Use locmemcache for ci

TagSerializer.get_children() built a full nested TagSerializer(many=True)
for every tag, even when it had zero children, likely the common case for
most tags and maybe even most installs. Constructing a DRF ModelSerializer isn't
free (field introspection, deepcopy of declared fields, i18n lookups
all re-run per instantiation), so this scaled GET /api/tags/ linearly
with tag count in pure Python overhead, unrelated to SQL query count.

.github/workflows/issue-bot.yml

@@ -72,7 +72,7 @@ jobs:
                 'You are welcome to open a new issue that describes the problem you observed in your own words.'
               : 'This issue was automatically closed because it was not opened using our bug report form. ' +
                 'Issues have to be created through the form so that the details we need to investigate are included.\n\n' +
-                `If the problem is still there, please [open a new issue](${newIssue}) using the form. No other action is needed here.\n\n' +
+                `If the problem is still there, please [open a new issue](${newIssue}) using the form. No other action is needed here.\n\n` +
                 'If any part of your report was written by an AI tool or agent, you must say so: undisclosed AI-generated ' +
                 `contributions are a violation of our [Code of Conduct](${codeOfConduct}).`;
 

.github/workflows/pr-bot.yml

@@ -25,6 +25,10 @@ jobs:
   pr-bot:
     name: Automated PR Bot
     runs-on: ubuntu-latest
+    # Runs after Anti-slop so the welcome comment can see whether the PR was closed
+    # instead of racing it. Still runs if that job fails, so labeling is not lost.
+    needs: Anti-slop
+    if: ${{ !cancelled() }}
     permissions:
       contents: read
       pull-requests: write
@@ -99,8 +103,25 @@ jobs:
         uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
         with:
           script: |
-            const pr = context.payload.pull_request;
-            const user = pr.user.login;
+            const user = context.payload.pull_request.user.login;
+
+            // Re-read the PR: Anti-slop may have closed and labeled it after the webhook
+            const { data: pr } = await github.rest.pulls.get({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              pull_number: context.payload.pull_request.number,
+            });
+
+            if (pr.state === 'closed') {
+              core.info('Skipping comment: PR is already closed');
+              return;
+            }
+
+            const labels = pr.labels.map((label) => (typeof label === 'string' ? label : label.name));
+            if (labels.includes('ai')) {
+              core.info('Skipping comment: PR is labeled ai');
+              return;
+            }
 
             const { data: members } = await github.rest.orgs.listMembers({
               org: 'paperless-ngx',

src-ui/messages.xlf

@@ -501,15 +501,43 @@
           <context context-type="linenumber">30</context>
         </context-group>
       </trans-unit>
+      <trans-unit id="7057826840809102816" datatype="html">
+        <source>This value overrides <x id="INTERPOLATION" equiv-text="{{option.config_key}}"/>, which is set outside Paperless.</source>
+        <context-group purpose="location">
+          <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
+          <context context-type="linenumber">35</context>
+        </context-group>
+      </trans-unit>
+      <trans-unit id="7221396516204435584" datatype="html">
+        <source><x id="INTERPOLATION" equiv-text="{{option.config_key}}"/> is set outside Paperless. Enter a value here to override it.</source>
+        <context-group purpose="location">
+          <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
+          <context context-type="linenumber">37</context>
+        </context-group>
+      </trans-unit>
+      <trans-unit id="8318849619178340389" datatype="html">
+        <source>Use the externally configured value</source>
+        <context-group purpose="location">
+          <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
+          <context context-type="linenumber">42</context>
+        </context-group>
+      </trans-unit>
+      <trans-unit id="6032629623003430385" datatype="html">
+        <source>Reset to external</source>
+        <context-group purpose="location">
+          <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
+          <context context-type="linenumber">43</context>
+        </context-group>
+      </trans-unit>
       <trans-unit id="7808756054397155068" datatype="html">
         <source>Reset</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
-          <context context-type="linenumber">34</context>
+          <context context-type="linenumber">46</context>
         </context-group>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
-          <context context-type="linenumber">35</context>
+          <context context-type="linenumber">47</context>
         </context-group>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/settings/settings.component.html</context>
@@ -520,7 +548,7 @@
         <source>Enable</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
-          <context context-type="linenumber">43</context>
+          <context context-type="linenumber">56</context>
         </context-group>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/common/profile-edit-dialog/profile-edit-dialog.component.html</context>
@@ -531,7 +559,7 @@
         <source>Cancel</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
-          <context context-type="linenumber">67,68</context>
+          <context context-type="linenumber">80,81</context>
         </context-group>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/settings/settings.component.html</context>
@@ -610,7 +638,7 @@
         <source>Save</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.html</context>
-          <context context-type="linenumber">70,71</context>
+          <context context-type="linenumber">83,84</context>
         </context-group>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/settings/settings.component.html</context>
@@ -681,42 +709,42 @@
         <source>Error retrieving config</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">117</context>
+          <context context-type="linenumber">118</context>
         </context-group>
       </trans-unit>
       <trans-unit id="1172622527269118932" datatype="html">
         <source>Invalid JSON</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">143</context>
+          <context context-type="linenumber">144</context>
         </context-group>
       </trans-unit>
       <trans-unit id="5103146006962696736" datatype="html">
         <source>Configuration updated</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">187</context>
+          <context context-type="linenumber">193</context>
         </context-group>
       </trans-unit>
       <trans-unit id="1664963291286452273" datatype="html">
         <source>An error occurred updating configuration</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">192</context>
+          <context context-type="linenumber">198</context>
         </context-group>
       </trans-unit>
       <trans-unit id="2653081282186526824" datatype="html">
         <source>File successfully updated</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">214</context>
+          <context context-type="linenumber">220</context>
         </context-group>
       </trans-unit>
       <trans-unit id="5902783625859504265" datatype="html">
         <source>An error occurred uploading file</source>
         <context-group purpose="location">
           <context context-type="sourcefile">src/app/components/admin/config/config.component.ts</context>
-          <context context-type="linenumber">219</context>
+          <context context-type="linenumber">225</context>
         </context-group>
       </trans-unit>
       <trans-unit id="4804785061014590286" datatype="html">

src-ui/src/app/components/admin/config/config.component.html

@@ -23,17 +23,30 @@
                                     <div class="col">
                                         <div class="card bg-light">
                                             <div class="card-body">
-                                                <div class="card-title d-flex align-items-center">
+                                                <div class="card-title d-flex align-items-center flex-wrap">
                                                     <h6 class="mb-0">
                                                       {{option.title}}
                                                     </h6>
                                                     <a class="btn btn-sm btn-link" title="Read the documentation about this setting" i18n-title [href]="getDocsUrl(option.config_key)" target="_blank" referrerpolicy="no-referrer">
                                                         <i-bs name="info-circle"></i-bs>
                                                     </a>
+                                                    @if (isExternallyConfigured(option.config_key)) {
+                                                      @if (isSet(option.key)) {
+                                                        <span class="badge rounded-pill bg-body-secondary text-dark fw-normal" title="This value overrides {{option.config_key}}, which is set outside Paperless." i18n-title>Overrides external</span>
+                                                      } @else {
+                                                        <span class="badge rounded-pill bg-body-secondary text-dark fw-normal" title="{{option.config_key}} is set outside Paperless. Enter a value here to override it." i18n-title>Set externally</span>
+                                                      }
+                                                    }
                                                     @if (isSet(option.key)) {
-                                                      <button type="button" class="btn btn-sm btn-link text-danger ms-auto pe-0" title="Reset" i18n-title (click)="resetOption(option.key)">
-                                                          <i-bs class="me-1" name="x"></i-bs><ng-container i18n>Reset</ng-container>
-                                                      </button>
+                                                      @if (isExternallyConfigured(option.config_key)) {
+                                                        <button type="button" class="btn btn-sm btn-link text-danger ms-auto pe-0" title="Use the externally configured value" i18n-title (click)="resetOption(option.key)">
+                                                            <i-bs class="me-1" name="x"></i-bs><ng-container i18n>Reset to external</ng-container>
+                                                        </button>
+                                                      } @else {
+                                                        <button type="button" class="btn btn-sm btn-link text-danger ms-auto pe-0" title="Reset" i18n-title (click)="resetOption(option.key)">
+                                                            <i-bs class="me-1" name="x"></i-bs><ng-container i18n>Reset</ng-container>
+                                                        </button>
+                                                      }
                                                     }
                                                 </div>
                                                 <div class="mb-n3">

src-ui/src/app/components/admin/config/config.component.spec.ts

@@ -163,6 +163,19 @@ describe('ConfigComponent', () => {
     expect(component.configForm.get('barcodes_enabled').value).toBeNull()
   })
 
+  it('should identify externally configured options', () => {
+    component.externallyConfiguredVariables = new Set([
+      'PAPERLESS_OCR_LANGUAGE',
+    ])
+
+    expect(
+      component.isExternallyConfigured('PAPERLESS_OCR_LANGUAGE')
+    ).toBeTruthy()
+    expect(
+      component.isExternallyConfigured('PAPERLESS_OCR_OUTPUT_TYPE')
+    ).toBeFalsy()
+  })
+
   it('should group options into sections within a category, or not', () => {
     const sections = component.getCategorySections(ConfigCategory.OCR)
     expect(sections).toEqual([null, ConfigSection.RemoteOCR])

src-ui/src/app/components/admin/config/config.component.ts

@@ -69,6 +69,7 @@ export class ConfigComponent
   public configForm = new FormGroup({})
 
   public errors = {}
+  public externallyConfiguredVariables = new Set<string>()
 
   get optionCategories(): string[] {
     return Object.values(ConfigCategory)
@@ -152,6 +153,9 @@ export class ConfigComponent
   }
 
   private initialize(config: PaperlessConfig) {
+    this.externallyConfiguredVariables = new Set(
+      config.externally_configured_variables ?? []
+    )
     if (!this.store) {
       this.store = new BehaviorSubject(config)
 
@@ -162,7 +166,9 @@ export class ConfigComponent
           this.configForm.patchValue(state, { emitEvent: false })
         })
 
-      this.isDirty$ = dirtyCheck(this.configForm, this.store.asObservable())
+      this.isDirty$ = dirtyCheck(this.configForm, this.store.asObservable(), {
+        excludeKeys: ['externally_configured_variables'],
+      })
     }
     this.configForm.patchValue(config)
 
@@ -227,6 +233,10 @@ export class ConfigComponent
     return this.configForm.get(key).value != null
   }
 
+  public isExternallyConfigured(configKey: string): boolean {
+    return this.externallyConfiguredVariables.has(configKey)
+  }
+
   public resetOption(key: string) {
     this.configForm.get(key).setValue(null)
   }

src-ui/src/app/data/paperless-config.ts

@@ -422,6 +422,7 @@ export const PaperlessConfigOptions: ConfigOption[] = [
 ]
 
 export interface PaperlessConfig extends ObjectWithId {
+  externally_configured_variables: string[]
   output_type: OutputTypeConfig
   pages: number
   language: string

src/documents/caching.py

@@ -3,13 +3,16 @@ from __future__ import annotations
 import hashlib
 import logging
 import pickle
+import time
 import uuid
 from binascii import hexlify
 from collections import OrderedDict
 from dataclasses import dataclass
+from hashlib import sha256
 from typing import TYPE_CHECKING
 from typing import Any
 from typing import Final
+from uuid import uuid4
 
 from django.conf import settings
 from django.core.cache import cache
@@ -21,6 +24,7 @@ from paperless.signed_pickle import signed_pickle_dumps
 from paperless.signed_pickle import signed_pickle_loads
 
 if TYPE_CHECKING:
+    from django.contrib.auth.models import User
     from django.core.cache.backends.base import BaseCache
 
     from documents.classifier import DocumentClassifier
@@ -59,6 +63,9 @@ CLASSIFIER_MODIFIED_KEY: Final[str] = "classifier_modified"
 #          validated separately, so candidate-anchored 1001 results are stale
 LLM_CACHE_CLASSIFIER_VERSION: Final[int] = 1002
 
+# How often a request waiting on llm generation re-checks the cache
+LLM_SUGGESTION_POLL_INTERVAL: Final[float] = 0.5
+
 CACHE_1_MINUTE: Final[int] = 60
 CACHE_5_MINUTES: Final[int] = 5 * CACHE_1_MINUTE
 CACHE_50_MINUTES: Final[int] = 50 * CACHE_1_MINUTE
@@ -262,6 +269,68 @@ def get_llm_suggestion_cache(
     return None
 
 
+def retrieve_llm_suggestions(
+    document: Document,
+    user: User | None,
+    output_language: str | None,
+    *,
+    backend: str,
+    lock_timeout: int,
+) -> dict:
+    """Return cached LLM suggestions, generating them once across workers."""
+    # Lazy import to avoid pulling in the whole AI stuff
+    from paperless_ai.ai_classifier import get_ai_document_classification
+    from paperless_ai.exceptions import LLMTimeoutError
+
+    lock_key = (
+        f"{get_suggestion_cache_key(document.pk)}_llm_lock_"
+        f"{sha256(backend.encode()).hexdigest()}"
+    )
+    waited = False
+
+    while True:
+        cached = get_llm_suggestion_cache(document.pk, backend=backend)
+        if cached is not None:
+            refresh_suggestions_cache(document.pk)
+            return cached.suggestions
+
+        lock_token = uuid4().hex
+        if cache.add(lock_key, lock_token, lock_timeout):
+            if waited:
+                # The generation we were waiting on has ended without caching
+                # anything so it either failed or outlived its lock. Give up
+                # rather than re-running it
+                cache.delete(lock_key)
+                raise LLMTimeoutError
+
+            try:
+                # The cache may have been populated while acquiring the lock.
+                cached = get_llm_suggestion_cache(document.pk, backend=backend)
+                if cached is not None:
+                    refresh_suggestions_cache(document.pk)
+                    return cached.suggestions
+
+                suggestions = get_ai_document_classification(
+                    document,
+                    user,
+                    output_language,
+                )
+                set_llm_suggestions_cache(
+                    document.pk,
+                    suggestions,
+                    backend=backend,
+                )
+                return suggestions
+            finally:
+                # Don't remove lock if this one expired while generation was still running
+                if cache.get(lock_key) == lock_token:
+                    cache.delete(lock_key)
+
+        waited = True
+        # Another worker is generating suggestions, poll to avoid another LLM request
+        time.sleep(LLM_SUGGESTION_POLL_INTERVAL)
+
+
 def set_llm_suggestions_cache(
     document_id: int,
     suggestions: dict,

src/documents/filters.py

@@ -12,7 +12,6 @@ from typing import TYPE_CHECKING
 from typing import Any
 
 from django.contrib.contenttypes.models import ContentType
-from django.core.exceptions import FieldError
 from django.db.models import Case
 from django.db.models import CharField
 from django.db.models import Count
@@ -53,6 +52,7 @@ from documents.models import StoragePath
 from documents.models import Tag
 from documents.permissions import permitted_document_ids
 from documents.permissions import permitted_object_ids
+from documents.versioning import annotate_effective_content
 
 if TYPE_CHECKING:
     from collections.abc import Callable
@@ -182,14 +182,9 @@ class TitleContentFilter(Filter):
             logger.warning(
                 "Deprecated document filter parameter 'title_content' used; use `text` instead.",
             )
-            try:
-                return qs.filter(
-                    Q(title__icontains=value) | Q(effective_content__icontains=value),
-                )
-            except FieldError:
-                return qs.filter(
-                    Q(title__icontains=value) | Q(content__icontains=value),
-                )
+            return annotate_effective_content(qs).filter(
+                Q(title__icontains=value) | Q(effective_content__icontains=value),
+            )
         else:
             return qs
 
@@ -200,14 +195,9 @@ class EffectiveContentFilter(Filter):
         value = value.strip() if isinstance(value, str) else value
         if not value:
             return qs
-        try:
-            return qs.filter(
-                **{f"effective_content__{self.lookup_expr}": value},
-            )
-        except FieldError:
-            return qs.filter(
-                **{f"content__{self.lookup_expr}": value},
-            )
+        return annotate_effective_content(qs).filter(
+            **{f"effective_content__{self.lookup_expr}": value},
+        )
 
 
 @extend_schema_field(serializers.BooleanField)

src/documents/serialisers.py

@@ -674,6 +674,9 @@ class TagSerializer(MatchingModelSerializer, OwnedObjectSerializer):
             ordering = ordering or (Lower("name"),)
             children = children.order_by(*ordering)
 
+        if not children:
+            return []
+
         serializer = TagSerializer(
             children,
             many=True,

src/documents/tests/test_api_app_config.py

@@ -35,7 +35,8 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
         THEN:
             - Existing config
         """
-        response = self.client.get(self.ENDPOINT, format="json")
+        with patch.dict("os.environ", {}, clear=True):
+            response = self.client.get(self.ENDPOINT, format="json")
 
         self.assertEqual(response.status_code, status.HTTP_200_OK)
 
@@ -45,6 +46,7 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
             response.data[0],
             {
                 "id": 1,
+                "externally_configured_variables": [],
                 "output_type": None,
                 "pages": None,
                 "language": None,
@@ -91,6 +93,31 @@ class TestApiAppConfig(DirectoriesMixin, APITestCase):
             },
         )
 
+    def test_api_get_config_reports_external_configuration_without_values(self) -> None:
+        with patch.dict(
+            "os.environ",
+            {
+                "PAPERLESS_OCR_LANGUAGE": "eng",
+                "PAPERLESS_REMOTE_OCR_API_KEY": "secret-value",
+                "PAPERLESS_FUTURE_SETTING": "future-value",
+                "UNRELATED_SETTING": "unrelated-value",
+            },
+            clear=True,
+        ):
+            response = self.client.get(self.ENDPOINT, format="json")
+
+        self.assertCountEqual(
+            response.data[0]["externally_configured_variables"],
+            [
+                "PAPERLESS_FUTURE_SETTING",
+                "PAPERLESS_OCR_LANGUAGE",
+                "PAPERLESS_REMOTE_OCR_API_KEY",
+            ],
+        )
+        self.assertNotContains(response, "secret-value")
+        self.assertNotContains(response, "future-value")
+        self.assertNotContains(response, "UNRELATED_SETTING")
+
     def test_api_get_ui_settings_with_config(self) -> None:
         """
         GIVEN:

src/documents/tests/test_api_document_versions.py

@@ -2,14 +2,12 @@ from __future__ import annotations
 
 import datetime
 from typing import TYPE_CHECKING
-from unittest import TestCase
 from unittest import mock
 
 from auditlog.models import LogEntry  # type: ignore[import-untyped]
 from django.contrib.auth.models import Permission
 from django.contrib.auth.models import User
 from django.contrib.contenttypes.models import ContentType
-from django.core.exceptions import FieldError
 from django.core.files.uploadedfile import SimpleUploadedFile
 from django.test import TestCase as DjangoTestCase
 from django.utils import timezone
@@ -22,6 +20,7 @@ from documents.filters import TitleContentFilter
 from documents.models import Document
 from documents.tests.utils import DirectoriesMixin
 from documents.tests.utils import read_streaming_response
+from documents.versioning import annotate_effective_content
 from documents.views import DocumentSelectionMixin
 
 if TYPE_CHECKING:
@@ -892,32 +891,104 @@ class TestDocumentVersioningApi(DirectoriesMixin, APITestCase):
         )
 
 
-class TestVersionAwareFilters(TestCase):
-    def test_title_content_filter_falls_back_to_content(self) -> None:
-        queryset = mock.Mock()
-        fallback_queryset = mock.Mock()
-        queryset.filter.side_effect = [FieldError("missing field"), fallback_queryset]
+class TestVersionAwareFilters(DjangoTestCase):
+    """
+    The filters annotate effective_content themselves rather than relying on
+    the caller's queryset carrying it, so they stay version-aware on a plain
+    Document queryset (e.g. the bulk-edit "select all matching" path).
+    """
 
-        result = TitleContentFilter().filter(queryset, " latest ")
+    def setUp(self) -> None:
+        super().setUp()
+        self.root = Document.objects.create(
+            title="root",
+            checksum="root",
+            mime_type="application/pdf",
+            content="superseded-content",
+        )
+        Document.objects.create(
+            title="version",
+            checksum="version",
+            mime_type="application/pdf",
+            root_document=self.root,
+            version_index=1,
+            content="latest-content",
+        )
+        self.unversioned = Document.objects.create(
+            title="unversioned",
+            checksum="unversioned",
+            mime_type="application/pdf",
+            content="latest-content",
+        )
 
-        self.assertIs(result, fallback_queryset)
-        self.assertEqual(queryset.filter.call_count, 2)
-
-    def test_effective_content_filter_falls_back_to_content_lookup(self) -> None:
-        queryset = mock.Mock()
-        fallback_queryset = mock.Mock()
-        queryset.filter.side_effect = [FieldError("missing field"), fallback_queryset]
-
-        result = EffectiveContentFilter(lookup_expr="icontains").filter(
-            queryset,
+    def test_title_content_filter_matches_latest_version_content(self) -> None:
+        result = TitleContentFilter().filter(
+            Document.objects.filter(root_document__isnull=True),
             " latest ",
         )
 
-        self.assertIs(result, fallback_queryset)
-        first_kwargs = queryset.filter.call_args_list[0].kwargs
-        second_kwargs = queryset.filter.call_args_list[1].kwargs
-        self.assertEqual(first_kwargs, {"effective_content__icontains": "latest"})
-        self.assertEqual(second_kwargs, {"content__icontains": "latest"})
+        self.assertCountEqual(
+            [doc.id for doc in result],
+            [self.root.id, self.unversioned.id],
+        )
+
+    def test_effective_content_filter_matches_latest_version_content(self) -> None:
+        result = EffectiveContentFilter(lookup_expr="icontains").filter(
+            Document.objects.filter(root_document__isnull=True),
+            " latest ",
+        )
+
+        self.assertCountEqual(
+            [doc.id for doc in result],
+            [self.root.id, self.unversioned.id],
+        )
+
+    def test_effective_content_filter_ignores_superseded_content(self) -> None:
+        result = EffectiveContentFilter(lookup_expr="icontains").filter(
+            Document.objects.filter(root_document__isnull=True),
+            "superseded",
+        )
+
+        self.assertEqual(list(result), [])
+
+    def test_filters_reuse_an_existing_annotation(self) -> None:
+        """
+        Annotating twice under the same alias is an error, so an already
+        annotated queryset (the search path) has to be left alone.
+        """
+        annotated = annotate_effective_content(
+            Document.objects.filter(root_document__isnull=True),
+        )
+        self.assertIs(annotate_effective_content(annotated), annotated)
+
+        result = EffectiveContentFilter(lookup_expr="icontains").filter(
+            annotated,
+            "latest",
+        )
+
+        self.assertCountEqual(
+            [doc.id for doc in result],
+            [self.root.id, self.unversioned.id],
+        )
+
+    def test_bulk_selection_does_not_match_superseded_content(self) -> None:
+        """
+        Bulk edit's "select all matching" builds its own queryset, so before
+        the filters annotated for themselves it matched the root document's
+        superseded content -- selecting documents the list view, filtered by
+        the same term, does not show.
+        """
+        user = User.objects.create_superuser(username="bulk_selection")
+
+        selected = DocumentSelectionMixin()._resolve_document_ids(
+            user=user,
+            validated_data={
+                "all": True,
+                "filters": {"content__icontains": "superseded"},
+            },
+        )
+
+        self.assertEqual(selected, [])
 
     def test_effective_content_filter_returns_input_for_empty_values(self) -> None:
         queryset = mock.Mock()

src/documents/tests/test_api_documents.py

@@ -2608,7 +2608,7 @@ 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_api_search.py

@@ -1947,6 +1947,29 @@ class TestDocumentSearchApi(DirectoriesMixin, APITestCase):
         self.assertEqual(len(response.data["documents"]), 1)
         self.assertEqual(response.data["documents"][0]["id"], title_match.id)
 
+    def test_global_search_returns_latest_version_content(self) -> None:
+        root = Document.objects.create(
+            title="bank statement",
+            content="superseded content",
+            checksum="GSV1",
+            pk=23,
+        )
+        Document.objects.create(
+            title="bank statement v2",
+            content="latest content",
+            checksum="GSV2",
+            pk=24,
+            root_document=root,
+            version_index=1,
+        )
+
+        self.client.force_authenticate(self.user)
+
+        response = self.client.get("/api/search/?query=bank&db_only=true")
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        returned = {doc["id"]: doc["content"] for doc in response.data["documents"]}
+        self.assertEqual(returned.get(root.id), "latest content")
+
     def test_global_search_filters_owned_mail_objects(self) -> None:
         user1 = User.objects.create_user("mail-search-user")
         user2 = User.objects.create_user("other-mail-search-user")

src/documents/tests/test_caching.py

@@ -1,7 +1,17 @@
+from concurrent.futures import ThreadPoolExecutor
+from threading import Event
+from threading import Lock
+from uuid import uuid4
+
+import pytest
+from django.core.cache.backends.locmem import LocMemCache
+
 from documents.caching import StoredLRUCache
+from documents.caching import retrieve_llm_suggestions
 from paperless.signed_pickle import HMAC_SIZE
 from paperless.signed_pickle import signed_pickle_dumps
 from paperless.signed_pickle import signed_pickle_loads
+from paperless_ai.exceptions import LLMTimeoutError
 
 
 def test_lru_cache_entries() -> None:
@@ -56,3 +66,131 @@ def test_stored_lru_cache_rejects_tampered_data(mocker) -> None:
     cache.load()
 
     assert cache.get("x") is None
+
+
+def test_llm_suggestions_are_generated_once_for_concurrent_requests(mocker) -> None:
+    mocker.patch(
+        "documents.caching.cache",
+        LocMemCache(uuid4().hex, {}),
+    )
+    generation_started = Event()
+    finish_generation = Event()
+    waiter_started = Event()
+    release_waiter = Event()
+    call_lock = Lock()
+    calls = 0
+    suggestions = {"title": "Generated once"}
+    document = mocker.Mock(pk=42)
+    user = mocker.Mock()
+
+    def generate(*args) -> dict:
+        nonlocal calls
+        with call_lock:
+            calls += 1
+        generation_started.set()
+        assert finish_generation.wait(timeout=2)
+        return suggestions
+
+    def wait_for_generation(_interval: float) -> None:
+        waiter_started.set()
+        assert release_waiter.wait(timeout=2)
+
+    mock_get_classification = mocker.patch(
+        "paperless_ai.ai_classifier.get_ai_document_classification",
+        side_effect=generate,
+    )
+    mocker.patch("documents.caching.time.sleep", side_effect=wait_for_generation)
+
+    with ThreadPoolExecutor(max_workers=2) as executor:
+        first = executor.submit(
+            retrieve_llm_suggestions,
+            document,
+            user,
+            None,
+            backend="ollama:model",
+            lock_timeout=10,
+        )
+        assert generation_started.wait(timeout=2)
+        second = executor.submit(
+            retrieve_llm_suggestions,
+            document,
+            user,
+            None,
+            backend="ollama:model",
+            lock_timeout=10,
+        )
+        assert waiter_started.wait(timeout=2)
+        finish_generation.set()
+
+        assert first.result(timeout=2) == suggestions
+        release_waiter.set()
+        assert second.result(timeout=2) == suggestions
+
+    assert calls == 1
+    mock_get_classification.assert_called_once_with(document, user, None)
+
+
+def test_llm_suggestions_waiter_does_not_rerun_a_failed_generation(mocker) -> None:
+    """
+    A request queued behind a generation that fails should give up, not take
+    its turn at re-running a query that just failed.
+    """
+    mocker.patch(
+        "documents.caching.cache",
+        LocMemCache(uuid4().hex, {}),
+    )
+    generation_started = Event()
+    fail_generation = Event()
+    waiter_started = Event()
+    release_waiter = Event()
+    call_lock = Lock()
+    calls = 0
+    document = mocker.Mock(pk=43)
+    user = mocker.Mock()
+
+    def generate(*args) -> dict:
+        nonlocal calls
+        with call_lock:
+            calls += 1
+        generation_started.set()
+        assert fail_generation.wait(timeout=2)
+        raise ValueError("Unknown model")
+
+    def wait_for_generation(_interval: float) -> None:
+        waiter_started.set()
+        assert release_waiter.wait(timeout=2)
+
+    mocker.patch(
+        "paperless_ai.ai_classifier.get_ai_document_classification",
+        side_effect=generate,
+    )
+    mocker.patch("documents.caching.time.sleep", side_effect=wait_for_generation)
+
+    with ThreadPoolExecutor(max_workers=2) as executor:
+        first = executor.submit(
+            retrieve_llm_suggestions,
+            document,
+            user,
+            None,
+            backend="ollama:model",
+            lock_timeout=10,
+        )
+        assert generation_started.wait(timeout=2)
+        second = executor.submit(
+            retrieve_llm_suggestions,
+            document,
+            user,
+            None,
+            backend="ollama:model",
+            lock_timeout=10,
+        )
+        assert waiter_started.wait(timeout=2)
+        fail_generation.set()
+
+        with pytest.raises(ValueError, match="Unknown model"):
+            first.result(timeout=2)
+        release_waiter.set()
+        with pytest.raises(LLMTimeoutError):
+            second.result(timeout=2)
+
+    assert calls == 1

src/documents/tests/test_views.py

@@ -446,7 +446,7 @@ 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",
@@ -496,7 +496,7 @@ 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",
@@ -534,7 +534,7 @@ 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",
@@ -573,7 +573,7 @@ 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",
@@ -609,7 +609,7 @@ 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",
@@ -681,7 +681,7 @@ 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",
@@ -710,7 +710,7 @@ 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",
@@ -737,7 +737,7 @@ 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",
@@ -775,7 +775,7 @@ 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",
@@ -814,7 +814,7 @@ 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",

src/documents/versioning.py

@@ -27,10 +27,13 @@ def versions_newest_first(documents: QuerySet[Document]) -> QuerySet[Document]:
 
 def annotate_effective_content(documents: QuerySet[Document]) -> QuerySet[Document]:
     """
-    Annotates documents with the content of their newest version, falling back
-    to their own, so get_effective_content() can answer from the row rather
-    than querying for the versions of each document
+    Annotates documents with the content of their newest version unless the
+    queryset already carries the annotation, falling back to their own, so
+    get_effective_content() can answer from the row rather than querying for
+    the versions of each document.
     """
+    if "effective_content" in documents.query.annotations:
+        return documents
     return documents.annotate(
         effective_content=Coalesce(
             Subquery(

src/documents/views.py

@@ -116,7 +116,7 @@ from documents.caching import get_suggestion_cache
 from documents.caching import refresh_llm_suggestions_cache
 from documents.caching import refresh_metadata_cache
 from documents.caching import refresh_suggestions_cache
-from documents.caching import set_llm_suggestions_cache
+from documents.caching import retrieve_llm_suggestions
 from documents.caching import set_metadata_cache
 from documents.caching import set_suggestions_cache
 from documents.classifier import load_classifier
@@ -232,6 +232,7 @@ from documents.tasks import train_classifier
 from documents.tasks import update_document_parent_tags
 from documents.utils import get_boolean
 from documents.versioning import VersionResolutionError
+from documents.versioning import annotate_effective_content
 from documents.versioning import get_latest_version_for_root
 from documents.versioning import get_request_version_param
 from documents.versioning import get_root_document
@@ -248,7 +249,6 @@ 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 LLMTimeoutError
@@ -1574,10 +1574,13 @@ class DocumentViewSet(
             llm_suggestions = cached_llm_suggestions.suggestions
         else:
             try:
-                llm_suggestions = get_ai_document_classification(
-                    doc,
-                    request.user,
-                    output_language,
+                llm_suggestions = retrieve_llm_suggestions(
+                    document=doc,
+                    user=request.user,
+                    output_language=output_language,
+                    backend=llm_cache_backend,
+                    # Classification, localization + 30s
+                    lock_timeout=(2 * ai_config.llm_request_timeout) + 30,
                 )
             except ValueError as exc:
                 logger.exception(
@@ -1602,11 +1605,6 @@ class DocumentViewSet(
                     {"ai": [_("AI backend request timed out.")]},
                     status=status.HTTP_503_SERVICE_UNAVAILABLE,
                 )
-            set_llm_suggestions_cache(
-                doc.pk,
-                llm_suggestions,
-                backend=llm_cache_backend,
-            )
 
         tags_choice: TaxonomyChoiceDict = llm_suggestions["tags"]
         correspondents_choice: TaxonomyChoiceDict = llm_suggestions["correspondents"]
@@ -3632,8 +3630,13 @@ class GlobalSearchView(PassUserMixin):
         OBJECT_LIMIT = 3
         docs = []
         if request.user.has_perm("documents.view_document"):
-            all_docs = Document.objects.filter(
-                id__in=permitted_document_ids(request.user),
+            # Never more than OBJECT_LIMIT rows come back here, so annotating
+            # is cheap -- and without it these results show the root
+            # document's superseded content.
+            all_docs = annotate_effective_content(
+                Document.objects.filter(
+                    id__in=permitted_document_ids(request.user),
+                ),
             )
             if db_only:
                 docs = all_docs.filter(title__icontains=query)[:OBJECT_LIMIT]

src/locale/en_US/LC_MESSAGES/django.po

@@ -2,7 +2,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version: paperless-ngx\n"
 "Report-Msgid-Bugs-To: \n"
-"POT-Creation-Date: 2026-09-07 20:47+0000\n"
+"POT-Creation-Date: 2026-09-08 15:56+0000\n"
 "PO-Revision-Date: 2022-02-17 04:17\n"
 "Last-Translator: \n"
 "Language-Team: English\n"
@@ -21,39 +21,39 @@ msgstr ""
 msgid "Documents"
 msgstr ""
 
-#: documents/filters.py:473
+#: documents/filters.py:463
 msgid "Value must be valid JSON."
 msgstr ""
 
-#: documents/filters.py:492
+#: documents/filters.py:482
 msgid "Invalid custom field query expression"
 msgstr ""
 
-#: documents/filters.py:502
+#: documents/filters.py:492
 msgid "Invalid expression list. Must be nonempty."
 msgstr ""
 
-#: documents/filters.py:523
+#: documents/filters.py:513
 msgid "Invalid logical operator {op!r}"
 msgstr ""
 
-#: documents/filters.py:537
+#: documents/filters.py:527
 msgid "Maximum number of query conditions exceeded."
 msgstr ""
 
-#: documents/filters.py:601
+#: documents/filters.py:591
 msgid "{name!r} is not a valid custom field."
 msgstr ""
 
-#: documents/filters.py:638
+#: documents/filters.py:628
 msgid "{data_type} does not support query expr {expr!r}."
 msgstr ""
 
-#: documents/filters.py:757 documents/models.py:136
+#: documents/filters.py:747 documents/models.py:136
 msgid "Maximum nesting depth exceeded."
 msgstr ""
 
-#: documents/filters.py:1119
+#: documents/filters.py:1109
 msgid "Custom field not found"
 msgstr ""
 
@@ -1631,49 +1631,49 @@ msgstr ""
 msgid "workflow runs"
 msgstr ""
 
-#: documents/serialisers.py:524 documents/serialisers.py:878
-#: documents/serialisers.py:2838 documents/views.py:314 documents/views.py:2624
+#: documents/serialisers.py:524 documents/serialisers.py:881
+#: documents/serialisers.py:2841 documents/views.py:315 documents/views.py:2625
 #: paperless_mail/serialisers.py:156
 msgid "Insufficient permissions."
 msgstr ""
 
-#: documents/serialisers.py:714
+#: documents/serialisers.py:717
 msgid "Invalid color."
 msgstr ""
 
-#: documents/serialisers.py:2315
+#: documents/serialisers.py:2318
 #, python-format
 msgid "File type %(type)s not supported"
 msgstr ""
 
-#: documents/serialisers.py:2359
+#: documents/serialisers.py:2362
 #, python-format
 msgid "Custom field id must be an integer: %(id)s"
 msgstr ""
 
-#: documents/serialisers.py:2366
+#: documents/serialisers.py:2369
 #, python-format
 msgid "Custom field with id %(id)s does not exist"
 msgstr ""
 
-#: documents/serialisers.py:2383 documents/serialisers.py:2393
+#: documents/serialisers.py:2386 documents/serialisers.py:2396
 msgid ""
 "Custom fields must be a list of integers or an object mapping ids to values."
 msgstr ""
 
-#: documents/serialisers.py:2388
+#: documents/serialisers.py:2391
 msgid "Some custom fields don't exist or were specified twice."
 msgstr ""
 
-#: documents/serialisers.py:2535
+#: documents/serialisers.py:2538
 msgid "Invalid variable detected."
 msgstr ""
 
-#: documents/serialisers.py:2894
+#: documents/serialisers.py:2897
 msgid "Duplicate document identifiers are not allowed."
 msgstr ""
 
-#: documents/serialisers.py:2924 documents/views.py:4626
+#: documents/serialisers.py:2927 documents/views.py:4632
 #, python-format
 msgid "Documents not found: %(ids)s"
 msgstr ""
@@ -1941,36 +1941,36 @@ msgstr ""
 msgid "Unable to parse URI {value}"
 msgstr ""
 
-#: documents/views.py:307 documents/views.py:2621
+#: documents/views.py:308 documents/views.py:2622
 msgid "Invalid more_like_id"
 msgstr ""
 
-#: documents/views.py:1591
+#: documents/views.py:1592
 msgid "Invalid AI configuration."
 msgstr ""
 
-#: documents/views.py:1602
+#: documents/views.py:1603
 msgid "AI backend request timed out."
 msgstr ""
 
-#: documents/views.py:2446 documents/views.py:2767
+#: documents/views.py:2447 documents/views.py:2768
 msgid "Specify only one of text, title_search, query, or more_like_id."
 msgstr ""
 
-#: documents/views.py:4639
+#: documents/views.py:4645
 #, python-format
 msgid "Insufficient permissions to share document %(id)s."
 msgstr ""
 
-#: documents/views.py:4685
+#: documents/views.py:4691
 msgid "Bundle is already being processed."
 msgstr ""
 
-#: documents/views.py:4749
+#: documents/views.py:4755
 msgid "The share link bundle is still being prepared. Please try again later."
 msgstr ""
 
-#: documents/views.py:4763
+#: documents/views.py:4769
 msgid "The share link bundle is unavailable."
 msgstr ""
 

src/paperless/serialisers.py

@@ -1,4 +1,5 @@
 import logging
+import os
 from io import BytesIO
 
 import magic
@@ -212,6 +213,7 @@ class ProfileSerializer(PasswordValidationMixin, serializers.ModelSerializer[Use
 class ApplicationConfigurationSerializer(
     serializers.ModelSerializer[ApplicationConfiguration],
 ):
+    externally_configured_variables = serializers.SerializerMethodField()
     user_args = serializers.JSONField(binary=True, allow_null=True)
     barcode_tag_mapping = serializers.JSONField(binary=True, allow_null=True)
     llm_api_key = ObfuscatedPasswordField(
@@ -227,6 +229,12 @@ class ApplicationConfigurationSerializer(
 
     OBFUSCATED_FIELDS = ("llm_api_key", "remote_ocr_api_key")
 
+    def get_externally_configured_variables(
+        self,
+        instance: ApplicationConfiguration,
+    ) -> list[str]:
+        return sorted(name for name in os.environ if name.startswith("PAPERLESS_"))
+
     def run_validation(self, data):
         # Empty strings treated as None to avoid unexpected behavior
         if "user_args" in data and data["user_args"] == "":