Handles the rebase from the trash changes

If the filter will exceed SQLite IN limits, load to a temporary table and use a subquery to filter instead
When a user is unrestricted chatting, still exclude trashed documents using a 'NOT IN' SQL statement. Wire that up where we need it
2026-09-03 00:17:16 +00:00 · 2026-09-02 10:11:23 -07:00 · 2026-09-02 10:07:13 -07:00 · 2026-09-02 10:00:36 -07:00 · 2026-09-02 08:14:41 -07:00 · 2026-09-02 08:14:23 -07:00
80 changed files with 1490 additions and 860 deletions
@@ -61,7 +61,7 @@ def replace_with_symlinks(
    total_duplicates = 0
    space_saved = 0

-    for file_list in duplicate_groups.values():
+    for file_hash, file_list in duplicate_groups.items():
        # Keep the first file as the original, replace others with symlinks
        original_file = file_list[0]
        duplicates = file_list[1:]
@@ -186,121 +186,68 @@ line-ending = "lf"
 # https://docs.astral.sh/ruff/rules/
 select = [ "E4", "E7", "E9", "F" ]
 extend-select = [
-  "ASYNC",   # https://docs.astral.sh/ruff/rules/#flake8-async-async
-  "B002",    # https://docs.astral.sh/ruff/rules/#flake8-bugbear-b
-  "B003",
-  "B004",
-  "B005",
-  "B006",
-  "B008",
-  "B009",
-  "B010",
-  "B012",
-  "B013",
-  "B014",
-  "B015",
-  "B016",
-  "B017",
-  "B018",
-  "B019",
-  "B020",
-  "B021",
-  "B022",
-  "B023",
-  "B025",
-  "B026",
-  "B029",
-  "B030",
-  "B031",
-  "B032",
-  "B033",
-  "B035",
-  "B039",
-  "C4",      # https://docs.astral.sh/ruff/rules/#flake8-comprehensions-c4
-  "COM",     # https://docs.astral.sh/ruff/rules/#flake8-commas-com
-  "D419",    # https://docs.astral.sh/ruff/rules/#pydocstyle-d
-  "DJ",      # https://docs.astral.sh/ruff/rules/#flake8-django-dj
-  "DTZ",     # https://docs.astral.sh/ruff/rules/#flake8-datetimez-dtz
-  "EXE",     # https://docs.astral.sh/ruff/rules/#flake8-executable-exe
-  "FA",      # https://docs.astral.sh/ruff/rules/#flake8-future-annotations-fa
-  "FBT",     # https://docs.astral.sh/ruff/rules/#flake8-boolean-trap-fbt
-  "FLY",     # https://docs.astral.sh/ruff/rules/#flynt-fly
-  "FURB",    # https://docs.astral.sh/ruff/rules/#refurb-furb
-  "G",       # https://docs.astral.sh/ruff/rules/#flake8-logging-format-g
-  "I",       # https://docs.astral.sh/ruff/rules/#isort-i
-  "ICN",     # https://docs.astral.sh/ruff/rules/#flake8-import-conventions-icn
-  "INP",     # https://docs.astral.sh/ruff/rules/#flake8-no-pep420-inp
-  "INT",     # https://docs.astral.sh/ruff/rules/#flake8-gettext-int
-  "ISC",     # https://docs.astral.sh/ruff/rules/#flake8-implicit-str-concat-isc
-  "LOG",     # https://docs.astral.sh/ruff/rules/#flake8-logging-log
-  "N999",    # https://docs.astral.sh/ruff/rules/#pep8-naming-n
-  "PERF101", # https://docs.astral.sh/ruff/rules/#perflint-perf
-  "PERF102",
-  "PERF402",
-  "PGH005",  # https://docs.astral.sh/ruff/rules/#pygrep-hooks-pgh
-  "PIE",     # https://docs.astral.sh/ruff/rules/#flake8-pie-pie
-  "PLC",     # https://docs.astral.sh/ruff/rules/#pylint-pl
-  "PLE",     # https://docs.astral.sh/ruff/rules/#error-ple
-  "PLR0124", # https://docs.astral.sh/ruff/rules/#refactor-plr
-  "PLR0133",
-  "PLR0206",
-  "PLR0402",
-  "PLR1704",
-  "PLR1708",
-  "PLR1711",
-  "PLR1716",
-  "PLR1722",
-  "PLR1730",
-  "PLR1733",
-  "PLR1736",
-  "PLR2044",
-  "PLW",     # https://docs.astral.sh/ruff/rules/#warning-plw
-  "PT010",   # https://docs.astral.sh/ruff/rules/#flake8-pytest-style-pt
-  "PT014",
-  "PT020",
-  "PT025",
-  "PT026",
-  "PT031",
-  "PTH",     # https://docs.astral.sh/ruff/rules/#flake8-use-pathlib-pth
-  "Q",       # https://docs.astral.sh/ruff/rules/#flake8-quotes-q
-  "RSE",     # https://docs.astral.sh/ruff/rules/#flake8-raise-rse
-  "RUF",     # https://docs.astral.sh/ruff/rules/#ruff-specific-rules-ruf
-  "S102",    # https://docs.astral.sh/ruff/rules/#flake8-bandit-s
-  "S110",
-  "S112",
-  "S113",
-  "SIM",     # https://docs.astral.sh/ruff/rules/#flake8-simplify-sim
-  "T100",    # https://docs.astral.sh/ruff/rules/#flake8-debugger-t10
-  "T20",     # https://docs.astral.sh/ruff/rules/#flake8-print-t20
-  "TC",      # https://docs.astral.sh/ruff/rules/#flake8-type-checking-tc
-  "TID",     # https://docs.astral.sh/ruff/rules/#flake8-tidy-imports-tid
-  "TRY002",  # https://docs.astral.sh/ruff/rules/#tryceratops-try
-  "TRY004",
-  "TRY201",
-  "TRY203",
-  "TRY401",
-  "UP",      # https://docs.astral.sh/ruff/rules/#pyupgrade-up
-  "W",       # https://docs.astral.sh/ruff/rules/#pycodestyle-e-w
-  "YTT",     # https://docs.astral.sh/ruff/rules/#flake8-2020-ytt
+  "COM",  # https://docs.astral.sh/ruff/rules/#flake8-commas-com
+  "DJ",   # https://docs.astral.sh/ruff/rules/#flake8-django-dj
+  "EXE",  # https://docs.astral.sh/ruff/rules/#flake8-executable-exe
+  "FBT",  # https://docs.astral.sh/ruff/rules/#flake8-boolean-trap-fbt
+  "FLY",  # https://docs.astral.sh/ruff/rules/#flynt-fly
+  "G201", # https://docs.astral.sh/ruff/rules/#flake8-logging-format-g
+  "I",    # https://docs.astral.sh/ruff/rules/#isort-i
+  "ICN",  # https://docs.astral.sh/ruff/rules/#flake8-import-conventions-icn
+  "INP",  # https://docs.astral.sh/ruff/rules/#flake8-no-pep420-inp
+  "ISC",  # https://docs.astral.sh/ruff/rules/#flake8-implicit-str-concat-isc
+  "PIE",  # https://docs.astral.sh/ruff/rules/#flake8-pie-pie
+  "PLC",  # https://docs.astral.sh/ruff/rules/#pylint-pl
+  "PLE",  # https://docs.astral.sh/ruff/rules/#pylint-pl
+  "PTH",  # https://docs.astral.sh/ruff/rules/#flake8-use-pathlib-pth
+  "Q",    # https://docs.astral.sh/ruff/rules/#flake8-quotes-q
+  "RSE",  # https://docs.astral.sh/ruff/rules/#flake8-raise-rse
+  "RUF",  # https://docs.astral.sh/ruff/rules/#ruff-specific-rules-ruf
+  "SIM",  # https://docs.astral.sh/ruff/rules/#flake8-simplify-sim
+  "T20",  # https://docs.astral.sh/ruff/rules/#flake8-print-t20
+  "TC",   # https://docs.astral.sh/ruff/rules/#flake8-type-checking-tc
+  "TID",  # https://docs.astral.sh/ruff/rules/#flake8-tidy-imports-tid
+  "UP",   # https://docs.astral.sh/ruff/rules/#pyupgrade-up
+  "W",    # https://docs.astral.sh/ruff/rules/#pycodestyle-e-w
 ]
 ignore = [
  "DJ001",
  "PLC0415",
  "RUF012",
  "SIM105",
-  "G004",    # Logging statement uses f-string - good to do, but a large diff
 ]
 # Migrations
-per-file-ignores."*/migrations/*.py" = []
+per-file-ignores."*/migrations/*.py" = [
+  "E501",
+  "SIM",
+  "T201",
+]
 # Testing
 per-file-ignores."*/tests/*.py" = [
-  "DTZ",
+  "E501",
  "SIM117",
 ]
+per-file-ignores.".github/scripts/*.py" = [
+  "E501",
+  "INP001",
+  "SIM117",
+]
+# Docker specific
+per-file-ignores."docker/rootfs/usr/local/bin/wait-for-redis.py" = [
+  "INP001",
+  "T201",
+]
+per-file-ignores."docker/wait-for-redis.py" = [
+  "INP001",
+  "T201",
+]
+per-file-ignores."src/documents/models.py" = [
+  "SIM115",
+]
 isort.force-single-line = true

 [tool.codespell]
-ignore-words-list = "criterias,afterall,valeu,ureue,equest,ure,assertIn,Oktober,commitish"
+ignore-words-list = "criterias,afterall,valeu,ureue,equest,ure,assertIn,Oktober,commitish,NIN,nin"
 skip = """\
  src-ui/src/locale/*,src-ui/pnpm-lock.yaml,src-ui/e2e/*,src/paperless_mail/tests/samples/*,src/paperless/tests/samples\
  /mail/*,src/documents/tests/samples/*,*.po,*.json\
@@ -41,6 +41,8 @@ export class TrashComponent
  private modalService = inject(NgbModal)
  private settingsService = inject(SettingsService)
  private router = inject(Router)
+  private readonly emptyTrashDelaySetting =
+    this.settingsService.getSignal<number>(SETTINGS_KEYS.EMPTY_TRASH_DELAY)

  readonly documentsInTrash = signal<Document[]>([])
  readonly selectedDocuments = signal<Set<number>>(new Set())
@@ -200,8 +202,7 @@ export class TrashComponent
  }

  getDaysRemaining(document: Document): number {
-    this.settingsService.trackChanges()
-    const delay = this.settingsService.get(SETTINGS_KEYS.EMPTY_TRASH_DELAY)
+    const delay = this.emptyTrashDelaySetting()
    const diff = new Date().getTime() - new Date(document.deleted_at).getTime()
    const days = Math.ceil(diff / (1000 * 3600 * 24))
    return delay - days
@@ -193,6 +193,23 @@ describe('AppFrameComponent', () => {
    expect(savedViewSpy).toHaveBeenCalled()
  })

+  it('should update reinitialized signal-backed settings without manual change detection', async () => {
+    settingsService.initializeSettings().subscribe()
+    httpTestingController
+      .expectOne(`${environment.apiBaseUrl}ui_settings/`)
+      .flush({
+        settings: { app_title: 'Reactive title' },
+        user: {},
+        permissions: [],
+      })
+
+    await fixture.whenStable()
+
+    expect(
+      fixture.nativeElement.querySelector('.brand-title').textContent
+    ).toBe('Reactive title')
+  })
+
  it('should check for update if enabled', () => {
    const updateCheckSpy = jest.spyOn(remoteVersionService, 'checkForUpdates')
    updateCheckSpy.mockImplementation(() => {
@@ -98,6 +98,29 @@ export class AppFrameComponent
  readonly isMenuCollapsed = signal(true)
  readonly slimSidebarAnimating = signal(false)
  readonly mobileSearchHidden = signal(false)
+  private readonly versionSetting = this.settingsService.getSignal<string>(
+    SETTINGS_KEYS.VERSION
+  )
+  private readonly appTitleSetting = this.settingsService.getSignal<string>(
+    SETTINGS_KEYS.APP_TITLE
+  )
+  private readonly appLogoSetting = this.settingsService.getSignal<string>(
+    SETTINGS_KEYS.APP_LOGO
+  )
+  private readonly slimSidebarSetting = this.settingsService.getSignal<boolean>(
+    SETTINGS_KEYS.SLIM_SIDEBAR
+  )
+  private readonly attributesSectionsCollapsedSetting =
+    this.settingsService.getSignal<CollapsibleSection[]>(
+      SETTINGS_KEYS.ATTRIBUTES_SECTIONS_COLLAPSED
+    )
+  private readonly aiEnabledSetting = this.settingsService.getSignal<boolean>(
+    SETTINGS_KEYS.AI_ENABLED
+  )
+  private readonly sidebarViewsShowCountSetting =
+    this.settingsService.getSignal<boolean>(
+      SETTINGS_KEYS.SIDEBAR_VIEWS_SHOW_COUNT
+    )
  private lastScrollY: number = 0

  constructor() {
@@ -191,33 +214,23 @@ export class AppFrameComponent
  }

  get versionString(): string {
-    this.settingsService.trackChanges()
-    return `${environment.appTitle} v${this.settingsService.get(SETTINGS_KEYS.VERSION)}${environment.tag === 'prod' ? '' : ` #${environment.tag}`}`
+    return `${environment.appTitle} v${this.versionSetting()}${environment.tag === 'prod' ? '' : ` #${environment.tag}`}`
  }

  get appTitle(): string {
-    this.settingsService.trackChanges()
-    return (
-      this.settingsService.get(SETTINGS_KEYS.APP_TITLE) || environment.appTitle
-    )
+    return this.appTitleSetting() || environment.appTitle
  }

  get customAppTitle(): string {
-    this.settingsService.trackChanges()
-    return this.settingsService.get(SETTINGS_KEYS.APP_TITLE)
+    return this.appTitleSetting()
  }

  get hasCustomBranding(): boolean {
-    this.settingsService.trackChanges()
-    return !!(
-      this.settingsService.get(SETTINGS_KEYS.APP_TITLE)?.length ||
-      this.settingsService.get(SETTINGS_KEYS.APP_LOGO)?.length
-    )
+    return !!(this.appTitleSetting()?.length || this.appLogoSetting()?.length)
  }

  get customAppLogo(): string {
-    this.settingsService.trackChanges()
-    const logo = this.settingsService.get(SETTINGS_KEYS.APP_LOGO)
+    const logo = this.appLogoSetting()
    return logo?.length
      ? environment.apiBaseUrl.replace(/\/api\/$/, logo)
      : null
@@ -262,8 +275,7 @@ export class AppFrameComponent
  }

  get slimSidebarEnabled(): boolean {
-    this.settingsService.trackChanges()
-    return this.settingsService.get(SETTINGS_KEYS.SLIM_SIDEBAR)
+    return this.slimSidebarSetting()
  }

  set slimSidebarEnabled(enabled: boolean) {
@@ -286,10 +298,9 @@ export class AppFrameComponent
  }

  get attributesSectionsCollapsed(): boolean {
-    this.settingsService.trackChanges()
-    return this.settingsService
-      .get(SETTINGS_KEYS.ATTRIBUTES_SECTIONS_COLLAPSED)
-      ?.includes(CollapsibleSection.ATTRIBUTES)
+    return this.attributesSectionsCollapsedSetting()?.includes(
+      CollapsibleSection.ATTRIBUTES
+    )
  }

  set attributesSectionsCollapsed(collapsed: boolean) {
@@ -312,8 +323,7 @@ export class AppFrameComponent
  }

  get aiEnabled(): boolean {
-    this.settingsService.trackChanges()
-    return this.settingsService.get(SETTINGS_KEYS.AI_ENABLED)
+    return this.aiEnabledSetting()
  }

  @HostListener('window:resize')
@@ -480,9 +490,8 @@ export class AppFrameComponent
  }

  get showSidebarCounts(): boolean {
-    this.settingsService.trackChanges()
    return (
-      this.settingsService.get(SETTINGS_KEYS.SIDEBAR_VIEWS_SHOW_COUNT) &&
+      this.sidebarViewsShowCountSetting() &&
      !this.settingsService.organizingSidebarSavedViews()
    )
  }
@@ -81,6 +81,10 @@ export class GlobalSearchComponent implements OnInit {
  private hotkeyService = inject(HotKeyService)
  private settingsService = inject(SettingsService)
  private locationStrategy = inject(LocationStrategy)
+  private readonly searchFullTypeSetting =
+    this.settingsService.getSignal<GlobalSearchType>(
+      SETTINGS_KEYS.SEARCH_FULL_TYPE
+    )

  public DataType = DataType
  readonly query = signal<string>(null)
@@ -97,11 +101,7 @@ export class GlobalSearchComponent implements OnInit {
  @ViewChildren('secondaryButton') secondaryButtons: QueryList<ElementRef>

  get useAdvancedForFullSearch(): boolean {
-    this.settingsService.trackChanges()
-    return (
-      this.settingsService.get(SETTINGS_KEYS.SEARCH_FULL_TYPE) ===
-      GlobalSearchType.ADVANCED
-    )
+    return this.searchFullTypeSetting() === GlobalSearchType.ADVANCED
  }

  constructor() {
@@ -196,6 +196,16 @@ describe('WorkflowEditDialogComponent', () => {
    fixture.detectChanges()
  })

+  function setActionSettings({
+    email = true,
+    remoteOcr = true,
+    ai = true,
+  } = {}) {
+    settingsService.set(SETTINGS_KEYS.EMAIL_ENABLED, email)
+    settingsService.set(SETTINGS_KEYS.REMOTE_OCR_CONFIGURED, remoteOcr)
+    settingsService.set(SETTINGS_KEYS.AI_ENABLED, ai)
+  }
+
  it('should support create and edit modes, support adding triggers and actions on new workflow', () => {
    component.dialogMode.set(EditDialogMode.CREATE)
    const createTitleSpy = jest.spyOn(component, 'getCreateTitle')
@@ -218,7 +228,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should return source options, type options, type name, schedule date field options', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()
    component.ngOnInit()
    expect(component.sourceOptions).toEqual(DOCUMENT_SOURCE_OPTIONS)
    expect(component.triggerTypeOptions).toEqual(WORKFLOW_TYPE_OPTIONS)
@@ -242,7 +252,7 @@ describe('WorkflowEditDialogComponent', () => {
    )

    // Email, remote OCR and AI all disabled
-    jest.spyOn(settingsService, 'get').mockReturnValue(false)
+    setActionSettings({ email: false, remoteOcr: false, ai: false })
    component.ngOnInit()
    expect(component.actionTypeOptions).toEqual(
      WORKFLOW_ACTION_OPTIONS.filter(
@@ -255,7 +265,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should offer remote OCR only for consumption workflows', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()

    // A consumption trigger makes the action reachable
    component.object = {
@@ -285,7 +295,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should offer remote OCR on a trigger added to a new workflow', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()
    component.ngOnInit()

    // Nothing for the action to apply to yet
@@ -311,7 +321,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should keep remote OCR listed when an action already uses it', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()

    // Otherwise changing the trigger would silently blank the selection
    component.object = {
@@ -329,9 +339,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should not offer remote OCR when no engine is configured', () => {
-    jest
-      .spyOn(settingsService, 'get')
-      .mockImplementation((key) => key !== SETTINGS_KEYS.REMOTE_OCR_CONFIGURED)
+    setActionSettings({ remoteOcr: false })

    component.object = {
      name: 'Workflow 1',
@@ -348,7 +356,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should offer apply AI suggestions unless every trigger is consumption', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()

    // Consumption runs before the document has been parsed, so there would be
    // no content to make suggestions from
@@ -382,7 +390,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should keep apply AI suggestions listed when an action already uses it', () => {
-    jest.spyOn(settingsService, 'get').mockReturnValue(true)
+    setActionSettings()

    // Otherwise changing the trigger would silently blank the selection
    component.object = {
@@ -400,9 +408,7 @@ describe('WorkflowEditDialogComponent', () => {
  })

  it('should not offer apply AI suggestions when AI is disabled', () => {
-    jest
-      .spyOn(settingsService, 'get')
-      .mockImplementation((key) => key !== SETTINGS_KEYS.AI_ENABLED)
+    setActionSettings({ ai: false })

    component.object = {
      name: 'Workflow 1',
@@ -537,6 +537,13 @@ export class WorkflowEditDialogComponent
  readonly dateCustomFields = computed(() =>
    this.customFields()?.filter((f) => f.data_type === CustomFieldDataType.Date)
  )
+  private readonly emailEnabledSetting =
+    this.settingsService.getSignal<boolean>(SETTINGS_KEYS.EMAIL_ENABLED)
+  private readonly remoteOcrConfiguredSetting =
+    this.settingsService.getSignal<boolean>(SETTINGS_KEYS.REMOTE_OCR_CONFIGURED)
+  private readonly aiEnabledSetting = this.settingsService.getSignal<boolean>(
+    SETTINGS_KEYS.AI_ENABLED
+  )

  expandedItem: number = null

@@ -589,7 +596,7 @@ export class WorkflowEditDialogComponent
  private getAllowedActionTypes() {
    let allowed = WORKFLOW_ACTION_OPTIONS

-    if (!this.settingsService.get(SETTINGS_KEYS.EMAIL_ENABLED)) {
+    if (!this.emailEnabledSetting()) {
      allowed = allowed.filter((a) => a.id !== WorkflowActionType.Email)
    }

@@ -597,7 +604,7 @@ export class WorkflowEditDialogComponent
    // offered for workflows that run at consumption.
    const formWorkflow: Workflow = this.objectForm?.value
    const remoteOcrUsable =
-      this.settingsService.get(SETTINGS_KEYS.REMOTE_OCR_CONFIGURED) &&
+      this.remoteOcrConfiguredSetting() &&
      (formWorkflow?.triggers?.some(
        (trigger) => trigger.type === WorkflowTriggerType.Consumption
      ) ||
@@ -612,7 +619,7 @@ export class WorkflowEditDialogComponent
    // once every trigger is consumption, so it stays offered on a workflow
    // that has no triggers yet.
    const aiSuggestionsUsable =
-      this.settingsService.get(SETTINGS_KEYS.AI_ENABLED) &&
+      this.aiEnabledSetting() &&
      (!formWorkflow?.triggers?.length ||
        formWorkflow.triggers.some(
          (trigger) => trigger.type !== WorkflowTriggerType.Consumption
@@ -1362,7 +1369,6 @@ export class WorkflowEditDialogComponent
  }

  get actionTypeOptions() {
-    this.settingsService.trackChanges()
    // Computed on read rather than cached
    return this.getAllowedActionTypes()
  }
@@ -839,7 +839,9 @@ describe('FilterableDropdownComponent & FilterableDropdownSelectionModel', () =>
    selectionModel.items = [memoRoot]
    selectionModel.documentCounts = [{ id: memoRoot.id, document_count: 9 }]

-    const getRootDocCount = (selectionModel as any).createRootDocCounter()
+    const getRootDocCount = (selectionModel as any).createRootDocCounter(
+      selectionModel.items
+    )

    expect(getRootDocCount(memoRoot.id)).toEqual(9)
    selectionModel.documentCounts = []
@@ -855,7 +857,9 @@ describe('FilterableDropdownComponent & FilterableDropdownSelectionModel', () =>
    selectionModel.items = [rootWithoutSelection]
    selectionModel.documentCounts = []

-    const getRootDocCount = (selectionModel as any).createRootDocCounter()
+    const getRootDocCount = (selectionModel as any).createRootDocCounter(
+      selectionModel.items
+    )

    expect(getRootDocCount(rootWithoutSelection.id)).toEqual(4)
  })
@@ -865,7 +869,9 @@ describe('FilterableDropdownComponent & FilterableDropdownSelectionModel', () =>
    selectionModel.items = [rootWithoutCounts]
    selectionModel.documentCounts = []

-    const getRootDocCount = (selectionModel as any).createRootDocCounter()
+    const getRootDocCount = (selectionModel as any).createRootDocCounter(
+      selectionModel.items
+    )

    expect(getRootDocCount(rootWithoutCounts.id)).toEqual(0)
  })
@@ -966,7 +972,7 @@ describe('FilterableDropdownComponent & FilterableDropdownSelectionModel', () =>
    component.selectionModel['temporarySelectionStates'].set(id, state)
    const changedSpy = jest.spyOn(component.selectionModel.changed, 'next')
    component.selectionModel.exclude(id)
-    expect(component.selectionModel.temporaryLogicalOperator).toBe(
+    expect(component.selectionModel.temporaryLogicalOperator()).toBe(
      LogicalOperator.And
    )
    expect(component.selectionModel['temporarySelectionStates'].get(id)).toBe(
@@ -64,43 +64,56 @@ export class FilterableDropdownSelectionModel {

  manyToOne = false
  singleSelect = false
-  private _logicalOperator: LogicalOperator = LogicalOperator.And
-  temporaryLogicalOperator: LogicalOperator = this._logicalOperator
-  private _intersection: Intersection = Intersection.Include
-  temporaryIntersection: Intersection = this._intersection

-  private _documentCounts: SelectionDataItem[] = []
+  private readonly _logicalOperator = signal(LogicalOperator.And)
+  readonly temporaryLogicalOperator = signal(LogicalOperator.And)
+  private readonly _intersection = signal(Intersection.Include)
+  readonly temporaryIntersection = signal(Intersection.Include)
+  private readonly _documentCounts = signal<SelectionDataItem[]>([])
+  private readonly _items = signal<MatchingModel[]>([])
+  private readonly _selectionStates = signal(
+    new Map<number, ToggleableItemState>()
+  )
+  private readonly _temporarySelectionStates = signal(
+    new Map<number, ToggleableItemState>()
+  )
+
  public documentCountSortingEnabled = false

+  private get selectionStates(): ReadonlyMap<number, ToggleableItemState> {
+    return this._selectionStates()
+  }
+
+  private get temporarySelectionStates(): ReadonlyMap<
+    number,
+    ToggleableItemState
+  > {
+    return this._temporarySelectionStates()
+  }
+
  public set documentCounts(counts: SelectionDataItem[]) {
-    this._documentCounts = counts
+    this._documentCounts.set(counts)
    if (this.documentCountSortingEnabled) {
-      this.sortItems()
+      this._items.set(this.sortItems(this.items))
    }
  }

-  private _items: MatchingModel[] = []
  get items(): MatchingModel[] {
-    return this._items
+    return this._items()
  }

  set items(items: MatchingModel[]) {
    if (items) {
-      this._items = Array.from(items)
-      this.sortItems()
-      this.setNullItem()
+      this._items.set(this.withNullItem(this.sortItems(Array.from(items))))
    }
  }

-  private setNullItem() {
+  private withNullItem(items: MatchingModel[]): MatchingModel[] {
    if (this.manyToOne && this.logicalOperator === LogicalOperator.Or) {
-      if (this._items[0]?.id === null) {
-        this._items.shift()
-      }
-      return
+      return items[0]?.id === null ? items.slice(1) : items
    }

-    const item = {
+    const nullItem = {
      name: $localize`:Filter drop down element to filter for documents with no correspondent/type/tag assigned:Not assigned`,
      id:
        this.manyToOne || this.intersection === Intersection.Include
@@ -108,22 +121,17 @@ export class FilterableDropdownSelectionModel {
          : NEGATIVE_NULL_FILTER_VALUE,
    }

-    if (
-      this._items[0]?.id === null ||
-      this._items[0]?.id === NEGATIVE_NULL_FILTER_VALUE
-    ) {
-      this._items[0] = item
-    } else if (this._items) {
-      this._items.unshift(item)
-    }
+    return items[0]?.id === null || items[0]?.id === NEGATIVE_NULL_FILTER_VALUE
+      ? [nullItem, ...items.slice(1)]
+      : [nullItem, ...items]
  }

  constructor(manyToOne: boolean = false) {
    this.manyToOne = manyToOne
  }

-  private sortItems() {
-    this._items.sort((a, b) => {
+  private sortItems(items: MatchingModel[]): MatchingModel[] {
+    const sorted = [...items].sort((a, b) => {
      if (
        (a.id == null && b.id != null) ||
        (a.id == NEGATIVE_NULL_FILTER_VALUE &&
@@ -154,13 +162,13 @@ export class FilterableDropdownSelectionModel {
      ) {
        return -1
      } else if (
-        this._documentCounts.length &&
+        this._documentCounts().length &&
        this.getDocumentCount(b.id) === 0 &&
        this.getDocumentCount(a.id) > this.getDocumentCount(b.id)
      ) {
        return -1
      } else if (
-        this._documentCounts.length &&
+        this._documentCounts().length &&
        this.getDocumentCount(a.id) === 0 &&
        this.getDocumentCount(a.id) < this.getDocumentCount(b.id)
      ) {
@@ -170,15 +178,11 @@ export class FilterableDropdownSelectionModel {
      }
    })

-    if (this._documentCounts.length) {
-      this.promoteBranchesWithDocumentCounts()
-    }
+    return this._documentCounts().length
+      ? this.promoteBranchesWithDocumentCounts(sorted)
+      : sorted
  }

-  private selectionStates = new Map<number, ToggleableItemState>()
-
-  private temporarySelectionStates = new Map<number, ToggleableItemState>()
-
  getSelectedItems() {
    return this.items.filter(
      (i) =>
@@ -194,30 +198,33 @@ export class FilterableDropdownSelectionModel {
  }

  set(id: number, state: ToggleableItemState, fireEvent = true) {
+    const states = new Map(this.temporarySelectionStates)
    if (state == ToggleableItemState.NotSelected) {
-      this.temporarySelectionStates.delete(id)
+      states.delete(id)
    } else {
-      this.temporarySelectionStates.set(id, state)
+      states.set(id, state)
    }
+    this._temporarySelectionStates.set(states)
    if (fireEvent) {
      this.changed.next(this)
    }
  }

  toggle(id: number, fireEvent = true) {
-    let state = this.temporarySelectionStates.get(id)
+    const states = new Map(this.temporarySelectionStates)
+    let state = states.get(id)
    if (
      state == undefined ||
      (state != ToggleableItemState.Selected &&
        state != ToggleableItemState.Excluded)
    ) {
      if (this.manyToOne || this.singleSelect) {
-        this.temporarySelectionStates.set(id, ToggleableItemState.Selected)
+        states.set(id, ToggleableItemState.Selected)

        if (this.singleSelect) {
-          for (let key of this.temporarySelectionStates.keys()) {
+          for (let key of states.keys()) {
            if (key != id) {
-              this.temporarySelectionStates.delete(key)
+              states.delete(key)
            }
          }
        }
@@ -233,25 +240,26 @@ export class FilterableDropdownSelectionModel {
        ) {
          newState = ToggleableItemState.NotSelected
        }
-        this.temporarySelectionStates.set(id, newState)
+        states.set(id, newState)
      }
    } else if (
      state == ToggleableItemState.Selected ||
      state == ToggleableItemState.Excluded
    ) {
-      this.temporarySelectionStates.delete(id)
-      this.clearDescendantSelections(id)
+      states.delete(id)
+      this.clearDescendantSelections(states, id)
    }

    if (!id) {
-      for (let key of this.temporarySelectionStates.keys()) {
+      for (let key of states.keys()) {
        if (key) {
-          this.temporarySelectionStates.delete(key)
+          states.delete(key)
        }
      }
    } else {
-      this.temporarySelectionStates.delete(null)
+      states.delete(null)
    }
+    this._temporarySelectionStates.set(states)

    if (fireEvent) {
      this.changed.next(this)
@@ -259,20 +267,21 @@ export class FilterableDropdownSelectionModel {
  }

  exclude(id: number, fireEvent: boolean = true) {
-    let state = this.temporarySelectionStates.get(id)
+    const states = new Map(this.temporarySelectionStates)
+    let state = states.get(id)
    if (id && (state == null || state != ToggleableItemState.Excluded)) {
-      this.temporaryLogicalOperator = this._logicalOperator = this.manyToOne
-        ? LogicalOperator.And
-        : LogicalOperator.Or
+      const operator = this.manyToOne ? LogicalOperator.And : LogicalOperator.Or
+      this.temporaryLogicalOperator.set(operator)
+      this._logicalOperator.set(operator)

      if (this.manyToOne || this.singleSelect) {
-        this.temporarySelectionStates.set(id, ToggleableItemState.Excluded)
-        this.clearDescendantSelections(id)
+        states.set(id, ToggleableItemState.Excluded)
+        this.clearDescendantSelections(states, id)

        if (this.singleSelect) {
-          for (let key of this.temporarySelectionStates.keys()) {
+          for (let key of states.keys()) {
            if (key != id) {
-              this.temporarySelectionStates.delete(key)
+              states.delete(key)
            }
          }
        }
@@ -287,17 +296,18 @@ export class FilterableDropdownSelectionModel {
        ) {
          newState = ToggleableItemState.NotSelected
        }
-        this.temporarySelectionStates.set(id, newState)
+        states.set(id, newState)
        if (newState == ToggleableItemState.Excluded) {
-          this.clearDescendantSelections(id)
+          this.clearDescendantSelections(states, id)
        }
      }
    } else if (!id || state == ToggleableItemState.Excluded) {
-      this.temporarySelectionStates.delete(id)
+      states.delete(id)
      if (id) {
-        this.clearDescendantSelections(id)
+        this.clearDescendantSelections(states, id)
      }
    }
+    this._temporarySelectionStates.set(states)

    if (fireEvent) {
      this.changed.next(this)
@@ -308,9 +318,12 @@ export class FilterableDropdownSelectionModel {
    return this.selectionStates.get(id) || ToggleableItemState.NotSelected
  }

-  private clearDescendantSelections(id: number) {
+  private clearDescendantSelections(
+    states: Map<number, ToggleableItemState>,
+    id: number
+  ) {
    for (const descendantID of this.getDescendantIDs(id)) {
-      this.temporarySelectionStates.delete(descendantID)
+      states.delete(descendantID)
    }
  }

@@ -320,7 +333,7 @@ export class FilterableDropdownSelectionModel {

    while (queue.length) {
      const parentID = queue.shift()
-      for (const item of this._items) {
+      for (const item of this.items) {
        if (
          typeof item?.id === 'number' &&
          typeof (item as any)['parent'] === 'number' &&
@@ -336,12 +349,12 @@ export class FilterableDropdownSelectionModel {
  }

  get logicalOperator(): LogicalOperator {
-    return this.temporaryLogicalOperator
+    return this.temporaryLogicalOperator()
  }

  set logicalOperator(operator: LogicalOperator) {
-    this.temporaryLogicalOperator = operator
-    this.setNullItem()
+    this.temporaryLogicalOperator.set(operator)
+    this._items.set(this.withNullItem(this.items))
  }

  toggleOperator() {
@@ -349,12 +362,12 @@ export class FilterableDropdownSelectionModel {
  }

  get intersection(): Intersection {
-    return this.temporaryIntersection
+    return this.temporaryIntersection()
  }

  set intersection(intersection: Intersection) {
-    this.temporaryIntersection = intersection
-    this.setNullItem()
+    this.temporaryIntersection.set(intersection)
+    this._items.set(this.withNullItem(this.items))
  }

  toggleIntersection() {
@@ -364,18 +377,20 @@ export class FilterableDropdownSelectionModel {
        ? ToggleableItemState.Selected
        : ToggleableItemState.Excluded

-    this.temporarySelectionStates.forEach((state, key) => {
+    const states = new Map(this.temporarySelectionStates)
+    states.forEach((state, key) => {
      if (key === null && this.intersection === Intersection.Exclude) {
-        this.temporarySelectionStates.set(NEGATIVE_NULL_FILTER_VALUE, newState)
+        states.set(NEGATIVE_NULL_FILTER_VALUE, newState)
      } else if (
        key === NEGATIVE_NULL_FILTER_VALUE &&
        this.intersection === Intersection.Include
      ) {
-        this.temporarySelectionStates.set(null, newState)
+        states.set(null, newState)
      } else {
-        this.temporarySelectionStates.set(key, newState)
+        states.set(key, newState)
      }
    })
+    this._temporarySelectionStates.set(states)

    this.changed.next(this)
  }
@@ -395,10 +410,12 @@ export class FilterableDropdownSelectionModel {
  }

  clear(fireEvent = true) {
-    this.temporarySelectionStates.clear()
-    this.temporaryLogicalOperator = this._logicalOperator = LogicalOperator.And
-    this.temporaryIntersection = this._intersection = Intersection.Include
-    this.setNullItem()
+    this._temporarySelectionStates.set(new Map())
+    this.temporaryLogicalOperator.set(LogicalOperator.And)
+    this._logicalOperator.set(LogicalOperator.And)
+    this.temporaryIntersection.set(Intersection.Include)
+    this._intersection.set(Intersection.Include)
+    this._items.set(this.withNullItem(this.items))
    if (fireEvent) {
      this.changed.next(this)
    }
@@ -419,9 +436,9 @@ export class FilterableDropdownSelectionModel {
      )
    ) {
      return true
-    } else if (this.temporaryLogicalOperator !== this._logicalOperator) {
+    } else if (this.temporaryLogicalOperator() !== this._logicalOperator()) {
      return true
-    } else if (this.temporaryIntersection !== this._intersection) {
+    } else if (this.temporaryIntersection() !== this._intersection()) {
      return true
    } else {
      return false
@@ -438,23 +455,29 @@ export class FilterableDropdownSelectionModel {
  }

  getDocumentCount(id: number) {
-    return this._documentCounts.find((c) => c.id === id)?.document_count
+    return this._documentCounts().find((c) => c.id === id)?.document_count
  }

-  private promoteBranchesWithDocumentCounts() {
-    const parentById = this.buildParentById()
+  private promoteBranchesWithDocumentCounts(
+    items: MatchingModel[]
+  ): MatchingModel[] {
+    const parentById = this.buildParentById(items)
    const findRootId = this.createRootFinder(parentById)
-    const getRootDocCount = this.createRootDocCounter()
-    const summaries = this.buildBranchSummaries(findRootId, getRootDocCount)
+    const getRootDocCount = this.createRootDocCounter(items)
+    const summaries = this.buildBranchSummaries(
+      items,
+      findRootId,
+      getRootDocCount
+    )
    const orderedBranches = this.orderBranchesByPriority(summaries)

-    this._items = orderedBranches.flatMap((summary) => summary.items)
+    return orderedBranches.flatMap((summary) => summary.items)
  }

-  private buildParentById(): Map<number, number | null> {
+  private buildParentById(items: MatchingModel[]): Map<number, number | null> {
    const parentById = new Map<number, number | null>()

-    for (const item of this._items) {
+    for (const item of items) {
      if (typeof item?.id === 'number') {
        const parentValue = (item as any)['parent']
        parentById.set(
@@ -492,7 +515,9 @@ export class FilterableDropdownSelectionModel {
    return findRootId
  }

-  private createRootDocCounter(): (rootId: number) => number {
+  private createRootDocCounter(
+    items: MatchingModel[]
+  ): (rootId: number) => number {
    const docCountMemo = new Map<number, number>()

    return (rootId: number): number => {
@@ -507,7 +532,7 @@ export class FilterableDropdownSelectionModel {
        return explicit
      }

-      const rootItem = this._items.find((i) => i.id === rootId)
+      const rootItem = items.find((i) => i.id === rootId)
      const fallback =
        typeof (rootItem as any)?.['document_count'] === 'number'
          ? (rootItem as any)['document_count']
@@ -519,12 +544,13 @@ export class FilterableDropdownSelectionModel {
  }

  private buildBranchSummaries(
+    items: MatchingModel[],
    findRootId: (id: number) => number,
    getRootDocCount: (rootId: number) => number
  ): Map<string, BranchSummary> {
    const summaries = new Map<string, BranchSummary>()

-    for (const [index, item] of this._items.entries()) {
+    for (const [index, item] of items.entries()) {
      const { key, special, rootId } = this.describeBranchItem(
        item,
        index,
@@ -616,28 +642,23 @@ export class FilterableDropdownSelectionModel {
  }

  init(map: Map<number, ToggleableItemState>) {
-    this.temporarySelectionStates = map
+    this._temporarySelectionStates.set(new Map(map))
    this.apply()
  }

  apply() {
-    this.selectionStates.clear()
-    this.temporarySelectionStates.forEach((value, key) => {
-      this.selectionStates.set(key, value)
-    })
-    this._logicalOperator = this.temporaryLogicalOperator
-    this._intersection = this.temporaryIntersection
-    this.sortItems()
+    this._selectionStates.set(new Map(this.temporarySelectionStates))
+    this._logicalOperator.set(this.temporaryLogicalOperator())
+    this._intersection.set(this.temporaryIntersection())
+    this._items.set(this.sortItems(this.items))
  }

  reset(complete: boolean = false) {
-    this.temporarySelectionStates.clear()
    if (complete) {
-      this.selectionStates.clear()
+      this._selectionStates.set(new Map())
+      this._temporarySelectionStates.set(new Map())
    } else {
-      this.selectionStates.forEach((value, key) => {
-        this.temporarySelectionStates.set(key, value)
-      })
+      this._temporarySelectionStates.set(new Map(this.selectionStates))
    }
  }

@@ -7,7 +7,7 @@
    <div class="list-group list-group-flush">
      <button class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" (click)="setFilter(OwnerFilterType.NONE)" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.NONE) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.NONE) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -17,7 +17,7 @@
      </button>
      <button class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" (click)="setFilter(OwnerFilterType.SELF)" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.SELF) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.SELF) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -27,7 +27,7 @@
      </button>
      <button class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" (click)="setFilter(OwnerFilterType.NOT_SELF)" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.NOT_SELF) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.NOT_SELF) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -37,7 +37,7 @@
      </button>
      <button class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" (click)="setFilter(OwnerFilterType.SHARED_BY_ME)" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.SHARED_BY_ME) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.SHARED_BY_ME) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -47,7 +47,7 @@
      </button>
      <button class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" (click)="setFilter(OwnerFilterType.UNOWNED)" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.UNOWNED) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.UNOWNED) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -57,7 +57,7 @@
      </button>
      <button *pngxIfPermissions="{ action: PermissionAction.View, type: PermissionType.User }" class="list-group-item list-group-item-action d-flex align-items-center p-2 border-top-0 border-start-0 border-end-0 border-bottom" role="menuitem" [disabled]="disabled">
        <div class="selected-icon me-1">
-          @if (selectionModel.ownerFilter === OwnerFilterType.OTHERS) {
+          @if (selectionModel.ownerFilter() === OwnerFilterType.OTHERS) {
            <i-bs width="1em" height="1em" name="check"></i-bs>
          }
        </div>
@@ -65,7 +65,8 @@
          <ng-select
            name="user"
            class="user-select small"
-            [(ngModel)]="selectionModel.includeUsers"
+            [ngModel]="selectionModel.includeUsers()"
+            (ngModelChange)="selectionModel.includeUsers.set($event)"
            [disabled]="disabled"
            [clearable]="false"
            [items]="users()"
@@ -78,10 +79,10 @@
          </ng-select>
        </div>
      </button>
-      @if (selectionModel.ownerFilter === OwnerFilterType.NONE || selectionModel.ownerFilter === OwnerFilterType.NOT_SELF) {
+      @if (selectionModel.ownerFilter() === OwnerFilterType.NONE || selectionModel.ownerFilter() === OwnerFilterType.NOT_SELF) {
        <div class="list-group-item list-group-item-action d-flex align-items-center p-2 ps-3 border-bottom-0 border-start-0 border-end-0">
          <div class="form-check form-switch w-100">
-            <input type="checkbox" class="form-check-input" id="hideUnowned" [(ngModel)]="this.selectionModel.hideUnowned" (change)="onChange()" [disabled]="disabled">
+            <input type="checkbox" class="form-check-input" id="hideUnowned" [ngModel]="selectionModel.hideUnowned()" (ngModelChange)="selectionModel.hideUnowned.set($event)" (change)="onChange()" [disabled]="disabled">
            <label class="form-check-label w-100" for="hideUnowned"><small i18n>Hide unowned</small></label>
          </div>
        </div>
@@ -90,56 +90,56 @@ describe('PermissionsFilterDropdownComponent', () => {
    component.setFilter(OwnerFilterType.OTHERS)
    expect(component.isActive).toBeTruthy()
    component.setFilter(OwnerFilterType.NONE)
-    component.selectionModel.hideUnowned = true
+    component.selectionModel.hideUnowned.set(true)
    expect(component.isActive).toBeTruthy()
  })

  it('should describe concrete user filters honestly', () => {
-    component.selectionModel.ownerFilter = OwnerFilterType.SELF
-    component.selectionModel.userID = 1
+    component.selectionModel.ownerFilter.set(OwnerFilterType.SELF)
+    component.selectionModel.userID.set(1)
    expect(component.ownerFilterLabel).toEqual('Owned by user1')

-    component.selectionModel.ownerFilter = OwnerFilterType.NOT_SELF
-    component.selectionModel.excludeUsers = [1]
+    component.selectionModel.ownerFilter.set(OwnerFilterType.NOT_SELF)
+    component.selectionModel.excludeUsers.set([1])
    expect(component.ownerExclusionFilterLabel).toEqual('Not owned by user1')

-    component.selectionModel.ownerFilter = OwnerFilterType.SHARED_BY_ME
-    component.selectionModel.userID = 1
+    component.selectionModel.ownerFilter.set(OwnerFilterType.SHARED_BY_ME)
+    component.selectionModel.userID.set(1)
    expect(component.sharedByFilterLabel).toEqual('Shared by user1')
  })

  it('should describe concrete filters when usernames are unavailable', () => {
-    component.selectionModel.ownerFilter = OwnerFilterType.SELF
-    component.selectionModel.userID = 99
+    component.selectionModel.ownerFilter.set(OwnerFilterType.SELF)
+    component.selectionModel.userID.set(99)
    expect(component.ownerFilterLabel).toEqual('Owned by another user')

-    component.selectionModel.ownerFilter = OwnerFilterType.NOT_SELF
-    component.selectionModel.excludeUsers = [99]
+    component.selectionModel.ownerFilter.set(OwnerFilterType.NOT_SELF)
+    component.selectionModel.excludeUsers.set([99])
    expect(component.ownerExclusionFilterLabel).toEqual(
      'Not owned by another user'
    )

-    component.selectionModel.excludeUsers = [98, 99]
+    component.selectionModel.excludeUsers.set([98, 99])
    expect(component.ownerExclusionFilterLabel).toEqual(
      'Not owned by selected users'
    )

-    component.selectionModel.ownerFilter = OwnerFilterType.SHARED_BY_ME
-    component.selectionModel.userID = 99
+    component.selectionModel.ownerFilter.set(OwnerFilterType.SHARED_BY_ME)
+    component.selectionModel.userID.set(99)
    expect(component.sharedByFilterLabel).toEqual('Shared by another user')
  })

  it('should retain relative labels for filters bound to the current user', () => {
-    component.selectionModel.userID = currentUserID
+    component.selectionModel.userID.set(currentUserID)
    expect(component.ownerFilterLabel).toEqual('My documents')
    expect(component.sharedByFilterLabel).toEqual('Shared by me')

-    component.selectionModel.excludeUsers = [currentUserID]
+    component.selectionModel.excludeUsers.set([currentUserID])
    expect(component.ownerExclusionFilterLabel).toEqual('Shared with me')
  })

  it('should retain relative labels for inactive filter choices', () => {
-    component.selectionModel.ownerFilter = OwnerFilterType.NONE
+    component.selectionModel.ownerFilter.set(OwnerFilterType.NONE)

    expect(component.ownerFilterLabel).toEqual('My documents')
    expect(component.ownerExclusionFilterLabel).toEqual('Shared with me')
@@ -148,32 +148,41 @@ describe('PermissionsFilterDropdownComponent', () => {

  it('should support reset', () => {
    component.setFilter(OwnerFilterType.OTHERS)
-    expect(component.selectionModel.ownerFilter).not.toEqual(
+    expect(component.selectionModel.ownerFilter()).not.toEqual(
      OwnerFilterType.NONE
    )
    component.reset()
-    expect(component.selectionModel.ownerFilter).toEqual(OwnerFilterType.NONE)
+    expect(component.selectionModel.ownerFilter()).toEqual(OwnerFilterType.NONE)
  })

  it('should toggle owner filter type when users selected', () => {
-    component.selectionModel.ownerFilter = OwnerFilterType.NONE
+    component.selectionModel.ownerFilter.set(OwnerFilterType.NONE)

    // this would normally be done by select component
-    component.selectionModel.includeUsers = [12]
+    component.selectionModel.includeUsers.set([12])
    component.onUserSelect()
-    expect(component.selectionModel.ownerFilter).toEqual(OwnerFilterType.OTHERS)
+    expect(component.selectionModel.ownerFilter()).toEqual(
+      OwnerFilterType.OTHERS
+    )

    // this would normally be done by select component
-    component.selectionModel.includeUsers = null
+    component.selectionModel.includeUsers.set(null)
    component.onUserSelect()

-    expect(component.selectionModel.ownerFilter).toEqual(OwnerFilterType.NONE)
+    expect(component.selectionModel.ownerFilter()).toEqual(OwnerFilterType.NONE)
  })
  it('should emit a selection model depending on the type of owner filter set', () => {
-    component.selectionModel.ownerFilter = OwnerFilterType.NONE
+    const emitted = () => ({
+      excludeUsers: ownerFilterSetResult.excludeUsers(),
+      hideUnowned: ownerFilterSetResult.hideUnowned(),
+      includeUsers: ownerFilterSetResult.includeUsers(),
+      ownerFilter: ownerFilterSetResult.ownerFilter(),
+      userID: ownerFilterSetResult.userID(),
+    })
+    component.selectionModel.ownerFilter.set(OwnerFilterType.NONE)

    component.setFilter(OwnerFilterType.SELF)
-    expect(ownerFilterSetResult).toEqual({
+    expect(emitted()).toEqual({
      excludeUsers: [],
      hideUnowned: false,
      includeUsers: [],
@@ -182,7 +191,7 @@ describe('PermissionsFilterDropdownComponent', () => {
    })

    component.setFilter(OwnerFilterType.NOT_SELF)
-    expect(ownerFilterSetResult).toEqual({
+    expect(emitted()).toEqual({
      excludeUsers: [currentUserID],
      hideUnowned: false,
      includeUsers: [],
@@ -191,7 +200,7 @@ describe('PermissionsFilterDropdownComponent', () => {
    })

    component.setFilter(OwnerFilterType.NONE)
-    expect(ownerFilterSetResult).toEqual({
+    expect(emitted()).toEqual({
      excludeUsers: [],
      hideUnowned: false,
      includeUsers: [],
@@ -200,7 +209,7 @@ describe('PermissionsFilterDropdownComponent', () => {
    })

    component.setFilter(OwnerFilterType.SHARED_BY_ME)
-    expect(ownerFilterSetResult).toEqual({
+    expect(emitted()).toEqual({
      excludeUsers: [],
      hideUnowned: false,
      includeUsers: [],
@@ -209,7 +218,7 @@ describe('PermissionsFilterDropdownComponent', () => {
    })

    component.setFilter(OwnerFilterType.UNOWNED)
-    expect(ownerFilterSetResult).toEqual({
+    expect(emitted()).toEqual({
      excludeUsers: [],
      hideUnowned: false,
      includeUsers: [],
@@ -25,18 +25,18 @@ import { ComponentWithPermissions } from '../../with-permissions/with-permission
 import { ClearableBadgeComponent } from '../clearable-badge/clearable-badge.component'

 export class PermissionsSelectionModel {
-  ownerFilter: OwnerFilterType
-  hideUnowned: boolean
-  userID: number
-  includeUsers: number[]
-  excludeUsers: number[]
+  readonly ownerFilter = signal(OwnerFilterType.NONE)
+  readonly hideUnowned = signal(false)
+  readonly userID = signal<number>(null)
+  readonly includeUsers = signal<number[]>([])
+  readonly excludeUsers = signal<number[]>([])

  clear() {
-    this.ownerFilter = OwnerFilterType.NONE
-    this.userID = null
-    this.hideUnowned = false
-    this.includeUsers = []
-    this.excludeUsers = []
+    this.ownerFilter.set(OwnerFilterType.NONE)
+    this.userID.set(null)
+    this.hideUnowned.set(false)
+    this.includeUsers.set([])
+    this.excludeUsers.set([])
  }
 }

@@ -84,33 +84,31 @@ export class PermissionsFilterDropdownComponent extends ComponentWithPermissions

  readonly users = signal<User[]>([])

-  hideUnowned: boolean
-
  get isActive(): boolean {
    return (
-      this.selectionModel.ownerFilter !== OwnerFilterType.NONE ||
-      this.selectionModel.hideUnowned
+      this.selectionModel.ownerFilter() !== OwnerFilterType.NONE ||
+      this.selectionModel.hideUnowned()
    )
  }

  get ownerFilterLabel(): string {
    if (
-      this.selectionModel?.ownerFilter !== OwnerFilterType.SELF ||
-      this.selectionModel?.userID === this.settingsService.currentUser()?.id
+      this.selectionModel?.ownerFilter() !== OwnerFilterType.SELF ||
+      this.selectionModel?.userID() === this.settingsService.currentUser()?.id
    ) {
      return $localize`My documents`
    }

-    const username = this.getUsername(this.selectionModel?.userID)
+    const username = this.getUsername(this.selectionModel?.userID())
    return username
      ? $localize`Owned by ${username}`
      : $localize`Owned by another user`
  }

  get ownerExclusionFilterLabel(): string {
-    const excludedUsers = this.selectionModel?.excludeUsers ?? []
+    const excludedUsers = this.selectionModel?.excludeUsers() ?? []
    if (
-      this.selectionModel?.ownerFilter !== OwnerFilterType.NOT_SELF ||
+      this.selectionModel?.ownerFilter() !== OwnerFilterType.NOT_SELF ||
      (excludedUsers.length === 1 &&
        excludedUsers[0] === this.settingsService.currentUser()?.id)
    ) {
@@ -130,13 +128,13 @@ export class PermissionsFilterDropdownComponent extends ComponentWithPermissions

  get sharedByFilterLabel(): string {
    if (
-      this.selectionModel?.ownerFilter !== OwnerFilterType.SHARED_BY_ME ||
-      this.selectionModel?.userID === this.settingsService.currentUser()?.id
+      this.selectionModel?.ownerFilter() !== OwnerFilterType.SHARED_BY_ME ||
+      this.selectionModel?.userID() === this.settingsService.currentUser()?.id
    ) {
      return $localize`Shared by me`
    }

-    const username = this.getUsername(this.selectionModel?.userID)
+    const username = this.getUsername(this.selectionModel?.userID())
    return username
      ? $localize`Shared by ${username}`
      : $localize`Shared by another user`
@@ -169,34 +167,36 @@ export class PermissionsFilterDropdownComponent extends ComponentWithPermissions
  }

  setFilter(type: OwnerFilterType) {
-    this.selectionModel.ownerFilter = type
-    if (this.selectionModel.ownerFilter === OwnerFilterType.SELF) {
-      this.selectionModel.includeUsers = []
-      this.selectionModel.excludeUsers = []
-      this.selectionModel.userID = this.settingsService.currentUser().id
-      this.selectionModel.hideUnowned = false
-    } else if (this.selectionModel.ownerFilter === OwnerFilterType.NOT_SELF) {
-      this.selectionModel.userID = null
-      this.selectionModel.includeUsers = []
-      this.selectionModel.excludeUsers = [this.settingsService.currentUser().id]
-      this.selectionModel.hideUnowned = false
-    } else if (this.selectionModel.ownerFilter === OwnerFilterType.NONE) {
-      this.selectionModel.userID = null
-      this.selectionModel.includeUsers = []
-      this.selectionModel.excludeUsers = []
-      this.selectionModel.hideUnowned = false
+    this.selectionModel.ownerFilter.set(type)
+    if (this.selectionModel.ownerFilter() === OwnerFilterType.SELF) {
+      this.selectionModel.includeUsers.set([])
+      this.selectionModel.excludeUsers.set([])
+      this.selectionModel.userID.set(this.settingsService.currentUser().id)
+      this.selectionModel.hideUnowned.set(false)
+    } else if (this.selectionModel.ownerFilter() === OwnerFilterType.NOT_SELF) {
+      this.selectionModel.userID.set(null)
+      this.selectionModel.includeUsers.set([])
+      this.selectionModel.excludeUsers.set([
+        this.settingsService.currentUser().id,
+      ])
+      this.selectionModel.hideUnowned.set(false)
+    } else if (this.selectionModel.ownerFilter() === OwnerFilterType.NONE) {
+      this.selectionModel.userID.set(null)
+      this.selectionModel.includeUsers.set([])
+      this.selectionModel.excludeUsers.set([])
+      this.selectionModel.hideUnowned.set(false)
    } else if (
-      this.selectionModel.ownerFilter === OwnerFilterType.SHARED_BY_ME
+      this.selectionModel.ownerFilter() === OwnerFilterType.SHARED_BY_ME
    ) {
-      this.selectionModel.userID = this.settingsService.currentUser()?.id
-      this.selectionModel.includeUsers = []
-      this.selectionModel.excludeUsers = []
-      this.selectionModel.hideUnowned = false
-    } else if (this.selectionModel.ownerFilter === OwnerFilterType.UNOWNED) {
-      this.selectionModel.userID = null
-      this.selectionModel.includeUsers = []
-      this.selectionModel.excludeUsers = []
-      this.selectionModel.hideUnowned = false
+      this.selectionModel.userID.set(this.settingsService.currentUser()?.id)
+      this.selectionModel.includeUsers.set([])
+      this.selectionModel.excludeUsers.set([])
+      this.selectionModel.hideUnowned.set(false)
+    } else if (this.selectionModel.ownerFilter() === OwnerFilterType.UNOWNED) {
+      this.selectionModel.userID.set(null)
+      this.selectionModel.includeUsers.set([])
+      this.selectionModel.excludeUsers.set([])
+      this.selectionModel.hideUnowned.set(false)
    }
    this.onChange()
  }
@@ -206,11 +206,11 @@ export class PermissionsFilterDropdownComponent extends ComponentWithPermissions
  }

  onUserSelect() {
-    if (this.selectionModel.includeUsers?.length) {
-      this.selectionModel.ownerFilter = OwnerFilterType.OTHERS
-    } else {
-      this.selectionModel.ownerFilter = OwnerFilterType.NONE
-    }
+    this.selectionModel.ownerFilter.set(
+      this.selectionModel.includeUsers()?.length
+        ? OwnerFilterType.OTHERS
+        : OwnerFilterType.NONE
+    )
    this.onChange()
  }

@@ -1209,24 +1209,53 @@ describe('DocumentDetailComponent', () => {
    expect(fixture.debugElement.queryAll(By.css('textarea.rtl'))).not.toBeNull()
  })

-  it('should display built-in pdf viewer if not disabled', () => {
+  it('should display built-in pdf viewer if not disabled', async () => {
    initNormally()
-    component.document().archived_file_name = 'file.pdf'
+    component.document.update((document) => ({
+      ...document,
+      archived_file_name: 'file.pdf',
+    }))
    settingsService.set(SETTINGS_KEYS.USE_NATIVE_PDF_VIEWER, false)
    expect(component.useNativePdfViewer).toBeFalsy()
-    fixture.detectChanges()
+    await fixture.whenStable()
    expect(fixture.debugElement.query(By.css('pngx-pdf-viewer'))).not.toBeNull()
  })

  it('should display native pdf viewer if enabled', () => {
    initNormally()
-    component.document().archived_file_name = 'file.pdf'
+    component.document.update((document) => ({
+      ...document,
+      archived_file_name: 'file.pdf',
+    }))
    settingsService.set(SETTINGS_KEYS.USE_NATIVE_PDF_VIEWER, true)
    expect(component.useNativePdfViewer).toBeTruthy()
    fixture.detectChanges()
    expect(fixture.debugElement.query(By.css('object'))).not.toBeNull()
  })

+  it('should reflect signal-backed document detail display settings', () => {
+    settingsService.set(SETTINGS_KEYS.DOCUMENT_EDITING_OVERLAY_THUMBNAIL, false)
+    settingsService.set(SETTINGS_KEYS.DOCUMENT_DETAILS_HIDDEN_FIELDS, [
+      component.DocumentDetailFieldID.Correspondent,
+    ])
+
+    expect(component.showThumbnailOverlay).toBeFalsy()
+    expect(
+      component.isFieldHidden(component.DocumentDetailFieldID.Correspondent)
+    ).toBeTruthy()
+    expect(
+      component.isFieldHidden(component.DocumentDetailFieldID.DocumentType)
+    ).toBeFalsy()
+
+    settingsService.set(SETTINGS_KEYS.DOCUMENT_EDITING_OVERLAY_THUMBNAIL, true)
+    settingsService.set(SETTINGS_KEYS.DOCUMENT_DETAILS_HIDDEN_FIELDS, [])
+
+    expect(component.showThumbnailOverlay).toBeTruthy()
+    expect(
+      component.isFieldHidden(component.DocumentDetailFieldID.Correspondent)
+    ).toBeFalsy()
+  })
+
  it('should attempt to retrieve metadata', () => {
    const metadataSpy = jest.spyOn(documentService, 'getMetadata')
    metadataSpy.mockReturnValue(of({ has_archive_version: true }))
@@ -1685,7 +1714,10 @@ describe('DocumentDetailComponent', () => {

  it('should change preview element by render type', () => {
    initNormally()
-    component.document().archived_file_name = 'file.pdf'
+    component.document.update((document) => ({
+      ...document,
+      archived_file_name: 'file.pdf',
+    }))
    fixture.detectChanges()
    expect(component.archiveContentRenderType).toEqual(
      component.ContentRenderType.PDF
@@ -1694,8 +1726,11 @@ describe('DocumentDetailComponent', () => {
      fixture.debugElement.query(By.css('pdf-viewer-container'))
    ).not.toBeUndefined()

-    component.document().archived_file_name = undefined
-    component.document().mime_type = 'text/plain'
+    component.document.update((document) => ({
+      ...document,
+      archived_file_name: undefined,
+      mime_type: 'text/plain',
+    }))
    fixture.detectChanges()
    expect(component.archiveContentRenderType).toEqual(
      component.ContentRenderType.Text
@@ -1704,7 +1739,10 @@ describe('DocumentDetailComponent', () => {
      fixture.debugElement.query(By.css('div.preview-sticky'))
    ).not.toBeUndefined()

-    component.document().mime_type = 'image/jpeg'
+    component.document.update((document) => ({
+      ...document,
+      mime_type: 'image/jpeg',
+    }))
    fixture.detectChanges()
    expect(component.archiveContentRenderType).toEqual(
      component.ContentRenderType.Image
@@ -1712,9 +1750,12 @@ describe('DocumentDetailComponent', () => {
    expect(
      fixture.debugElement.query(By.css('.preview-sticky img'))
    ).not.toBeUndefined()
-    ;((component.document().mime_type =
-      'application/vnd.openxmlformats-officedocument.wordprocessingml.document'),
-      fixture.detectChanges())
+    component.document.update((document) => ({
+      ...document,
+      mime_type:
+        'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+    }))
+    fixture.detectChanges()
    expect(component.archiveContentRenderType).toEqual(
      component.ContentRenderType.Other
    )
@@ -227,6 +227,19 @@ export class DocumentDetailComponent
  private deviceDetectorService = inject(DeviceDetectorService)
  private savedViewService = inject(SavedViewService)
  private readonly websocketStatusService = inject(WebsocketStatusService)
+  private readonly useNativePdfViewerSetting = this.settings.getSignal<boolean>(
+    SETTINGS_KEYS.USE_NATIVE_PDF_VIEWER
+  )
+  private readonly aiEnabledSetting = this.settings.getSignal<boolean>(
+    SETTINGS_KEYS.AI_ENABLED
+  )
+  private readonly showThumbnailOverlaySetting =
+    this.settings.getSignal<boolean>(
+      SETTINGS_KEYS.DOCUMENT_EDITING_OVERLAY_THUMBNAIL
+    )
+  private readonly hiddenFieldsSetting = this.settings.getSignal<
+    DocumentDetailFieldID[]
+  >(SETTINGS_KEYS.DOCUMENT_DETAILS_HIDDEN_FIELDS)

  @ViewChild('inputTitle')
  titleInput: TextComponent
@@ -333,8 +346,7 @@ export class DocumentDetailComponent
  }

  get useNativePdfViewer(): boolean {
-    this.settings.trackChanges()
-    return this.settings.get(SETTINGS_KEYS.USE_NATIVE_PDF_VIEWER)
+    return this.useNativePdfViewerSetting()
  }

  get isMobile(): boolean {
@@ -342,12 +354,10 @@ export class DocumentDetailComponent
  }

  get aiEnabled(): boolean {
-    this.settings.trackChanges()
-    return this.settings.get(SETTINGS_KEYS.AI_ENABLED)
+    return this.aiEnabledSetting()
  }

  get archiveContentRenderType(): ContentRenderType {
-    this.settings.trackChanges()
    const hasArchiveVersion =
      this.metadata()?.has_archive_version ??
      !!this.document()?.archived_file_name
@@ -359,22 +369,17 @@ export class DocumentDetailComponent
  }

  get originalContentRenderType(): ContentRenderType {
-    this.settings.trackChanges()
    return this.getRenderType(
      this.metadata()?.original_mime_type || this.document()?.mime_type
    )
  }

  get showThumbnailOverlay(): boolean {
-    this.settings.trackChanges()
-    return this.settings.get(SETTINGS_KEYS.DOCUMENT_EDITING_OVERLAY_THUMBNAIL)
+    return this.showThumbnailOverlaySetting()
  }

  isFieldHidden(fieldId: DocumentDetailFieldID): boolean {
-    this.settings.trackChanges()
-    return this.settings
-      .get(SETTINGS_KEYS.DOCUMENT_DETAILS_HIDDEN_FIELDS)
-      .includes(fieldId)
+    return this.hiddenFieldsSetting().includes(fieldId)
  }

  private getRenderType(mimeType: string): ContentRenderType {
@@ -121,6 +121,8 @@ export class DocumentListComponent
  settingsService = inject(SettingsService)
  private hotKeyService = inject(HotKeyService)
  permissionService = inject(PermissionsService)
+  private readonly notesEnabledSetting =
+    this.settingsService.getSignal<boolean>(SETTINGS_KEYS.NOTES_ENABLED)

  DisplayField = DisplayField
  DisplayMode = DisplayMode
@@ -574,8 +576,7 @@ export class DocumentListComponent
  }

  get notesEnabled(): boolean {
-    this.settingsService.trackChanges()
-    return this.settingsService.get(SETTINGS_KEYS.NOTES_ENABLED)
+    return this.notesEnabledSetting()
  }

  resetFilters() {
@@ -621,6 +621,43 @@ describe('FilterEditorComponent', () => {
    component.toggleTag(2) // coverage
  })

+  it('should reflect ingested tag filter rules in the dropdown toggle', () => {
+    const dropdown = fixture.debugElement.query(
+      By.css('pngx-filterable-dropdown')
+    )
+    const toggle = dropdown.nativeElement.querySelector('#dropdown_tags')
+    expect(toggle.classList.contains('btn-primary')).toBeFalsy()
+    expect(
+      dropdown.nativeElement.querySelector('pngx-clearable-badge')
+    ).toBeNull()
+
+    // switching to a view with a tag filter
+    component.filterRules = [
+      {
+        rule_type: FILTER_HAS_TAGS_ALL,
+        value: '2',
+      },
+    ]
+    fixture.detectChanges()
+    expect(toggle.classList.contains('btn-primary')).toBeTruthy()
+    expect(
+      dropdown.nativeElement.querySelector('pngx-clearable-badge')
+    ).not.toBeNull()
+
+    // and back to a view without one
+    component.filterRules = [
+      {
+        rule_type: FILTER_HAS_CORRESPONDENT_ANY,
+        value: '12',
+      },
+    ]
+    fixture.detectChanges()
+    expect(toggle.classList.contains('btn-primary')).toBeFalsy()
+    expect(
+      dropdown.nativeElement.querySelector('pngx-clearable-badge')
+    ).toBeNull()
+  })
+
  it('should ingest filter rules for has any tags', () => {
    expect(component.tagSelectionModel.getSelectedItems()).toHaveLength(0)
    component.filterRules = [
@@ -1078,7 +1115,7 @@ describe('FilterEditorComponent', () => {
  })

  it('should ingest filter rules for owner', () => {
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.NONE
    )
    component.filterRules = [
@@ -1087,15 +1124,38 @@ describe('FilterEditorComponent', () => {
        value: '100',
      },
    ]
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.SELF
    )
-    expect(component.permissionsSelectionModel.hideUnowned).toBeFalsy()
-    expect(component.permissionsSelectionModel.userID).toEqual(100)
+    expect(component.permissionsSelectionModel.hideUnowned()).toBeFalsy()
+    expect(component.permissionsSelectionModel.userID()).toEqual(100)
+  })
+
+  it('should reflect ingested owner filter rules in the dropdown toggle', () => {
+    const dropdown = fixture.debugElement.query(
+      By.css('pngx-permissions-filter-dropdown')
+    )
+    const toggle = dropdown.nativeElement.querySelector('button')
+    expect(toggle.classList.contains('btn-primary')).toBeFalsy()
+
+    // switching to a view with an owner filter
+    component.filterRules = [
+      {
+        rule_type: FILTER_OWNER,
+        value: '100',
+      },
+    ]
+    fixture.detectChanges()
+    expect(toggle.classList.contains('btn-primary')).toBeTruthy()
+
+    // and back to a view without one
+    component.filterRules = []
+    fixture.detectChanges()
+    expect(toggle.classList.contains('btn-primary')).toBeFalsy()
  })

  it('should ingest filter rules for owner is others', () => {
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.NONE
    )
    component.filterRules = [
@@ -1104,14 +1164,14 @@ describe('FilterEditorComponent', () => {
        value: '50',
      },
    ]
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.OTHERS
    )
-    expect(component.permissionsSelectionModel.includeUsers).toContain(50)
+    expect(component.permissionsSelectionModel.includeUsers()).toContain(50)
  })

  it('should ingest filter rules for owner does not include others', () => {
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.NONE
    )
    component.filterRules = [
@@ -1120,14 +1180,14 @@ describe('FilterEditorComponent', () => {
        value: '50',
      },
    ]
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.NOT_SELF
    )
-    expect(component.permissionsSelectionModel.excludeUsers).toContain(50)
+    expect(component.permissionsSelectionModel.excludeUsers()).toContain(50)
  })

  it('should ingest filter rules for owner is null', () => {
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.NONE
    )
    component.filterRules = [
@@ -1136,10 +1196,10 @@ describe('FilterEditorComponent', () => {
        value: 'true',
      },
    ]
-    expect(component.permissionsSelectionModel.ownerFilter).toEqual(
+    expect(component.permissionsSelectionModel.ownerFilter()).toEqual(
      OwnerFilterType.UNOWNED
    )
-    expect(component.permissionsSelectionModel.hideUnowned).toBeFalsy()
+    expect(component.permissionsSelectionModel.hideUnowned()).toBeFalsy()
  })

  it('should ingest filter rules for owner is not null', () => {
@@ -1149,14 +1209,14 @@ describe('FilterEditorComponent', () => {
        value: 'false',
      },
    ]
-    expect(component.permissionsSelectionModel.hideUnowned).toBeTruthy()
+    expect(component.permissionsSelectionModel.hideUnowned()).toBeTruthy()
    component.filterRules = [
      {
        rule_type: FILTER_OWNER_ISNULL,
        value: '0',
      },
    ]
-    expect(component.permissionsSelectionModel.hideUnowned).toBeTruthy()
+    expect(component.permissionsSelectionModel.hideUnowned()).toBeTruthy()
  })

  it('should ingest filter rules for shared by me', () => {
@@ -1166,7 +1226,7 @@ describe('FilterEditorComponent', () => {
        value: '2',
      },
    ]
-    expect(component.permissionsSelectionModel.userID).toEqual(2)
+    expect(component.permissionsSelectionModel.userID()).toEqual(2)
  })

  // GET filterRules
@@ -1932,7 +1992,10 @@ describe('FilterEditorComponent', () => {
        value: '1',
      },
    ])
-    component.permissionsSelectionModel.excludeUsers.push(2)
+    component.permissionsSelectionModel.excludeUsers.update((users) => [
+      ...users,
+      2,
+    ])
    fixture.detectChanges()
    expect(component.filterRules).toEqual([
      {
@@ -1982,8 +2045,11 @@ describe('FilterEditorComponent', () => {
    // TODO: mock input in code
    // userSelect.query(By.css('input')).nativeElement.value = '3'
    // userSelect.triggerEventHandler('change')
-    component.permissionsSelectionModel.ownerFilter = OwnerFilterType.OTHERS
-    component.permissionsSelectionModel.includeUsers.push(3)
+    component.permissionsSelectionModel.ownerFilter.set(OwnerFilterType.OTHERS)
+    component.permissionsSelectionModel.includeUsers.update((users) => [
+      ...users,
+      3,
+    ])
    fixture.detectChanges()
    expect(component.filterRules).toEqual([
      {
@@ -2003,7 +2069,7 @@ describe('FilterEditorComponent', () => {
    ownerToggle.nativeElement.checked = true
    // ownerToggle.triggerEventHandler('change')
    // TODO: ngModel isn't doing this here
-    component.permissionsSelectionModel.hideUnowned = true
+    component.permissionsSelectionModel.hideUnowned.set(true)
    fixture.detectChanges()
    expect(component.filterRules).toEqual([
      {
@@ -735,38 +735,50 @@ export class FilterEditorComponent
          this._textFilter = rule.value
          break
        case FILTER_OWNER:
-          this.permissionsSelectionModel.ownerFilter = OwnerFilterType.SELF
-          this.permissionsSelectionModel.hideUnowned = false
+          this.permissionsSelectionModel.ownerFilter.set(OwnerFilterType.SELF)
+          this.permissionsSelectionModel.hideUnowned.set(false)
          if (rule.value)
-            this.permissionsSelectionModel.userID = parseInt(rule.value, 10)
+            this.permissionsSelectionModel.userID.set(
+              Number.parseInt(rule.value, 10)
+            )
          break
        case FILTER_OWNER_ANY:
-          this.permissionsSelectionModel.ownerFilter = OwnerFilterType.OTHERS
+          this.permissionsSelectionModel.ownerFilter.set(OwnerFilterType.OTHERS)
          if (rule.value)
-            this.permissionsSelectionModel.includeUsers.push(
-              parseInt(rule.value, 10)
-            )
+            this.permissionsSelectionModel.includeUsers.update((users) => [
+              ...users,
+              Number.parseInt(rule.value, 10),
+            ])
          break
        case FILTER_OWNER_DOES_NOT_INCLUDE:
-          this.permissionsSelectionModel.ownerFilter = OwnerFilterType.NOT_SELF
+          this.permissionsSelectionModel.ownerFilter.set(
+            OwnerFilterType.NOT_SELF
+          )
          if (rule.value)
-            this.permissionsSelectionModel.excludeUsers.push(
-              parseInt(rule.value, 10)
-            )
+            this.permissionsSelectionModel.excludeUsers.update((users) => [
+              ...users,
+              Number.parseInt(rule.value, 10),
+            ])
          break
        case FILTER_SHARED_BY_USER:
-          this.permissionsSelectionModel.ownerFilter =
+          this.permissionsSelectionModel.ownerFilter.set(
            OwnerFilterType.SHARED_BY_ME
+          )
          if (rule.value)
-            this.permissionsSelectionModel.userID = parseInt(rule.value, 10)
+            this.permissionsSelectionModel.userID.set(
+              Number.parseInt(rule.value, 10)
+            )
          break
        case FILTER_OWNER_ISNULL:
          if (rule.value === 'true' || rule.value === '1') {
-            this.permissionsSelectionModel.hideUnowned = false
-            this.permissionsSelectionModel.ownerFilter = OwnerFilterType.UNOWNED
+            this.permissionsSelectionModel.hideUnowned.set(false)
+            this.permissionsSelectionModel.ownerFilter.set(
+              OwnerFilterType.UNOWNED
+            )
          } else {
-            this.permissionsSelectionModel.hideUnowned =
+            this.permissionsSelectionModel.hideUnowned.set(
              rule.value === 'false' || rule.value === '0'
+            )
            break
          }
      }
@@ -1074,34 +1086,35 @@ export class FilterEditorComponent
        })
      }
    }
-    if (this.permissionsSelectionModel.ownerFilter == OwnerFilterType.SELF) {
+    if (this.permissionsSelectionModel.ownerFilter() == OwnerFilterType.SELF) {
      filterRules.push({
        rule_type: FILTER_OWNER,
-        value: this.permissionsSelectionModel.userID.toString(),
+        value: this.permissionsSelectionModel.userID().toString(),
      })
    } else if (
-      this.permissionsSelectionModel.ownerFilter == OwnerFilterType.NOT_SELF
+      this.permissionsSelectionModel.ownerFilter() == OwnerFilterType.NOT_SELF
    ) {
      filterRules.push({
        rule_type: FILTER_OWNER_DOES_NOT_INCLUDE,
-        value: this.permissionsSelectionModel.excludeUsers?.join(','),
+        value: this.permissionsSelectionModel.excludeUsers()?.join(','),
      })
    } else if (
-      this.permissionsSelectionModel.ownerFilter == OwnerFilterType.OTHERS
+      this.permissionsSelectionModel.ownerFilter() == OwnerFilterType.OTHERS
    ) {
      filterRules.push({
        rule_type: FILTER_OWNER_ANY,
-        value: this.permissionsSelectionModel.includeUsers?.join(','),
+        value: this.permissionsSelectionModel.includeUsers()?.join(','),
      })
    } else if (
-      this.permissionsSelectionModel.ownerFilter == OwnerFilterType.SHARED_BY_ME
+      this.permissionsSelectionModel.ownerFilter() ==
+      OwnerFilterType.SHARED_BY_ME
    ) {
      filterRules.push({
        rule_type: FILTER_SHARED_BY_USER,
-        value: this.permissionsSelectionModel.userID.toString(),
+        value: this.permissionsSelectionModel.userID().toString(),
      })
    } else if (
-      this.permissionsSelectionModel.ownerFilter == OwnerFilterType.UNOWNED
+      this.permissionsSelectionModel.ownerFilter() == OwnerFilterType.UNOWNED
    ) {
      filterRules.push({
        rule_type: FILTER_OWNER_ISNULL,
@@ -1109,7 +1122,7 @@ export class FilterEditorComponent
      })
    }

-    if (this.permissionsSelectionModel.hideUnowned) {
+    if (this.permissionsSelectionModel.hideUnowned()) {
      filterRules.push({
        rule_type: FILTER_OWNER_ISNULL,
        value: 'false',
@@ -210,6 +210,48 @@ describe('SettingsService', () => {
    expect(settingsService.get(SETTINGS_KEYS.THEME_COLOR)).toEqual('#000000')
  })

+  it('provides stable signals that update when settings change', () => {
+    const req = httpTestingController.expectOne(
+      `${environment.apiBaseUrl}ui_settings/`
+    )
+    req.flush(ui_settings)
+
+    const notesEnabled = settingsService.getSignal<boolean>(
+      SETTINGS_KEYS.NOTES_ENABLED
+    )
+
+    expect(notesEnabled()).toBeTruthy()
+    expect(
+      settingsService.getSignal<boolean>(SETTINGS_KEYS.NOTES_ENABLED)
+    ).toBe(notesEnabled)
+
+    settingsService.set(SETTINGS_KEYS.NOTES_ENABLED, false)
+
+    expect(notesEnabled()).toBeFalsy()
+  })
+
+  it('updates setting signals when settings are reinitialized', () => {
+    let req = httpTestingController.expectOne(
+      `${environment.apiBaseUrl}ui_settings/`
+    )
+    req.flush(ui_settings)
+    const appTitle = settingsService.getSignal<string>(SETTINGS_KEYS.APP_TITLE)
+
+    settingsService.initializeSettings().subscribe()
+    req = httpTestingController.expectOne(
+      `${environment.apiBaseUrl}ui_settings/`
+    )
+    req.flush({
+      ...ui_settings,
+      settings: {
+        ...ui_settings.settings,
+        app_title: 'Updated title',
+      },
+    })
+
+    expect(appTitle()).toBe('Updated title')
+  })
+
  it('sets django cookie for languages', () => {
    httpTestingController
      .expectOne(`${environment.apiBaseUrl}ui_settings/`)
@@ -2,6 +2,8 @@ import { HttpClient } from '@angular/common/http'
 import {
  DOCUMENT,
  EventEmitter,
+  Signal,
+  computed,
  inject,
  Injectable,
  LOCALE_ID,
@@ -297,6 +299,7 @@ export class SettingsService {

  private settings: Record<string, any> = {}
  private readonly settingsVersion = signal(0)
+  private readonly settingSignals = new Map<string, Signal<unknown>>()
  readonly currentUser = signal<User>(undefined)

  public settingsSaved: EventEmitter<any> = new EventEmitter()
@@ -326,10 +329,6 @@ export class SettingsService {
    return !UNSAFE_OBJECT_KEYS.has(key)
  }

-  public trackChanges(): void {
-    this.settingsVersion()
-  }
-
  private assignSafeSettings(source: Record<string, any>) {
    if (!source || typeof source !== 'object' || Array.isArray(source)) {
      return
@@ -339,6 +338,7 @@ export class SettingsService {
      if (!this.isSafeObjectKey(key)) continue
      this.settings[key] = source[key]
    }
+    this.settingsVersion.update((version) => version + 1)
  }

  // this is called by the app initializer in app.module
@@ -594,6 +594,18 @@ export class SettingsService {
    }
  }

+  getSignal<T = any>(key: string): Signal<T> {
+    let settingSignal = this.settingSignals.get(key)
+    if (!settingSignal) {
+      settingSignal = computed(() => {
+        this.settingsVersion()
+        return this.get(key)
+      })
+      this.settingSignals.set(key, settingSignal)
+    }
+    return settingSignal as Signal<T>
+  }
+
  set(key: string, value: any) {
    // parse key:key:key into nested object
    let settingObj = this.settings
@@ -507,8 +507,8 @@ def rotate(
            logger.info(
                f"Queued new rotated version for document {pair.root_doc.id} by {degrees} degrees",
            )
-        except Exception:
-            logger.exception(f"Error rotating document {pair.root_doc.id}")
+        except Exception as e:
+            logger.exception(f"Error rotating document {pair.root_doc.id}: {e}")

    return "OK"

@@ -554,9 +554,9 @@ def merge(
            affected_docs.append(doc.id)
            if handoff_asn is None and doc.archive_serial_number is not None:
                handoff_asn = doc.archive_serial_number
-        except Exception:
+        except Exception as e:
            logger.exception(
-                f"Error merging document {doc.id}, it will not be included in the merge",
+                f"Error merging document {doc.id}, it will not be included in the merge: {e}",
            )
    if len(affected_docs) == 0:
        logger.warning("No documents were merged")
@@ -805,8 +805,8 @@ def split(
            else:
                group(consume_tasks).delay()

-    except Exception:
-        logger.exception(f"Error splitting document {doc.id}")
+    except Exception as e:
+        logger.exception(f"Error splitting document {doc.id}: {e}")

    return "OK"

@@ -858,8 +858,8 @@ def delete_pages(
        logger.info(
            f"Queued new version for document {pair.root_doc.id} after deleting pages {pages}",
        )
-    except Exception:
-        logger.exception(f"Error deleting pages from document {pair.root_doc.id}")
+    except Exception as e:
+        logger.exception(f"Error deleting pages from document {pair.root_doc.id}: {e}")

    return "OK"

@@ -986,7 +986,7 @@ def edit_pdf(
                group(consume_tasks).delay()

    except Exception as e:
-        logger.exception(f"Error editing document {pair.root_doc.id}")
+        logger.exception(f"Error editing document {pair.root_doc.id}: {e}")
        raise ValueError(
            f"An error occurred while editing the document: {e}",
        ) from e
@@ -1097,7 +1097,7 @@ def remove_password(

        except Exception as e:
            logger.exception(
-                f"Error removing password from document {pair.root_doc.id}",
+                f"Error removing password from document {pair.root_doc.id}: {e}",
            )
            raise ValueError(
                f"An error occurred while removing the password: {e}",
@@ -72,8 +72,8 @@ def load_classifier(*, raise_exception: bool = False) -> DocumentClassifier | No
        Path(settings.MODEL_FILE).unlink()
        classifier = None
        if raise_exception:
-            raise
-    except ClassifierModelCorruptError:
+            raise e
+    except ClassifierModelCorruptError as e:
        # there's something wrong with the model file.
        logger.exception(
            "Unrecoverable error while loading document "
@@ -82,17 +82,17 @@ def load_classifier(*, raise_exception: bool = False) -> DocumentClassifier | No
        Path(settings.MODEL_FILE).unlink()
        classifier = None
        if raise_exception:
-            raise
-    except OSError:
+            raise e
+    except OSError as e:
        logger.exception("IO error while loading document classification model")
        classifier = None
        if raise_exception:
-            raise
-    except Exception:  # pragma: no cover
+            raise e
+    except Exception as e:  # pragma: no cover
        logger.exception("Unknown error while loading document classification model")
        classifier = None
        if raise_exception:
-            raise
+            raise e

    return classifier

@@ -217,7 +217,7 @@ class ConsumerPluginMixin:
            current_progress,
            max_progress,
            document_id=document_id,
-            owner_id=self.metadata.owner_id or None,
+            owner_id=self.metadata.owner_id if self.metadata.owner_id else None,
            users_can_view=(self.metadata.view_users or [])
            + (self.metadata.change_users or []),
            groups_can_view=(self.metadata.view_groups or [])
@@ -675,7 +675,9 @@ class ConsumerPlugin(
                            document=document,
                            logging_group=self.logging_group,
                            classifier=classifier,
-                            original_file=self.unmodified_original or self.working_copy,
+                            original_file=self.unmodified_original
+                            if self.unmodified_original
+                            else self.working_copy,
                        )

                        # After everything is in the database, copy the files into
@@ -856,7 +858,7 @@ class ConsumerPlugin(
        else:
            stats = Path(self.input_doc.original_file).stat()
            create_date = timezone.make_aware(
-                datetime.datetime.fromtimestamp(stats.st_mtime),  # noqa: DTZ006 - make_aware() requires a naive datetime
+                datetime.datetime.fromtimestamp(stats.st_mtime),
            )
            self.log.debug(f"Creation date from st_mtime: {create_date}")

@@ -970,7 +972,7 @@ class ConsumerPlugin(
        try:
            copy_basic_file_stats(source, target)
        except Exception:  # pragma: no cover
-            self.log.debug("Unable to copy file stats from %s to %s", source, target)
+            pass


 class ConsumerPreflightPlugin(
@@ -78,9 +78,7 @@ class CollatePlugin(NoCleanupPluginMixin, NoSetupPluginMixin, ConsumeTaskPlugin)
            stats = staging.stat()
            # if the file is older than the timeout, we don't consider
            # it valid
-            if (
-                dt.datetime.now(tz=dt.UTC).timestamp() - stats.st_mtime
-            ) > TIMEOUT_SECONDS:
+            if (dt.datetime.now().timestamp() - stats.st_mtime) > TIMEOUT_SECONDS:
                logger.warning("Outdated double sided staging file exists, deleting it")
                staging.unlink()
            else:
@@ -136,7 +134,7 @@ class CollatePlugin(NoCleanupPluginMixin, NoSetupPluginMixin, ConsumeTaskPlugin)
            shutil.move(pdf_file, staging)
            # update access to modification time so we know if the file
            # is outdated when another file gets uploaded
-            timestamp = dt.datetime.now(tz=dt.UTC).timestamp()
+            timestamp = dt.datetime.now().timestamp()
            os.utime(staging, (timestamp, timestamp))
            logger.info(
                "Got scan with odd numbered pages of double-sided scan, moved it to %s",
@@ -734,7 +734,7 @@ class CustomFieldQueryParser:
        )

        # Check if any of the requested IDs are missing.
-        missing_ids = set(value) - {link.document_id for link in links}
+        missing_ids = set(value) - set(link.document_id for link in links)
        if missing_ids:
            # The result should be an empty set in this case.
            return Q(id__in=[])
@@ -631,25 +631,23 @@ class Command(BaseCommand):
                ):
                    # Process each change
                    for change_type, path in changes:
-                        resolved_path = Path(path).resolve()
+                        path = Path(path).resolve()
                        if change_type == Change.deleted:
                            # Consumed (or otherwise removed); a later file
                            # reusing this name must not be skipped as
                            # already-queued.
-                            queued.discard(resolved_path)
-                        if not resolved_path.is_file():
+                            queued.discard(path)
+                        if not path.is_file():
                            continue
-                        if resolved_path in queued:
+                        if path in queued:
                            # Already queued and awaiting consumption; a stray
                            # event (NAS metadata touch, AV scan, etc.) while
                            # the file sits on disk mid-consumption must not
                            # cause it to be queued a second time (GH #13511).
-                            logger.debug(
-                                f"Ignoring event for queued file: {resolved_path}",
-                            )
+                            logger.debug(f"Ignoring event for queued file: {path}")
                            continue
-                        logger.debug(f"Event: {change_type.name} for {resolved_path}")
-                        tracker.track(resolved_path, change_type)
+                        logger.debug(f"Event: {change_type.name} for {path}")
+                        tracker.track(path, change_type)

                    # Check for stable files
                    for stable_path in tracker.get_stable_files():
@@ -30,10 +30,6 @@ if TYPE_CHECKING:
 logger = logging.getLogger("paperless.matching")


-class UnsupportedWorkflowTriggerTypeError(Exception):
-    pass
-
-
 def log_reason(
    matching_model: MatchingModel | WorkflowTrigger,
    document: Document,
@@ -695,9 +691,7 @@ def document_matches_workflow(
                )
            else:
                # New trigger types need to be explicitly checked above
-                raise UnsupportedWorkflowTriggerTypeError(
-                    f"Trigger type {trigger_type} not yet supported",
-                )
+                raise Exception(f"Trigger type {trigger_type} not yet supported")

            if trigger_matched:
                logger.info(f"Document matched {trigger} from {workflow}")
@@ -75,7 +75,7 @@ def recompute_checksums(apps, schema_editor):
        if updated_fields:
            batch.append(doc)

-        processed += 1  # noqa: SIM113
+        processed += 1

        if len(batch) >= _BATCH_SIZE:
            Document.objects.bulk_update(batch, ["checksum", "archive_checksum"])
@@ -377,7 +377,7 @@ class Document(SoftDeleteModel, ModelWithOwner):  # type: ignore[django-manager-
        from documents.versioning import versions_newest_first

        if hasattr(self, "effective_content"):
-            return self.effective_content
+            return getattr(self, "effective_content")

        if self.root_document_id is not None or self.pk is None:
            return self.content
@@ -41,7 +41,7 @@ def get_default_file_extension(mime_type: str) -> str:
            return supported[mime_type]

    ext = mimetypes.guess_extension(mime_type)
-    return ext or ""
+    return ext if ext else ""


 def is_file_ext_supported(ext: str) -> bool:
@@ -110,7 +110,7 @@ def run_convert(
    args += ["-define", "pdf:use-cropbox=true"] if use_cropbox else []
    args += [str(input_file), str(output_file)]

-    logger.debug("Execute: %s", " ".join(args), extra={"group": logging_group})
+    logger.debug("Execute: " + " ".join(args), extra={"group": logging_group})

    try:
        run_subprocess(args, environment, logger)
@@ -43,8 +43,8 @@ def _discover_parser_class() -> type[DateParserPluginBase]:
                valid_plugins.append(ep)
            else:
                logger.warning(f"Plugin {ep.name} does not subclass DateParser.")
-        except Exception:
-            logger.exception(f"Unable to load date parser plugin {ep.name}")
+        except Exception as e:
+            logger.exception(f"Unable to load date parser plugin {ep.name}: {e}")

    if not valid_plugins:
        return RegexDateParserPlugin
@@ -91,8 +91,8 @@ class DateParserPluginBase(ABC):
                },
                locales=self.config.languages,
            )
-        except Exception:
-            logger.exception(f"Error while parsing date string '{date_string}'")
+        except Exception as e:
+            logger.exception(f"Error while parsing date string '{date_string}': {e}")
            return None

    def _filter_date(
@@ -59,10 +59,11 @@ def safe_regex_match(pattern: str, text: str, *, flags: int = 0):
    try:
        validate_regex_pattern(pattern)
        compiled = regex.compile(pattern, flags=flags)
-    except (regex.error, ValueError):
+    except (regex.error, ValueError) as exc:
        logger.exception(
-            "Error while processing regular expression %s",
+            "Error while processing regular expression %s: %s",
            textwrap.shorten(pattern, width=80, placeholder="…"),
+            exc,
        )
        return None

@@ -85,10 +86,11 @@ def safe_regex_sub(pattern: str, repl: str, text: str, *, flags: int = 0) -> str
    try:
        validate_regex_pattern(pattern)
        compiled = regex.compile(pattern, flags=flags)
-    except (regex.error, ValueError):
+    except (regex.error, ValueError) as exc:
        logger.exception(
-            "Error while processing regular expression %s",
+            "Error while processing regular expression %s: %s",
            textwrap.shorten(pattern, width=80, placeholder="…"),
+            exc,
        )
        return None

@@ -1142,7 +1142,7 @@ def get_backend() -> TantivyBackend:
    Returns:
        Thread-safe singleton TantivyBackend instance
    """
-    global _backend, _backend_path  # noqa: PLW0603 - module-level singleton, no class to hold this state
+    global _backend, _backend_path

    current_path: Path = settings.INDEX_DIR

@@ -1173,7 +1173,7 @@ def reset_backend() -> None:
    Forces creation of a new backend instance on the next get_backend() call.
    Used for test isolation and when switching between different index directories.
    """
-    global _backend, _backend_path  # noqa: PLW0603 - module-level singleton, no class to hold this state
+    global _backend, _backend_path

    with _backend_lock:
        if _backend is not None:
@@ -240,7 +240,7 @@ def parse_user_query(
            DEFAULT_SEARCH_FIELDS,
            field_boosts=_FIELD_BOOSTS,
            # (prefix=True, distance=1, transposition_cost_one=True) — edit-distance fuzziness
-            fuzzy_fields=dict.fromkeys(DEFAULT_SEARCH_FIELDS, (True, 1, True)),
+            fuzzy_fields={f: (True, 1, True) for f in DEFAULT_SEARCH_FIELDS},
        )
        # 0.1 boost keeps fuzzy hits ranked below exact matches (intentional)
        clauses.append((tantivy.Occur.Should, tantivy.Query.boost_query(fuzzy, 0.1)))
@@ -434,7 +434,7 @@ class OwnedObjectSerializer(
            return set()

        ctype = ContentType.objects.get_for_model(first_obj)
-        object_pks = [obj.pk for obj in objects]
+        object_pks = list(obj.pk for obj in objects)
        pk_type = type(first_obj.pk)

        def get_pks_for_permission_type(model):
@@ -730,7 +730,7 @@ class TagSerializer(MatchingModelSerializer, OwnedObjectSerializer):
                self.instance.clean()
            except ValidationError as e:
                logger.debug("Tag parent validation failed: %s", e)
-                raise
+                raise e
            finally:
                self.instance.tn_parent = original_parent
        else:
@@ -740,7 +740,7 @@ class TagSerializer(MatchingModelSerializer, OwnedObjectSerializer):
                temp.clean()
            except ValidationError as e:
                logger.debug("Tag parent validation failed: %s", e)
-                raise
+                raise e

        return super().validate(attrs)

@@ -1150,7 +1150,7 @@ class DocumentSerializer(
    def to_representation(self, instance):
        doc = super().to_representation(instance)
        if "content" in self.fields and hasattr(instance, "effective_content"):
-            doc["content"] = instance.effective_content or ""
+            doc["content"] = getattr(instance, "effective_content") or ""
        if self.truncate_content and "content" in self.fields:
            doc["content"] = doc.get("content")[0:550]
        return doc
@@ -1860,8 +1860,8 @@ class BulkEditSerializer(
        if isinstance(custom_fields, dict):
            try:
                ids = [int(i[0]) for i in custom_fields.items()]
-            except Exception:
-                logger.exception("Error validating custom fields")
+            except Exception as e:
+                logger.exception(f"Error validating custom fields: {e}")
                raise serializers.ValidationError(
                    f"{name} must be a list of integers or a dict of id:value pairs, see the log for details",
                )
@@ -2059,12 +2059,13 @@ class BulkEditSerializer(
            for doc in docs:
                if "-" in doc:
                    pages.append(
-                        list(
-                            range(
+                        [
+                            x
+                            for x in range(
                                int(doc.split("-")[0]),
                                int(doc.split("-")[1]) + 1,
-                            ),
-                        ),
+                            )
+                        ],
                    )
                else:
                    pages.append([int(doc)])
@@ -2925,7 +2926,7 @@ class ShareLinkBundleSerializer(OwnedObjectSerializer):
        return share_link_bundle

    def get_document_count(self, obj: ShareLinkBundle) -> int:
-        return obj.document_total or obj.documents.count()
+        return getattr(obj, "document_total") or obj.documents.count()


 class BulkEditObjectsSerializer(SerializerWithPerms, SetPermissionsMixin):
@@ -637,7 +637,7 @@ def update_filename_and_move_files(
                #  so this is not the end of the world.
                # B: if moving the original file failed, nothing has changed
                #  anyway.
-                logger.exception("Error reverting document changes")
+                pass

            # restore old values on the instance
            instance.filename = old_filename
@@ -1102,11 +1102,10 @@ def _extract_input_data(
                if v is None or k.startswith("_"):
                    continue
                if isinstance(v, datetime.date):
-                    override_dict[k] = v.isoformat()
+                    v = v.isoformat()
                elif isinstance(v, Path):
-                    override_dict[k] = str(v)
-                else:
-                    override_dict[k] = v
+                    v = str(v)
+                override_dict[k] = v
            if override_dict:
                data["overrides"] = override_dict
        return data
@@ -217,9 +217,9 @@ def consume_file(
                overrides.filename or input_doc.original_file.name,
                self.request.id,
            ) as status_mgr,
-            TemporaryDirectory(dir=settings.SCRATCH_DIR) as tmp_dir_name,
+            TemporaryDirectory(dir=settings.SCRATCH_DIR) as tmp_dir,
        ):
-            tmp_dir = Path(tmp_dir_name)
+            tmp_dir = Path(tmp_dir)
            msg = None
            for plugin_class in plugins:
                plugin_name = plugin_class.NAME
@@ -261,7 +261,7 @@ def consume_file(
                    )

                except Exception as e:
-                    logger.exception(f"{plugin_name} failed")
+                    logger.exception(f"{plugin_name} failed: {e}")
                    status_mgr.send_progress(
                        ProgressStatusOptions.FAILED,
                        f"{e}",
@@ -495,8 +495,8 @@ def empty_trash(doc_ids=None) -> None:
                content_type=ContentType.objects.get_for_model(Document),
                object_id__in=deleted_document_ids,
            ).delete()
-    except Exception:  # pragma: no cover
-        logger.exception("Error while emptying trash")
+    except Exception as e:  # pragma: no cover
+        logger.exception(f"Error while emptying trash: {e}")
    finally:
        models.signals.post_delete.disconnect(
            cleanup_document_deletion,
@@ -832,8 +832,9 @@ def build_share_link_bundle(bundle_id: int) -> None:
        logger.info("Built share link bundle %s", bundle.pk)
    except Exception as exc:
        logger.exception(
-            "Failed to build share link bundle %s",
+            "Failed to build share link bundle %s: %s",
            bundle_id,
+            exc,
        )
        bundle.status = ShareLinkBundle.Status.FAILED
        bundle.last_error = {
@@ -78,10 +78,6 @@ class PlaceholderString(str):
    def __ne__(self, other) -> bool:
        return not self.__eq__(other)

-    def __hash__(self) -> int:
-        # Equal to both "-none-" and "none", so hash to a single canonical value
-        return hash("-none-")
-

 NO_VALUE_PLACEHOLDER = PlaceholderString("-none-")

@@ -138,9 +138,9 @@ def parse_w_workflow_placeholders(

        # We're good!
        return rendered_template
-    except UndefinedError:
+    except UndefinedError as e:
        # The undefined class logs this already for us
-        raise
+        raise e
    except TemplateSyntaxError as e:
        logger.warning(f"Template syntax error in title generation: {e}")
    except SecurityError as e:
@@ -150,5 +150,5 @@ def parse_w_workflow_placeholders(
        logger.warning(
            f"Invalid title format '{text}', workflow not applied: {e}",
        )
-        raise
+        raise e
    return None
@@ -296,7 +296,7 @@ class TestRegexDateParser:

            # simulate parse failure for malformed input
            if "99/99/9999" in date_string or "bad date" in date_string:
-                raise Exception("parse failed for malformed date")  # noqa: TRY002 - simulates a generic parser failure
+                raise Exception("parse failed for malformed date")

            return None

@@ -57,13 +57,13 @@ class MultiprocessCommand(PaperlessCommand):

    def handle(self, *args, **options):
        items = list(range(5))
-        results = list(
-            self.process_parallel(
-                _double_value,
-                items,
-                description="Processing...",
-            ),
-        )
+        results = []
+        for result in self.process_parallel(
+            _double_value,
+            items,
+            description="Processing...",
+        ):
+            results.append(result)
        successes = sum(1 for r in results if r.success)
        self.stdout.write(f"Successes: {successes}")

@@ -66,7 +66,7 @@ class TestWriteBatchLockRetry:
        )
        mock_sleep = mocker.patch(
            "documents.search._backend.time.sleep",
-            side_effect=sleep_values.append,
+            side_effect=lambda s: sleep_values.append(s),
        )

        # Should not raise — 4th attempt succeeds
@@ -111,7 +111,7 @@ class TestWriteBatchLockRetry:
        sleep_values: list[float] = []
        mocker.patch(
            "documents.search._backend.time.sleep",
-            side_effect=sleep_values.append,
+            side_effect=lambda s: sleep_values.append(s),
        )
        for _ in range(50):
            sleep_values.clear()
@@ -1003,8 +1003,8 @@ class TestBulkEditAPI(DirectoriesMixin, APITestCase):
            for correspondent in response.data[field]:
                self.assertEqual(correspondent["document_count"], 0)
            self.assertCountEqual(
-                (c["id"] for c in response.data[field]),
-                (c["id"] for c in Entity.objects.values("id")),
+                map(lambda c: c["id"], response.data[field]),
+                map(lambda c: c["id"], Entity.objects.values("id")),
            )

    def test_api_selection_data(self) -> None:
@@ -1,11 +1,18 @@
 from __future__ import annotations

+from typing import TYPE_CHECKING
 from unittest import mock

+import pytest
+from django.contrib.auth.models import Permission
 from django.contrib.auth.models import User
 from rest_framework import status
+from rest_framework.test import APIClient
 from rest_framework.test import APITestCase

+if TYPE_CHECKING:
+    from pytest_mock import MockerFixture
+

 class TestChatStreamingViewInputValidation(APITestCase):
    def setUp(self) -> None:
@@ -42,3 +49,73 @@ class TestChatStreamingViewInputValidation(APITestCase):
                format="json",
            )
        assert resp.status_code == status.HTTP_400_BAD_REQUEST
+
+
+@pytest.mark.django_db
+class TestChatStreamingViewUnrestrictedFlag:
+    """The document id filter may only be skipped (``unrestricted=True``) for
+    an active superuser, never for a regular user -- regardless of what
+    permissions that user holds.
+    """
+
+    @pytest.fixture
+    def mocked_stream_chat(self, mocker: MockerFixture) -> mock.MagicMock:
+        """AI enabled, with stream_chat_with_documents patched so the view
+        never touches the real vector store; returns the patched callable so
+        tests can inspect how it was called.
+        """
+        mocker.patch("documents.views.AIConfig").return_value.ai_enabled = True
+        return mocker.patch(
+            "documents.views.stream_chat_with_documents",
+            return_value=iter(()),
+        )
+
+    @pytest.fixture
+    def viewer_client(self, user_client: APIClient, regular_user: User) -> APIClient:
+        """The conftest regular-user client, granted the global
+        view_document permission -- the minimum ViewDocumentsPermissions
+        needs to reach the view at all. Model-level only: says nothing
+        about which documents (if any) this user can actually see.
+        """
+        regular_user.user_permissions.add(
+            *Permission.objects.filter(codename="view_document"),
+        )
+        return user_client
+
+    @pytest.mark.parametrize(
+        ("client_fixture", "expected_unrestricted"),
+        [
+            pytest.param("admin_client", True, id="superuser_is_unrestricted"),
+            pytest.param("viewer_client", False, id="regular_user_is_restricted"),
+        ],
+    )
+    def test_unrestricted_only_for_superuser(
+        self,
+        request: pytest.FixtureRequest,
+        mocked_stream_chat: mock.MagicMock,
+        client_fixture: str,
+        *,
+        expected_unrestricted: bool,
+    ) -> None:
+        """
+        GIVEN:
+            - A superuser, or a regular user holding the global
+              view_document permission (but no object-level document access)
+        WHEN:
+            - They post a chat question with no document_id
+        THEN:
+            - stream_chat_with_documents is called with unrestricted=True
+              only for the superuser; the regular user is always
+              unrestricted=False, regardless of their permissions
+        """
+        client: APIClient = request.getfixturevalue(client_fixture)
+
+        client.post(
+            "/api/documents/chat/",
+            data={"q": "What's in these documents?"},
+            format="json",
+        )
+
+        assert (
+            mocked_stream_chat.call_args.kwargs["unrestricted"] is expected_unrestricted
+        )
@@ -18,8 +18,8 @@ class MockOpenIDProvider:

    def get_brands(self):
        default_servers = [
-            {"id": "yahoo", "name": "Yahoo", "openid_url": "http://me.yahoo.com"},
-            {"id": "hyves", "name": "Hyves", "openid_url": "http://hyves.nl"},
+            dict(id="yahoo", name="Yahoo", openid_url="http://me.yahoo.com"),
+            dict(id="hyves", name="Hyves", openid_url="http://hyves.nl"),
        ]
        return default_servers

@@ -205,12 +205,12 @@ class TestBarcode(
            - Barcode is detected on page 1 (zero indexed)
        """

-        for test_filename in [
+        for test_file in [
            "patch-code-t-middle-reverse.pdf",
            "patch-code-t-middle-distorted.pdf",
            "patch-code-t-middle-fuzzy.pdf",
        ]:
-            test_file = self.BARCODE_SAMPLE_DIR / test_filename
+            test_file = self.BARCODE_SAMPLE_DIR / test_file

            with self.get_reader(test_file) as reader:
                reader.detect()
@@ -777,7 +777,7 @@ class TestPDFActions(DirectoriesMixin, TestCase):
        sig.set.return_value.apply_async.side_effect = Exception("boom")
        mock_consume_file.return_value = sig

-        with self.assertRaises(Exception):  # noqa: B017 - mock injects a bare Exception
+        with self.assertRaises(Exception):
            bulk_edit.merge(doc_ids, delete_originals=True)

        self.doc1.refresh_from_db()
@@ -1318,7 +1318,7 @@ class TestPDFActions(DirectoriesMixin, TestCase):
        sig.apply_async.side_effect = Exception("boom")
        mock_chord.return_value = sig

-        with self.assertRaises(Exception):  # noqa: B017 - mock injects a bare Exception
+        with self.assertRaises(Exception):
            bulk_edit.edit_pdf(doc_ids, operations, delete_original=True)

        self.doc2.refresh_from_db()
@@ -1430,7 +1430,7 @@ class TestPDFActions(DirectoriesMixin, TestCase):
            {"page": 9999},  # invalid page, forces error during PDF load
        ]
        with self.assertLogs("paperless.bulk_edit", level="ERROR"):
-            with self.assertRaises(ValueError):
+            with self.assertRaises(Exception):
                bulk_edit.edit_pdf(doc_ids, operations)
        mock_group.assert_not_called()
        mock_consume_file.assert_not_called()
@@ -806,7 +806,7 @@ class TestClassifier(DirectoriesMixin, TestCase):

        Path(settings.MODEL_FILE).touch()
        mock_load.side_effect = Exception()
-        with self.assertRaises(Exception):  # noqa: B017 - mock injects a bare Exception
+        with self.assertRaises(Exception):
            load_classifier(raise_exception=True)


@@ -137,7 +137,7 @@ class FaultyParser(_BaseNewStyleParser):

 class FaultyGenericExceptionParser(_BaseNewStyleParser):
    def parse(self, document_path, mime_type, *, produce_archive: bool = True) -> None:
-        raise Exception("Generic exception.")  # noqa: TRY002 - deliberately not a ParseError
+        raise Exception("Generic exception.")


 def fake_magic_from_file(file, *, mime=False):  # NOSONAR
@@ -1356,7 +1356,7 @@ class PreConsumeTestCase(DirectoriesMixin, GetConsumerMixin, TestCase):
            script_calls = [
                call
                for call in m.call_args_list
-                if call.args and call.args[0] and call.args[0][0] != "pdftotext"
+                if call.args and call.args[0] and call.args[0][0] not in ("pdftotext",)
            ]
            self.assertEqual(script_calls, [])

@@ -44,7 +44,6 @@ from documents import tasks
 from documents.data_models import ConsumableDocument
 from documents.data_models import DocumentMetadataOverrides
 from documents.data_models import DocumentSource
-from documents.matching import UnsupportedWorkflowTriggerTypeError
 from documents.matching import document_matches_workflow
 from documents.matching import existing_document_matches_workflow
 from documents.matching import prefilter_documents_by_workflowtrigger
@@ -2852,13 +2851,7 @@ class TestWorkflows(
        doc = Document.objects.create(
            title="test",
        )
-        self.assertRaises(
-            UnsupportedWorkflowTriggerTypeError,
-            document_matches_workflow,
-            doc,
-            w,
-            99,
-        )
+        self.assertRaises(Exception, document_matches_workflow, doc, w, 99)

    def test_removal_action_document_updated_workflow(self) -> None:
        """
@@ -21,32 +21,28 @@ def uri_validator(value: str, allowed_schemes: set[str] | None = None) -> None:
        parts = urlparse(value)
        if not parts.scheme:
            raise ValidationError(
-                _("Unable to parse URI %(value)s, missing scheme"),
+                _(f"Unable to parse URI {value}, missing scheme"),
                params={"value": value},
            )
        elif not parts.netloc and not parts.path:
            raise ValidationError(
-                _("Unable to parse URI %(value)s, missing net location or path"),
+                _(f"Unable to parse URI {value}, missing net location or path"),
                params={"value": value},
            )

        if allowed_schemes and parts.scheme not in allowed_schemes:
            raise ValidationError(
                _(
-                    "URI scheme '%(scheme)s' is not allowed. Allowed schemes: %(allowed_schemes)s",
+                    f"URI scheme '{parts.scheme}' is not allowed. Allowed schemes: {', '.join(allowed_schemes)}",
                ),
-                params={
-                    "value": value,
-                    "scheme": parts.scheme,
-                    "allowed_schemes": ", ".join(allowed_schemes),
-                },
+                params={"value": value, "scheme": parts.scheme},
            )

    except ValidationError:
        raise
    except Exception as e:
        raise ValidationError(
-            _("Unable to parse URI %(value)s"),
+            _(f"Unable to parse URI {value}"),
            params={"value": value},
        ) from e

@@ -180,6 +180,7 @@ from documents.permissions import has_system_status_permission
 from documents.permissions import permitted_document_ids
 from documents.permissions import permitted_object_ids
 from documents.permissions import set_permissions_for_object
+from documents.permissions import user_is_unrestricted
 from documents.plugins.date_parsing import get_date_parser
 from documents.schema import generate_object_with_permissions_schema
 from documents.search import SearchHit
@@ -1449,7 +1450,7 @@ class DocumentViewSet(
        try:
            lang = detect(doc.content)
        except Exception:
-            logger.debug("Unable to detect language for document %s", doc.pk)
+            pass
        meta["lang"] = lang

        return Response(meta)
@@ -1487,12 +1488,13 @@ class DocumentViewSet(
            with get_date_parser() as date_parser:
                gen = date_parser.parse(doc.filename, doc.content)
                dates = sorted(
-                    set(
-                        itertools.islice(
+                    {
+                        i
+                        for i in itertools.islice(
                            gen,
                            settings.NUMBER_OF_SUGGESTED_DATES,
-                        ),
-                    ),
+                        )
+                    },
                )

        resp_data = {
@@ -1580,16 +1582,21 @@ class DocumentViewSet(
            except ValueError as exc:
                logger.exception(
                    "Invalid AI configuration while generating suggestions for "
-                    "document %s",
+                    "document %s: %s",
                    doc.pk,
+                    exc,
+                    exc_info=True,
                )
                raise ValidationError(
                    {"ai": [_("Invalid AI configuration.")]},
                ) from exc
-            except LLMTimeoutError:
+            except LLMTimeoutError as exc:
                logger.exception(
-                    "AI backend timed out while generating suggestions for document %s",
+                    "AI backend timed out while generating suggestions for "
+                    "document %s: %s",
                    doc.pk,
+                    exc,
+                    exc_info=True,
                )
                return Response(
                    {"ai": [_("AI backend request timed out.")]},
@@ -2062,7 +2069,7 @@ class DocumentViewSet(
            doc_name, doc_data = serializer.validated_data.get("document")
            version_label = serializer.validated_data.get("version_label")

-            t = int(mktime(datetime.now().timetuple()))  # noqa: DTZ005 - mktime() requires a local time tuple
+            t = int(mktime(datetime.now().timetuple()))

            settings.SCRATCH_DIR.mkdir(parents=True, exist_ok=True)

@@ -2323,10 +2330,12 @@ class ChatStreamingView(GenericAPIView[Any]):
                return HttpResponseForbidden("Insufficient permissions")

            documents = Document.objects.filter(pk=document.pk)
+            unrestricted = False
        else:
            documents = Document.objects.filter(
                id__in=permitted_document_ids(request.user),
            )
+            unrestricted = user_is_unrestricted(request.user)

        output_language = get_llm_output_language(
            ai_config=ai_config,
@@ -2337,6 +2346,7 @@ class ChatStreamingView(GenericAPIView[Any]):
            stream_chat_with_documents(
                query_str=question,
                documents=documents,
+                unrestricted=unrestricted,
                output_language=output_language,
            ),
            content_type="text/event-stream",
@@ -3332,7 +3342,7 @@ class PostDocumentView(GenericAPIView[Any]):
        cf = serializer.validated_data.get("custom_fields")
        from_webui = serializer.validated_data.get("from_webui")

-        t = int(mktime(datetime.now().timetuple()))  # noqa: DTZ005 - mktime() requires a local time tuple
+        t = int(mktime(datetime.now().timetuple()))

        settings.SCRATCH_DIR.mkdir(parents=True, exist_ok=True)

@@ -4142,7 +4152,7 @@ class UiSettingsView(GenericAPIView[Any]):
            user_resp["last_name"] = user.last_name

        # strip <app_label>.
-        roles = (re.sub(r"^\w+.", "", perm) for perm in user.get_all_permissions())
+        roles = map(lambda perm: re.sub(r"^\w+.", "", perm), user.get_all_permissions())
        return Response(
            {
                "user": user_resp,
@@ -5180,11 +5190,11 @@ class SystemStatusView(PassUserMixin):
                f"{m.app}.{m.name}"
                for m in MigrationRecorder.Migration.objects.all().order_by("id")
            ]
-        except Exception:  # pragma: no cover
+        except Exception as e:  # pragma: no cover
            applied_migrations = []
            db_status = "ERROR"
            logger.exception(
-                "System status detected a possible problem while connecting to the database",
+                f"System status detected a possible problem while connecting to the database: {e}",
            )
            db_error = "Error connecting to database, check logs for more detail."

@@ -5200,10 +5210,10 @@ class SystemStatusView(PassUserMixin):
            try:
                client.ping()
                redis_status = "OK"
-            except Exception:
+            except Exception as e:
                redis_status = "ERROR"
                logger.exception(
-                    "System status detected a possible problem while connecting to redis",
+                    f"System status detected a possible problem while connecting to redis: {e}",
                )
                redis_error = "Error connecting to redis, check logs for more detail."

@@ -5233,10 +5243,10 @@ class SystemStatusView(PassUserMixin):
                else:
                    celery_active = "WARNING"
                    celery_error = "Celery worker responded unexpectedly."
-        except Exception:
+        except Exception as e:
            celery_active = "ERROR"
            logger.exception(
-                "System status detected a possible problem while connecting to celery",
+                f"System status detected a possible problem while connecting to celery: {e}",
            )
            celery_error = "Error connecting to celery, check logs for more detail."

@@ -5251,15 +5261,13 @@ class SystemStatusView(PassUserMixin):
            index_dir = settings.INDEX_DIR
            mtimes = [p.stat().st_mtime for p in index_dir.iterdir() if p.is_file()]
            index_last_modified = (
-                make_aware(datetime.fromtimestamp(max(mtimes)))  # noqa: DTZ006 - make_aware() requires a naive datetime
-                if mtimes
-                else None
+                make_aware(datetime.fromtimestamp(max(mtimes))) if mtimes else None
            )
-        except Exception:
+        except Exception as e:
            index_status = "ERROR"
            index_error = "Error opening index, check logs for more detail."
            logger.exception(
-                "System status detected a possible problem while opening the index",
+                f"System status detected a possible problem while opening the index: {e}",
            )
            index_last_modified = None

@@ -66,7 +66,7 @@ def build_workflow_action_context(
        else None
    )

-    filename = document.original_file or ""
+    filename = document.original_file if document.original_file else ""
    return {
        "title": overrides.title
        if overrides and overrides.title
@@ -179,9 +179,9 @@ def execute_email_action(
            f"Sent {n_messages} notification email(s) to {action.email.to}",
            extra={"group": logging_group},
        )
-    except Exception:
+    except Exception as e:
        logger.exception(
-            "Error occurred sending notification email",
+            f"Error occurred sending notification email: {e}",
            extra={"group": logging_group},
        )

@@ -265,9 +265,9 @@ def execute_webhook_action(
            f"Webhook to {action.webhook.url} queued",
            extra={"group": logging_group},
        )
-    except Exception:
+    except Exception as e:
        logger.exception(
-            "Error occurred sending webhook",
+            f"Error occurred sending webhook: {e}",
            extra={"group": logging_group},
        )

@@ -47,7 +47,7 @@ def resolve_date(dates: list[str]) -> date | None:
    """
    for value in dates:
        try:
-            return datetime.strptime(value, "%Y-%m-%d").date()  # noqa: DTZ007 - only the calendar date is used, time/tz is discarded
+            return datetime.strptime(value, "%Y-%m-%d").date()
        except (TypeError, ValueError):
            logger.debug("Ignoring unparsable suggested date %s", value)
    return None
@@ -70,6 +70,6 @@ def send_webhook(
        logger.error(
            f"Failed attempt sending webhook to {url}: {e}",
        )
-        raise
+        raise e
    finally:
        transport.close()
@@ -241,7 +241,7 @@ def check_v3_minimum_upgrade_version(
        return []

    logger = logging.getLogger(__name__)
-    last_applied = max(applied) if applied else "(none)"
+    last_applied = sorted(applied)[-1] if applied else "(none)"
    logger.error(
        "V3 upgrade check failed: last applied documents migration is %r. "
        "Expected '1075_workflowaction_order' (v2.20.15). "
@@ -341,7 +341,6 @@ def get_tesseract_langs():
    proc = subprocess.run(
        [shutil.which("tesseract"), "--list-langs"],
        capture_output=True,
-        check=False,
    )

    # Decode bytes to string, split on newlines, trim out the header
@@ -84,7 +84,7 @@ def get_parser_registry() -> ParserRegistry:
    ParserRegistry
        The shared registry singleton.
    """
-    global _registry, _discovery_complete  # noqa: PLW0603 - module-level singleton, no class to hold this state
+    global _registry, _discovery_complete

    with _lock:
        if _registry is None:
@@ -113,7 +113,7 @@ def init_builtin_parsers() -> None:
    -------
    None
    """
-    global _registry  # noqa: PLW0603 - module-level singleton, no class to hold this state
+    global _registry

    with _lock:
        if _registry is None:
@@ -137,7 +137,7 @@ def reset_parser_registry() -> None:
    -------
    None
    """
-    global _registry, _discovery_complete  # noqa: PLW0603 - module-level singleton, no class to hold this state
+    global _registry, _discovery_complete

    _registry = None
    _discovery_complete = False
@@ -78,7 +78,7 @@ class RemoteEngineConfig:
    def engine_is_valid(self) -> bool:
        """Return True when the engine is known and fully configured."""
        return (
-            self.engine == "azureai"
+            self.engine in ("azureai",)
            and self.api_key is not None
            and not (self.engine == "azureai" and self.endpoint is None)
        )
@@ -505,7 +505,7 @@ class RemoteDocumentParser:
            return result.content

        except Exception as e:
-            logger.exception("Azure AI Vision parsing failed")
+            logger.exception("Azure AI Vision parsing failed: %s", e)
            raise ParseError(f"Azure AI Vision parsing failed: {e}") from e

        finally:
@@ -306,9 +306,8 @@ def extract_pdf_metadata(

    for key, value in meta.items():
        if isinstance(value, list):
-            str_value = " ".join(str(e) for e in value)
-        else:
-            str_value = str(value)
+            value = " ".join(str(e) for e in value)
+        value = str(value)

        try:
            m = namespace_pattern.match(key)
@@ -330,7 +329,7 @@ def extract_pdf_metadata(
                    namespace=namespace,
                    prefix=meta.REVERSE_NS[namespace],
                    key=key_value,
-                    value=str_value,
+                    value=value,
                ),
            )
        except Exception as e:
@@ -294,7 +294,7 @@ if _CHANNELS_BACKEND.startswith("channels_redis."):
 ###############################################################################

 EMAIL_HOST: Final[str] = os.getenv("PAPERLESS_EMAIL_HOST", "localhost")
-EMAIL_PORT: Final[int] = get_int_from_env("PAPERLESS_EMAIL_PORT", 25)
+EMAIL_PORT: Final[int] = int(os.getenv("PAPERLESS_EMAIL_PORT", 25))
 EMAIL_HOST_USER: Final[str] = os.getenv("PAPERLESS_EMAIL_HOST_USER", "")
 EMAIL_HOST_PASSWORD: Final[str] = os.getenv("PAPERLESS_EMAIL_HOST_PASSWORD", "")
 DEFAULT_FROM_EMAIL: Final[str] = os.getenv("PAPERLESS_EMAIL_FROM", EMAIL_HOST_USER)
@@ -381,9 +381,8 @@ ACCOUNT_SESSION_REMEMBER = get_bool_from_env(
    "True",
 )
 SESSION_EXPIRE_AT_BROWSER_CLOSE = not ACCOUNT_SESSION_REMEMBER
-SESSION_COOKIE_AGE = get_int_from_env(
-    "PAPERLESS_SESSION_COOKIE_AGE",
-    60 * 60 * 24 * 7 * 3,
+SESSION_COOKIE_AGE = int(
+    os.getenv("PAPERLESS_SESSION_COOKIE_AGE", 60 * 60 * 24 * 7 * 3),
 )
 # https://docs.djangoproject.com/en/5.1/ref/settings/#std-setting-SESSION_ENGINE
 SESSION_ENGINE = "django.contrib.sessions.backends.cached_db"
@@ -396,6 +395,7 @@ if AUTO_LOGIN_USERNAME:


 def _parse_remote_user_settings() -> str:
+    global MIDDLEWARE, AUTHENTICATION_BACKENDS, REST_FRAMEWORK
    enable = get_bool_from_env("PAPERLESS_ENABLE_HTTP_REMOTE_USER")
    enable_api = get_bool_from_env("PAPERLESS_ENABLE_HTTP_REMOTE_USER_API")
    if enable or enable_api:
@@ -454,6 +454,7 @@ if ALLOWED_HOSTS != ["*"]:


 def _parse_paperless_url():
+    global CSRF_TRUSTED_ORIGINS, CORS_ALLOWED_ORIGINS, ALLOWED_HOSTS
    url = os.getenv("PAPERLESS_URL")
    if url:
        CSRF_TRUSTED_ORIGINS.append(url)
@@ -613,8 +614,8 @@ USE_TZ = True

 LOGGING_DIR.mkdir(parents=True, exist_ok=True)

-LOGROTATE_MAX_SIZE = get_int_from_env("PAPERLESS_LOGROTATE_MAX_SIZE", 1024 * 1024)
-LOGROTATE_MAX_BACKUPS = get_int_from_env("PAPERLESS_LOGROTATE_MAX_BACKUPS", 20)
+LOGROTATE_MAX_SIZE = os.getenv("PAPERLESS_LOGROTATE_MAX_SIZE", 1024 * 1024)
+LOGROTATE_MAX_BACKUPS = os.getenv("PAPERLESS_LOGROTATE_MAX_BACKUPS", 20)

 LOGGING = {
    "version": 1,
@@ -810,15 +811,9 @@ IGNORABLE_FILES: Final[list[str]] = [
    "Thumbs.db",
 ]

-CONSUMER_POLLING_INTERVAL = get_float_from_env(
-    "PAPERLESS_CONSUMER_POLLING_INTERVAL",
-    0.0,
-)
+CONSUMER_POLLING_INTERVAL = float(os.getenv("PAPERLESS_CONSUMER_POLLING_INTERVAL", 0))

-CONSUMER_STABILITY_DELAY = get_float_from_env(
-    "PAPERLESS_CONSUMER_STABILITY_DELAY",
-    5.0,
-)
+CONSUMER_STABILITY_DELAY = float(os.getenv("PAPERLESS_CONSUMER_STABILITY_DELAY", 5))

 CONSUMER_DELETE_DUPLICATES = get_bool_from_env("PAPERLESS_CONSUMER_DELETE_DUPLICATES")

@@ -111,7 +111,7 @@ def parse_dict_from_str(
            return False

    settings: dict[str, Any] = copy.deepcopy(defaults) if defaults else {}
-    _type_map = type_map or {}
+    _type_map = type_map if type_map else {}

    if not env_str:
        return settings
@@ -114,17 +114,17 @@ def test_cache_hit_when_enabled() -> None:
    assert settings.CACHALOT_TIMEOUT == 1

    # Read a table to populate the cache
-    list(Tag.objects.values_list("id", flat=True))
+    list(list(Tag.objects.values_list("id", flat=True)))

    # Invalidate the cache then read the database, there should be DB hit
    invalidate_db_cache()
    with CaptureQueriesContext(connection) as ctx:
-        list(Tag.objects.values_list("id", flat=True))
+        list(list(Tag.objects.values_list("id", flat=True)))
    assert len(ctx)

    # Doing the same request again should hit the cache, not the DB
    with CaptureQueriesContext(connection) as ctx:
-        list(Tag.objects.values_list("id", flat=True))
+        list(list(Tag.objects.values_list("id", flat=True)))
    assert not len(ctx)

    # Wait the end of TTL
@@ -133,7 +133,7 @@ def test_cache_hit_when_enabled() -> None:

    # Read the DB again. The DB should be hit because the cache has expired
    with CaptureQueriesContext(connection) as ctx:
-        list(Tag.objects.values_list("id", flat=True))
+        list(list(Tag.objects.values_list("id", flat=True)))
    assert len(ctx)

    # Invalidate the cache at the end of test
@@ -149,7 +149,7 @@ def test_cache_is_disabled_by_default() -> None:
    # Read the table multiple times: the DB should always be hit without cache
    for _ in range(3):
        with CaptureQueriesContext(connection) as ctx:
-            list(Tag.objects.values_list("id", flat=True))
+            list(list(Tag.objects.values_list("id", flat=True)))
        assert len(ctx)

    # Invalidate the cache at the end of test
@@ -59,7 +59,7 @@ def test_ocr_to_dateparser_languages_exception(
            raise RuntimeError("Simulated error")

    with caplog.at_level(logging.WARNING):
-        monkeypatch.setattr(utils, "LocaleDataLoader", DummyLoader)
+        monkeypatch.setattr(utils, "LocaleDataLoader", lambda: DummyLoader())
        result = utils.ocr_to_dateparser_languages("eng+fra")
        assert result == []
        assert (
@@ -8,7 +8,8 @@ from documents.models import Document
 from paperless.config import AIConfig
 from paperless_ai.client import AIClient
 from paperless_ai.db import db_connection_released
-from paperless_ai.indexing import _document_id_filters
+from paperless_ai.indexing import document_id_filters
+from paperless_ai.indexing import exclude_document_ids_filter
 from paperless_ai.indexing import get_rag_prompt_helper
 from paperless_ai.indexing import load_or_build_index
 from paperless_ai.indexing import read_store
@@ -95,22 +96,27 @@ def _format_chat_metadata_trailer(references: list[dict[str, int | str]]) -> str
 def stream_chat_with_documents(
    query_str: str,
    documents: QuerySet[Document],
+    *,
+    unrestricted: bool = False,
    output_language: str | None = None,
 ):
    try:
        yield from _stream_chat_with_documents(
            query_str,
            documents,
+            unrestricted=unrestricted,
            output_language=output_language,
        )
-    except Exception:
-        logger.exception("Failed to stream document chat response")
+    except Exception as e:
+        logger.exception("Failed to stream document chat response: %s", e)
        yield CHAT_ERROR_MESSAGE


 def _stream_chat_with_documents(
    query_str: str,
    documents: QuerySet[Document],
+    *,
+    unrestricted: bool = False,
    output_language: str | None = None,
 ):
    if not documents.exists():
@@ -123,9 +129,20 @@ def _stream_chat_with_documents(
    from llama_index.core.retrievers import VectorIndexRetriever

    config = AIConfig()
-    filters = _document_id_filters(
-        str(pk) for pk in documents.values_list("pk", flat=True)
-    )
+    if unrestricted:
+        # Exclude trashed ids (usually few) instead of an IN filter over the
+        # full permitted set, which risks the vector store's bound parameter
+        # limit (_MAX_IN_VALUES) on large installs. Trashed documents stay
+        # indexed until permanent deletion (delete_document_from_llm_index
+        # hangs off post_delete, not trash), so must be excluded explicitly.
+        trashed_ids = Document.global_objects.filter(
+            deleted_at__isnull=False,
+        ).values_list("pk", flat=True)
+        filters = exclude_document_ids_filter(str(pk) for pk in trashed_ids)
+    else:
+        filters = document_id_filters(
+            str(pk) for pk in documents.values_list("pk", flat=True)
+        )

    # Hold the shared read lock for the whole operation: the query engine
    # retrieves from the vector store again during synthesis, so the connection
@@ -362,7 +362,7 @@ def _embed_nodes(nodes: list["BaseNode"], embed_model) -> None:
        node.embedding = emb


-def _document_id_filters(doc_ids):
+def document_id_filters(doc_ids):
    """Return a MetadataFilters IN filter scoped to ``doc_ids``."""
    from llama_index.core.vector_stores.types import FilterOperator
    from llama_index.core.vector_stores.types import MetadataFilter
@@ -396,6 +396,23 @@ def _exclude_document_id_filter(document_id: int | str):
    )


+def exclude_document_ids_filter(doc_ids):
+    """Return a MetadataFilters NIN filter excluding every id in ``doc_ids``."""
+    from llama_index.core.vector_stores.types import FilterOperator
+    from llama_index.core.vector_stores.types import MetadataFilter
+    from llama_index.core.vector_stores.types import MetadataFilters
+
+    return MetadataFilters(
+        filters=[
+            MetadataFilter(
+                key="document_id",
+                operator=FilterOperator.NIN,
+                value=list(doc_ids),
+            ),
+        ],
+    )
+
+
 def update_llm_index(
    *,
    iter_wrapper: IterWrapper[Document] = identity,
@@ -660,7 +677,7 @@ def retrieve_similar_nodes(

    filter_parts = []
    if allowed_document_ids is not None:
-        filter_parts.extend(_document_id_filters(allowed_document_ids).filters)
+        filter_parts.extend(document_id_filters(allowed_document_ids).filters)
    if document.pk is not None:
        filter_parts.extend(_exclude_document_id_filter(document.pk).filters)

@@ -154,6 +154,35 @@ class DocumentMetaTable:
        }


+class PermittedIdsTable:
+    """Per-connection scratch space for an oversized IN-filter id list.
+
+    A literal ``IN (?,?,...)`` list binds one SQL parameter per id, capped by
+    SQLite's own SQLITE_MAX_VARIABLE_NUMBER (see _MAX_IN_VALUES in
+    vector_store.py). Loading the ids into a TEMP TABLE and filtering via a
+    subquery instead has no such limit. TEMP tables live in a
+    connection-private namespace -- never visible to another connection,
+    even under this identical name -- so this is safe under the vector
+    store's one-connection-per-request model without any extra locking or
+    per-call naming scheme.
+    """
+
+    TABLE_NAME = "permitted_document_ids"
+
+    @staticmethod
+    def load(conn: sqlite3.Connection, ids: Iterable[int]) -> None:
+        """Replace this connection's scratch table with ``ids``."""
+        conn.execute(f"DROP TABLE IF EXISTS temp.{PermittedIdsTable.TABLE_NAME}")
+        conn.execute(
+            f"CREATE TEMP TABLE {PermittedIdsTable.TABLE_NAME} "
+            "(id INTEGER PRIMARY KEY)",
+        )
+        conn.executemany(
+            f"INSERT INTO {PermittedIdsTable.TABLE_NAME} (id) VALUES (?)",
+            ((i,) for i in ids),
+        )
+
+
 class IndexMetaTable:
    """Typed accessors over index_meta's key/value rows -- replaces
    PaperlessSqliteVecVectorStore._meta_get_on/_meta_set_on, which returned
@@ -164,7 +164,7 @@ def test_get_ai_document_classification_failure(mock_run_llm_query, mock_documen
    """
    mock_run_llm_query.side_effect = Exception("LLM query failed")

-    with pytest.raises(Exception):  # noqa: B017 - mock injects a bare Exception
+    with pytest.raises(Exception):
        get_ai_document_classification(mock_document)


@@ -1,9 +1,14 @@
+from __future__ import annotations
+
 import json
+from typing import TYPE_CHECKING
+from typing import Any
 from unittest.mock import MagicMock
 from unittest.mock import patch

 import pytest
 from django.db.models.signals import post_init
+from django.utils import timezone
 from llama_index.core import settings as llama_settings
 from llama_index.core.embeddings.mock_embed_model import MockEmbedding
 from llama_index.core.schema import TextNode
@@ -18,6 +23,11 @@ from paperless_ai.chat import _build_chat_prompt
 from paperless_ai.chat import _build_refine_prompt
 from paperless_ai.chat import stream_chat_with_documents

+if TYPE_CHECKING:
+    from pathlib import Path
+
+    import pytest_mock
+

@pytest.fixture(autouse=True)
 def patch_embed_model():
@@ -310,8 +320,40 @@ def test_stream_chat_unexpected_failure_returns_generic_error(caplog) -> None:
        assert "private provider detail" in caplog.text


+def _retriever_filter_values(captured_filters: list[Any]) -> list[str]:
+    """The value list of the single MetadataFilter the retriever received."""
+    assert captured_filters, "VectorIndexRetriever was never constructed"
+    filt = captured_filters[0]
+    assert filt is not None, "Retriever must receive a MetadataFilters"
+    return filt.filters[0].value
+
+
@pytest.mark.django_db
 class TestStreamChatRetrieval:
+    @pytest.fixture
+    def captured_filters(self, mocker: pytest_mock.MockerFixture) -> list[Any]:
+        """Stub out the AI client and the retriever, capturing the ``filters``
+        kwarg of every VectorIndexRetriever construction.
+
+        VectorIndexRetriever is imported inside _stream_chat_with_documents,
+        so it is patched at the llama_index source for the lazy import to
+        pick it up.
+        """
+        captured: list[Any] = []
+        retriever = mocker.MagicMock()
+        retriever.retrieve.return_value = []
+
+        def capture_retriever(*args, **kwargs) -> pytest_mock.MockType:
+            captured.append(kwargs.get("filters"))
+            return retriever
+
+        mocker.patch("paperless_ai.chat.AIClient")
+        mocker.patch(
+            "llama_index.core.retrievers.VectorIndexRetriever",
+            side_effect=capture_retriever,
+        )
+        return captured
+
    def test_no_nodes_yields_no_content_message(
        self,
        temp_llm_index_dir,
@@ -329,9 +371,9 @@ class TestStreamChatRetrieval:

    def test_chat_filter_contains_only_requested_document_ids(
        self,
-        temp_llm_index_dir,
-        mock_embed_model,
-        mocker,
+        temp_llm_index_dir: Path,
+        mock_embed_model: pytest_mock.MockType,
+        captured_filters: list[Any],
    ) -> None:
        """The MetadataFilter passed to the retriever must be scoped to the
        requested documents only — content from other indexed documents must
@@ -342,22 +384,6 @@ class TestStreamChatRetrieval:
        indexing.llm_index_add_or_update_document(included)
        indexing.llm_index_add_or_update_document(excluded)

-        # VectorIndexRetriever is imported inside _stream_chat_with_documents;
-        # patch it at the llama_index source so the lazy import picks it up.
-        captured_filters = []
-        mock_retriever = mocker.MagicMock()
-        mock_retriever.retrieve.return_value = []
-
-        def capture_retriever(*args, **kwargs):
-            captured_filters.append(kwargs.get("filters"))
-            return mock_retriever
-
-        mocker.patch("paperless_ai.chat.AIClient")
-        mocker.patch(
-            "llama_index.core.retrievers.VectorIndexRetriever",
-            side_effect=capture_retriever,
-        )
-
        list(
            chat.stream_chat_with_documents(
                "question?",
@@ -365,13 +391,78 @@ class TestStreamChatRetrieval:
            ),
        )

-        assert captured_filters, "VectorIndexRetriever was never constructed"
-        filt = captured_filters[0]
-        assert filt is not None, "Retriever must receive a MetadataFilters"
-        filter_values = filt.filters[0].value
+        filter_values = _retriever_filter_values(captured_filters)
        assert str(included.pk) in filter_values
        assert str(excluded.pk) not in filter_values

+    def test_unrestricted_chat_excludes_nothing_when_no_documents_are_trashed(
+        self,
+        temp_llm_index_dir: Path,
+        mock_embed_model: pytest_mock.MockType,
+        captured_filters: list[Any],
+    ) -> None:
+        """
+        GIVEN:
+            - A document indexed in the vector store, nothing trashed
+        WHEN:
+            - stream_chat_with_documents is called with unrestricted=True
+        THEN:
+            - The retriever receives a NOT IN filter excluding zero ids, so
+              the whole index is effectively searched -- and no IN-list is
+              built from the full permitted set, which is what risks the
+              vector store's safety limit on large installs
+        """
+        document = DocumentFactory.create(content="indexed document content")
+        indexing.llm_index_add_or_update_document(document)
+
+        list(
+            chat.stream_chat_with_documents(
+                "question?",
+                Document.objects.filter(pk=document.pk),
+                unrestricted=True,
+            ),
+        )
+
+        assert _retriever_filter_values(captured_filters) == []
+
+    def test_unrestricted_chat_excludes_trashed_documents(
+        self,
+        temp_llm_index_dir: Path,
+        mock_embed_model: pytest_mock.MockType,
+        captured_filters: list[Any],
+    ) -> None:
+        """
+        GIVEN:
+            - Two indexed documents, one of them trashed -- trashed documents
+              stay in the vector index until permanently deleted, since
+              delete_document_from_llm_index is wired to post_delete
+        WHEN:
+            - stream_chat_with_documents is called with unrestricted=True
+        THEN:
+            - The retriever receives a NOT IN filter excluding the trashed
+              document's id, so an unrestricted caller (e.g. a superuser)
+              never has trashed content surfaced in a chat answer
+        """
+        kept = DocumentFactory.create(content="kept document content")
+        trashed = DocumentFactory.create(content="trashed document content")
+        indexing.llm_index_add_or_update_document(kept)
+        indexing.llm_index_add_or_update_document(trashed)
+        Document.global_objects.filter(pk=trashed.pk).update(
+            deleted_at=timezone.now(),
+        )
+
+        list(
+            chat.stream_chat_with_documents(
+                "question?",
+                Document.objects.filter(pk=kept.pk),
+                unrestricted=True,
+            ),
+        )
+
+        filter_values = _retriever_filter_values(captured_filters)
+        assert str(trashed.pk) in filter_values
+        assert str(kept.pk) not in filter_values
+
    @pytest.mark.django_db
    def test_get_document_references_only_queries_referenced_documents(
        self,
@@ -21,6 +21,5 @@ class TestLazyAiImports:
            capture_output=True,
            text=True,
            cwd=_SRC_DIR,
-            check=False,
        )
        assert result.returncode == 0, result.stdout + result.stderr
@@ -9,6 +9,7 @@ from paperless_ai.tables import DocumentChunksTable
 from paperless_ai.tables import DocumentMetaRow
 from paperless_ai.tables import DocumentMetaTable
 from paperless_ai.tables import IndexMetaTable
+from paperless_ai.tables import PermittedIdsTable


@pytest.fixture
@@ -338,3 +339,85 @@ class TestIndexMetaTable:
        IndexMetaTable.increment_total_inserts(conn, 100)
        IndexMetaTable.reset_total_inserts(conn, 7)
        assert IndexMetaTable.get_total_inserts(conn) == 7
+
+
+class TestPermittedIdsTable:
+    def _loaded_ids(self, conn: sqlite3.Connection) -> list[int]:
+        return [
+            row["id"]
+            for row in conn.execute(
+                f"SELECT id FROM {PermittedIdsTable.TABLE_NAME} ORDER BY id",
+            )
+        ]
+
+    def test_load_then_read_back_all_ids(self, conn: sqlite3.Connection) -> None:
+        """
+        GIVEN:
+            - A bare sqlite3 connection
+        WHEN:
+            - load() is called with a set of ids
+        THEN:
+            - Every id is present in the TEMP TABLE, and only those ids
+        """
+        PermittedIdsTable.load(conn, [3, 1, 2])
+        assert self._loaded_ids(conn) == [1, 2, 3]
+
+    def test_load_replaces_previous_contents(self, conn: sqlite3.Connection) -> None:
+        """
+        GIVEN:
+            - A connection whose PermittedIdsTable already holds one id set
+        WHEN:
+            - load() is called again with a different id set
+        THEN:
+            - Only the new ids are present -- a connection reused across
+              multiple queries in one request never leaks a stale filter
+        """
+        PermittedIdsTable.load(conn, [1, 2, 3])
+        PermittedIdsTable.load(conn, [4, 5])
+        assert self._loaded_ids(conn) == [4, 5]
+
+    def test_load_is_connection_private(self) -> None:
+        """
+        GIVEN:
+            - Two separate connections
+        WHEN:
+            - Each loads PermittedIdsTable with a different id set, under
+              the identical TABLE_NAME
+        THEN:
+            - Each connection sees only its own ids -- TEMP TABLE is
+              connection-private, so concurrent requests never collide or
+              cross-contaminate despite sharing the same table name (the
+              vector store opens one connection per request; see
+              PaperlessSqliteVecVectorStore)
+        """
+        conn_a = sqlite3.connect(":memory:")
+        conn_a.row_factory = sqlite3.Row
+        conn_b = sqlite3.connect(":memory:")
+        conn_b.row_factory = sqlite3.Row
+        try:
+            PermittedIdsTable.load(conn_a, [1, 2, 3])
+            PermittedIdsTable.load(conn_b, [4, 5, 6])
+            assert self._loaded_ids(conn_a) == [1, 2, 3]
+            assert self._loaded_ids(conn_b) == [4, 5, 6]
+        finally:
+            conn_a.close()
+            conn_b.close()
+
+    def test_load_handles_more_ids_than_a_bound_parameter_list_could(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
+        """
+        GIVEN:
+            - An id count over SQLite's own bound-parameter limit
+              (SQLITE_MAX_VARIABLE_NUMBER, 32766 by default) -- more than a
+              literal IN(?,?,...) list could ever bind in one statement
+        WHEN:
+            - load() is called with that many ids
+        THEN:
+            - Every id is loaded without error, since executemany() binds
+              one row at a time rather than one statement with N parameters
+        """
+        ids = list(range(40_000))
+        PermittedIdsTable.load(conn, ids)
+        assert self._loaded_ids(conn) == ids
@@ -1,5 +1,6 @@
 import inspect
 import sqlite3
+from collections.abc import Callable
 from collections.abc import Generator
 from pathlib import Path

@@ -17,6 +18,7 @@ from paperless_ai.migrations import Migration
 from paperless_ai.migrations import m0001_v1_to_v2
 from paperless_ai.tables import DocumentChunksTable
 from paperless_ai.tables import DocumentMetaTable
+from paperless_ai.tables import PermittedIdsTable
 from paperless_ai.vector_store import _MAX_IN_VALUES
 from paperless_ai.vector_store import DB_FILENAME
 from paperless_ai.vector_store import DEFAULT_TABLE_NAME
@@ -97,6 +99,18 @@ def _ne_filter(document_id: int):
    )


+def _nin_filter(document_ids: list[int]):
+    return MetadataFilters(
+        filters=[
+            MetadataFilter(
+                key="document_id",
+                operator=FilterOperator.NIN,
+                value=document_ids,
+            ),
+        ],
+    )
+
+
 class TestCrud:
    def test_add_then_query_returns_node(self, store) -> None:
        node = make_node("n1", 1)
@@ -267,8 +281,23 @@ class TestCrud:


 class TestBuildWhere:
-    def test_ne_filter_translates_to_not_equal_clause(self) -> None:
-        where, params = _build_where(_ne_filter(1))
+    @pytest.fixture
+    def conn(self) -> Generator[sqlite3.Connection, None, None]:
+        """A bare connection, sufficient for _build_where(): it only ever
+        touches the connection via PermittedIdsTable, which needs no vec0
+        extension loaded.
+        """
+        connection = sqlite3.connect(":memory:")
+        try:
+            yield connection
+        finally:
+            connection.close()
+
+    def test_ne_filter_translates_to_not_equal_clause(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
+        where, params = _build_where(conn, _ne_filter(1))
        assert where == "(document_id != ?)"
        assert params == [1]

@@ -280,7 +309,60 @@ class TestBuildWhere:
            "b1",
        ]

-    def test_fails_closed_when_no_filter_is_translatable(self) -> None:
+    def test_nin_filter_translates_to_not_in_clause(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
+        where, params = _build_where(conn, _nin_filter([1, 2]))
+        assert where == "(document_id NOT IN (?,?))"
+        assert params == [1, 2]
+
+    def test_query_with_nin_filter_excludes_matching_documents(self, store) -> None:
+        store.add([make_node("a1", 1), make_node("b1", 2), make_node("c1", 3)])
+        assert sorted(
+            _query(store, [0.0] * DIM, top_k=5, filters=_nin_filter([1, 2])).ids,
+        ) == ["c1"]
+
+    def test_empty_in_filter_excludes_everything(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
+        """
+        GIVEN:
+            - An IN filter with an empty value list
+        WHEN:
+            - _build_where() translates it to SQL
+        THEN:
+            - It excludes everything (the opposite of an empty NOT IN
+              filter) -- an empty inclusion list must never widen results
+        """
+        where, params = _build_where(conn, _in_filter([]))
+        assert where == "(1 = 0)"
+        assert params == []
+
+    def test_empty_nin_filter_excludes_nothing(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
+        """
+        GIVEN:
+            - A NOT IN filter with an empty value list -- e.g. an
+              unrestricted chat caller when nothing is currently trashed
+        WHEN:
+            - _build_where() translates it to SQL
+        THEN:
+            - It excludes nothing (unlike an empty IN filter, which
+              excludes everything) -- an empty exclusion list must never
+              narrow results
+        """
+        where, params = _build_where(conn, _nin_filter([]))
+        assert where == "(1 = 1)"
+        assert params == []
+
+    def test_fails_closed_when_no_filter_is_translatable(
+        self,
+        conn: sqlite3.Connection,
+    ) -> None:
        # A nested MetadataFilters is not a MetadataFilter, so it is skipped.
        # With no translatable clauses, the function must fail closed rather
        # than emit "()" (invalid SQL) and never widen document access.
@@ -293,35 +375,88 @@ class TestBuildWhere:
                ),
            ],
        )
-        where, params = _build_where(MetadataFilters(filters=[nested]))
+        where, params = _build_where(conn, MetadataFilters(filters=[nested]))
        assert where == "1 = 0"
        assert params == []

-    def test_fails_closed_when_in_filter_exceeds_max_values(
+    @pytest.mark.parametrize(
+        ("build_filter", "sql_op"),
+        [(_in_filter, "IN"), (_nin_filter, "NOT IN")],
+        ids=["in", "nin"],
+    )
+    def test_filter_over_max_values_uses_permitted_ids_table(
        self,
-        caplog: pytest.LogCaptureFixture,
+        conn: sqlite3.Connection,
+        build_filter: Callable[[list[str]], MetadataFilters],
+        sql_op: str,
    ) -> None:
        """
        GIVEN:
-            - An IN filter with more values than _MAX_IN_VALUES (SQLite's
-              own bound-parameter limit is 32766; this guard sits below
-              that with headroom for the query's other bound parameters)
+            - An IN or NOT IN filter with more values than _MAX_IN_VALUES
+              (SQLite's own bound-parameter limit is 32766; this threshold
+              sits below that with headroom for the query's other bound
+              parameters)
        WHEN:
            - _build_where() translates it to SQL
        THEN:
-            - It fails closed ("1 = 0", no params) instead of building an
-              IN clause SQLite would reject, and logs a warning -- this
-              filter scopes document access, so refusing to build it must
-              never widen the scope to "everything" by accident
+            - It builds a subquery against PermittedIdsTable's TEMP TABLE,
+              loaded with every id, instead of a literal list SQLite would
+              reject past its own limit -- true for NOT IN too (e.g. an
+              install with an enormous trash), not just IN
        """
-        oversized = _in_filter([str(i) for i in range(_MAX_IN_VALUES + 1)])
+        ids = list(range(_MAX_IN_VALUES + 1))
+        oversized = build_filter([str(i) for i in ids])

-        with caplog.at_level("WARNING"):
-            where, params = _build_where(oversized)
+        where, params = _build_where(conn, oversized)

-        assert where == "(1 = 0)"
+        assert where == (
+            f"(document_id {sql_op} (SELECT id FROM {PermittedIdsTable.TABLE_NAME}))"
+        )
        assert params == []
-        assert "document_id" in caplog.text
+        loaded = [
+            row[0]
+            for row in conn.execute(
+                f"SELECT id FROM {PermittedIdsTable.TABLE_NAME} ORDER BY id",
+            )
+        ]
+        assert loaded == ids
+
+    @pytest.mark.parametrize(
+        ("build_filter", "expected_ids"),
+        [(_in_filter, ["b1", "c1"]), (_nin_filter, ["a1"])],
+        ids=["in", "nin"],
+    )
+    def test_query_and_get_nodes_scope_correctly_when_filter_exceeds_max_values(
+        self,
+        store: PaperlessSqliteVecVectorStore,
+        mocker: MockerFixture,
+        build_filter: Callable[[list[int]], MetadataFilters],
+        expected_ids: list[str],
+    ) -> None:
+        """
+        GIVEN:
+            - _MAX_IN_VALUES lowered so a small IN/NOT IN filter exceeds it
+        WHEN:
+            - query() and get_nodes() are called with that filter
+        THEN:
+            - Both still correctly scope results -- the PermittedIdsTable
+              temp-table path behaves identically to the literal
+              IN(...)/NOT IN(...) path it replaces above the threshold
+        """
+        mocker.patch("paperless_ai.vector_store._MAX_IN_VALUES", 1)
+        store.add(
+            [
+                make_node("a1", 1, seed=0.0),
+                make_node("b1", 2, seed=1.0),
+                make_node("c1", 3, seed=2.0),
+            ],
+        )
+
+        result = _query(store, [0.0] * DIM, top_k=10, filters=build_filter([2, 3]))
+        nodes = store.get_nodes(filters=build_filter([2, 3]))
+
+        assert sorted(result.ids) == expected_ids
+        assert sorted(n.node_id for n in nodes) == expected_ids

    def test_query_with_untranslatable_filter_returns_no_rows(
        self,
@@ -30,6 +30,7 @@ from paperless_ai.tables import DocumentChunksTable
 from paperless_ai.tables import DocumentMetaRow
 from paperless_ai.tables import DocumentMetaTable
 from paperless_ai.tables import IndexMetaTable
+from paperless_ai.tables import PermittedIdsTable

 logger = logging.getLogger("paperless_ai.vector_store")

@@ -75,14 +76,12 @@ class _Row(NamedTuple):
    embedding: bytes


-# _build_where(): the largest IN value list translated into bound SQL
-# parameters. SQLite's own hard limit (SQLITE_MAX_VARIABLE_NUMBER) is 32766
-# by default; this leaves headroom below that for the query's other bound
-# parameters (the embedding blob, k, and any NE clause) and for the limit
-# itself to move. An IN filter this large should not happen in practice --
-# callers are expected to pass None (no filter) rather than every id when
-# the filter would not actually narrow anything -- so this is a guard
-# against a future regression, not a normal code path.
+# _build_where(): the largest IN value list translated into a literal
+# IN (?,?,...) clause. SQLite's own hard limit (SQLITE_MAX_VARIABLE_NUMBER)
+# is 32766 by default; this leaves headroom below that for the query's other
+# bound parameters (the embedding blob, k, and any NE clause) and for the
+# limit itself to move. Above this threshold _build_where() switches to
+# PermittedIdsTable instead of failing closed -- see its docstring.
 _MAX_IN_VALUES = 32700


@@ -106,13 +105,21 @@ def _vec0_params(rows: list[_Row]) -> list[tuple[str, int, str, bytes]]:
    return [(r.chunk_id, r.document_id, r.node_content, r.embedding) for r in rows]


-def _build_where(filters: MetadataFilters | None) -> tuple[str, list[int]]:
-    """Translate the EQ / IN / NE filters we use into a parameterized SQL
-    clause on vec0 metadata columns. Returns ("", []) when there is nothing
-    to filter. document_id is vec0's only filterable column and is INTEGER;
-    every value is coerced via int() here so callers (which today still pass
-    strings in places, e.g. indexing.py's MetadataFilter construction) don't
-    have to be individually correct -- vec0 doesn't coerce types itself.
+def _build_where(
+    conn: sqlite3.Connection,
+    filters: MetadataFilters | None,
+) -> tuple[str, list[int]]:
+    """Translate the EQ / IN / NIN / NE filters we use into a parameterized
+    SQL clause on vec0 metadata columns. Returns ("", []) when there is
+    nothing to filter. document_id is vec0's only filterable column and is
+    INTEGER; every value is coerced via int() here so callers (which today
+    still pass strings in places, e.g. indexing.py's MetadataFilter
+    construction) don't have to be individually correct -- vec0 doesn't
+    coerce types itself.
+
+    ``conn`` is only used for an IN/NOT IN filter over _MAX_IN_VALUES: it
+    loads the ids into PermittedIdsTable's TEMP TABLE on that connection
+    rather than binding them as SQL parameters.
    """
    if filters is None or not filters.filters:
        return "", []
@@ -125,28 +132,28 @@ def _build_where(filters: MetadataFilters | None) -> tuple[str, list[int]]:
            continue
        if f.key not in _FILTER_COLUMNS:  # pragma: no cover - we build the keys
            raise NotImplementedError(f"Unsupported filter column: {f.key}")
-        if f.operator == FilterOperator.IN:
+        if f.operator in (FilterOperator.IN, FilterOperator.NIN):
+            is_in = f.operator == FilterOperator.IN
+            sql_op = "IN" if is_in else "NOT IN"
            values = [int(v) for v in f.value]  # type: ignore[union-attr]
-            if not values:  # pragma: no cover
-                clauses.append("1 = 0")
+            if not values:
+                # An empty IN list matches nothing; an empty NOT IN list
+                # excludes nothing, so it matches everything.
+                clauses.append("1 = 0" if is_in else "1 = 1")
                continue
            if len(values) > _MAX_IN_VALUES:
-                # Fail closed (see the empty-clauses case below) rather than
-                # let SQLite raise "too many SQL variables" past its own
-                # limit: this filter scopes document access, so an IN list
-                # too large to safely bind must match no rows, never widen
-                # the scope to "everything" by accident.
-                logger.warning(
-                    "Refusing to build an IN filter on %r with %d values "
-                    "(over the %d-value safety limit); returning no rows.",
-                    f.key,
-                    len(values),
-                    _MAX_IN_VALUES,
+                # A literal list this large would exceed SQLite's own
+                # bound-parameter limit. Load the ids into a TEMP TABLE on
+                # this connection instead and filter via subquery, which has
+                # no such limit -- see PermittedIdsTable. Applies to NOT IN
+                # too (e.g. an install with an enormous trash), not just IN.
+                PermittedIdsTable.load(conn, values)
+                clauses.append(
+                    f"{f.key} {sql_op} (SELECT id FROM {PermittedIdsTable.TABLE_NAME})",
                )
-                clauses.append("1 = 0")
                continue
            placeholders = ",".join("?" for _ in values)
-            clauses.append(f"{f.key} IN ({placeholders})")
+            clauses.append(f"{f.key} {sql_op} ({placeholders})")
            params.extend(values)
        elif f.operator == FilterOperator.EQ:
            clauses.append(f"{f.key} = ?")
@@ -154,7 +161,7 @@ def _build_where(filters: MetadataFilters | None) -> tuple[str, list[int]]:
        elif f.operator == FilterOperator.NE:
            clauses.append(f"{f.key} != ?")
            params.append(int(f.value))
-        else:  # pragma: no cover - we only ever build EQ/IN/NE filters
+        else:  # pragma: no cover - we only ever build EQ/IN/NIN/NE filters
            raise NotImplementedError(f"Unsupported filter operator: {f.operator}")
    if not clauses:
        # Filters were requested but none could be translated. Fail closed
@@ -482,7 +489,7 @@ class PaperlessSqliteVecVectorStore(BasePydanticVectorStore):
            )
        if not self.table_exists():
            return []
-        where, params = _build_where(filters)
+        where, params = _build_where(self._conn, filters)
        sql = "SELECT node_content, embedding FROM " + DEFAULT_TABLE_NAME
        if where:
            sql += " WHERE " + where
@@ -498,7 +505,7 @@ class PaperlessSqliteVecVectorStore(BasePydanticVectorStore):
        if query.query_embedding is None:  # pragma: no cover
            return VectorStoreQueryResult(nodes=[], similarities=[], ids=[])
        top_k = query.similarity_top_k if query.similarity_top_k is not None else 10
-        where, params = _build_where(query.filters)
+        where, params = _build_where(self._conn, query.filters)
        sql = (
            "SELECT id, node_content, embedding, distance FROM "
            + DEFAULT_TABLE_NAME
@@ -7,6 +7,7 @@ import ssl
 import tempfile
 import traceback
 import unicodedata
+from datetime import date
 from datetime import timedelta
 from fnmatch import fnmatch
 from pathlib import Path
@@ -405,7 +406,7 @@ def make_criterias(rule: MailRule, *, supports_gmail_labels: bool):
    Returns criteria to be applied to MailBox.fetch for the given rule.
    """

-    maximum_age = timezone.localdate() - timedelta(days=rule.maximum_age)
+    maximum_age = date.today() - timedelta(days=rule.maximum_age)
    criterias = {}
    if rule.maximum_age > 0:
        criterias["date_gte"] = maximum_age
@@ -722,9 +723,9 @@ class MailAccountHandler(LoggingMixin):
                                f"Rule {rule}: Stopping processing rules due to stop_processing flag",
                            )
                            break
-                    except Exception:
+                    except Exception as e:
                        self.log.exception(
-                            f"Rule {rule}: Error while processing rule",
+                            f"Rule {rule}: Error while processing rule: {e}",
                        )
        except MailError:
            raise
@@ -766,8 +767,8 @@ class MailAccountHandler(LoggingMixin):
                    self.log.info(f"Located folder: {folder_info.name}")
            except Exception as e:
                self.log.error(
-                    "Exception during folder listing, unable to provide list folders: %s",
-                    str(e),
+                    "Exception during folder listing, unable to provide list folders: "
+                    + str(e),
                )

            raise MailError(
@@ -873,9 +874,9 @@ class MailAccountHandler(LoggingMixin):

                total_processed_files += processed_files
                mails_processed += 1
-            except Exception:
+            except Exception as e:
                self.log.exception(
-                    f"Rule {rule}: Error while processing mail {message.uid}",
+                    f"Rule {rule}: Error while processing mail {message.uid}: {e}",
                )

        self.log.debug(f"Rule {rule}: Processed {mails_processed} matching mail(s)")
@@ -11,10 +11,6 @@ from imap_tools import MailMessage
 from documents.loggers import LoggingMixin


-class MailDecryptionError(Exception):
-    pass
-
-
 class MailMessagePreprocessor(abc.ABC):
    """
    Defines the interface for preprocessors that alter messages before they are handled in MailAccountHandler
@@ -73,7 +69,7 @@ class MailMessageDecryptor(MailMessagePreprocessor, LoggingMixin):
                f"Message decryption failed with status message "
                f"{decrypted_raw_message.status}",
            )
-            raise MailDecryptionError(
+            raise Exception(
                f"Decryption failed: {decrypted_raw_message.status}, {decrypted_raw_message.stderr}",
            )
        self.log.debug("Message decrypted successfully.")
@@ -50,7 +50,7 @@ class ProcessedMailFactory(DjangoModelFactory[ProcessedMail]):

    rule = factory.SubFactory(MailRuleFactory)
    folder = "INBOX"
-    uid = factory.Sequence(str)
+    uid = factory.Sequence(lambda n: str(n))
    subject = factory.Faker("sentence", nb_words=4)
    received = factory.LazyFunction(timezone.now)
    processed = factory.LazyFunction(timezone.now)
@@ -214,7 +214,7 @@ class BogusMailBox(AbstractContextManager):
            )
            self.messages = list(filter(lambda m: m.uid not in uid_list, self.messages))
        else:
-            raise Exception  # noqa: TRY002 - test double simulating a generic mailbox failure
+            raise Exception


 def fake_magic_from_buffer(buffer, *, mime=False):
@@ -14,7 +14,6 @@ from imap_tools import MailMessage

 from paperless_mail.mail import MailAccountHandler
 from paperless_mail.models import MailRule
-from paperless_mail.preprocessor import MailDecryptionError
 from paperless_mail.preprocessor import MailMessageDecryptor
 from paperless_mail.tests.factories import MailAccountFactory
 from paperless_mail.tests.test_mail import TestMail
@@ -83,9 +82,7 @@ class MessageEncryptor:
            armor=True,
        )
        if not encrypted_data.ok:
-            raise Exception(  # noqa: TRY002 - test fixture setup, not production code
-                f"Encryption failed: {encrypted_data.stderr}",
-            )
+            raise Exception(f"Encryption failed: {encrypted_data.stderr}")
        encrypted_email_content = encrypted_data.data

        new_email = MIMEMultipart("encrypted", protocol="application/pgp-encrypted")
@@ -187,11 +184,7 @@ class TestMailMessageGpgDecryptor(TestMail):
                EMAIL_GNUPG_HOME=empty_gpg_home,
            ):
                message_decryptor = MailMessageDecryptor()
-                self.assertRaises(
-                    MailDecryptionError,
-                    message_decryptor.run,
-                    encrypted_message,
-                )
+                self.assertRaises(Exception, message_decryptor.run, encrypted_message)
        finally:
            # Clean up the temporary GPG home used only by this test
            try:
@@ -1,3 +1,4 @@
+import datetime
 import logging
 from datetime import timedelta
 from http import HTTPStatus
@@ -86,7 +87,7 @@ class MailAccountViewSet(PassUserMixin, ModelViewSet[MailAccount]):
    @action(methods=["post"], detail=False)
    def test(self, request):
        logger = logging.getLogger("paperless_mail")
-        request.data["name"] = timezone.now().isoformat()
+        request.data["name"] = datetime.datetime.now().isoformat()
        serializer = self.get_serializer(data=request.data)
        serializer.is_valid(raise_exception=True)
        existing_account = None
Author	SHA1	Message	Date
stumpylog	2f440cade2	Handles the rebase from the trash changes	2026-09-02 10:11:23 -07:00
stumpylog	61397749b7	If the filter will exceed SQLite IN limits, load to a temporary table and use a subquery to filter instead	2026-09-02 10:07:13 -07:00
stumpylog	3666234f51	When a user is unrestricted chatting, still exclude trashed documents using a 'NOT IN' SQL statement. Wire that up where we need it	2026-09-02 10:00:36 -07:00
stumpylog	ec7745e71f	Minor improvements from a Claude review	2026-09-02 08:14:41 -07:00
stumpylog	98e87d91ad	Fix: skip vector store document id filter for unrestricted chat users ChatStreamingView built an IN filter from every permitted document id for the "chat over all documents" case, which exceeds the vector store's SQLite bound-parameter safety limit on installs with more than ~32700 documents, silently returning no context. For a user who can see every document (an active superuser), that filter never narrows anything, so skip it and let the retriever search the whole index instead.	2026-09-02 08:14:23 -07:00
GitHub Actions	912c6eb52e	Auto translate strings	2026-09-01 22:07:30 +00:00
shamoon	73ef14f37a	Fix/chore: refactor some signal-backed conversion technical debt (#13902 )	2026-09-01 15:05:58 -07:00