless_ai.ai_classifier import get_llm_output_language
-from paperless_ai.chat import stream_chat_with_documents
 from paperless_ai.exceptions import LLMProviderError
 from paperless_ai.exceptions import LLMTimeoutError
-from paperless_ai.matching import extract_unmatched_names
-from paperless_ai.matching import match_correspondents_by_name
-from paperless_ai.matching import match_document_types_by_name
-from paperless_ai.matching import match_storage_paths_by_name
-from paperless_ai.matching import match_tags_by_name
-from paperless_ai.matching import resolve_correspondent_ids
-from paperless_ai.matching import resolve_document_type_ids
-from paperless_ai.matching import resolve_storage_path_ids
-from paperless_ai.matching import resolve_tag_ids
 from paperless_mail.models import MailAccount
 from paperless_mail.models import MailRule
 from paperless_mail.oauth import PaperlessMailOAuth2Manager
@@ -1613,6 +1601,18 @@ class DocumentViewSet(
         if not ai_config.ai_enabled:
             return HttpResponseBadRequest("AI is required for this feature")
 
+        from paperless_ai.ai_classifier import get_ai_document_classification
+        from paperless_ai.ai_classifier import get_llm_output_language
+        from paperless_ai.matching import extract_unmatched_names
+        from paperless_ai.matching import match_correspondents_by_name
+        from paperless_ai.matching import match_document_types_by_name
+        from paperless_ai.matching import match_storage_paths_by_name
+        from paperless_ai.matching import match_tags_by_name
+        from paperless_ai.matching import resolve_correspondent_ids
+        from paperless_ai.matching import resolve_document_type_ids
+        from paperless_ai.matching import resolve_storage_path_ids
+        from paperless_ai.matching import resolve_tag_ids
+
         output_language = get_llm_output_language(
             ai_config=ai_config,
             user=request.user,
@@ -2446,6 +2446,9 @@ class ChatStreamingView(GenericAPIView[Any]):
             )
             unrestricted = user_is_unrestricted(request.user)
 
+        from paperless_ai.ai_classifier import get_llm_output_language
+        from paperless_ai.chat import stream_chat_with_documents
+
         output_language = get_llm_output_language(
             ai_config=ai_config,
             user=request.user,

src/paperless/asgi.py

@@ -17,7 +17,7 @@ from paperless.urls import websocket_urlpatterns  # noqa: E402
 
 application = ProtocolTypeRouter(
     {
-        "http": get_asgi_application(),
+        "http": django_asgi_app,
         "websocket": AuthMiddlewareStack(URLRouter(websocket_urlpatterns)),
     },
 )

src/paperless/parsers/mail.py

@@ -36,9 +36,6 @@ from humanize import naturalsize
 from imap_tools import MailAttachment
 from imap_tools import MailMessage
 from tika_client import TikaClient
-from turbohtml.clean import Linkify
-from turbohtml.clean import linkify
-from turbohtml.migration.bleach import clean
 
 from documents.parsers import ParseError
 from documents.parsers import make_thumbnail_from_pdf
@@ -622,6 +619,10 @@ class MailDocumentParser:
 
         def clean_html(text: str) -> str:
             """Attempt to clean, escape, and linkify the given HTML string."""
+            from turbohtml.clean import Linkify
+            from turbohtml.clean import linkify
+            from turbohtml.migration.bleach import clean
+
             if isinstance(text, list):
                 text = "\n".join([str(e) for e in text])
             if not isinstance(text, str):

src/paperless/parsers/tesseract.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import functools
 import importlib.resources
 import logging
 import os
@@ -39,9 +40,12 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger("paperless.parsing.tesseract")
 
-_SRGB_ICC_DATA: Final[bytes] = (
-    importlib.resources.files("ocrmypdf.data").joinpath("sRGB.icc").read_bytes()
-)
+
+@functools.cache
+def _srgb_icc_data() -> bytes:
+    # Read on first use, importing ocrmypdf.data pulls in the whole ocrmypdf package
+    return importlib.resources.files("ocrmypdf.data").joinpath("sRGB.icc").read_bytes()
+
 
 _SUPPORTED_MIME_TYPES: Final[dict[str, str]] = {
     "application/pdf": ".pdf",
@@ -403,7 +407,7 @@ class RasterisedDocumentParser:
         pdfa_path = Path(self.tempdir) / "archive.pdf"
         try:
             with pikepdf.open(plain_pdf_path) as pdf:
-                cs = pdf.make_stream(_SRGB_ICC_DATA)
+                cs = pdf.make_stream(_srgb_icc_data())
                 cs["/N"] = 3
                 output_intent = pikepdf.Dictionary(
                     Type=pikepdf.Name("/OutputIntent"),

src/paperless/settings/custom.py

@@ -6,7 +6,6 @@ from pathlib import Path
 from typing import Any
 
 from celery.schedules import crontab
-from dateparser.languages.loader import LocaleDataLoader
 
 from paperless.settings.parsers import get_choice_from_env
 from paperless.settings.parsers import get_int_from_env
@@ -352,6 +351,8 @@ def parse_dateparser_languages(languages: str | None) -> list[str]:
             )
             language_list.append("zh")
 
+    from dateparser.languages.loader import LocaleDataLoader
+
     return list(LocaleDataLoader().get_locale_map(locales=language_list))
 
 

src/paperless/tests/test_utils.py

@@ -59,7 +59,10 @@ def test_ocr_to_dateparser_languages_exception(
             raise RuntimeError("Simulated error")
 
     with caplog.at_level(logging.WARNING):
-        monkeypatch.setattr(utils, "LocaleDataLoader", lambda: DummyLoader())
+        monkeypatch.setattr(
+            "dateparser.languages.loader.LocaleDataLoader",
+            lambda: DummyLoader(),
+        )
         result = utils.ocr_to_dateparser_languages("eng+fra")
         assert result == []
         assert (

src/paperless/utils.py

@@ -1,7 +1,5 @@
 import logging
 
-from dateparser.languages.loader import LocaleDataLoader
-
 logger = logging.getLogger("paperless.utils")
 
 OCR_TO_DATEPARSER_LANGUAGES = {
@@ -125,6 +123,8 @@ def ocr_to_dateparser_languages(ocr_languages: str) -> list[str]:
     - If a language cannot be mapped or validated, it is skipped with a warning.
     - Returns a list of valid locales, or an empty list if none could be converted.
     """
+    from dateparser.languages.loader import LocaleDataLoader
+
     loader = LocaleDataLoader()
     result = []
     try:

src/paperless/validators.py

@@ -3,7 +3,6 @@ from io import BytesIO
 from django.conf import settings
 from django.core.exceptions import ValidationError
 from django.core.files.uploadedfile import UploadedFile
-from lxml import etree
 from PIL import Image
 
 ALLOWED_SVG_TAGS: set[str] = {
@@ -186,6 +185,7 @@ def reject_dangerous_svg(file: UploadedFile) -> None:
     Raises ValidationError if unsafe content is found.
     See GHSA-6p53-hqqw-8j62
     """
+    from lxml import etree
 
     try:
         parser = etree.XMLParser(resolve_entities=False)

src/paperless/views.py

@@ -49,7 +49,6 @@ from paperless.serialisers import GroupSerializer
 from paperless.serialisers import PaperlessAuthTokenSerializer
 from paperless.serialisers import ProfileSerializer
 from paperless.serialisers import UserSerializer
-from paperless_ai.indexing import llm_index_exists
 
 
 class PaperlessObtainAuthTokenView(ObtainAuthToken):
@@ -471,6 +470,8 @@ class ApplicationConfigurationViewSet(ModelViewSet[ApplicationConfiguration]):
             or old_llm_embedding_model != new_llm_embedding_model
             or old_llm_context_size != new_llm_context_size
         )
+        from paperless_ai.indexing import llm_index_exists
+
         rebuild_needed = new_ai_index_enabled and (
             not llm_index_exists() or embedding_config_changed
         )