Fix: drop redundant change perms checks on cards open button (#14601 )

Chore: fix a couple e2e warnings / flakiness (#14596 )
Chore: Skip repeating Django system checks in the Docker consumer, scheduler and worker (#14598 )
2026-10-31 00:35:19 +00:00 · 2026-10-30 16:16:48 -07:00 · 2026-10-30 15:45:46 -07:00 · 2026-10-30 22:37:37 +00:00 · 2026-10-30 22:26:54 +00:00 · 2026-10-30 22:25:42 +00:00
57 changed files with 290 additions and 385 deletions
@@ -10,9 +10,11 @@ if [[ -n "${PAPERLESS_CONSUMER_DISABLE}" ]]; then
 else
 	cd ${PAPERLESS_SRC_DIR}

+	# Django system checks already ran once in init-system-checks, skip repeating them
+
 	if [[ -n "${USER_IS_NON_ROOT}" ]]; then
-		exec python3 manage.py document_consumer
+		exec python3 manage.py document_consumer --skip-checks
 	else
-		exec s6-setuidgid paperless python3 manage.py document_consumer
+		exec s6-setuidgid paperless python3 manage.py document_consumer --skip-checks
 	fi
 fi
@@ -3,8 +3,9 @@

 cd ${PAPERLESS_SRC_DIR}

+# Django system checks already ran once in init-system-checks, skip repeating them
 if [[ -n "${USER_IS_NON_ROOT}" ]]; then
-	exec /usr/local/bin/celery --app paperless beat --loglevel INFO
+	exec /usr/local/bin/celery --skip-checks --app paperless beat --loglevel INFO
 else
-	exec s6-setuidgid paperless /usr/local/bin/celery --app paperless beat --loglevel INFO
+	exec s6-setuidgid paperless /usr/local/bin/celery --skip-checks --app paperless beat --loglevel INFO
 fi
@@ -3,8 +3,9 @@

 cd ${PAPERLESS_SRC_DIR}

+# Django system checks already ran once in init-system-checks, skip repeating them
 if [[ -n "${USER_IS_NON_ROOT}" ]]; then
-	exec /usr/local/bin/celery --app paperless worker --loglevel INFO --without-mingle --without-gossip
+	exec /usr/local/bin/celery --skip-checks --app paperless worker --loglevel INFO --without-mingle --without-gossip
 else
-	exec s6-setuidgid paperless /usr/local/bin/celery --app paperless worker --loglevel INFO --without-mingle --without-gossip
+	exec s6-setuidgid paperless /usr/local/bin/celery --skip-checks --app paperless worker --loglevel INFO --without-mingle --without-gossip
 fi
@@ -43,6 +43,7 @@ def configure_environment(instance_root: Path) -> None:


 def seed_database() -> None:
+    from django.contrib.auth.models import Permission
    from django.contrib.auth.models import User
    from django.core.management import call_command
    from django.utils import timezone
@@ -63,11 +64,18 @@ def seed_database() -> None:
        username="playwright",
        password="playwright",  # NOSONAR
    )
-    User.objects.create_user(
+    viewer = User.objects.create_user(
        username="viewer",
        password="viewer",  # NOSONAR
    )

+    viewer.user_permissions.add(
+        Permission.objects.get(
+            content_type__app_label="documents",
+            codename="view_uisettings",
+        ),
+    )
+
    inbox = Tag.objects.create(name="Inbox", is_inbox_tag=True, owner=admin)
    quick_filter = Tag.objects.create(name="Another Sample Tag", owner=admin)
    Tag.objects.create(name="TagWithPartial", owner=admin)
@@ -80,12 +80,21 @@ test('should show a list of notes', async ({ page }) => {

 test('should support quick filters', async ({ page }) => {
  await page.goto('/documents/1/details')
+  // filter button is available before tags have loaded
+  const selectedTags = page.locator('pngx-input-tags .ng-value pngx-tag')
+  await expect(selectedTags).toHaveCount(2)
+  await expect(
+    selectedTags.getByText('Another Sample Tag', { exact: true })
+  ).toBeVisible()
+  await expect(selectedTags.getByText('Inbox', { exact: true })).toBeVisible()
  await page
    .getByRole('button', { name: 'Filter documents with these Tags' })
    .click()
-  await expect(page).toHaveURL(
-    /tags__id__all=2,1&sort=created&reverse=1&page=1/
-  )
+  await expect(page).toHaveURL((url) => {
+    const tags = url.searchParams.get('tags__id__all')?.split(',').sort()
+    return url.pathname === '/documents' && tags?.join(',') === '1,2'
+  })
+  await expect(page.locator('pngx-document-list')).toHaveText(/one document/i)
 })

 test('should finish reloading the preview after a remote document update', async ({
@@ -6,8 +6,15 @@ test.use({
  },
 })

-test('should not allow user to edit settings', async ({ page }) => {
+test.beforeEach(async ({ page }) => {
+  const settingsLoaded = page.waitForResponse((response) =>
+    response.url().endsWith('/api/ui_settings/')
+  )
  await page.goto('/dashboard')
+  expect((await settingsLoaded).status()).toBe(200)
+})
+
+test('should not allow user to edit settings', async ({ page }) => {
  await expect(page.getByRole('link', { name: 'Settings' })).not.toBeAttached()
  await page.goto('/settings')
  await expect(page.locator('body')).toHaveText(
@@ -16,7 +23,6 @@ test('should not allow user to edit settings', async ({ page }) => {
 })

 test('should not allow user to view documents', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(
    page.locator('nav').getByRole('link', { name: 'Documents' })
  ).not.toBeAttached()
@@ -31,7 +37,6 @@ test('should not allow user to view documents', async ({ page }) => {
 })

 test('should not allow user to view correspondents', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(
    page.getByRole('link', { name: 'Attributes' })
  ).not.toBeAttached()
@@ -42,7 +47,6 @@ test('should not allow user to view correspondents', async ({ page }) => {
 })

 test('should not allow user to view tags', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(
    page.getByRole('link', { name: 'Attributes' })
  ).not.toBeAttached()
@@ -53,7 +57,6 @@ test('should not allow user to view tags', async ({ page }) => {
 })

 test('should not allow user to view document types', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(
    page.getByRole('link', { name: 'Attributes' })
  ).not.toBeAttached()
@@ -64,7 +67,6 @@ test('should not allow user to view document types', async ({ page }) => {
 })

 test('should not allow user to view storage paths', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(
    page.getByRole('link', { name: 'Attributes' })
  ).not.toBeAttached()
@@ -75,7 +77,6 @@ test('should not allow user to view storage paths', async ({ page }) => {
 })

 test('should not allow user to view logs', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(page.getByRole('link', { name: 'Logs' })).not.toBeAttached()
  await page.goto('/logs')
  await expect(page.locator('body')).toHaveText(
@@ -84,7 +85,6 @@ test('should not allow user to view logs', async ({ page }) => {
 })

 test('should not allow user to view tasks', async ({ page }) => {
-  await page.goto('/dashboard')
  await expect(page.getByRole('link', { name: 'Tasks' })).not.toBeAttached()
  await page.goto('/tasks')
  await expect(page.locator('body')).toHaveText(
@@ -10334,18 +10334,18 @@
          <context context-type="linenumber">294,295</context>
        </context-group>
      </trans-unit>
-      <trans-unit id="2179847500064178686" datatype="html">
-        <source>Edit document</source>
+      <trans-unit id="3420321797707163677" datatype="html">
+        <source>Preview document</source>
        <context-group purpose="location">
          <context context-type="sourcefile">src/app/components/document-list/document-list.component.html</context>
          <context context-type="linenumber">328</context>
        </context-group>
      </trans-unit>
-      <trans-unit id="3420321797707163677" datatype="html">
-        <source>Preview document</source>
+      <trans-unit id="2179847500064178686" datatype="html">
+        <source>Edit document</source>
        <context-group purpose="location">
          <context context-type="sourcefile">src/app/components/document-list/document-list.component.html</context>
-          <context context-type="linenumber">329</context>
+          <context context-type="linenumber">331</context>
        </context-group>
      </trans-unit>
      <trans-unit id="4512084577073831437" datatype="html">
@@ -465,7 +465,7 @@
    @switch (archiveContentRenderType) {
      @case (ContentRenderType.PDF) {
        @if (!useNativePdfViewer) {
-          <div class="preview-sticky pdf-viewer-container">
+          <div class="preview-sticky pdf-viewer-container" tabindex="-1">
            <pngx-pdf-viewer
              [src]="pdfSource()"
              [sourceRevision]="previewRevision()"
@@ -477,7 +477,7 @@
              (loadError)="onError($event)"
              (afterLoadComplete)="pdfPreviewLoaded($event)">
            </pngx-pdf-viewer>
-            <div class="position-absolute bottom-0 start-50 translate-middle-x z-1 mb-3 d-flex gap-2 p-1 bg-body border rounded shadow">
+            <div class="preview-controls position-absolute bottom-0 start-50 translate-middle-x z-1 mb-3 d-flex gap-2 p-1 bg-body border rounded shadow">
              @if (previewNumPages()) {
                <div class="input-group input-group-sm w-auto flex-nowrap">
                  <input class="form-control flex-grow-0 w-auto" type="number" min="1" [max]="previewNumPages()" [ngModelOptions]="{standalone: true}" [ngModel]="previewCurrentPage()" (ngModelChange)="previewCurrentPage.set($event)" aria-label="Page" i18n-aria-label />
@@ -12,6 +12,22 @@
    width: 100%;
    height: 100%;
  }
+
+  &:focus {
+    outline: none;
+  }
+
+  .preview-controls {
+    opacity: 0;
+    pointer-events: none;
+    transition: opacity 0.15s ease-in-out;
+  }
+
+  &:hover .preview-controls,
+  &:focus-within .preview-controls {
+    opacity: 1;
+    pointer-events: auto;
+  }
 }

 .btn-group .dropdown-toggle-split {
@@ -72,7 +72,7 @@
              <a class="btn btn-sm btn-outline-secondary" (click)="clickMoreLike.emit()">
                <i-bs name="diagram-3" class="me-1"></i-bs><span class="d-none d-md-inline" i18n>More like this</span>
              </a>
-              <a routerLink="/documents/{{document().id}}" class="btn btn-sm btn-outline-secondary" *pngxIfPermissions="{ action: PermissionAction.Change, type: PermissionType.Document }">
+              <a routerLink="/documents/{{document().id}}" class="btn btn-sm btn-outline-secondary">
                <i-bs name="file-earmark-richtext" class="me-1"></i-bs><span class="d-none d-md-inline" i18n>Open</span>
              </a>
              <pngx-preview-popup [document]="document()" #popupPreview>
@@ -10,7 +10,6 @@ import {
  NgbTooltipModule,
 } from '@ng-bootstrap/ng-bootstrap'
 import { NgxBootstrapIconsModule, allIcons } from 'ngx-bootstrap-icons'
-import { IfPermissionsDirective } from 'src/app/directives/if-permissions.directive'
 import { CustomDatePipe } from 'src/app/pipes/custom-date.pipe'
 import { DocumentTitlePipe } from 'src/app/pipes/document-title.pipe'
 import { IsNumberPipe } from 'src/app/pipes/is-number.pipe'
@@ -52,7 +51,6 @@ describe('DocumentCardLargeComponent', () => {
        DocumentCardLargeComponent,
        DocumentTitlePipe,
        CustomDatePipe,
-        IfPermissionsDirective,
        SafeUrlPipe,
        IsNumberPipe,
        PreviewPopupComponent,
@@ -20,7 +20,6 @@ import {
  Document,
 } from 'src/app/data/document'
 import { SETTINGS_KEYS } from 'src/app/data/ui-settings'
-import { IfPermissionsDirective } from 'src/app/directives/if-permissions.directive'
 import { CorrespondentNamePipe } from 'src/app/pipes/correspondent-name.pipe'
 import { CustomDatePipe } from 'src/app/pipes/custom-date.pipe'
 import { DocumentTitlePipe } from 'src/app/pipes/document-title.pipe'
@@ -51,7 +50,6 @@ import { LoadingComponentWithPermissions } from '../../loading-component/loading
    CorrespondentNamePipe,
    DocumentTypeNamePipe,
    StoragePathNamePipe,
-    IfPermissionsDirective,
    CustomDatePipe,
    RouterModule,
    NgbTooltipModule,
@@ -144,7 +144,7 @@
      <div class="d-flex justify-content-between align-items-center">
        <div class="btn-group w-100">
           @if (document()) {
-            <a routerLink="/documents/{{document().id}}" class="btn btn-sm btn-outline-secondary" title="Open" i18n-title *pngxIfPermissions="{ action: PermissionAction.Change, type: PermissionType.Document }" i18n-title>
+            <a routerLink="/documents/{{document().id}}" class="btn btn-sm btn-outline-secondary" title="Open" i18n-title>
              <i-bs name="file-earmark-richtext"></i-bs>
            </a>
            <pngx-preview-popup [document]="document()" #popupPreview>
@@ -20,7 +20,6 @@ import {
  Document,
 } from 'src/app/data/document'
 import { SETTINGS_KEYS } from 'src/app/data/ui-settings'
-import { IfPermissionsDirective } from 'src/app/directives/if-permissions.directive'
 import { CorrespondentNamePipe } from 'src/app/pipes/correspondent-name.pipe'
 import { CustomDatePipe } from 'src/app/pipes/custom-date.pipe'
 import { DocumentTitlePipe } from 'src/app/pipes/document-title.pipe'
@@ -51,7 +50,6 @@ import { LoadingComponentWithPermissions } from '../../loading-component/loading
    CorrespondentNamePipe,
    DocumentTypeNamePipe,
    StoragePathNamePipe,
-    IfPermissionsDirective,
    CustomDatePipe,
    RouterModule,
    NgbTooltipModule,
@@ -208,7 +208,7 @@
                  i18n>Correspondent</th>
              }
              @if (activeDisplayFields.includes(DisplayField.TITLE)) {
-                <th class="cursor-pointer"
+                <th class="cursor-pointer ps-5"
                  pngxSortable="title"
                  title="Sort by title" i18n-title
                  [currentSortField]="list.sortField"
@@ -322,13 +322,13 @@
                  </td>
                }
                @if (activeDisplayFields.includes(DisplayField.TITLE) || activeDisplayFields.includes(DisplayField.TAGS)) {
-                  <td width="30%">
+                  <td width="30%" class="position-relative" [class.ps-5]="activeDisplayFields.includes(DisplayField.TITLE)">
                    @if (activeDisplayFields.includes(DisplayField.TITLE)) {
                      <div class="d-inline-block" (mouseleave)="popupPreview.close()">
-                        <a routerLink="/documents/{{d.id}}" title="Edit document" i18n-title style="overflow-wrap: anywhere;">{{d.title | documentTitle}}</a>
-                        <pngx-preview-popup [document]="d" linkClasses="btn btn-sm btn-link text-secondary" linkTitle="Preview document" (click)="$event.stopPropagation()" i18n-linkTitle #popupPreview>
+                        <pngx-preview-popup class="position-absolute start-0" [document]="d" linkClasses="btn btn-sm btn-link text-secondary" linkTitle="Preview document" (click)="$event.stopPropagation()" i18n-linkTitle #popupPreview>
                          <i-bs name="eye"></i-bs>
                        </pngx-preview-popup>
+                        <a routerLink="/documents/{{d.id}}" title="Edit document" i18n-title style="overflow-wrap: anywhere;">{{d.title | documentTitle}}</a>
                      </div>
                    }
                    @if (activeDisplayFields.includes(DisplayField.TAGS)) {
@@ -87,6 +87,10 @@ $paperless-card-breakpoints: (
  }
 }

+.table .ps-5 {
+  padding-left: 2.2rem !important;
+}
+
 a {
  cursor: pointer;
 }
@@ -77,30 +77,6 @@ def _clear_django_caches() -> None:
        cache.clear()


-@pytest.fixture(autouse=True)
-def _keep_connection_open_in_atomic(monkeypatch: pytest.MonkeyPatch) -> None:
-    """Make ``close_old_connections()`` a no-op inside the test's transaction.
-
-    The Celery signal handlers call it, as a worker would between tasks. Tests
-    run inside an open atomic block, where Postgres really closes the
-    connection, taking the test transaction and every later test on the
-    worker with it. SQLite in-memory ignores the close, which hid this.
-    """
-    from django.db.backends.base.base import BaseDatabaseWrapper
-
-    original = BaseDatabaseWrapper.close_if_unusable_or_obsolete
-
-    def close_unless_atomic(self: BaseDatabaseWrapper) -> None:
-        if not self.in_atomic_block:
-            original(self)
-
-    monkeypatch.setattr(
-        BaseDatabaseWrapper,
-        "close_if_unusable_or_obsolete",
-        close_unless_atomic,
-    )
-
-
@pytest.fixture
 def paperless_dirs(
    tmp_path: Path,
@@ -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")
@@ -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()

@@ -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:
@@ -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,
@@ -3,6 +3,7 @@ from __future__ import annotations
 import logging
 import re
 from fnmatch import fnmatch
+from fnmatch import translate as fnmatch_translate
 from typing import TYPE_CHECKING

 from rest_framework import serializers
@@ -547,48 +548,6 @@ def existing_document_matches_workflow(
    return (True, None)


-def glob_to_database_regex(pattern: str) -> str:
-    """
-    Translates an fnmatch style glob into a regular expression which the
-    database can evaluate, matching the whole value.
-
-    fnmatch.translate is not usable here: it emits syntax such as (?s:...),
-    atomic groups and named group backreferences, which PostgreSQL rejects.
-    Only constructs common to Python, PostgreSQL and MariaDB are produced.
-    """
-    parts: list[str] = []
-    i = 0
-    n = len(pattern)
-    while i < n:
-        char = pattern[i]
-        i += 1
-        if char == "*":
-            parts.append(".*")
-        elif char == "?":
-            parts.append(".")
-        elif char == "[":
-            j = i
-            if j < n and pattern[j] == "!":
-                j += 1
-            if j < n and pattern[j] == "]":
-                j += 1
-            while j < n and pattern[j] != "]":
-                j += 1
-            if j >= n:
-                parts.append(r"\[")
-            else:
-                stuff = pattern[i:j].replace("\\", r"\\").replace("[", r"\[")
-                i = j + 1
-                if stuff.startswith("!"):
-                    stuff = "^" + stuff[1:]
-                elif stuff.startswith("^"):
-                    stuff = "\\" + stuff
-                parts.append(f"[{stuff}]")
-        else:
-            parts.append(re.escape(char))
-    return "^" + "".join(parts) + "$"
-
-
 def prefilter_documents_by_workflowtrigger(
    documents: QuerySet[Document],
    trigger: WorkflowTrigger,
@@ -669,9 +628,8 @@ def prefilter_documents_by_workflowtrigger(
        documents = documents.annotate(**annotations).filter(custom_field_q)

    if trigger.filter_filename:
-        documents = documents.filter(
-            original_filename__iregex=glob_to_database_regex(trigger.filter_filename),
-        )
+        regex = fnmatch_translate(trigger.filter_filename).lstrip("^").rstrip("$")
+        documents = documents.filter(original_filename__iregex=regex)

    return documents

@@ -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)
@@ -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,
@@ -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)


@@ -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,
@@ -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(
@@ -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(()),
        )

@@ -1062,7 +1062,7 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
                "custom_fields": [
                    {
                        "field": custom_field_doclink.id,
-                        "value": [doc2.id, doc3.id, doc4.id],
+                        "value": [2, 3, 4],
                    },
                ],
            },
@@ -1073,13 +1073,13 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
        self.assertEqual(CustomFieldInstance.objects.count(), 4)
        _cf_5 = doc2.custom_fields.first()
        assert _cf_5 is not None
-        self.assertEqual(_cf_5.value, [doc1.id])
+        self.assertEqual(_cf_5.value, [1])
        _cf_6 = doc3.custom_fields.first()
        assert _cf_6 is not None
-        self.assertEqual(_cf_6.value, [doc1.id])
+        self.assertEqual(_cf_6.value, [1])
        _cf_7 = doc4.custom_fields.first()
        assert _cf_7 is not None
-        self.assertEqual(_cf_7.value, [doc1.id])
+        self.assertEqual(_cf_7.value, [1])

        # Add links appends if necessary
        resp = self.client.patch(
@@ -1088,7 +1088,7 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
                "custom_fields": [
                    {
                        "field": custom_field_doclink.id,
-                        "value": [doc1.id, doc4.id],
+                        "value": [1, 4],
                    },
                ],
            },
@@ -1098,7 +1098,7 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
        self.assertEqual(resp.status_code, status.HTTP_200_OK)
        _cf_8 = doc4.custom_fields.first()
        assert _cf_8 is not None
-        self.assertEqual(_cf_8.value, [doc1.id, doc3.id])
+        self.assertEqual(_cf_8.value, [1, 3])

        # Remove one of the links, removed on other doc
        resp = self.client.patch(
@@ -1107,7 +1107,7 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
                "custom_fields": [
                    {
                        "field": custom_field_doclink.id,
-                        "value": [doc2.id, doc3.id],
+                        "value": [2, 3],
                    },
                ],
            },
@@ -1117,13 +1117,13 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
        self.assertEqual(resp.status_code, status.HTTP_200_OK)
        _cf_9 = doc2.custom_fields.first()
        assert _cf_9 is not None
-        self.assertEqual(_cf_9.value, [doc1.id])
+        self.assertEqual(_cf_9.value, [1])
        _cf_10 = doc3.custom_fields.first()
        assert _cf_10 is not None
-        self.assertEqual(_cf_10.value, [doc1.id, doc4.id])
+        self.assertEqual(_cf_10.value, [1, 4])
        _cf_11 = doc4.custom_fields.first()
        assert _cf_11 is not None
-        self.assertEqual(_cf_11.value, [doc3.id])
+        self.assertEqual(_cf_11.value, [3])

        # Removes the field entirely
        resp = self.client.patch(
@@ -1140,10 +1140,10 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
        self.assertEqual(_cf_12.value, [])
        _cf_13 = doc3.custom_fields.first()
        assert _cf_13 is not None
-        self.assertEqual(_cf_13.value, [doc4.id])
+        self.assertEqual(_cf_13.value, [4])
        _cf_14 = doc4.custom_fields.first()
        assert _cf_14 is not None
-        self.assertEqual(_cf_14.value, [doc3.id])
+        self.assertEqual(_cf_14.value, [3])

        # If field exists on target doc but value is None
        doc5 = Document.objects.create(
@@ -1170,7 +1170,7 @@ class TestCustomFieldsAPI(DirectoriesMixin, APITestCase):
        self.assertEqual(resp.status_code, status.HTTP_200_OK)
        _cf_15 = doc5.custom_fields.first()
        assert _cf_15 is not None
-        self.assertEqual(_cf_15.value, [doc1.id])
+        self.assertEqual(_cf_15.value, [1])

    def test_documentlink_patch_requires_change_permission_on_target_documents(
        self,
@@ -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(
@@ -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,
@@ -6,7 +6,6 @@ from unittest.mock import patch
 from urllib.parse import quote

 import time_machine
-from django.db import connection
 from django.test import override_settings
 from django.utils import timezone
 from rest_framework.test import APITestCase
@@ -21,13 +20,6 @@ from paperless_testing.dirs import DirectoriesMixin
 from paperless_testing.factories import UserFactory


-def _exact_equal(value: str | None, expected: str) -> bool:
-    """`exact` follows the column collation, which is case insensitive on MariaDB."""
-    if connection.vendor == "mysql" and value is not None:
-        return value.casefold() == expected.casefold()
-    return value == expected
-
-
 class DocumentWrapper:
    """
    Allows Pythonic access to the custom fields associated with the wrapped document.
@@ -295,8 +287,7 @@ class TestCustomFieldsSearch(DirectoriesMixin, APITestCase):
        self._assert_query_match_predicate(
            ["string_field", "exact", "paperless"],
            lambda document: (
-                "string_field" in document
-                and _exact_equal(document["string_field"], "paperless")
+                "string_field" in document and document["string_field"] == "paperless"
            ),
        )

@@ -305,10 +296,7 @@ class TestCustomFieldsSearch(DirectoriesMixin, APITestCase):
            ["string_field", "in", ["paperless", "Paperless"]],
            lambda document: (
                "string_field" in document
-                and any(
-                    _exact_equal(document["string_field"], expected)
-                    for expected in ("paperless", "Paperless")
-                )
+                and document["string_field"] in ("paperless", "Paperless")
            ),
        )

@@ -640,23 +628,22 @@ class TestCustomFieldsSearch(DirectoriesMixin, APITestCase):
    # Subset check (document link field only)                   #
    # ==========================================================#
    def test_document_link_contains(self) -> None:
-        linked = {self.documents[1].id, self.documents[2].id}
        # Document link field "contains" performs a subset check.
        self._assert_query_match_predicate(
-            ["documentlink_field", "contains", sorted(linked)],
+            ["documentlink_field", "contains", [1, 2]],
            lambda document: (
                "documentlink_field" in document
                and document["documentlink_field"] is not None
-                and set(document["documentlink_field"]) >= linked
+                and set(document["documentlink_field"]) >= {1, 2}
            ),
        )
        # The order of IDs don't matter - this is the same as above.
        self._assert_query_match_predicate(
-            ["documentlink_field", "contains", sorted(linked, reverse=True)],
+            ["documentlink_field", "contains", [2, 1]],
            lambda document: (
                "documentlink_field" in document
                and document["documentlink_field"] is not None
-                and set(document["documentlink_field"]) >= linked
+                and set(document["documentlink_field"]) >= {1, 2}
            ),
        )

@@ -773,7 +760,7 @@ class TestCustomFieldsSearch(DirectoriesMixin, APITestCase):
            lambda document: (
                not (
                    "string_field" in document
-                    and _exact_equal(document["string_field"], "paperless")
+                    and document["string_field"] == "paperless"
                )
            ),
        )
@@ -944,7 +944,7 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
        self.assertEqual(response.status_code, status.HTTP_200_OK)

        results = response.data["results"]
-        self.assertEqual({r["id"] for r in results}, {d1.id, d2.id})
+        self.assertEqual({r["id"] for r in results}, {1, 2})

    @mock.patch("documents.search._backend.parse_user_query")
    def test_search_parser_bug_surfaces_as_500_not_400(self, m) -> None:
@@ -1801,7 +1801,6 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
    def test_search_sorting(self) -> None:
        u1 = UserFactory(username="user1")
        u2 = UserFactory(username="user2")
-        u3 = UserFactory(username="user3")
        c1 = Correspondent.objects.create(name="corres Ax")
        c2 = Correspondent.objects.create(name="corres Cx")
        c3 = Correspondent.objects.create(name="corres Bx")
@@ -1827,7 +1826,6 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
            content="test",
            archive_serial_number=1,
            title="1",
-            owner=u3,
        )
        Note.objects.create(
            note="This is a note.",
@@ -1880,14 +1878,14 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
            search_query("&ordering=-num_notes"),
            [d1.id, d3.id, d2.id],
        )
-        # owner sort: ORM orders by owner_id (integer)
+        # owner sort: ORM orders by owner_id (integer); NULLs first in SQLite ASC
        self.assertListEqual(
            search_query("&ordering=owner"),
-            [d1.id, d2.id, d3.id],
+            [d3.id, d1.id, d2.id],
        )
        self.assertListEqual(
            search_query("&ordering=-owner"),
-            [d3.id, d2.id, d1.id],
+            [d2.id, d1.id, d3.id],
        )

    def test_search_ordering_by_score(self) -> None:
@@ -5,7 +5,6 @@ from datetime import timedelta
 from pathlib import Path
 from unittest import mock

-from django.db import connection
 from django.test import override_settings
 from django.utils import timezone
 from rest_framework import status
@@ -69,7 +68,7 @@ class TestSystemStatus(APITestCase):
        self.assertEqual(response.data["install_type"], "bare-metal")
        self.assertIsNotNone(response.data["storage"]["total"])
        self.assertIsNotNone(response.data["storage"]["available"])
-        self.assertEqual(response.data["database"]["type"], connection.vendor)
+        self.assertEqual(response.data["database"]["type"], "sqlite")
        self.assertIsNotNone(response.data["database"]["url"])
        self.assertEqual(response.data["database"]["status"], "OK")
        self.assertIsNone(response.data["database"]["error"])
@@ -1031,7 +1031,9 @@ class TestPDFActions(DirectoriesMixin, TestCase):
        with self.assertLogs("paperless.bulk_edit", level="ERROR") as cm:
            bulk_edit.merge(doc_ids)
            error_str = cm.output[0]
-            expected_str = f"Error merging document {self.doc2.id}, it will not be included in the merge"
+            expected_str = (
+                "Error merging document 2, it will not be included in the merge"
+            )
            self.assertIn(expected_str, error_str)

        mock_consume_file.assert_not_called()
@@ -1191,7 +1193,7 @@ class TestPDFActions(DirectoriesMixin, TestCase):
        with self.assertLogs("paperless.bulk_edit", level="ERROR") as cm:
            bulk_edit.split(doc_ids, pages)
            error_str = cm.output[0]
-            expected_str = f"Error splitting document {self.doc2.id}"
+            expected_str = "Error splitting document 2"
            self.assertIn(expected_str, error_str)

        mock_consume_file.assert_not_called()
@@ -248,8 +248,8 @@ class TestConsumer(
        )
        self.assertIsNone(document.correspondent)
        self.assertIsNone(document.document_type)
-        self.assertEqual(document.filename, f"{document.pk:07}.pdf")
-        self.assertEqual(document.archive_filename, f"{document.pk:07}.pdf")
+        self.assertEqual(document.filename, "0000001.pdf")
+        self.assertEqual(document.archive_filename, "0000001.pdf")

        self.assertIsFile(document.source_path)

@@ -1127,17 +1127,17 @@ class TestConsumer(
        with self.get_consumer(settings.CONSUMPTION_DIR / "simple.png") as consumer:
            consumer.run()

-            doc1 = Document.objects.latest("pk")
+            doc1 = Document.objects.filter(pk=1).first()

        with self.get_consumer(settings.CONSUMPTION_DIR / "simple.pdf") as consumer:
            consumer.run()

-            doc2 = Document.objects.latest("pk")
+            doc2 = Document.objects.filter(pk=2).first()

        with self.get_consumer(settings.CONSUMPTION_DIR / "simple.png.pdf") as consumer:
            consumer.run()

-            doc3 = Document.objects.latest("pk")
+            doc3 = Document.objects.filter(pk=3).first()

        self.assertEqual(doc1.filename, "simple.png")
        self.assertEqual(doc1.archive_filename, "simple.pdf")
@@ -242,7 +242,9 @@ class TestDocumentListEffectiveContentAnnotation:
        # Only the list query itself reads a content column: no extra query
        # for the skipped prefetch, none for a per-instance fallback
        content_queries = [
-            query for query in ctx.captured_queries if _content_column() in query["sql"]
+            query
+            for query in ctx.captured_queries
+            if '"documents_document"."content"' in query["sql"]
        ]
        assert len(content_queries) == 1

@@ -363,11 +365,6 @@ def _get_document_queries(
    return [q for q in ctx.captured_queries if '"django_content_type"' not in q["sql"]]


-def _content_column() -> str:
-    quote = connection.ops.quote_name
-    return f"{quote('documents_document')}.{quote('content')}"
-
-
 def _get_effective_content_fallback_queries(
    ctx: CaptureQueriesContext,
 ) -> list[dict[str, str]]:
@@ -382,7 +379,7 @@ def _get_effective_content_fallback_queries(
    return [
        q
        for q in ctx.captured_queries
-        if q["sql"].startswith(f"SELECT {_content_column()} FROM")
+        if q["sql"].startswith('SELECT "documents_document"."content" FROM')
    ]


@@ -760,9 +760,6 @@ class TestFileHandlingWithArchive(DirectoriesMixin, FileSystemAssertsMixin, Test
            archive_checksum="B",
        )

-        # Without a format the files are renamed to match the primary key
-        original = doc.source_path
-        archive = doc.archive_path
        self.assertIsFile(original)
        self.assertIsFile(archive)
        self.assertIsFile(doc.source_path)
@@ -981,9 +978,6 @@ class TestFileHandlingWithArchive(DirectoriesMixin, FileSystemAssertsMixin, Test
            archive_filename="0000001.pdf",
        )

-        # Without a format the files are renamed to match the primary key
-        original = doc.source_path
-        archive = doc.archive_path
        self.assertIsFile(original)
        self.assertIsFile(archive)
        self.assertIsFile(doc.source_path)
@@ -151,20 +151,6 @@ class TestExportImport(

        super().setUp()

-    def _install_sample_files(self) -> None:
-        documents_dir = Path(self.dirs.media_dir) / "documents"
-        shutil.rmtree(documents_dir)
-        install_document_samples(documents_dir)
-        # Thumbnails are located by primary key, which the database sequence
-        # does not reset between tests on PostgreSQL
-        for number, document in enumerate(
-            (self.d1, self.d2, self.d3, self.d4),
-            start=1,
-        ):
-            source = documents_dir / "thumbnails" / f"{number:07}.webp"
-            if source != document.thumbnail_path:
-                shutil.copy(source, document.thumbnail_path)
-
    def _get_document_from_manifest(self, manifest, id):
        f = list(
            filter(
@@ -218,7 +204,8 @@ class TestExportImport(
        return manifest

    def test_exporter(self, *, use_filename_format=False) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        num_permission_objects = Permission.objects.count()

@@ -321,7 +308,8 @@ class TestExportImport(
            self.assertEqual(messages.total_issue_count, 0)

    def test_exporter_with_filename_format(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        with override_settings(
            FILENAME_FORMAT="{created_year}/{correspondent}/{title}",
@@ -329,7 +317,8 @@ class TestExportImport(
            self.test_exporter(use_filename_format=True)

    def test_exporter_includes_share_links_and_bundles(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        share_link = ShareLink.objects.create(
            slug="share-link-slug",
@@ -427,7 +416,8 @@ class TestExportImport(
            )

    def test_update_export_changed_time(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        self._do_export()
        self.assertIsFile(self.target / "manifest.json")
@@ -462,7 +452,8 @@ class TestExportImport(
        self.assertEqual(st_mtime_3, st_mtime_4)

    def test_update_export_changed_checksum(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        self._do_export()

@@ -488,7 +479,8 @@ class TestExportImport(
        self.assertIsFile(self.target / "manifest.json")

    def test_update_export_deleted_document(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        manifest = self._do_export()

@@ -519,7 +511,8 @@ class TestExportImport(

    @override_settings(FILENAME_FORMAT="{title}/{correspondent}")
    def test_update_export_changed_location(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        self._do_export(use_filename_format=True)
        self.assertIsFile(self.target / "wow1" / "c.pdf")
@@ -560,7 +553,8 @@ class TestExportImport(
            - Zipfile is created
            - Zipfile contains exported files
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        args = ["document_exporter", self.target, "--zip"]

@@ -588,7 +582,8 @@ class TestExportImport(
            - Zipfile is created
            - Zipfile contains exported files
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        args = ["document_exporter", self.target, "--zip", "--use-filename-format"]

@@ -623,7 +618,8 @@ class TestExportImport(
            - Zipfile contains exported files
            - The existing file and directory in target are removed
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        # Create stuff in target directory
        existing_file = self.target / "test.txt"
@@ -718,7 +714,8 @@ class TestExportImport(
            - Manifest.json doesn't contain information about archive files
            - Documents can be imported again
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        manifest = self._do_export()
        has_archive = False
@@ -760,7 +757,8 @@ class TestExportImport(
            - Manifest.json doesn't contain information about thumbnails
            - Documents can be imported again
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        manifest = self._do_export()
        has_thumbnail = False
@@ -804,7 +802,8 @@ class TestExportImport(
            - Main manifest.json file doesn't contain information about documents
            - Documents can be imported again
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        manifest = self._do_export(split_manifest=True)
        has_document = False
@@ -828,7 +827,8 @@ class TestExportImport(
            self.assertEqual(CustomFieldInstance.objects.count(), 1)

    def _export_import_barcodes(self, *, split_manifest: bool) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")
        DocumentBarcodeFactory(document=self.d1, value="https://example.com")
        DocumentBarcodeFactory(document=self.d2, page=2, value="DE8937")

@@ -863,7 +863,8 @@ class TestExportImport(
        THEN:
            - Documents can be imported again
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        self._do_export(use_folder_prefix=True)

@@ -889,7 +890,8 @@ class TestExportImport(
        THEN:
            - Documents can be imported again
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        self._do_export(use_folder_prefix=True, split_manifest=True)

@@ -915,7 +917,8 @@ class TestExportImport(
            - ContentType & Permission objects are not deleted, db transaction rolled back
        """

-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        num_content_type_objects = ContentType.objects.count()
        num_permission_objects = Permission.objects.count()
@@ -951,7 +954,8 @@ class TestExportImport(
            self.assertEqual(Permission.objects.count(), num_permission_objects + 1)

    def test_exporter_with_auditlog_disabled(self) -> None:
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        with override_settings(
            AUDIT_LOG_ENABLED=False,
@@ -970,7 +974,8 @@ class TestExportImport(
            - The soft-deleted document, note, and custom field instance
              survive the round-trip with deleted_at preserved
        """
-        self._install_sample_files()
+        shutil.rmtree(Path(self.dirs.media_dir) / "documents")
+        install_document_samples(Path(self.dirs.media_dir) / "documents")

        # d1 has self.note and self.cfi1 attached via setUp
        self.d1.delete()
@@ -19,10 +19,6 @@ class TestFuzzyMatchCommand(TestCase):
        call_command(
            "document_fuzzy_match",
            "--no-progress-bar",
-            # The default scales with the CPU count and forks, which closes
-            # the test's database connection on PostgreSQL
-            "--processes",
-            "1",
            *args,
            stdout=stdout,
            stderr=stderr,
@@ -425,29 +425,26 @@ class TestRetaggerStoragePath(DirectoriesMixin):
@pytest.mark.management
@pytest.mark.django_db
 class TestRetaggerIdRange(DirectoriesMixin):
+    @pytest.mark.usefixtures("documents")
    @pytest.mark.parametrize(
-        ("range_end", "expected_count"),
+        ("id_range_args", "expected_count"),
        [
-            pytest.param("second", 1, id="narrow_range_limits_scope"),
-            pytest.param("extra", 2, id="wide_range_tags_all_matches"),
+            pytest.param(["1", "2"], 1, id="narrow_range_limits_scope"),
+            pytest.param(["1", "9999"], 2, id="wide_range_tags_all_matches"),
        ],
    )
    def test_id_range_limits_scope(
        self,
        tags: TagTuple,
-        documents: DocumentTuple,
-        range_end: str,
+        id_range_args: list[str],
        expected_count: int,
    ) -> None:
-        first, second, *_ = documents
-        extra = DocumentFactory(content="NOT the first document")
-        end = {"second": second, "extra": extra}[range_end]
+        DocumentFactory(content="NOT the first document")
        call_command(
            "document_retagger",
            "--tags",
            "--id-range",
-            str(first.pk),
-            str(end.pk),
+            *id_range_args,
            skip_checks=True,
        )
        tag_first, *_ = tags
@@ -1,68 +0,0 @@
-import re
-from fnmatch import fnmatch
-
-import pytest
-
-from documents.matching import glob_to_database_regex
-from documents.models import Document
-from paperless_testing.factories import DocumentFactory
-
-NAMES = [
-    "invoice.pdf",
-    "Invoice.PDF",
-    "invoice_2024.pdf",
-    "a.b.c",
-    "a-b",
-    "sample [1].pdf",
-    "sample 1.pdf",
-    "x^y",
-    "back\\slash",
-    "dir/file.pdf",
-    "",
-]
-
-PATTERNS = [
-    "*",
-    "invoice.pdf",
-    "*.pdf",
-    "*invoice*pdf",
-    "*a*b*c*",
-    "invoice_????.pdf",
-    "sample [[]1].pdf",
-    "sample [1-2].pdf",
-    "sample [!2].pdf",
-    "sample [^2].pdf",
-    "sample [",
-    "a.b.c",
-    "x^y",
-    "back\\slash",
-    "[]]",
-]
-
-
-@pytest.mark.parametrize("pattern", PATTERNS)
-def test_regex_agrees_with_fnmatch(pattern: str) -> None:
-    regex = glob_to_database_regex(pattern)
-    for name in NAMES:
-        expected = fnmatch(name.lower(), pattern.lower())
-        actual = re.search(regex, name, re.IGNORECASE) is not None
-        assert actual is expected, f"{pattern!r} vs {name!r}"
-
-
-@pytest.mark.parametrize("pattern", ["*", "*a*b*c*", "a*b*c*d*e"])
-def test_regex_avoids_syntax_postgresql_rejects(pattern: str) -> None:
-    regex = glob_to_database_regex(pattern)
-    assert not any(marker in regex for marker in ("(?", "\\Z"))
-
-
-@pytest.mark.django_db
-def test_database_filter_matches_glob() -> None:
-    matching = DocumentFactory(original_filename="Invoice_2024.PDF")
-    DocumentFactory(original_filename="invoice_2024.txt")
-    DocumentFactory(original_filename="receipt.pdf")
-
-    found = Document.objects.filter(
-        original_filename__iregex=glob_to_database_regex("*invoice*.pdf"),
-    )
-
-    assert list(found) == [matching]
@@ -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"])

@@ -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]
@@ -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,
@@ -174,8 +174,8 @@ class TestWorkflows(
        w.save()

        self.assertEqual(w.__str__(), "Workflow: Workflow 1")
-        self.assertEqual(trigger.__str__(), f"WorkflowTrigger {trigger.pk}")
-        self.assertEqual(action.__str__(), f"WorkflowAction {action.pk}")
+        self.assertEqual(trigger.__str__(), "WorkflowTrigger 1")
+        self.assertEqual(action.__str__(), "WorkflowAction 1")

        test_file = shutil.copy(
            self.SIMPLE_PDF,
@@ -433,7 +433,7 @@ class TestWorkflows(
                list(document.tags.all()),
                [self.t1, self.t2, self.t3],
            )
-            self.assertCountEqual(
+            self.assertEqual(
                list(
                    get_users_with_perms(
                        document,
@@ -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,
@@ -2,7 +2,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version: paperless-ngx\n"
 "Report-Msgid-Bugs-To: \n"
-"POT-Creation-Date: 2026-10-29 17:25+0000\n"
+"POT-Creation-Date: 2026-10-30 22:26+0000\n"
 "PO-Revision-Date: 2022-02-17 04:17\n"
 "Last-Translator: \n"
 "Language-Team: English\n"
@@ -21,43 +21,43 @@ msgstr ""
 msgid "Documents"
 msgstr ""

-#: documents/filters.py:337
+#: documents/filters.py:336
 msgid "Expected a date or a relative date such as \"today\" or \"in 30 days\"."
 msgstr ""

-#: documents/filters.py:513
+#: documents/filters.py:514
 msgid "Value must be valid JSON."
 msgstr ""

-#: documents/filters.py:534
+#: documents/filters.py:535
 msgid "Invalid custom field query expression"
 msgstr ""

-#: documents/filters.py:544
+#: documents/filters.py:545
 msgid "Invalid expression list. Must be nonempty."
 msgstr ""

-#: documents/filters.py:565
+#: documents/filters.py:566
 msgid "Invalid logical operator {op!r}"
 msgstr ""

-#: documents/filters.py:579
+#: documents/filters.py:580
 msgid "Maximum number of query conditions exceeded."
 msgstr ""

-#: documents/filters.py:643
+#: documents/filters.py:644
 msgid "{name!r} is not a valid custom field."
 msgstr ""

-#: documents/filters.py:680
+#: documents/filters.py:681
 msgid "{data_type} does not support query expr {expr!r}."
 msgstr ""

-#: documents/filters.py:801 documents/models.py:137
+#: documents/filters.py:802 documents/models.py:137
 msgid "Maximum nesting depth exceeded."
 msgstr ""

-#: documents/filters.py:1178
+#: documents/filters.py:1179
 msgid "Custom field not found"
 msgstr ""

@@ -1657,7 +1657,7 @@ msgid "workflow runs"
 msgstr ""

 #: documents/serialisers.py:483 documents/serialisers.py:840
-#: documents/serialisers.py:2872 documents/views.py:360 documents/views.py:2730
+#: documents/serialisers.py:2872 documents/views.py:348 documents/views.py:2733
 #: paperless_mail/serialisers.py:156
 msgid "Insufficient permissions."
 msgstr ""
@@ -1698,7 +1698,7 @@ msgstr ""
 msgid "Duplicate document identifiers are not allowed."
 msgstr ""

-#: documents/serialisers.py:2958 documents/views.py:4780
+#: documents/serialisers.py:2958 documents/views.py:4783
 #, python-format
 msgid "Documents not found: %(ids)s"
 msgstr ""
@@ -1949,11 +1949,11 @@ msgstr ""
 msgid ", "
 msgstr ""

-#: documents/views.py:315
+#: documents/views.py:303
 msgid "Specify only one of text, title_search, query, or more_like_id."
 msgstr ""

-#: documents/views.py:353 documents/views.py:2727
+#: documents/views.py:341 documents/views.py:2730
 msgid "Invalid more_like_id"
 msgstr ""

@@ -1969,20 +1969,20 @@ msgstr ""
 msgid "AI backend rejected the request. Check logs for details."
 msgstr ""

-#: documents/views.py:4796
+#: documents/views.py:4799
 #, python-format
 msgid "Insufficient permissions to share document %(id)s."
 msgstr ""

-#: documents/views.py:4843
+#: documents/views.py:4846
 msgid "Bundle is already being processed."
 msgstr ""

-#: documents/views.py:4909
+#: documents/views.py:4912
 msgid "The share link bundle is still being prepared. Please try again later."
 msgstr ""

-#: documents/views.py:4923
+#: documents/views.py:4926
 msgid "The share link bundle is unavailable."
 msgstr ""

@@ -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)),
    },
 )
@@ -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):
@@ -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"),
@@ -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))


@@ -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 (
@@ -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:
@@ -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)
@@ -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
        )
@@ -1,7 +1,6 @@
 from __future__ import annotations

 import json
-from contextlib import nullcontext
 from typing import TYPE_CHECKING
 from typing import Any
 from unittest.mock import MagicMock
@@ -38,13 +37,6 @@ def patch_embed_model():
    llama_settings.Settings.embed_model = None


-@pytest.fixture(autouse=True)
-def keep_db_connection():
-    # Closing the connection would roll back the test's transaction on PostgreSQL
-    with patch("paperless_ai.chat.db_connection_released", nullcontext):
-        yield
-
-
@pytest.fixture(autouse=True)
 def patch_embed_nodes():
    with patch(
Author	SHA1	Message	Date
shamoon	331fa13088	Fix: drop redundant change perms checks on cards open button (#14601 )	2026-10-30 16:16:48 -07:00
shamoon	ebddd5687c	Chore: fix a couple e2e warnings / flakiness (#14596 )	2026-10-30 15:45:46 -07:00
Trenton H	2925b8de81	Chore: Skip repeating Django system checks in the Docker consumer, scheduler and worker (#14598 ) The init-system-checks step already runs them once and all three services depend on init-complete. Measured idle anon PSS saved on Python 3.14: consumer 14 MiB, beat 26 MiB, worker 27 MiB.	2026-10-30 22:37:37 +00:00
GitHub Actions	5e397e1222	Auto translate strings	2026-10-30 22:26:54 +00:00
Trenton H	6f99ad9469	Chore: Improve idle memory via more lazy imports (#14599 ) * Chore: Import dateparser lazily so processes that never parse dates skip it * Chore: Import turbohtml lazily in the mail parser * Chore: Import lxml lazily in the SVG validator * Chore: Import paperless_ai.indexing lazily in document tasks * Chore: Import paperless_ai helpers lazily in views * Chore: Read the OCRmyPDF sRGB profile on first use instead of at import * Chore: Import pikepdf and img2pdf lazily so startup does not load them * Chore: Reuse the initialized Django ASGI application for HTTP routing	2026-10-30 22:25:42 +00:00
shamoon	a5d3247ca7	Tweakhancement: make new pdf controls mostly hidden (#14595 )	2026-10-30 10:22:50 -07:00
GitHub Actions	8a67481a9d	Auto translate strings	2026-10-30 16:59:59 +00:00
shamoon	652f964f5d	Tweak: move preview button in doc list (#14594 )	2026-10-30 09:58:29 -07:00