Commit Graph
100 Commits
Author SHA1 Message Date
G30 cce3b68265 fix: enforce a single open user profile preview across ProfilePreview instances (#27578) 2026-07-27 01:21:57 -04:00
G30 085d11eef2 chore: drop redundant background repaints so surfaces inherit their parent (#27576)
* chore: drop redundant background repaints so surfaces inherit their parent

Four spots repaint the exact color their parent surface already provides
(bg-white / dark:bg-gray-900 rows inside same-colored pages and modals,
and the selectClass dark repaint inside the connection modals — the
sibling input const is already fully transparent). Visually identical in
stock light and dark; removing them lets instance theming show through
instead of leaving opaque boxes:

- .tiptap tr (app.css) — table rows in notes/editors
- Edit User Group Users tab body rows (common Modal surface)
- AddToolServerModal + AddTerminalServerModal selectClass dark repaint

The matching repaints inside the ModelUsage/UserUsage components are
not part of this change — those files were dead code and were removed
entirely in #27574.

* chore: catch remaining redundant surface repaints missed in the first pass

Same rule as the previous commit — every one of these repaints the exact
color its parent surface already provides, so removal is stock-identical
in light and dark while letting instance theming show through:

- Analytics Dashboard's inline Model Usage / User Activity row markup
  (the Analytics tab renders these tables from Dashboard.svelte itself;
  the unreferenced ModelUsage/UserUsage component files were removed
  in #27574)
- Evaluations Feedbacks + Leaderboard body rows (settings modal surface)
- admin UserList body rows (app page surface)
- chat markdown tables (MarkdownTokens): thead and body rows — unlike
  the tiptap header (gray-850 contrast, untouched), this thead painted
  the page's own color
- CitationsModal source rows (common Modal surface)
- AddConnectionModal selectClass dark repaint — third copy of the same
  const already fixed in AddToolServerModal / AddTerminalServerModal
2026-07-27 00:44:51 -04:00
G30 8295f2dacc chore: remove dead admin Analytics ModelUsage and UserUsage components (#27574)
Nothing in the tree imports either component; the admin Analytics tab
renders its own inline copies of both tables directly from
Dashboard.svelte. Both files landed with the dashboard in a4ad34841
(feat: analytics frontend dashboard) but were never wired into it.

The remaining name matches elsewhere (the getUserUsage API and
UserUsage* types in src/lib/apis/users/index.ts, consumed by
chat/Settings/Usage.svelte, plus the backend usage endpoints) belong to
the unrelated per-user usage feature and are untouched.
2026-07-27 00:35:08 -04:00
G30andClaude Fable 5 cb64068893 feat: add default file upload mode user setting (#20900)
* feat: add default upload mode setting

Add user setting to configure the default upload mode for files, allowing users to choose between "Using Entire Document" (full context) and "Using Focused Retrieval" (RAG processing) as the default behavior.

* i18n: sync locale catalogs for the new upload mode strings

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore: re-trigger CI (previous run hit the pre-existing Node heap OOM, see #27254)

* fix: apply the default upload mode at upload time so the payload carries it

The previous approach only pre-set the modal toggle's visual state on
mount; item.context is written solely by the Switch's on:change, so the
sent files kept context: undefined and the backend never saw 'full'. It
also showed a misleading ON state for legacy context-less files, since
FileItemModal mounts with every FileItem chip render.

Stamp context on the fileItem in uploadFileHandler instead (before
...itemData, so callers passing an explicit context still win) and
revert the FileItemModal hunk — the modal already renders from
item.context alone.

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-07-26 22:50:03 -05:00
G30 bda49ccdb6 fix(ui): close sidebar on mobile when opening Calendar from user menu (#26979)
Every other navigation entry in the user menu (Settings, Admin Panel,
Archived Chats, Workspace, Notes, Automations, Playground, Sign Out)
collapses the sidebar on mobile after navigating, but the Calendar entry
was missing this handling, leaving the sidebar open over the Calendar
page on mobile. Add the same mobile guard used by the sibling entries.
2026-07-26 18:47:12 -04:00
G30 11d72c1ce2 fix(ui): replace history entry on redirect so browser back navigation works (#27478) 2026-07-26 18:44:23 -04:00
G30 79695a1d14 fix: persist modelIdx so duplicate side-by-side models don't collapse on reload (#26980)
When the same model is selected multiple times in a side-by-side chat,
each response is created with a distinct modelIdx (0,1,2,3) that
identifies its column. The backend now owns message persistence, but it
built the assistant placeholders without modelIdx, so the field was
never saved. On reload MultiResponseMessages groups responses by
modelIdx and falls back to grouping by model id when modelIdx is
missing; with duplicate models that fallback lumps every response into
each column, so all columns render the first response (and show a bogus
'1/N' pager).

Send modelIdx with each message_ids entry from the frontend and persist
it on the assistant placeholders in both the new-chat and existing-chat
paths. The message_ids list is now forwarded for every send (not just
multi-model ones) so single-column regenerations in a duplicate-model
chat also keep their column identity across reloads.
2026-07-26 18:44:03 -04:00
G30andClaude Opus 4.8 86bf927d08 fix(ui): stack automation modal footer on mobile (#27027)
The automation create/edit modal footer laid the schedule/model dropdowns
and the Cancel/Create actions in a single justify-between row. On narrow
modals the fixed-width actions left too little room for the dropdowns,
which wrapped to two stacked lines with Cancel squeezed in the middle.

Stack the footer vertically on mobile (dropdowns row, then a right-aligned
actions row) and restore the horizontal layout at sm and up.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-26 18:43:45 -04:00
G30 db92ef292f fix: unarchive chats moved into folders and refresh sidebar folders after menu moves (#27485) 2026-07-26 18:43:34 -04:00
G30 71f8b6d5b4 feat: add a master OAuth / OIDC enable toggle in Authentication settings (#26988)
The OAuth / OIDC section in Admin Settings > Authentication had no
enable/disable switch, unlike the LDAP section above it. Add one that
persists via the existing Save flow and actually gates OAuth sign-in,
mirroring how the LDAP toggle works.

- config: new ENABLE_OAUTH persistent config ('oauth.enable'), defaulting
  to True so existing deployments with a provider configured keep working.
- oauth: expose ENABLE_OAUTH via the OAuth runtime config and reject the
  login and callback handlers with 404 when it is disabled.
- /api/config: report no OAuth providers when disabled so the login page
  hides the OAuth buttons (and cannot auto-redirect), without clearing the
  admin's provider configuration.
- auths: expose ENABLE_OAUTH through the admin OAuth config get/update
  endpoints (OAuthConfigForm + OAUTH_CONFIG_KEYS).
- Authentication.svelte: bind the OAuth / OIDC header Switch to the
  persisted oauthConfig.ENABLE_OAUTH and collapse the section when off,
  matching the LDAP header (size, weight, alignment).
2026-07-26 18:43:21 -04:00
G30 771540f3de feat: sortable title and updated at columns on the knowledge workspace page (#27457) 2026-07-26 18:19:31 -04:00
G30 c9c97835bf fix(ui): keep dropdowns positioned and stable as their content resizes (#27460) 2026-07-26 18:17:53 -04:00
G30 527b0d3b7f fix(ui): stop the chat input attach menus from fetching twice on open (#27461) 2026-07-26 18:14:06 -04:00
G30 202f47ece8 fix(ui): stop sidebar chat rows flickering while the pointer moves across them (#27474) 2026-07-26 17:57:44 -04:00
G30 878cebac07 fix(ui): persist ollama connection deletions immediately (#27483) 2026-07-26 17:38:49 -04:00
G30 4ac22b89fd fix: don't wipe sidebar folder chat list when a refresh overlaps an in-flight fetch (#27535) 2026-07-26 17:28:15 -04:00
G30 29499cb4ba perf: dedupe folder refetches and chat-list sweeps on sidebar folder selection (#27540) 2026-07-26 17:27:33 -04:00
G30 4d576c1aa2 chore: remove no-op stopPropagation statement from folder title click handler (#27542) 2026-07-26 17:27:12 -04:00
G30 65209b0235 fix: enforce a single open chat hover preview across sidebar chat items (#27549) 2026-07-26 17:26:49 -04:00
G30 ef197de0d7 fix: open settings deep links on client-side navigation and open Add Terminal settings directly (#27552) 2026-07-26 17:26:34 -04:00
G30 06d2189b26 fix(chat): keep sidebar chat selection in sync with the active chat (#26977)
When navigating from a chat to a non-chat route (e.g. the admin panel),
the previously-viewed chat stayed selected in the sidebar and
deleting/archiving it wrongly redirected back to the new-chat page.
Cloning a chat also left the source chat highlighted alongside the new
clone, so two chats appeared selected at once.

Two independent sources kept the stale selection:

- The chatId store was never cleared when the Chat component unmounted,
  so $chatId still pointed at the last-viewed chat (this drove the
  delete/archive redirect). Clear chatId/chatTitle in Chat's onDestroy.
- The sidebar's optimistic selectedChatId highlight, set on click, was
  only cleared on window blur (hence it appeared to fix itself after a
  tab switch) and never followed programmatic navigation. Bind it to the
  chatId store so it tracks the active chat for leave, delete and clone.
2026-07-24 01:45:37 -05:00
G30 e212e3c7f4 fix(ui): resolve workspace model ID to base model on delete from model selector (#26819) 2026-07-24 00:03:31 -05:00
G30 1dc4fd3e9d fix(ui): hide redundant edit pencil on mobile in admin Models list (#27178)
On narrow viewports the admin Models list rows previously collapsed into
unusable vertical stacks; the Models page redesign on dev has since
absorbed the truncation fixes this branch carried (min-w-0 chain, real
truncate on the name, shrink-0 action group, inline access label).

The one remaining gap: the per-row edit pencil duplicates the row tap
(both open the model editor) while costing scarce horizontal space on
mobile. Hide it below the sm breakpoint; it remains on sm+ screens.
2026-07-23 23:58:07 -05:00
G30andClaude Fable 5 866e8582d1 ci: auto-label bug reports created outside the issue form (#27256)
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-07-23 23:07:25 -04:00
G30 3132f11e55 feat: add setting to toggle keyboard shortcuts on/off (#27300) 2026-07-23 22:14:56 -04:00
G30 4f823774ab fix: migrate deprecated authlib.jose import to joserfc in oauth utils (#27310) 2026-07-23 21:54:59 -04:00
G30 4fca375ad4 chore: remove unused ArchivedChatsModal component (#27308) 2026-07-23 21:54:45 -04:00
G30 e62e4eb9fe fix(ui): contain archived chats and analytics content within the settings modal (#27306) 2026-07-23 13:34:35 -05:00
G30 c584a4270c perf(ui): non-blocking model and tool server loading for unreachable connections (#26289)
- Move setModels() and setToolServers() out of the loaded gate in
  +layout.svelte so the page renders immediately instead of blocking
  behind slow/unreachable connection timeouts
- Track failed Ollama backend URLs from /api/tags and skip them in the
  subsequent /api/ps fan-out, eliminating a redundant second timeout
- Silence the autocomplete toast that fires during the transient empty
  model state
2026-06-29 12:34:00 -05:00
G30 6fd8c599c1 fix(ui): workspace models bulk actions and search pagination (#26274) 2026-06-29 12:14:15 -05:00
G30 dc1bc41d2e fix(ui): fix pinned note sidebar drag-and-drop into chat context (#26384) 2026-06-29 12:05:39 -05:00
G30 8c7428122b fix(rag): allow special tokens in document text during chunk measurement (#26210)
Documents (especially AI/LLM documentation) legitimately contain special token strings like <|endoftext|> as literal text. The tiktoken encoder raises a ValueError when encountering these during chunk size measurement in merge_docs_to_target_size(), preventing the entire file from being indexed. Pass disallowed_special=() to encoding.encode() to treat all text as normal content.
2026-06-29 10:43:24 -05:00
G30 5ffd4e53c3 fix(ui): prevent long default group name from overflowing admin settings layout (#25685) 2026-06-29 04:59:01 -05:00
G30 4fa3a74827 fix(chat): prevent sortable sidebar drags from triggering file upload overlay (#25675)
Add a custom MIME type (application/x-open-webui-drag) to intentional
chat and folder drag sources. The onDragOver handler in MessageInput
now checks for this type instead of the generic text/plain, which
SortableJS also sets during reorder operations for pinned menu items
(Notes, Workspace) and pinned Models.
2026-06-29 04:58:30 -05:00
G30andTim Baek f53ec857c0 feat: support folder, note, and model drag-and-drop from sidebar as context references (#25771)
* feat(ui): add drag-to-reorder for pinned notes in sidebar

* feat(chat): support folder, note, and model drag-and-drop from sidebar as context references

---------

Co-authored-by: Tim Baek <tim@openwebui.com>
2026-06-29 04:41:16 -05:00
G30 ea6d02da0f fix(ui): use correct field name for prompt clone title suffix (#25800) 2026-06-29 04:25:21 -05:00
G30 ee11069ef2 perf(ollama): offload multi-GB model file I/O with asyncio.to_thread (#25829)
The Ollama model upload and download handlers perform three stages
of sync file I/O on multi-GB model files inside async handlers:

1. Persist upload — file.file.read() + write() in a loop
2. SHA-256 hash — calculate_sha256() reads the file sequentially
3. Read for blob push — open().read() loads entire model into memory

All three stages block the event loop for the duration of the I/O.
For a 4GB model, each stage freezes the event loop for 10+ seconds
while every other user's request stalls.

Wrap each blocking stage in asyncio.to_thread() in both handlers:

- upload_model(): persist upload, calculate_sha256, blob read
- download_file_stream(): calculate_sha256, blob read

Benchmark (full pipeline: write + SHA-256 + read back, 3 trials):
- 200MB: max jitter 145ms → 1ms (145x improvement)
- 500MB: max jitter 1,026ms → 1ms (1,026x improvement)

File I/O is pure I/O — no GIL contention. asyncio.to_thread()
completely eliminates event loop blocking.
2026-06-29 03:48:48 -05:00
G30 b05be8a907 fix(notes): refresh pinned notes sidebar when a note is deleted (#25640) 2026-06-29 03:48:22 -05:00
G30 7b1aa749eb perf(retrieval): make URL fetch and file hash non-blocking with asyncio.to_thread (#25822)
- Wrap get_content_from_url() in asyncio.to_thread() inside
  get_sources_from_items() to prevent sync requests.get() from blocking
  the event loop when users attach URL sources to chat messages.
  The same function was already properly wrapped at retrieval.py:1839.

- Wrap hashlib.sha256(contents).hexdigest() in asyncio.to_thread()
  inside upload_file_handler() to prevent CPU-bound hashing from
  blocking the event loop during file uploads (44ms/100MB, scales
  linearly with file size).
2026-06-29 03:46:24 -05:00
G30 a8e5f0a54d fix(ui): prevent space key from opening file/directory during rename in knowledge base (#25627) 2026-06-29 03:37:09 -05:00
G30 ed2d7d4acd fix(i18n): standardize title case for settings toggle labels (#25765) 2026-06-29 03:30:40 -05:00
G30 70b89d01c2 feat(ui): add drag-to-reorder for pinned notes in sidebar (#25677) 2026-06-29 03:22:33 -05:00
G30 78d276b4ff fix(ui): update groups count to reflect filtered search results (#25689) 2026-06-29 03:21:31 -05:00
G30 6e14e446cb fix(api): handle orphaned shared_chat rows when unsharing (#25632) 2026-06-29 03:20:46 -05:00
G30 4712544d5e feat(search): add chat context menu to search modal (#25490) 2026-06-29 03:16:03 -05:00
G30 525c115a3a fix(ui): remove unnecessary margin causing scrollbar on About settings page (#25802) 2026-06-29 02:54:00 -05:00
G30andTim Baek 6fdf9b4340 perf(auth): make password hashing non-blocking and batch CSV user import (#25804)
Co-authored-by: Tim Baek <tim@openwebui.com>
2026-06-29 02:45:39 -05:00
G30 18e6a10778 fix(ui): update model share URL to new community post endpoint (#25801) 2026-06-29 02:44:28 -05:00
G30andTim Baek 6ea591491e perf(images): offload validate_url() DNS resolution with asyncio.to_thread (#25825)
validate_url() calls socket.getaddrinfo() for SSRF protection, which
blocks the event loop for 100-700ms per DNS lookup. This affects:

- Image generation (get_image_data) — every external image URL
- Image editing (load_url_image) — every external image URL
- OAuth profile pictures (_process_picture_url) — every login
- Webhook delivery (post_webhook) — every notification
- Image base64 conversion (get_image_base64_from_url) — chat images

Wrap all 5 async call sites in asyncio.to_thread() so DNS resolution
runs in the thread pool. The event loop remains free to serve other
requests during the lookup.

Benchmark (3 domains, 3 trials averaged):
- BEFORE: max jitter 479ms, 1 blocked ping per trial
- AFTER:  max jitter 1ms, 0 blocked pings (324x improvement)

Co-authored-by: Tim Baek <tim@openwebui.com>
2026-06-29 02:31:24 -05:00
G30 71d6212ab8 fix(redis): use await asyncio.sleep() instead of time.sleep() in async generator (#25823)
time.sleep() inside the _wrap_async_gen() async generator blocks the
entire event loop during Sentinel failover retries. The sibling method
_wrap_async_call() already correctly uses await asyncio.sleep().
2026-06-29 02:25:46 -05:00
G30 ff25ccca65 feat: add repeat/recurrence dropdown to calendar event modal (#25865)
* feat(calendar): add repeat/recurrence dropdown to event modal

* fix(calendar): anchor recurring event expansion to event start date

The expand_recurring_event() utility used the view range start as dtstart,
causing FREQ=WEEKLY events to land on the wrong day of the week. Use the
event's original start date instead so recurrence patterns stay correct.
2026-06-29 02:16:19 -05:00
G30 342539f1e1 perf(audio): make ML model loading non-blocking with asyncio.to_thread (#25806) 2026-06-29 02:14:15 -05:00
G30 964a098a4b fix(chat): use code block language tag to detect Vega-Lite specs for rendering (#25843) 2026-06-29 02:07:11 -05:00
G30 e69ce6e1c6 perf(channels): batch N+1 queries for reactions and thread replies (#25831)
Replace per-message database queries with batch IN-clause queries in
channel message handlers. This eliminates the N+1 query pattern that
caused ~102 queries per channel page load (50 messages × 2 queries each).

Changes:
- Add get_reactions_by_message_ids() to MessageTable: single query
  fetches all reactions for multiple messages using IN clause with
  User JOIN, returns dict[message_id, list[Reactions]]
- Add get_thread_reply_counts_by_message_ids() to MessageTable: single
  GROUP BY aggregate query returns (count, max_created_at) per parent,
  replacing full object loads just to call len()
- Refactor get_channel_messages(): 102 → 4 queries per page
- Refactor get_pinned_channel_messages(): 22 → 3 queries per page
- Refactor get_channel_thread_messages(): 53 → 4 queries per page
- Refactor send_notification(): N+1 membership check → batch set lookup
2026-06-29 02:05:16 -05:00
G30 2a9a023172 fix(ui): add Edit option to prompts 3-dots menu (#25789) 2026-06-29 02:02:39 -05:00
G30 f923edcaaa feat(ui): add Clone option to automations 3-dots menu (#25790) 2026-06-29 02:00:23 -05:00
G30 3c67774eb3 fix(auth): enforce settings.interface permission on /user/settings/update endpoint (#25996) 2026-06-29 01:53:20 -05:00
G30 677e164f29 feat(permissions): add workspace.skills_import and workspace.skills_export permissions (#25921) 2026-06-29 01:36:27 -05:00
G30 da0fd0da0d fix(ui): add max height and scroll to chat menu tags section (#26031) 2026-06-29 01:36:03 -05:00
G30 0502494e9a fix(ui): fix keyboard shortcuts modal Enter key emoji and menu capitalization (#26173)
- Append Unicode text variation selector U+FE0E to the ↩ character in
  ShortcutItem.svelte to prevent iOS/iPadOS rendering it as a colorful
  emoji instead of a plain text glyph.
- Fix 'Keyboard shortcuts' → 'Keyboard Shortcuts' (capital S) in the
  user menu to match the modal title and standard title case.
2026-06-29 00:59:07 -05:00
G30 08f64f7908 fix(calendar): prevent all-day event date drift on save (#25864) 2026-06-25 18:40:25 -04:00
G30 b2809e6293 perf(ui): hoist marked.use() to module scope to prevent extension stacking (#25837) 2026-06-17 03:09:09 +02:00
G30 beb9bf60e4 feat(memory): default user memory toggle to admin's global ENABLE_MEMORIES config (#25909) 2026-06-17 03:08:43 +02:00
G30 2f9b28a57d fix(ui): run initChatList API calls concurrently in Promise.all (#25838) 2026-06-17 03:08:02 +02:00
G30 819ad1d904 fix(chat): convert recursive findClosingDelimiter to iterative loop to prevent stack overflow (#25845) 2026-06-17 03:07:12 +02:00
G30 9fe3a00dba perf(chat): cache KaTeX rendered HTML in reactive statement to avoid redundant renders (#25847) 2026-06-17 03:06:56 +02:00
G30 4584adf900 feat(ui): add 'Unshare All Shared Chats' button to Shared Chats modal (#25848) 2026-06-17 03:06:41 +02:00
G30 dfdb76cc46 fix(ui): prevent Code Editor drawer from collapsing when content is empty (#25855) 2026-06-17 03:06:15 +02:00
G30 dee07d8a30 feat(ui): show total archived chat count in ChatsModal title (#25872) 2026-06-17 02:57:27 +02:00
G30 b4aef82401 feat(ui): show total file count in FilesModal title (#25873) 2026-06-17 02:55:13 +02:00
G30 57a5e43696 feat(permissions): add per-group features.user_webhooks permission (#25923) 2026-06-17 00:47:44 +02:00
G30 2b9b956dad fix(notes): sync pinned note state in sidebar and editor menu (#25918) 2026-06-17 00:42:20 +02:00
G30 25090dbf17 fix(chat): prevent false-positive unread indicators from title generation, pin/archive/folder actions, and new chat creation (#25912) 2026-06-17 00:41:44 +02:00
G30 bb2663aad1 fix(ui): gate all chat export formats by chat.export permission (#25914) 2026-06-17 00:41:18 +02:00
G30 3571db34e2 fix(ui): enforce allow_users permission in share chat modal (#25915) 2026-06-17 00:40:57 +02:00
G30 0726db7217 fix(analytics): remove duplicate onMount causing double API requests on dashboard load (#25935) 2026-06-17 00:29:07 +02:00
G30 9c6c375dfe fix(images): add null guard for config in onMount to prevent crash on API failure (#25933) 2026-06-17 00:28:42 +02:00
G30 3e3c5b6d78 fix(permissions): add admin bypass to memories permission checks (#25924) 2026-06-17 00:28:17 +02:00
G30 2bc91e8f52 fix(ui): add missing skills permission to workspace menu visibility check (#25925) 2026-06-17 00:27:25 +02:00
G30 e5ed45fb20 fix(ui): handle null chat in folder drop handler for cross-account drag (#25928) 2026-06-17 00:26:35 +02:00
G30 6b2d962cd6 feat(permissions): add reset to defaults button in permissions modals (#25931)
Adds a 'Reset to Defaults' button to both the Edit Default Permissions
modal and the Edit User Group permissions modal.

- Default permissions modal: resets to stock/env-var configuration
- User group modal: resets to current global default permissions

Backend: new GET /api/v1/users/default/permissions/defaults endpoint
that returns DEFAULT_USER_PERMISSIONS (env-var-based initial defaults).
2026-06-17 00:25:20 +02:00
G30 6f2e97aa58 fix(evaluations): prevent duplicate leaderboard API requests on navigation (#25934) 2026-06-17 00:16:27 +02:00
G30 966b3fdb57 fix(ui): deduplicate layout-level API requests (#25942) 2026-06-17 00:09:50 +02:00
G30 ee3a49a88d fix(ui): deduplicate store fetches with null-guards and mutex (#25943) 2026-06-17 00:07:11 +02:00
G30 1eecbc1ac2 fix(ui): collapse Code Execution settings when toggle is disabled (#25970) 2026-06-16 23:01:22 +02:00
G30 b93d560788 feat(ui): pre-select last model when adding a new model slot (#25974) 2026-06-16 22:56:56 +02:00
G30 84e4110538 fix(auth): enforce chat.tts permission on OpenAI /audio/speech proxy endpoint (#25993) 2026-06-16 22:54:35 +02:00
G30 bb66d435b7 fix(auth): enforce features.api_keys permission on GET and DELETE /api_key endpoints (#25992) 2026-06-16 22:48:44 +02:00
G30 1010fd3142 fix(knowledge): remove premature drag-and-drop upload toast (#25484) 2026-06-01 11:09:58 -07:00
G30 ceb1cbc009 fix(ui): guard JSON.parse(localStorage) calls with try/catch to prevent UI crashes (#25481) 2026-06-01 10:33:54 -07:00
G30 229e65b9f0 fix(ui): correct inverted high-contrast text colors for user message timestamp (#25461) 2026-06-01 09:26:18 -07:00
G30 bbd3e13094 fix(ui): use correct 'blur' event name instead of 'blur-sm' in window listeners (#25459) 2026-06-01 09:25:55 -07:00
G30 18b1a041dd fix(ui): enable custom parameters in user settings and admin model settings (#25200)
Add missing custom={true} prop to AdvancedParams in General.svelte and
ModelSettingsModal.svelte so the 'Add Custom Parameter' option appears
consistently across all advanced parameter surfaces.

Also forward custom_params in the General.svelte save handler so custom
parameters are persisted instead of silently dropped on save.
2026-05-31 15:02:32 -07:00
G30 1608a04fa9 fix(settings): correct presence_penalty and repeat_penalty saving wrong values (#25183)
Both presence_penalty and repeat_penalty in the saveHandler read from
params.frequency_penalty instead of their own values due to a
copy-paste error. This causes users adjusting either parameter to
silently save the frequency_penalty value instead.
2026-05-31 15:02:07 -07:00
G30 5112e0b62c fix: add null guards to channel Thread and PinnedMessagesModal components (#25209)
Thread.svelte: Add null check for messagesContainerElement in scrollToBottom()
to match the existing pattern in Channel.svelte. Prevents potential TypeError
when the DOM element is not yet bound during rapid thread switches.

PinnedMessagesModal.svelte: Move res.length check inside the if (res) block.
Previously, res.length was accessed unconditionally after a guarded block,
causing TypeError when the API call fails and the .catch() returns null.
2026-05-31 15:01:17 -07:00
G30 34b8844ccf fix(models): gracefully handle legacy svg profile_image_url in ModelMeta validator (#25173)
The SVG-XSS hardening introduced in f5f4b5895 correctly rejects
data:image/svg+xml URIs on new input, but also caused a
pydantic_core.ValidationError when reading pre-existing models from
the database that had SVG data URIs stored as their profile images.

This ValidationError propagated unhandled through _to_model_model and
get_all_models, crashing the entire /api/models endpoint with HTTP 500
and leaving users with no models available in the UI.

Fix:
- Wrap validate_profile_image_url() in a try/except ValueError inside
  ModelMeta.check_profile_image_url. Legacy entries are cleared to None
  with a warning log instead of raising — the /model/profile/image API
  endpoint already falls back to /static/favicon.png when the value is
  empty.
- Default ModelMeta.profile_image_url to None instead of hardcoding
  /static/favicon.png, since the serving endpoint handles the fallback.
- Add a per-model try/except in ModelsTable.get_all_models so that any
  future unexpected validation failure on a single record skips that
  model with an error log rather than aborting the entire list.
2026-05-31 14:55:07 -07:00
G30 9c9e4f1288 fix(types): add missing markdown rendering settings to Settings type (#25198) 2026-05-31 14:52:35 -07:00
G30 26b1529a45 fix(ui): add voice mode mute shortcut to keyboard shortcuts modal (#25193) 2026-05-31 14:52:20 -07:00
G30 c665d4a7c6 fix(ui): prevent long usernames from overflowing Edit User modal, User Preview modal, and sidebar (#25185)
Long usernames overflow the Edit User modal, User Preview modal header,
and the sidebar user area because the flex containers lack width
constraints.

- EditUserModal: add min-w-0 to the flex-1 container so the existing
  truncate class takes effect
- UserPreviewModal: add min-w-0 and truncate to the title container,
  flex-shrink-0 to the close button so it stays visible
- Sidebar: add truncate to the username display and flex-shrink-0 to
  the avatar container to prevent it from being squeezed
2026-05-31 14:51:56 -07:00
G30 c428ad0c1a fix(ui): include reasoning_tags in user settings advanced params save handler (#25204) 2026-05-31 14:50:41 -07:00