Compare commits

..
Author SHA1 Message Date
Wolfsblvt 06bde939fb Fix npm release publishing workflow (#6037)
Allow npm releases to be retried from an existing release tag through the trusted publishing workflow, and scope the `allow-directory` workaround to the publish command.
2026-09-14 21:01:34 +02:00
Wolfsblvt 7e8663cd9c Merge pull request #6036 from SillyTavern/staging
1.19.0
2026-09-14 19:55:46 +02:00
Wolfsblvt 61139aa9b6 Update release version number (#6035) 2026-09-14 19:44:18 +02:00
Wolfsblvt 5e8d463ed4 feat: add Claude Fable 5.1 and GPT-6 Astra (#6015) 2026-09-14 19:03:39 +02:00
F.B. c0a417b240 Add image embed for DeepSeek V4 Flash Vision Exp (#5964) (#5996)
Add deepseek-v4-flash-vision-exp to image embed whitelist.
Block image embed in non `user` msg as DS doesn't allow this.
2026-09-07 23:09:17 +02:00
X00LAandClaude Sonnet 5 cc053e8356 Fix CORS proxy forwarding raw zstd-compressed bytes to the client (#6007)
undici (Node's fetch) transparently decompresses gzip, deflate, and br,
but not zstd. When a browser's full Accept-Encoding header is forwarded
as-is to the target origin, a Cloudflare-fronted server may respond
with zstd, and corsProxyMiddleware pipes those still-compressed bytes
straight through — the client receives garbage instead of the expected
body (e.g. "SyntaxError: Unexpected token" when parsing JSON).

Cap the outgoing Accept-Encoding to the encodings undici can actually
decode, so the target always responds with something we can forward
correctly.

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
2026-09-07 23:07:13 +02:00
Siyuan ChenandWolfsblvt d6b244cd25 Fix: Avoid cpSync crash when renaming characters (#6012)
* Fix: Avoid cpSync crash when renaming characters

Add a no-op filter to fs.cpSync during character renaming.

This avoids Node.js's native copyDir fast path, which can terminate
the process on Windows with affected Node.js versions instead of
throwing a catchable filesystem error.

Refs: https://github.com/nodejs/node/issues/63970

* docs: explain character rename copy workaround

---------

Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
2026-09-07 23:05:18 +02:00
voodoohop e7c4f8ee98 fix: resolve Pollinations TTS model aliases (#6010) 2026-09-07 23:03:43 +02:00
QiuLGandQiuLG a8f7e0d900 Fix Quick Reply overwrite cancellation (#6005)
Co-authored-by: QiuLG <195722592+QiuLsG@users.noreply.github.com>
2026-09-07 23:00:44 +02:00
saitewasreset 4613abc26e Google AI Studio: Fetch all model pages (#5954)
Paginate the Gemini models endpoint before returning available
generateContent models, and reject incomplete model lists.
2026-09-01 00:58:51 +02:00
Jeremy SchoemakerandWolfsblvt dd326d244f Return a degraded preview instead of dropping a chat whose last line is unparseable (#5969)
* fix: return a degraded preview instead of {} when a chat's last line is unparseable

getChatInfo resolved an empty object when the last line of a chat .jsonl
failed to parse or lacked known fields (truncated write, external edit).
All three consumers (chat list, search, recents) drop entries without a
file_name, so an otherwise intact chat silently disappeared from the UI
while still occupying its name for uniqueness checks (#5940).

Resolve the already-built chatData instead: file_id, file_name, file_size
and mtime-based last_mes are known good from stat(), chat_items comes from
the line count, and mes falls back to the existing placeholder. Matcher
results from the intact lines are preserved for search.

Fixes #5940

* fix: exclude unreadable chat line from message count

Keep degraded previews aligned with the messages that can actually be loaded by excluding both the metadata line and broken trailing line.

---------

Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
2026-09-01 00:38:40 +02:00
Zhewen Tan ccd8287833 fix: preserve zero NovelAI Math1 temperature (#5965) 2026-08-31 23:36:06 +02:00
Em 30eaf26a43 Update TavernAI link in README (#5987)
Update TavernAI link in README
2026-08-27 01:33:51 +02:00
PeterGriffin 98af17a8a4 fix(chat): delete associated tool calls with messages (#5949) 2026-08-21 06:35:50 +02:00
primorLee 0a9ba91852 fix: parse boolean values in env migrations (#5950) 2026-08-21 06:33:08 +02:00
Christoph 814734d4af Fix duplicate tool call IDs if provider repeats id, type, name (#5948)
Before this fix, tool call IDs were duplicated repeatedly with streaming requests if the provider repeats the "id" field in each streaming chunk:

```
{
    role: 'tool',
    content: 'content',
    tool_call_id: 'chatcmpl-tool-aa62chatcmpl-tool-aa62chatcmpl-tool-aa62chatcmpl-tool-aa62'
},
```

After this fix:

```
{
    role: 'tool',
    content: 'content',
    tool_call_id: 'chatcmpl-tool-aa62'
},
```

Example of raw streaming chunks received from Fireworks AI exhibiting this behavior of repeating "id", "type", "name":

```json
{
  "id": "chatcmpl-b7de",
  "object": "chat.completion.chunk",
  "created": 1783459695,
  "model": "accounts/fireworks/models/glm-5p2",
  "choices": [
    {
      "index": 0,
      "delta": {
        "tool_calls": [
          {
            "index": 0,
            "id": "chatcmpl-tool-aa62",
            "type": "function",
            "function": {
              "name": "debug_tool",
              "arguments": ""
            },
            "name": null
          }
        ]
      },
      "finish_reason": null,
      "raw_output": null
    }
  ],
  "usage": null
}
{
  "id": "chatcmpl-b7de",
  "object": "chat.completion.chunk",
  "created": 1783459695,
  "model": "accounts/fireworks/models/glm-5p2",
  "choices": [
    {
      "index": 0,
      "delta": {
        "tool_calls": [
          {
            "index": 0,
            "id": "chatcmpl-tool-aa62",
            "type": "function",
            "function": {
              "name": null,
              "arguments": "{\"input\": \""
            },
            "name": null
          }
        ]
      },
      "finish_reason": null,
      "raw_output": null
    }
  ],
  "usage": null
}
{
  "id": "chatcmpl-b7de",
  "object": "chat.completion.chunk",
  "created": 1783459695,
  "model": "accounts/fireworks/models/glm-5p2",
  "choices": [
    {
      "index": 0,
      "delta": {
        "tool_calls": [
          {
            "index": 0,
            "id": "chatcmpl-tool-aa62",
            "type": "function",
            "function": {
              "name": null,
              "arguments": "First"
            },
            "name": null
          }
        ]
      },
      "finish_reason": null,
      "raw_output": null
    }
  ],
  "usage": null
}
```
2026-08-21 06:16:41 +02:00
Christoph 8992872765 Fireworks: Request and process reasoning (#5947)
This change enables reasoning effort selection for the Fireworks provider as documented at https://docs.fireworks.ai/guides/reasoning#using-reasoning_effort.
2026-08-19 23:47:16 +02:00
Christoph 558d72dde2 Fireworks AI: Set session affinity header for improved caching (#5826)
This change adds a chat-specific "x-session-affinity" HTTP header to requests to Fireworks that use the full prompt. Quiet requests like summarization requests that don't use the same prompt prefix are excluded.

This improves prompt caching by
1. Adding a stable "x-session-affinity" header, which Fireworks AI recommends to route requests to the same replica
2. Separating different chats into different sessions
3. Separating main LLM calls from quiet calls which have a completely different prompt prefix

See https://docs.fireworks.ai/guides/prompt-caching.

This uses an HMAC with the cookie secret instead of simply hashing the chat title to avoid leaking chat titles. Session affinity headers may not fall under zero-data retention in Fireworks or could be cached by proxies on the way. Chat titles are tiny, so the HMAC computation is instant.
2026-08-19 23:25:15 +02:00
WolfsblvtandClaude Fable 5 4ad7af3c06 fix: use provider error fields instead of statusText for HTTP 200 errors (#5945)
Providers can return an HTTP 200 response whose body carries an error
object without a message field, or the error as a plain string. The
error message fell back to response.statusText, which is 'OK' for any
200 response, producing the opaque 'Error: OK'. Error messages now fall
back to the error's code or type field, then the string form, and only
use statusText for non-2xx responses where it carries meaning. Applies
to the non-streaming request path and both streaming error toasts.

Fixes #5647

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-19 21:05:04 +02:00
WolfsblvtandClaude Fable 5 bcc42de2e9 Give non-ASCII chat names distinct backup keys and throttles (#5944)
* fix: give non-ASCII chat names distinct backup keys and throttles

Backup filenames replace all non-alphanumeric characters with
underscores, so names such as equal-length CJK ones collapsed to the
same key and shared a single backup quota: busy chats evicted the
backups of unrelated quiet ones. The backup key now appends a short
hash of the raw name whenever the name contains non-ASCII characters,
keeping keys distinct and stable while ASCII names keep their exact
current keys and retention.

The backup throttle was also keyed per user only, so rapid saves in
one chat deferred or swallowed the backup of another chat saved in the
same window. It is now keyed per user and chat.

Fixes #5780

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

* fix: replace stray NUL byte in backup throttle key separator

A literal 0x00 byte slipped into the template string separator, which
made Git treat the whole file as binary.

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

* chore: reunite backupChat docblock with its function

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

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-19 20:49:43 +02:00
WolfsblvtandClaude Fable 5 b4e162324b fix: mint a fresh integrity slug for new branches and checkpoints (#5943)
Branches and checkpoints inherited the parent chat's integrity UUID
through the chat_metadata spread in saveChat and saveGroupBookmarkChat,
so every member of a branch family shared one slug and the integrity
guard could not tell them apart. A save aimed at the wrong file within
the family would pass validation. New branches and checkpoints now get
their own slug at creation. Existing files are unaffected.

Fixes #5942

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-19 20:30:01 +02:00
Marcela Petra Ferraz de Novaes 2c61f3be1e fix: never resize avatars option does not work for group chat avatar (#5936)
* Check the never_resize_avatars setting to request or not cropping for group avatars.

* Change max width of thumbnail to 300px so square images are allowed.
2026-08-19 20:17:30 +02:00
48e7d605df fix(expressions): guard against null/undefined default-expression labels (#5938)
* fix(expressions): guard against null/undefined default-expression labels

Classifiers can return a falsy or literal "null"/"undefined" expression
label, and setDefaultEmojiForImage then requested the nonexistent
/img/default-expressions/null.png, producing a 404.

Fall back to DEFAULT_FALLBACK_EXPRESSION (joy) when the label is falsy,
a literal "null"/"undefined", or a custom expression without a default
asset. This is the only place the default-expressions URL is built, so the
guard covers all 404 paths.

Fixes #5863

* fix(expressions): fall back for any label without a default asset

Generalize the guard to a whitelist check against DEFAULT_EXPRESSIONS.
Every entry in that list has a matching default-expressions asset, so
any other label (falsy, literal null/undefined strings, custom
expressions, or unexpected classifier output) falls back to the
default expression instead of requesting a missing image.

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

---------

Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-19 20:13:23 +02:00
WolfsblvtandClaude Fable 5 12b4fc50aa tests: set extensions enabled override for discover suite
PR #5571 added an extensionsEnabledFeatureGuard middleware that reads
extensions.enabled via getConfigValue. The test environment has no
config.yaml, so getConfig() exits the process and kills the Jest
worker, failing the extensions-discover suite. Set the env var
override so the guard resolves without touching the config file.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-08-19 20:06:43 +02:00
Yongjie Guan 742d5601be Prevent proxy access keys from triggering password-manager prompts in Chromium-based browsers (#5913)
* fix: prevent proxy keys from triggering password manager

Render reverse proxy credentials as visually masked text fields with password-neutral DOM identifiers while preserving existing settings, preset, and request field names. Align the Azure OpenAI key input and add regression coverage.

* fix: preserve masked secret clipboard protection
2026-08-19 20:05:15 +02:00
4ed1372411 fix: correct malformed multimodal caption request for oobabooga (#5915)
Closes #4448

The 'ooba' special case in the caption-image route split the correctly
structured message (a single user message with a content array
containing text + image_url) into two separate messages, moving
image_url to a top-level property with an empty content array.

oobabooga's own multimodal image handling (modules/image_utils.py)
only looks for images inside content array items with
type: 'image_url', so this transformation caused the image to be
silently dropped, producing hallucinated or "I can't see an image"
captions - even though the request appeared to contain valid image
data.

Removing the special case lets 'ooba' use the same, already-correct
message structure as every other multimodal-capable API.

Verified against a local oobabooga (llama.cpp/GGUF loader) with a
vision-capable GGUF + mmproj pair: before the fix, captions ignored
the attached image; after the fix, captions accurately describe real
photos.

Co-authored-by: bbastex <bbastex@users.noreply.github.com>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-17 03:33:13 +02:00
Christoph 86b55c8696 Fireworks: Update model fetching because /v1/models is broken (#5831)
* Fireworks AI: Fetch model list from custom URL

Fireworks provides the model list on a custom URL in a custom format, see https://docs.fireworks.ai/api-reference/list-models. The standard /v1/models endpoint no longer works correctly and returns a wrong list of models:

```
Available models: [
  'accounts/fireworks/models/deepseek-v4-pro',
  'accounts/fireworks/models/flux-1-schnell-fp8',
  'accounts/fireworks/models/glm-5p1',
  'accounts/fireworks/models/glm-5p2',
  'accounts/fireworks/models/gpt-oss-120b',
  'accounts/fireworks/models/kimi-k2p5',
  'accounts/fireworks/models/kimi-k2p6'
]
```

Many new models like Deepseek V4 Pro, Kimi K2.7 Code are missing.

* Fireworks AI: Add "fast" models

Some models have a "fast" variant. The model list endpoint does not return this information, so for now the available fast versions are listed explicitly.

Fast models have the same parameters (max context etc.) as the regular ones. They are only faster and cost more.
2026-08-17 03:30:30 +02:00
Martin TibkeandMartin Tibke 1a7f88861d fix(tags): drop filters for tags with no assigned entities (#5911)
Filter lists only print tags that are assigned to at least one entity, so a
persisted filter on an unassigned tag has no element to toggle. "Clear all
filters" cannot reset it either, since it works by clicking the printed
elements. The result is a character list that stays empty on every app start,
showing only an "N characters hidden" notice with no visible cause.

removeMissingTagFilters() already dropped filters for deleted tags and for
empty open bogus folders (#5225). Generalize that check to any tag without
assigned entities, which covers plain tags as well, and clear the matching
account storage entry so the filter does not come back on the next reload.

Co-authored-by: Martin Tibke <martin@tibke.net>
2026-08-17 03:01:40 +02:00
WolfsblvtandClaude Fable 5 800caeb9e0 fix: skip extension folders without a manifest during discovery (#5934)
A manually deleted extension often leaves a folder husk on disk, for
example when Windows fails to remove locked .git files. The discover
endpoint listed any directory, so the husk kept reporting as an
installed extension: /extension-exists and /extension-state returned
true after a fresh load and the manifest fetch produced a 404 in the
console on every load. Discovery now skips folders without a
manifest.json and logs a warning naming the exact path, and the
client drops any discovered name whose manifest could not be loaded
so state queries cannot report ghost extensions.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-17 02:47:08 +02:00
f91050136c fix: show full Generic textgen model list in a select (#5897)
Browsers filter datalist suggestions to the current input value, so
Generic OpenAI-compatible text completion looked like a single model
while Chat Custom's select showed the full /v1/models list. Add an
Available Models select (select2 + reconnect) matching other textgen
backends, and keep the optional free-text Model ID field.

Co-authored-by: trahloc <trahloc@proton.me>
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-17 02:21:11 +02:00
f199580102 fix: remove GroupId requirement from MiniMax TTS (#5920)
* Update TTS model from Gemini 2.5 to 3.1 in TTS Extension (#5778)

* Update TTS model from Gemini 2.5 to 3.1 in TTS Extension

Added New versions and new Default model

* Add glm-5.2 option to model selection dropdown

* fix: remove GroupId requirement from MiniMax TTS

MiniMax's current API (sk-cp- keys) uses Bearer auth only. GroupId
is a legacy parameter from the pre-2025 API that the current API
tolerates/ignores. Requiring it blocks new users from setting up
MiniMax TTS.

- Backend: drop groupId from required params and URL
- Frontend: remove Group ID input field and all validation checks

Fixes #5909

---------

Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
Co-authored-by: Samueras <Samueras@jackoedv.de>
Co-authored-by: chen-jiying <chen-jiying@users.noreply.github.com>
2026-08-17 01:36:56 +02:00
eaf2858e43 fix: use exact ArrayBuffer slices for BYAF imports (#5917)
* Update TTS model from Gemini 2.5 to 3.1 in TTS Extension (#5778)

* Update TTS model from Gemini 2.5 to 3.1 in TTS Extension

Added New versions and new Default model

* Add glm-5.2 option to model selection dropdown

* Refactor file reading to use getArrayBufferSlice

* Update util.js

* Update util-pure.test.js

* test: add real-archive regression for BYAF pooled buffer handling

Builds an actual BYAF zip in memory, embeds it at an offset inside a
larger backing buffer to simulate Node's buffer pooling, and proves
through the real parser that the exact ArrayBuffer slice parses
correctly while the raw oversized backing buffer does not.

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

---------

Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
Co-authored-by: Samueras <Samueras@jackoedv.de>
Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-17 01:28:15 +02:00
1cfdcd82e1 fix: skip non-image ComfyUI outputs (#5910)
* fix: skip non-image ComfyUI outputs

* test: cover ComfyUI gif fallback and no-output error paths

Extends the skip-non-image regression with the gifs fallback that
animated workflows rely on and the error path when no output node
carries images or gifs.

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

---------

Co-authored-by: CaoBochun <caobochun@bytedance.com>
Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-17 01:19:49 +02:00
7228502f0b Add Claude Opus 5 / Sonnet 5 and Gemini 3.7 Flash / 3.6 Flash / 3.5 Flash-Lite (#5883)
* feature: add Claude Opus 5, Sonnet 5, and Gemini 3.6 Flash / 3.5 Flash-Lite

Enable adaptive thinking and strip rejected sampling/prefill fields so the new models work with Chat Completions and Image Captioning.

* feature: add Gemini 3.7 Flash

Strip rejected sampling/prefill fields and map min thinking to low so the model works with Chat Completions and Image Captioning.

* feat: refresh Claude and Gemini model lists on top of Claude 5 support

Builds on the Claude 5 and Gemini wiring in this branch: removes
retired models from the Claude and Gemini dropdowns and captioning
lists (claude-opus-4-1 verified retired with a live 404; Gemini 2.0
experimental models, 2.5 preview snapshots, and gemini-3-pro-preview
are shut down per Google's model docs), labels deprecated models
without touching option values so saved selections keep working, adds
claude-fable-5 to the captioning list, and moves new-install defaults
to claude-sonnet-5 and gemini-3.7-flash. Claude 5 generation, vision,
web search, and effort mapping verified live against the API; the
trailing-prefill test now uses a plain loop so the tests lint passes.

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

---------

Co-authored-by: Wolfsblvt <wolfsblvt@gmail.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-17 01:10:28 +02:00
WolfsblvtandClaude Fable 5 4f6977b779 feat: verify GPT-5.6 support and refresh OpenAI model list (#5932)
Verifies Sol, Terra, and Luna live against the API: generation,
vision, effort values (none/low/xhigh), the unlisted gpt-5.6 alias,
and the sampler contract (temperature and top_p are rejected even
with reasoning off, so the strict stripping stays). Removes models
OpenAI has shut down from the model dropdowns, labels announced
deprecations without touching option values, orders groups newest
first, and moves the new-install and captioning defaults off
gpt-4-turbo, which shuts down on October 23, 2026.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-17 00:37:46 +02:00
Samueras 8172dcd0ee Update TTS model from Gemini 2.5 to 3.1 in TTS Extension (#5778)
* Update TTS model from Gemini 2.5 to 3.1 in TTS Extension

Added New versions and new Default model

* Add glm-5.2 option to model selection dropdown
2026-07-07 19:36:20 +02:00
Cohee 51ad27fb86 Merge pull request #5591 from SillyTavern/staging
Staging
2026-05-03 18:45:46 +03:00
53 changed files with 1634 additions and 287 deletions
+1 -1
View File
@@ -75,7 +75,7 @@ but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.**
* [TavernAI](https://github.com/TavernAI/TavernAI) 1.2.8 by Humi: MIT License
* [TavernAI](https://github.com/TavernAI/TavernAI-v1) 1.2.8 by Humi: MIT License
* Portions of CncAnon's TavernAITurbo mod used with permission
* Visual Novel Mode inspired by the work of PepperTaco (<https://github.com/peppertaco/Tavern/>)
* Noto Sans font by Google (OFL license)
+19 -2
View File
@@ -1,4 +1,4 @@
# This workflow will run tests using node and then publish a package to GitHub Packages when a release is created
# Publish to npm for a new release or explicitly retry an existing release tag.
# For more information see: https://docs.github.com/en/actions/publishing-packages/publishing-nodejs-packages
name: Node.js Package
@@ -6,6 +6,16 @@ name: Node.js Package
on:
release:
types: [created]
workflow_dispatch:
inputs:
tag:
description: Existing release tag to publish
required: true
type: string
env:
PUBLISH_TAG: ${{ github.event_name == 'workflow_dispatch' && inputs.tag || github.event.release.tag_name }}
PUBLISH_REF: ${{ github.event_name == 'workflow_dispatch' && format('refs/tags/{0}', inputs.tag) || github.ref }}
permissions:
id-token: write # Required for OIDC
@@ -16,9 +26,13 @@ jobs:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3
with:
ref: ${{ env.PUBLISH_REF }}
- uses: actions/setup-node@3235b876344d2a9aa001b8d1453c930bba69e610 # v3
with:
node-version: 24
- name: Verify release tag matches package version
run: test "$(node -p "require('./package.json').version")" = "$PUBLISH_TAG"
- run: npm ci --omit=dev --ignore-scripts
publish-npm:
@@ -26,9 +40,12 @@ jobs:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3
with:
ref: ${{ env.PUBLISH_REF }}
- uses: actions/setup-node@3235b876344d2a9aa001b8d1453c930bba69e610 # v3
with:
node-version: 24
registry-url: https://registry.npmjs.org/
- run: npm ci --omit=dev --ignore-scripts
- run: npm publish
# https://github.com/npm/cli/issues/9755
- run: npm publish --allow-directory=all
+3 -3
View File
@@ -1,6 +1,6 @@
{
"chat_completion_source": "openai",
"openai_model": "gpt-4-turbo",
"openai_model": "gpt-5.6-terra",
"claude_model": "claude-sonnet-5",
"openrouter_model": "OR_Website",
"openrouter_use_fallback": false,
@@ -20,8 +20,8 @@
"custom_include_body": "",
"custom_exclude_body": "",
"custom_include_headers": "",
"google_model": "gemini-2.5-pro",
"vertexai_model": "gemini-2.5-pro",
"google_model": "gemini-3.7-flash",
"vertexai_model": "gemini-3.7-flash",
"temperature": 1,
"frequency_penalty": 0,
"presence_penalty": 0,
+2 -2
View File
@@ -1,12 +1,12 @@
{
"name": "sillytavern",
"version": "1.18.0",
"version": "1.19.0",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "sillytavern",
"version": "1.18.0",
"version": "1.19.0",
"license": "AGPL-3.0",
"dependencies": {
"@adobe/css-tools": "^4.4.4",
+1 -1
View File
@@ -115,7 +115,7 @@
"type": "git",
"url": "https://github.com/SillyTavern/SillyTavern.git"
},
"version": "1.18.0",
"version": "1.19.0",
"scripts": {
"init": "node src/server-init.js",
"start": "node server.js",
+78 -83
View File
@@ -2155,7 +2155,7 @@
</strong>
</div>
</div>
<div class="flex-container flexFlowColumn wide100p textAlignCenter marginTop10" data-source="openai,custom,claude,xai,makersuite,vertexai,aimlapi,openrouter,pollinations,perplexity,cometapi,electronhub,azure_openai,chutes,nanogpt,deepseek">
<div class="flex-container flexFlowColumn wide100p textAlignCenter marginTop10" data-source="openai,custom,claude,xai,makersuite,vertexai,aimlapi,openrouter,pollinations,perplexity,cometapi,electronhub,azure_openai,chutes,nanogpt,deepseek,fireworks">
<div class="flex-container oneline-dropdown" title="Constrains effort on reasoning for reasoning models.&#10;Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response." data-i18n="[title]Constrains effort on reasoning for reasoning models.">
<label for="openai_reasoning_effort">
<span data-i18n="Reasoning Effort">Reasoning Effort</span>
@@ -2172,6 +2172,9 @@
<div class="toggle-description justifyLeft marginBot5" data-source="openai,custom,xai,aimlapi,openrouter,pollinations,perplexity,cometapi,electronhub,azure_openai,chutes" data-i18n="OpenAI-compatible effort levels vary by model. Minimum and maximum may use model-specific aliases. Auto does not send an effort level.">
OpenAI-compatible effort levels vary by model. Minimum and maximum may use model-specific aliases. Auto does not send an effort level.
</div>
<div class="toggle-description justifyLeft marginBot5" data-source="fireworks" data-i18n="Fireworks supports low, medium, high, and max as distinct effort levels. Minimum is aliased to low. Auto does not send an effort level.">
Fireworks supports low, medium, high, and max as distinct effort levels. Minimum is aliased to low. Auto does not send an effort level.
</div>
<strong class="toggle-description justifyLeft marginBot5" data-source="openrouter">
Request model reasoning = Off with Reasoning Effort = Minimum disables reasoning entirely on models that support that, but can cause errors with some models.
</strong>
@@ -2593,8 +2596,19 @@
<small data-i18n="Example: http://127.0.0.1:5000">Example: http://127.0.0.1:5000</small>
<input id="generic_api_url_text" name="generic_api_url" class="text_pole wide100p" value="" autocomplete="off" data-server-history="generic">
</div>
<datalist id="generic_model_fill"></datalist>
<input id="generic_model_textgenerationwebui" list="generic_model_fill" class="text_pole wide100p" placeholder="Model ID (optional)" data-i18n="[placeholder]Model ID (optional)" type="text">
<h4 data-i18n="Enter a Model ID">Enter a Model ID</h4>
<div class="flex-container">
<input list="generic_model_fill" id="generic_model_textgenerationwebui" class="text_pole wide100p" placeholder="Model ID (optional)" data-i18n="[placeholder]Model ID (optional)" type="text" autocomplete="off">
<datalist id="generic_model_fill"></datalist>
</div>
<h4 data-i18n="Available Models">Available Models</h4>
<div class="flex-container">
<select id="generic_model_select" class="text_pole wide100p">
<option value="" data-i18n="-- Connect to the API --">
-- Connect to the API --
</option>
</select>
</div>
</div>
<div data-tg-type="ooba" class="flex-container flexFlowColumn">
<div class="flex-container flexFlowColumn">
@@ -2997,8 +3011,8 @@
<br>
</div>
<div class="flex-container width100p">
<input id="openai_proxy_password" type="password" class="text_pole flex1" placeholder="" form="openai_form" autocomplete="off" />
<div id="openai_proxy_password_show" title="Peek a password" data-i18n="[title]Peek a password" class="menu_button fa-solid fa-eye-slash fa-fw"></div>
<input id="openai_proxy_access_key" type="text" class="text_pole flex1 masked-secret" placeholder="" autocomplete="off" autocapitalize="none" autocorrect="off" spellcheck="false" />
<div id="openai_proxy_access_key_show" title="Peek a password" data-i18n="[title]Peek a password" class="menu_button fa-solid fa-eye-slash fa-fw"></div>
</div>
</div>
</div>
@@ -3052,6 +3066,9 @@
<div>
<h4 data-i18n="OpenAI Model">OpenAI Model</h4>
<select id="model_openai_select">
<optgroup label="GPT-6">
<option value="gpt-6-astra">gpt-6-astra</option>
</optgroup>
<optgroup label="GPT-5.6">
<option value="gpt-5.6">gpt-5.6</option>
<option value="gpt-5.6-sol">gpt-5.6-sol</option>
@@ -3071,12 +3088,12 @@
<option value="gpt-5.4-nano-2026-03-17">gpt-5.4-nano-2026-03-17</option>
</optgroup>
<optgroup label="GPT-5.3">
<option value="gpt-5.3-chat-latest">gpt-5.3-chat-latest</option>
<option value="gpt-5.3-chat-latest">gpt-5.3-chat-latest (deprecated)</option>
</optgroup>
<optgroup label="GPT-5.2">
<option value="gpt-5.2">gpt-5.2</option>
<option value="gpt-5.2-2025-12-11">gpt-5.2-2025-12-11</option>
<option value="gpt-5.2-chat-latest">gpt-5.2-chat-latest</option>
<option value="gpt-5.2-chat-latest">gpt-5.2-chat-latest (deprecated)</option>
</optgroup>
<optgroup label="GPT-5.1">
<option value="gpt-5.1">gpt-5.1</option>
@@ -3085,69 +3102,56 @@
</optgroup>
<optgroup label="GPT-5">
<option value="gpt-5">gpt-5</option>
<option value="gpt-5-2025-08-07">gpt-5-2025-08-07</option>
<option value="gpt-5-2025-08-07">gpt-5-2025-08-07 (deprecated)</option>
<option value="gpt-5-chat-latest">gpt-5-chat-latest</option>
<option value="gpt-5-mini">gpt-5-mini</option>
<option value="gpt-5-mini-2025-08-07">gpt-5-mini-2025-08-07</option>
<option value="gpt-5-mini-2025-08-07">gpt-5-mini-2025-08-07 (deprecated)</option>
<option value="gpt-5-nano">gpt-5-nano</option>
<option value="gpt-5-nano-2025-08-07">gpt-5-nano-2025-08-07</option>
</optgroup>
<optgroup label="GPT-4o">
<option value="gpt-4o">gpt-4o</option>
<option value="gpt-4o-2024-11-20">gpt-4o-2024-11-20</option>
<option value="gpt-4o-2024-08-06">gpt-4o-2024-08-06</option>
<option value="gpt-4o-2024-05-13">gpt-4o-2024-05-13</option>
<option value="chatgpt-4o-latest">chatgpt-4o-latest</option>
</optgroup>
<optgroup label="GPT-4o mini">
<option value="gpt-4o-mini">gpt-4o-mini</option>
<option value="gpt-4o-mini-2024-07-18">gpt-4o-mini-2024-07-18</option>
<option value="gpt-5-nano-2025-08-07">gpt-5-nano-2025-08-07 (deprecated)</option>
</optgroup>
<optgroup label="GPT-4.1">
<option value="gpt-4.1">gpt-4.1</option>
<option value="gpt-4.1-2025-04-14">gpt-4.1-2025-04-14</option>
<option value="gpt-4.1-mini">gpt-4.1-mini</option>
<option value="gpt-4.1-mini-2025-04-14">gpt-4.1-mini-2025-04-14</option>
<option value="gpt-4.1-nano">gpt-4.1-nano</option>
<option value="gpt-4.1-nano-2025-04-14">gpt-4.1-nano-2025-04-14</option>
<option value="gpt-4.1-nano">gpt-4.1-nano (deprecated)</option>
<option value="gpt-4.1-nano-2025-04-14">gpt-4.1-nano-2025-04-14 (deprecated)</option>
</optgroup>
<optgroup label="o1">
<option value="o1">o1</option>
<option value="o1-2024-12-17">o1-2024-12-17</option>
<option value="o1-mini">o1-mini</option>
<option value="o1-mini-2024-09-12">o1-mini-2024-09-12</option>
<option value="o1-preview">o1-preview</option>
<option value="o1-preview-2024-09-12">o1-preview-2024-09-12</option>
<optgroup label="o4">
<option value="o4-mini">o4-mini (deprecated)</option>
<option value="o4-mini-2025-04-16">o4-mini-2025-04-16 (deprecated)</option>
</optgroup>
<optgroup label="o3">
<option value="o3">o3</option>
<option value="o3-2025-04-16">o3-2025-04-16</option>
<option value="o3-mini">o3-mini</option>
<option value="o3-mini-2025-01-31">o3-mini-2025-01-31</option>
<option value="o3-2025-04-16">o3-2025-04-16 (deprecated)</option>
<option value="o3-mini">o3-mini (deprecated)</option>
<option value="o3-mini-2025-01-31">o3-mini-2025-01-31 (deprecated)</option>
</optgroup>
<optgroup label="o4">
<option value="o4-mini">o4-mini</option>
<option value="o4-mini-2025-04-16">o4-mini-2025-04-16</option>
<optgroup label="o1">
<option value="o1">o1 (deprecated)</option>
<option value="o1-2024-12-17">o1-2024-12-17 (deprecated)</option>
</optgroup>
<optgroup label="GPT-4.5">
<option value="gpt-4.5-preview">gpt-4.5-preview</option>
<option value="gpt-4.5-preview-2025-02-27">gpt-4.5-preview-2025-02-27</option>
<optgroup label="GPT-4o">
<option value="gpt-4o">gpt-4o</option>
<option value="gpt-4o-2024-11-20">gpt-4o-2024-11-20</option>
<option value="gpt-4o-2024-08-06">gpt-4o-2024-08-06</option>
<option value="gpt-4o-2024-05-13">gpt-4o-2024-05-13</option>
</optgroup>
<optgroup label="GPT-4o mini">
<option value="gpt-4o-mini">gpt-4o-mini</option>
<option value="gpt-4o-mini-2024-07-18">gpt-4o-mini-2024-07-18</option>
</optgroup>
<optgroup label="GPT-4 Turbo and GPT-4">
<option value="gpt-4-turbo">gpt-4-turbo</option>
<option value="gpt-4-turbo-2024-04-09">gpt-4-turbo-2024-04-09</option>
<option value="gpt-4-turbo-preview">gpt-4-turbo-preview</option>
<option value="gpt-4-0125-preview">gpt-4-0125-preview (2024)</option>
<option value="gpt-4-1106-preview">gpt-4-1106-preview (2023)</option>
<option value="gpt-4">gpt-4</option>
<option value="gpt-4-0613">gpt-4-0613 (2023)</option>
<option value="gpt-4-0314">gpt-4-0314 (2023)</option>
<option value="gpt-4-turbo">gpt-4-turbo (deprecated)</option>
<option value="gpt-4-turbo-2024-04-09">gpt-4-turbo-2024-04-09 (deprecated)</option>
<option value="gpt-4">gpt-4 (deprecated)</option>
<option value="gpt-4-0613">gpt-4-0613 (2023, deprecated)</option>
</optgroup>
<optgroup label="GPT-3.5 Turbo">
<option value="gpt-3.5-turbo">gpt-3.5-turbo</option>
<option value="gpt-3.5-turbo-0125">gpt-3.5-turbo-0125 (2024)</option>
<option value="gpt-3.5-turbo-1106">gpt-3.5-turbo-1106 (2023)</option>
<option value="gpt-3.5-turbo-instruct">gpt-3.5-turbo-instruct</option>
<option value="gpt-3.5-turbo">gpt-3.5-turbo (deprecated)</option>
<option value="gpt-3.5-turbo-0125">gpt-3.5-turbo-0125 (2024, deprecated)</option>
<option value="gpt-3.5-turbo-1106">gpt-3.5-turbo-1106 (2023, deprecated)</option>
<option value="gpt-3.5-turbo-instruct">gpt-3.5-turbo-instruct (deprecated)</option>
</optgroup>
<optgroup label="Other">
<option value="babbage-002">babbage-002</option>
@@ -3182,6 +3186,7 @@
<h4 data-i18n="Claude Model">Claude Model</h4>
<select id="model_claude_select">
<optgroup label="Versions">
<option value="claude-fable-5-1">claude-fable-5-1</option>
<option value="claude-fable-5">claude-fable-5</option>
<option value="claude-opus-5">claude-opus-5</option>
<option value="claude-sonnet-5">claude-sonnet-5</option>
@@ -3306,8 +3311,15 @@
<div>
<h4 data-i18n="Google Model">Google Model</h4>
<select id="model_google_select">
<optgroup label="Gemini 3.7">
<option value="gemini-3.7-flash">gemini-3.7-flash</option>
</optgroup>
<optgroup label="Gemini 3.6">
<option value="gemini-3.6-flash">gemini-3.6-flash</option>
</optgroup>
<optgroup label="Gemini 3.5">
<option value="gemini-3.5-flash">gemini-3.5-flash</option>
<option value="gemini-3.5-flash-lite">gemini-3.5-flash-lite</option>
</optgroup>
<optgroup label="Gemini 3.1">
<option value="gemini-3.1-pro-preview">gemini-3.1-pro-preview</option>
@@ -3315,40 +3327,20 @@
<option value="gemini-3.1-flash-image">gemini-3.1-flash-image</option>
</optgroup>
<optgroup label="Gemini 3.0">
<option value="gemini-3-pro-preview">gemini-3-pro-preview</option>
<option value="gemini-3-pro-image">gemini-3-pro-image</option>
<option value="gemini-3-flash-preview">gemini-3-flash-preview</option>
</optgroup>
<optgroup label="Gemini 2.5">
<option value="gemini-2.5-pro">gemini-2.5-pro</option>
<option value="gemini-2.5-pro-preview-06-05">gemini-2.5-pro-preview-06-05</option>
<option value="gemini-2.5-pro-preview-05-06">gemini-2.5-pro-preview-05-06</option>
<option value="gemini-2.5-pro-preview-03-25">gemini-2.5-pro-preview-03-25</option>
<option value="gemini-2.5-flash">gemini-2.5-flash</option>
<option value="gemini-2.5-flash-preview-09-2025">gemini-2.5-flash-preview-09-2025</option>
<option value="gemini-2.5-flash-preview-05-20">gemini-2.5-flash-preview-05-20</option>
<option value="gemini-2.5-flash-lite">gemini-2.5-flash-lite</option>
<option value="gemini-2.5-flash-lite-preview-09-2025">gemini-2.5-flash-lite-preview-09-2025</option>
<option value="gemini-2.5-flash-lite-preview-06-17">gemini-2.5-flash-lite-preview-06-17</option>
<option value="gemini-2.5-flash-image">gemini-2.5-flash-image</option>
<option value="gemini-2.5-flash-image-preview">gemini-2.5-flash-image-preview</option>
</optgroup>
<optgroup label="Gemini 2.0">
<option value="gemini-2.0-pro-exp-02-05">gemini-2.0-pro-exp-02-05 → 2.5-exp-03-25</option>
<option value="gemini-2.0-pro-exp">gemini-2.0-pro-exp → 2.5-exp-03-25</option>
<option value="gemini-exp-1206">gemini-exp-1206 → 2.5-exp-03-25</option>
<option value="gemini-2.0-flash-001">gemini-2.0-flash-001</option>
<option value="gemini-2.0-flash-exp-image-generation">gemini-2.0-flash-exp-image-generation</option>
<option value="gemini-2.0-flash-preview-image-generation">gemini-2.0-flash-preview-image-generation</option>
<option value="gemini-2.0-flash-exp">gemini-2.0-flash-exp</option>
<option value="gemini-2.0-flash">gemini-2.0-flash</option>
<option value="gemini-2.0-flash-thinking-exp-01-21">gemini-2.0-flash-thinking-exp-01-21 → 2.5-flash-preview-05-20</option>
<option value="gemini-2.0-flash-thinking-exp-1219">gemini-2.0-flash-thinking-exp-1219 → 2.5-flash-preview-05-20</option>
<option value="gemini-2.0-flash-thinking-exp">gemini-2.0-flash-thinking-exp → 2.5-flash-preview-05-20</option>
<option value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001</option>
<option value="gemini-2.0-flash-lite-preview-02-05">gemini-2.0-flash-lite-preview-02-05</option>
<option value="gemini-2.0-flash-lite-preview">gemini-2.0-flash-lite-preview</option>
<option value="gemini-2.0-flash-lite">gemini-2.0-flash-lite</option>
<option value="gemini-2.0-flash-001">gemini-2.0-flash-001 (deprecated)</option>
<option value="gemini-2.0-flash">gemini-2.0-flash (deprecated)</option>
<option value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001 (deprecated)</option>
<option value="gemini-2.0-flash-lite">gemini-2.0-flash-lite (deprecated)</option>
</optgroup>
<optgroup label="Gemma">
<option value="gemma-4-31b-it">gemma-4-31b-it</option>
@@ -3489,8 +3481,15 @@
<h4 data-i18n="Google Model">Google Model</h4>
<select id="model_vertexai_select">
<!-- data-mode="full" is for models that require a service account -->
<optgroup label="Gemini 3.7">
<option value="gemini-3.7-flash">gemini-3.7-flash</option>
</optgroup>
<optgroup label="Gemini 3.6">
<option value="gemini-3.6-flash">gemini-3.6-flash</option>
</optgroup>
<optgroup label="Gemini 3.5">
<option value="gemini-3.5-flash">gemini-3.5-flash</option>
<option value="gemini-3.5-flash-lite">gemini-3.5-flash-lite</option>
</optgroup>
<optgroup label="Gemini 3.1">
<option value="gemini-3.1-pro-preview">gemini-3.1-pro-preview</option>
@@ -3498,7 +3497,6 @@
<option value="gemini-3.1-flash-image">gemini-3.1-flash-image</option>
</optgroup>
<optgroup label="Gemini 3.0">
<option value="gemini-3-pro-preview">gemini-3-pro-preview</option>
<option value="gemini-3-pro-image">gemini-3-pro-image</option>
<option value="gemini-3-flash-preview">gemini-3-flash-preview</option>
</optgroup>
@@ -3507,14 +3505,11 @@
<option value="gemini-2.5-flash">gemini-2.5-flash</option>
<option value="gemini-2.5-flash-lite">gemini-2.5-flash-lite</option>
<option value="gemini-2.5-flash-image">gemini-2.5-flash-image</option>
<option value="gemini-2.5-flash-image-preview">gemini-2.5-flash-image-preview</option>
</optgroup>
<optgroup label="Gemini 2.0">
<option value="gemini-2.0-flash-exp" data-mode="full">gemini-2.0-flash-exp</option>
<option value="gemini-2.0-flash-preview-image-generation" data-mode="full">gemini-2.0-flash-preview-image-generation</option>
<option value="gemini-2.0-flash">gemini-2.0-flash</option>
<option value="gemini-2.0-flash-001">gemini-2.0-flash-001</option>
<option value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001</option>
<option value="gemini-2.0-flash">gemini-2.0-flash (deprecated)</option>
<option value="gemini-2.0-flash-001">gemini-2.0-flash-001 (deprecated)</option>
<option value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001 (deprecated)</option>
</optgroup>
</select>
</div>
@@ -3997,7 +3992,7 @@
<!-- Azure API Key -->
<h4><span data-i18n="Azure API Key">Azure API Key</span></h4>
<div class="flex-container">
<input id="api_key_azure_openai" data-setting="api_key_azure_openai" class="text_pole flex1" type="password" autocomplete="off">
<input id="api_key_azure_openai" data-setting="api_key_azure_openai" class="text_pole flex1" type="text" autocomplete="off">
<div title="Manage API keys" data-i18n="[title]Manage API keys" class="menu_button fa-solid fa-key fa-fw manage-api-keys" data-key="api_key_azure_openai"></div>
</div>
<div class="neutral_warning" data-i18n="For privacy reasons, your API key will be hidden after you click 'Connect'." data-for="api_key_azure_openai">
+36 -7
View File
@@ -604,6 +604,7 @@ export let is_send_press = false; //Send generation
export const isGenerating = () => (is_send_press || is_group_generating);
let this_del_mes = -1;
let deleteToolCallsInDeleteMode = true;
/** @type {string} */
let this_edit_mes_chname = '';
@@ -1610,13 +1611,32 @@ export async function deleteLastMessage() {
await eventSource.emit(event_types.MESSAGE_DELETED, chat.length);
}
function getMessageDeletionStartId(id, deleteToolCalls = true) {
const message = chat[id];
if (!deleteToolCalls || message?.is_user || message?.is_system) {
return id;
}
let startId = id;
while (startId > 0) {
const previousMessage = chat[startId - 1];
if (!previousMessage?.is_system || !Array.isArray(previousMessage.extra?.tool_invocations)) {
break;
}
startId--;
}
return startId;
}
/**
* Deletes a message from the chat by its ID, optionally asking for confirmation.
* @param {number} id The ID of the message to delete.
* @param {number} [swipeDeletionIndex] Deletes the swipe with that index.
* @param {boolean} [askConfirmation=false] Whether to ask for confirmation before deleting.
* @param {boolean} [deleteToolCalls=true] Whether to delete preceding tool-call messages.
*/
export async function deleteMessage(id, swipeDeletionIndex = undefined, askConfirmation = false) {
export async function deleteMessage(id, swipeDeletionIndex = undefined, askConfirmation = false, deleteToolCalls = true) {
const canDeleteSwipe = swipeDeletionIndex !== undefined && swipeDeletionIndex !== null;
if (canDeleteSwipe) {
if (swipeDeletionIndex < 0) {
@@ -1654,13 +1674,19 @@ export async function deleteMessage(id, swipeDeletionIndex = undefined, askConfi
return;
}
chat.splice(id, 1);
messageElement.remove();
const firstMessageId = getMessageDeletionStartId(id, deleteToolCalls);
const messageIds = Array.from({ length: id - firstMessageId + 1 }, (_, index) => id - index);
// Delete from the end so earlier indices remain stable.
for (const messageId of messageIds) {
chat.splice(messageId, 1);
chatElement.find(`.mes[mesid="${messageId}"]`).remove();
deleteItemizedPromptForMessage(messageId);
}
chat_metadata.tainted = true;
const startIndex = [0, minId].includes(id) ? id : null;
deleteItemizedPromptForMessage(id);
const startIndex = firstMessageId <= minId ? firstMessageId : null;
updateViewMessageIds(startIndex);
saveChatDebounced();
@@ -8166,7 +8192,7 @@ function updateMessage(div) {
return { mesBlock, text, mes, bias };
}
function openMessageDelete(fromSlashCommand) {
function openMessageDelete(fromSlashCommand, deleteToolCalls = true) {
closeMessageEditor();
hideSwipeButtons();
if (fromSlashCommand || (!is_send_press) || (selected_group && !is_group_generating)) {
@@ -8185,6 +8211,7 @@ function openMessageDelete(fromSlashCommand) {
is_group_generating: ${is_group_generating}`);
}
this_del_mes = -1;
deleteToolCallsInDeleteMode = deleteToolCalls;
is_delete_mode = true;
}
@@ -11229,6 +11256,7 @@ jQuery(async function () {
});
$(this).addClass('selected'); //sets the bg of the mes selected for deletion
var i = Number($(this).attr('mesid')); //checks the message ID in the chat
i = getMessageDeletionStartId(i, deleteToolCallsInDeleteMode);
this_del_mes = i;
//as long as the current message ID is less than the total chat length
while (i < chat.length) {
@@ -11553,6 +11581,7 @@ jQuery(async function () {
///////////// OPTIMIZED LISTENERS FOR LEFT SIDE OPTIONS POPUP MENU //////////////////////
$('#options [id]').on('click', async function (event, customData) {
const fromSlashCommand = customData?.fromSlashCommand || false;
const deleteToolCalls = customData?.deleteToolCalls ?? true;
var id = $(this).attr('id');
// Check whether a custom prompt was provided via custom data (for example through a slash command)
@@ -11631,7 +11660,7 @@ jQuery(async function () {
Generate('continue', buildOrFillAdditionalArgs());
}
} else if (id == 'option_delete_mes') {
setTimeout(() => openMessageDelete(fromSlashCommand), animation_duration);
setTimeout(() => openMessageDelete(fromSlashCommand, deleteToolCalls), animation_duration);
} else if (id == 'option_close_chat') {
await closeCurrentChat();
} else if (id === 'option_settings') {
+5 -2
View File
@@ -41,6 +41,7 @@ import { t } from './i18n.js';
import {
getUniqueName,
isTrueBoolean,
uuidv4,
} from './utils.js';
const bookmarkNameToken = 'Checkpoint #';
@@ -196,7 +197,8 @@ export async function createBranch(mesId, { swipeId = null } = {}) {
const lastMes = chat[mesId];
const mainChatName = (getCurrentChatDetails()).sessionName;
const newMetadata = { main_chat: mainChatName };
// Mint a fresh integrity slug so the branch is distinguishable from its parent (#5942)
const newMetadata = { main_chat: mainChatName, integrity: uuidv4() };
const selectedSwipeId = swipeId === null ? null : Number(swipeId);
if (selectedSwipeId !== null && (!Number.isInteger(selectedSwipeId) || selectedSwipeId < 0 || selectedSwipeId >= (lastMes?.swipes?.length ?? 0))) {
@@ -278,7 +280,8 @@ export async function createNewBookmark(mesId, { forceName = null } = {}) {
}
const mainChat = selected_group ? groups?.find(x => x.id == selected_group)?.chat_id : characters[this_chid].chat;
const newMetadata = { main_chat: mainChat };
// Mint a fresh integrity slug so the checkpoint is distinguishable from its parent (#5942)
const newMetadata = { main_chat: mainChat, integrity: uuidv4() };
await saveItemizedPrompts(name);
if (selected_group) {
+9
View File
@@ -1797,6 +1797,15 @@ export async function loadExtensionSettings(settings, versionChanged, enableAuto
extensionTypes = Object.fromEntries(extensions.map(x => [x.name, x.type]));
manifests = await getManifests(extensionNames);
// An extension without a loadable manifest cannot be used: drop it so state queries
// and the extensions list do not report a ghost entry (e.g. leftover folder on disk)
const namesWithoutManifest = extensionNames.filter(name => !Object.hasOwn(manifests, name));
if (namesWithoutManifest.length) {
console.warn('Ignoring extensions without a loadable manifest.json:', namesWithoutManifest);
extensionNames = extensionNames.filter(name => Object.hasOwn(manifests, name));
namesWithoutManifest.forEach(name => delete extensionTypes[name]);
}
if (versionChanged && enableAutoUpdate) {
await autoUpdateExtensions(false);
}
+2 -2
View File
@@ -42,7 +42,7 @@ function migrateSettings() {
if (extension_settings.caption.source === 'openai') {
extension_settings.caption.source = 'multimodal';
extension_settings.caption.multimodal_api = 'openai';
extension_settings.caption.multimodal_model = 'gpt-4-turbo';
extension_settings.caption.multimodal_model = 'gpt-5.6-luna';
}
if (['google', 'vertexai'].includes(extension_settings.caption.multimodal_api)) {
@@ -55,7 +55,7 @@ function migrateSettings() {
}
if (!extension_settings.caption.multimodal_model) {
extension_settings.caption.multimodal_model = 'gpt-4-turbo';
extension_settings.caption.multimodal_model = 'gpt-5.6-luna';
}
if (!extension_settings.caption.prompt) {
+13 -30
View File
@@ -54,6 +54,7 @@
<option data-type="cohere" value="c4ai-aya-vision-8b">c4ai-aya-vision-8b</option>
<option data-type="cohere" value="c4ai-aya-vision-32b">c4ai-aya-vision-32b</option>
<option data-type="cohere" value="command-a-vision-07-2025">command-a-vision-07-2025</option>
<option data-type="openai" value="gpt-6-astra">gpt-6-astra</option>
<option data-type="openai" value="gpt-5.6">gpt-5.6</option>
<option data-type="openai" value="gpt-5.6-sol">gpt-5.6-sol</option>
<option data-type="openai" value="gpt-5.6-terra">gpt-5.6-terra</option>
@@ -86,20 +87,17 @@
<option data-type="openai" value="gpt-4.1-mini-2025-04-14">gpt-4.1-mini-2025-04-14</option>
<option data-type="openai" value="gpt-4.1-nano">gpt-4.1-nano</option>
<option data-type="openai" value="gpt-4.1-nano-2025-04-14">gpt-4.1-nano-2025-04-14</option>
<option data-type="openai" value="gpt-4-vision-preview">gpt-4-vision-preview</option>
<option data-type="openai" value="gpt-4-turbo">gpt-4-turbo</option>
<option data-type="openai" value="gpt-4o">gpt-4o</option>
<option data-type="openai" value="gpt-4o-mini">gpt-4o-mini</option>
<option data-type="openai" value="gpt-4o-mini-2024-07-18">gpt-4o-mini-2024-07-18</option>
<option data-type="openai" value="chatgpt-4o-latest">chatgpt-4o-latest</option>
<option data-type="openai" value="o1">o1</option>
<option data-type="openai" value="o1-2024-12-17">o1-2024-12-17</option>
<option data-type="openai" value="o3">o3</option>
<option data-type="openai" value="o3-2025-04-16">o3-2025-04-16</option>
<option data-type="openai" value="o4-mini">o4-mini</option>
<option data-type="openai" value="o4-mini-2025-04-16">o4-mini-2025-04-16</option>
<option data-type="openai" value="gpt-4.5-preview">gpt-4.5-preview</option>
<option data-type="openai" value="gpt-4.5-preview-2025-02-27">gpt-4.5-preview-2025-02-27</option>
<option data-type="anthropic" value="claude-fable-5-1">claude-fable-5-1</option>
<option data-type="anthropic" value="claude-fable-5">claude-fable-5</option>
<option data-type="anthropic" value="claude-opus-5">claude-opus-5</option>
<option data-type="anthropic" value="claude-sonnet-5">claude-sonnet-5</option>
@@ -117,38 +115,22 @@
<option data-type="anthropic" value="claude-opus-4-20250514">claude-opus-4-20250514</option>
<option data-type="anthropic" value="claude-sonnet-4-0">claude-sonnet-4-0</option>
<option data-type="anthropic" value="claude-sonnet-4-20250514">claude-sonnet-4-20250514</option>
<option data-type="google" value="gemini-3.7-flash">gemini-3.7-flash</option>
<option data-type="google" value="gemini-3.6-flash">gemini-3.6-flash</option>
<option data-type="google" value="gemini-3.5-flash">gemini-3.5-flash</option>
<option data-type="google" value="gemini-3.5-flash-lite">gemini-3.5-flash-lite</option>
<option data-type="google" value="gemini-3.1-pro-preview">gemini-3.1-pro-preview</option>
<option data-type="google" value="gemini-3.1-flash-lite">gemini-3.1-flash-lite</option>
<option data-type="google" value="gemini-3.1-flash-image">gemini-3.1-flash-image</option>
<option data-type="google" value="gemini-3-pro-preview">gemini-3-pro-preview</option>
<option data-type="google" value="gemini-3-pro-image">gemini-3-pro-image</option>
<option data-type="google" value="gemini-3-flash-preview">gemini-3-flash-preview</option>
<option data-type="google" value="gemini-2.5-pro">gemini-2.5-pro</option>
<option data-type="google" value="gemini-2.5-pro-preview-06-05">gemini-2.5-pro-preview-06-05</option>
<option data-type="google" value="gemini-2.5-pro-preview-05-06">gemini-2.5-pro-preview-05-06</option>
<option data-type="google" value="gemini-2.5-pro-preview-03-25">gemini-2.5-pro-preview-03-25</option>
<option data-type="google" value="gemini-2.5-flash">gemini-2.5-flash</option>
<option data-type="google" value="gemini-2.5-flash-preview-09-2025">gemini-2.5-flash-preview-09-2025</option>
<option data-type="google" value="gemini-2.5-flash-preview-05-20">gemini-2.5-flash-preview-05-20</option>
<option data-type="google" value="gemini-2.5-flash-lite">gemini-2.5-flash-lite</option>
<option data-type="google" value="gemini-2.5-flash-lite-preview-09-2025">gemini-2.5-flash-lite-preview-09-2025</option>
<option data-type="google" value="gemini-2.5-flash-lite-preview-06-17">gemini-2.5-flash-lite-preview-06-17</option>
<option data-type="google" value="gemini-2.5-flash-image">gemini-2.5-flash-image</option>
<option data-type="google" value="gemini-2.5-flash-image-preview">gemini-2.5-flash-image-preview</option>
<option data-type="google" value="gemini-2.0-pro-exp-02-05">gemini-2.0-pro-exp-02-05 → 2.5-exp-03-25</option>
<option data-type="google" value="gemini-2.0-pro-exp">gemini-2.0-pro-exp → 2.5-exp-03-25</option>
<option data-type="google" value="gemini-exp-1206">gemini-exp-1206 → 2.5-exp-03-25</option>
<option data-type="google" value="gemini-2.0-flash-001">gemini-2.0-flash-001</option>
<option data-type="google" value="gemini-2.0-flash-exp-image-generation">gemini-2.0-flash-exp-image-generation</option>
<option data-type="google" value="gemini-2.0-flash-exp">gemini-2.0-flash-exp</option>
<option data-type="google" value="gemini-2.0-flash">gemini-2.0-flash</option>
<option data-type="google" value="gemini-2.0-flash-thinking-exp-01-21">gemini-2.0-flash-thinking-exp-01-21 → 2.5-flash-preview-5-20</option>
<option data-type="google" value="gemini-2.0-flash-thinking-exp-1219">gemini-2.0-flash-thinking-exp-1219 → 2.5-flash-preview-05-20</option>
<option data-type="google" value="gemini-2.0-flash-thinking-exp">gemini-2.0-flash-thinking-exp → 2.5-flash-preview-05-20</option>
<option data-type="google" value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001</option>
<option data-type="google" value="gemini-2.0-flash-lite-preview-02-05">gemini-2.0-flash-lite-preview-02-05</option>
<option data-type="google" value="gemini-2.0-flash-lite-preview">gemini-2.0-flash-lite-preview</option>
<option data-type="google" value="gemini-2.0-flash-001">gemini-2.0-flash-001 (deprecated)</option>
<option data-type="google" value="gemini-2.0-flash">gemini-2.0-flash (deprecated)</option>
<option data-type="google" value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001 (deprecated)</option>
<option data-type="google" value="learnlm-2.0-flash-experimental">learnlm-2.0-flash-experimental</option>
<option data-type="google" value="gemini-robotics-er-1.5-preview">gemini-robotics-er-1.5-preview</option>
<option data-type="google" value="gemma-4-31b-it">gemma-4-31b-it</option>
@@ -156,20 +138,21 @@
<option data-type="google" value="gemma-3-27b-it">gemma-3-27b-it</option>
<option data-type="google" value="gemma-3-12b-it">gemma-3-12b-it</option>
<option data-type="google" value="gemma-3-4b-it">gemma-3-4b-it</option>
<option data-type="vertexai" value="gemini-3.7-flash">gemini-3.7-flash</option>
<option data-type="vertexai" value="gemini-3.6-flash">gemini-3.6-flash</option>
<option data-type="vertexai" value="gemini-3.5-flash">gemini-3.5-flash</option>
<option data-type="vertexai" value="gemini-3.5-flash-lite">gemini-3.5-flash-lite</option>
<option data-type="vertexai" value="gemini-3.1-pro-preview">gemini-3.1-pro-preview</option>
<option data-type="vertexai" value="gemini-3.1-flash-lite">gemini-3.1-flash-lite</option>
<option data-type="vertexai" value="gemini-3.1-flash-image">gemini-3.1-flash-image</option>
<option data-type="vertexai" value="gemini-3-pro-preview">gemini-3-pro-preview</option>
<option data-type="vertexai" value="gemini-3-pro-image">gemini-3-pro-image</option>
<option data-type="vertexai" value="gemini-3-flash-preview">gemini-3-flash-preview</option>
<option data-type="vertexai" value="gemini-2.5-pro">gemini-2.5-pro</option>
<option data-type="vertexai" value="gemini-2.5-flash">gemini-2.5-flash</option>
<option data-type="vertexai" value="gemini-2.5-flash-lite">gemini-2.5-flash-lite</option>
<option data-type="vertexai" value="gemini-2.5-flash-image">gemini-2.5-flash-image</option>
<option data-type="vertexai" value="gemini-2.5-flash-image-preview">gemini-2.5-flash-image-preview</option>
<option data-type="vertexai" value="gemini-2.0-flash-001">gemini-2.0-flash-001</option>
<option data-type="vertexai" value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001</option>
<option data-type="vertexai" value="gemini-2.0-flash-001">gemini-2.0-flash-001 (deprecated)</option>
<option data-type="vertexai" value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001 (deprecated)</option>
<option data-type="groq" value="meta-llama/llama-4-scout-17b-16e-instruct">meta-llama/llama-4-scout-17b-16e-instruct</option>
<option data-type="groq" value="meta-llama/llama-4-maverick-17b-128e-instruct">meta-llama/llama-4-maverick-17b-128e-instruct</option>
<option data-type="ollama" value="ollama_current" data-i18n="currently_selected">[Currently selected]</option>
@@ -1717,8 +1717,10 @@ async function setExpression(spriteFolderName, expression, { force = false, over
* @param {string} expression - The expression label to use for the default image
*/
function setDefaultEmojiForImage(img, expression) {
if (extension_settings.expressions.custom?.includes(expression)) {
console.debug(`Can't set default emoji for a custom expression (${expression}). setting to ${DEFAULT_FALLBACK_EXPRESSION} instead.`);
// Classifiers can return a falsy, literal "null"/"undefined", or otherwise unknown label,
// and custom expressions have no default assets; only known labels have an image (#5863).
if (!DEFAULT_EXPRESSIONS.includes(expression)) {
console.debug(`Can't set default emoji for expression (${expression}). setting to ${DEFAULT_FALLBACK_EXPRESSION} instead.`);
expression = DEFAULT_FALLBACK_EXPRESSION;
}
@@ -381,7 +381,7 @@ export class SettingsUi {
if (name && name.length > 0) {
const oldQrs = QuickReplySet.get(name);
if (oldQrs) {
const replace = Popup.show.confirm('Replace existing Quick Reply set', `A Quick Reply Set named "${name}" already exists.<br>Do you want to overwrite the existing Quick Reply Set?<br>The existing set will be deleted. This cannot be undone.`);
const replace = await Popup.show.confirm('Replace existing Quick Reply set', `A Quick Reply Set named "${name}" already exists.<br>Do you want to overwrite the existing Quick Reply Set?<br>The existing set will be deleted. This cannot be undone.`);
if (replace) {
const idx = QuickReplySet.list.indexOf(oldQrs);
await this.doDeleteQrSet(oldQrs);
@@ -444,7 +444,7 @@ export class SettingsUi {
qrs.init();
const oldQrs = QuickReplySet.get(props.name);
if (oldQrs) {
const replace = Popup.show.confirm('Replace existing Quick Reply set', `A Quick Reply Set named "${name}" already exists.<br>Do you want to overwrite the existing Quick Reply Set?<br>The existing set will be deleted. This cannot be undone.`);
const replace = await Popup.show.confirm('Replace existing Quick Reply set', `A Quick Reply Set named "${props.name}" already exists.<br>Do you want to overwrite the existing Quick Reply Set?<br>The existing set will be deleted. This cannot be undone.`);
if (replace) {
const idx = QuickReplySet.list.indexOf(oldQrs);
await this.doDeleteQrSet(oldQrs);
+8 -14
View File
@@ -55,10 +55,6 @@ class MiniMaxTtsProvider {
<i class="fa-solid fa-key"></i>
<span>Click to set API Key</span>
</div>
<div id="minimax_group_id" class="menu_button menu_button_icon manage-api-keys" data-key="minimax_group_id">
<i class="fa-solid fa-key"></i>
<span>Click to set Group ID</span>
</div>
</div>
<div class="tts_block">
<label for="minimax_tts_api_host">API Host</label>
@@ -173,9 +169,8 @@ class MiniMaxTtsProvider {
constructor() {
this.handler = async function (/** @type {string} */ key) {
if (![SECRET_KEYS.MINIMAX, SECRET_KEYS.MINIMAX_GROUP_ID].includes(key)) return;
if (key !== SECRET_KEYS.MINIMAX) return;
$('#api_key_minimax').toggleClass('success', !!secret_state[SECRET_KEYS.MINIMAX]);
$('#minimax_group_id').toggleClass('success', !!secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]);
await this.onRefreshClick();
}.bind(this);
}
@@ -588,13 +583,12 @@ class MiniMaxTtsProvider {
}
$('#api_key_minimax').toggleClass('success', !!secret_state[SECRET_KEYS.MINIMAX]);
$('#minimax_group_id').toggleClass('success', !!secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]);
[event_types.SECRET_WRITTEN, event_types.SECRET_DELETED, event_types.SECRET_ROTATED].forEach(event => {
eventSource.on(event, this.handler);
});
// Only check ready status when API credentials are available
if (secret_state[SECRET_KEYS.MINIMAX] && secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]) {
if (secret_state[SECRET_KEYS.MINIMAX]) {
try {
await this.checkReady();
console.debug('MiniMax TTS: Settings loaded and ready');
@@ -608,8 +602,8 @@ class MiniMaxTtsProvider {
// Perform a simple readiness check
async checkReady() {
if (!secret_state[SECRET_KEYS.MINIMAX] || !secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]) {
const error = new Error('API Key and Group ID are required');
if (!secret_state[SECRET_KEYS.MINIMAX]) {
const error = new Error('API Key is required');
console.error('MiniMax TTS checkReady error:', error.message);
throw error;
}
@@ -709,8 +703,8 @@ class MiniMaxTtsProvider {
async fetchTtsVoiceObjects() {
try {
if (!secret_state[SECRET_KEYS.MINIMAX] || !secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]) {
console.warn('MiniMax TTS: API Key and Group ID required for fetching voices');
if (!secret_state[SECRET_KEYS.MINIMAX]) {
console.warn('MiniMax TTS: API Key required for fetching voices');
console.warn('Using all available voices (default + custom). Please check your API credentials');
return this.getAllVoices();
}
@@ -775,8 +769,8 @@ class MiniMaxTtsProvider {
async fetchTtsGeneration(inputText, voiceId, language = null) {
console.info(`Generating new MiniMax TTS for voice_id ${voiceId}`);
if (!secret_state[SECRET_KEYS.MINIMAX] || !secret_state[SECRET_KEYS.MINIMAX_GROUP_ID]) {
const error = new Error('API Key and Group ID are required');
if (!secret_state[SECRET_KEYS.MINIMAX]) {
const error = new Error('API Key is required');
console.error('MiniMax TTS fetchTtsGeneration error:', error.message);
throw error;
}
+3 -2
View File
@@ -1908,13 +1908,14 @@ async function uploadGroupAvatar(event) {
$('#dialogue_popup').addClass('large_dialogue_popup wide_dialogue_popup');
const croppedImage = await callGenericPopup('Set the crop position of the avatar image', POPUP_TYPE.CROP, '', { cropImage: result });
const croppedImage = power_user.never_resize_avatars ? result :
await callGenericPopup('Set the crop position of the avatar image', POPUP_TYPE.CROP, '', { cropImage: result });
if (!croppedImage) {
return;
}
let thumbnail = await createThumbnail(String(croppedImage), 200, 300);
let thumbnail = await createThumbnail(String(croppedImage), 300, 300);
//remove data:image/whatever;base64
thumbnail = thumbnail.replace(/^data:image\/[a-z]+;base64,/, '');
let _thisGroup = groups.find((x) => x.id == openGroupId);
+2 -2
View File
@@ -205,7 +205,7 @@ export function loadNovelPreset(preset) {
nai_settings.logit_bias = preset.logit_bias || [];
nai_settings.preamble = preset.preamble || default_preamble;
nai_settings.min_p = preset.min_p || 0;
nai_settings.math1_temp = preset.math1_temp || 1;
nai_settings.math1_temp = preset.math1_temp ?? 1;
nai_settings.math1_quad = preset.math1_quad || 0;
nai_settings.math1_quad_entropy_scale = preset.math1_quad_entropy_scale || 0;
nai_settings.extensions = preset.extensions || {};
@@ -259,7 +259,7 @@ export function loadNovelSettings(data, settings) {
nai_settings.order = settings.order || default_order;
nai_settings.logit_bias = settings.logit_bias || [];
nai_settings.min_p = settings.min_p || 0;
nai_settings.math1_temp = settings.math1_temp || 1;
nai_settings.math1_temp = settings.math1_temp ?? 1;
nai_settings.math1_quad = settings.math1_quad || 0;
nai_settings.math1_quad_entropy_scale = settings.math1_quad_entropy_scale || 0;
nai_settings.extensions = settings.extensions || {};
+97 -28
View File
@@ -14,6 +14,7 @@ import {
extension_prompt_roles,
extension_prompt_types,
Generate,
getCurrentChatId,
getExtensionPrompt,
getExtensionPromptMaxDepth,
getMediaDisplay,
@@ -376,7 +377,7 @@ export const settingsToUpdate = {
prompts: ['', 'prompts', false, false],
prompt_order: ['', 'prompt_order', false, false],
show_external_models: ['#openai_show_external_models', 'show_external_models', true, true],
proxy_password: ['#openai_proxy_password', 'proxy_password', false, true],
proxy_password: ['#openai_proxy_access_key', 'proxy_password', false, true],
assistant_prefill: ['#claude_assistant_prefill', 'assistant_prefill', false, false],
assistant_impersonation: ['#claude_assistant_impersonation', 'assistant_impersonation', false, false],
use_sysprompt: ['#use_sysprompt', 'use_sysprompt', true, false],
@@ -436,10 +437,10 @@ const default_settings = {
personality_format: default_personality_format,
sort_models: 'alphabetically',
group_models: false,
openai_model: 'gpt-4-turbo',
openai_model: 'gpt-5.6-terra',
claude_model: 'claude-sonnet-5',
google_model: 'gemini-2.5-pro',
vertexai_model: 'gemini-2.5-pro',
google_model: 'gemini-3.7-flash',
vertexai_model: 'gemini-3.7-flash',
ai21_model: 'jamba-large',
mistralai_model: 'mistral-large-latest',
cohere_model: 'command-r-plus',
@@ -1622,6 +1623,21 @@ export async function prepareOpenAIMessages({
return [chat, promptManager.tokenHandler.counts];
}
/**
* Extracts a human-readable message from a Chat Completion error response.
* Providers can return an error object without a message field, or the error as
* a plain string. statusText is only used for non-2xx responses, since for an
* HTTP 200 carrying an error payload it is just "OK" (#5647).
* @param {object} data Parsed response body
* @param {Response} response Fetch response
* @returns {string} Error message to display and throw
*/
function getChatCompletionErrorMessage(data, response) {
const error = data?.error ?? data?.detail?.error;
const message = typeof error === 'string' ? error : (error?.message || error?.code || error?.type);
return String(message || (!response.ok && response.statusText) || t`Unknown error`);
}
/**
* Handles errors during streaming requests.
* @param {Response} response
@@ -1644,7 +1660,7 @@ export function tryParseStreamingError(response, decoded, { quiet = false } = {}
// if trying to fix "[object Object]" displayed to users, start here
if (data.error) {
!quiet && toastr.error(data.error.message || response.statusText, 'Chat Completion API');
!quiet && toastr.error(getChatCompletionErrorMessage(data, response), 'Chat Completion API');
throw new Error(data);
}
@@ -1654,7 +1670,7 @@ export function tryParseStreamingError(response, decoded, { quiet = false } = {}
}
if (data.detail) {
!quiet && toastr.error(data.detail?.error?.message || response.statusText, 'Chat Completion API');
!quiet && toastr.error(getChatCompletionErrorMessage(data, response), 'Chat Completion API');
throw new Error(data);
}
} catch {
@@ -2289,14 +2305,12 @@ function saveModelList(data) {
if (oai_settings.chat_completion_source === chat_completion_sources.FIREWORKS) {
$('#model_fireworks_select').empty();
model_list.sort((a, b) => (a?.name || a?.id || '').localeCompare(b?.name || b?.id || ''));
model_list.forEach((model) => {
if (!model?.supports_chat) {
return;
}
$('#model_fireworks_select').append(
$('<option>', {
value: model.id,
text: model.id,
text: model.name || model.id,
}));
});
@@ -2546,6 +2560,7 @@ function getReasoningEffort(settings = null, model = null) {
chat_completion_sources.ELECTRONHUB,
chat_completion_sources.CHUTES,
chat_completion_sources.DEEPSEEK,
chat_completion_sources.FIREWORKS,
];
if (!reasoningEffortSources.includes(settings.chat_completion_source)) {
@@ -2567,6 +2582,17 @@ function getReasoningEffort(settings = null, model = null) {
}
}
if (settings.chat_completion_source === chat_completion_sources.FIREWORKS) {
switch (settings.reasoning_effort) {
case reasoning_effort_types.auto:
return undefined;
case reasoning_effort_types.min:
return reasoning_effort_types.low;
default:
return settings.reasoning_effort;
}
}
if (settings.chat_completion_source === chat_completion_sources.CUSTOM && /^koboldcpp\/(.+)$/.test(model)) {
switch (settings.reasoning_effort) {
case reasoning_effort_types.auto:
@@ -2605,6 +2631,10 @@ function getReasoningEffort(settings = null, model = null) {
return reasoning_effort_types.low;
case reasoning_effort_types.max:
if ([chat_completion_sources.OPENAI, chat_completion_sources.AZURE_OPENAI].includes(settings.chat_completion_source)
&& /^gpt-6-astra/.test(model)) {
return reasoning_effort_types.max;
}
if ([chat_completion_sources.OPENAI, chat_completion_sources.AZURE_OPENAI].includes(settings.chat_completion_source)
&& /^gpt-5\.6/.test(model)) {
// GPT-5.6 reserves "max" effort for the Responses API.
@@ -2665,6 +2695,20 @@ export async function createGenerationParameters(settings, model, type, messages
}
messages = messages.filter(msg => msg && typeof msg === 'object');
// DeepSeek only accepts image blocks in user messages. Media can also be
// attached to system and assistant messages by the shared inlining path.
const isDeepSeekVisionModel = typeof model === 'string' && model.toLowerCase().includes('deepseek-v4-flash-vision-exp');
if (isDeepSeekVisionModel) {
messages = messages.flatMap((message) => {
if (!['system', 'assistant'].includes(message.role) || !Array.isArray(message.content)) {
return [message];
}
const content = message.content.filter(block => block?.type !== 'image_url');
return content.length > 0 ? [{ ...message, content }] : [];
});
}
// "OpenAI-like" sources
const gptSources = [
chat_completion_sources.OPENAI,
@@ -2854,6 +2898,10 @@ export async function createGenerationParameters(settings, model, type, messages
generate_data.nanogpt_payg_override = settings.nanogpt_payg_override;
}
if (settings.chat_completion_source === chat_completion_sources.FIREWORKS && type !== 'quiet') {
generate_data.chat_id = getCurrentChatId();
}
if ([chat_completion_sources.MAKERSUITE, chat_completion_sources.VERTEXAI].includes(settings.chat_completion_source)) {
const stopStringsLimit = 5;
generate_data.top_k = Number(settings.top_k_openai);
@@ -3046,9 +3094,18 @@ export async function createGenerationParameters(settings, model, type, messages
}
}
// Claude Fable and Claude 5 generation models removed sampling parameters and reject
// them with HTTP 400, including via OpenAI-compatible proxies. Unanchored to also match
// prefixed ids like 'anthropic/claude-fable-5'.
if (gptSources.includes(settings.chat_completion_source) && /gpt-6-astra/.test(model)) {
generate_data.max_completion_tokens = generate_data.max_tokens;
delete generate_data.max_tokens;
delete generate_data.temperature;
delete generate_data.top_p;
delete generate_data.logprobs;
delete generate_data.top_logprobs;
}
// Claude Fable / Claude 5 models removed sampling parameters and reject them with HTTP 400,
// including via OpenAI-compatible proxies. Unanchored to also match prefixed ids
// like 'anthropic/claude-fable-5' or 'anthropic/claude-opus-5'.
if (/claude-(fable|opus-5|sonnet-5)/.test(model)) {
delete generate_data.temperature;
delete generate_data.top_p;
@@ -3056,7 +3113,7 @@ export async function createGenerationParameters(settings, model, type, messages
delete generate_data.frequency_penalty;
delete generate_data.presence_penalty;
// Keep reasoning_effort for the native Claude source, where the backend maps it to
// adaptive thinking; proxies may translate it into a thinking budget that Fable rejects.
// adaptive thinking; proxies may translate it into a thinking budget that these models reject.
if (settings.chat_completion_source !== chat_completion_sources.CLAUDE) {
delete generate_data.reasoning_effort;
}
@@ -3137,7 +3194,7 @@ async function sendOpenAIRequest(type, messages, signal, { jsonSchema = null } =
checkModerationError(data);
if (data.error) {
const message = data.error.message || response.statusText || t`Unknown error`;
const message = getChatCompletionErrorMessage(data, response);
toastr.error(message, t`API returned an error`);
throw new Error(message);
}
@@ -3229,7 +3286,7 @@ export function getStreamingReply(data, state, { chatCompletionSource = null, ov
}
});
return data.choices?.[0]?.delta?.content ?? data.choices?.[0]?.message?.content ?? data.choices?.[0]?.text ?? '';
} else if ([chat_completion_sources.CUSTOM, chat_completion_sources.POLLINATIONS, chat_completion_sources.AIMLAPI, chat_completion_sources.MOONSHOT, chat_completion_sources.COMETAPI, chat_completion_sources.ELECTRONHUB, chat_completion_sources.NANOGPT, chat_completion_sources.ZAI, chat_completion_sources.SILICONFLOW, chat_completion_sources.CHUTES, chat_completion_sources.WORKERS_AI].includes(chat_completion_source)) {
} else if ([chat_completion_sources.CUSTOM, chat_completion_sources.POLLINATIONS, chat_completion_sources.AIMLAPI, chat_completion_sources.MOONSHOT, chat_completion_sources.COMETAPI, chat_completion_sources.ELECTRONHUB, chat_completion_sources.NANOGPT, chat_completion_sources.ZAI, chat_completion_sources.SILICONFLOW, chat_completion_sources.CHUTES, chat_completion_sources.WORKERS_AI, chat_completion_sources.FIREWORKS].includes(chat_completion_source)) {
if (show_thoughts) {
state.reasoning +=
data.choices?.filter(x => x?.delta?.reasoning_content)?.[0]?.delta?.reasoning_content ??
@@ -5035,6 +5092,7 @@ function getMaxContextOpenAI(value) {
/** @type {[RegExp, number][]} */
const contextMap = [
[/^gpt-6-astra/, max_1050k],
[/^gpt-5\.6/, max_1050k],
[/^gpt-5\.[45]/, max_1mil],
[/^gpt-5/, max_400k],
@@ -5672,7 +5730,7 @@ async function onModelChange() {
if (oai_settings.chat_completion_source == chat_completion_sources.CLAUDE) {
if (oai_settings.max_context_unlocked) {
$('#openai_max_context').attr('max', unlocked_max);
} else if (/^claude-(sonnet-4-5|sonnet-4-6|opus-4-6|opus-4-7|opus-4-8|fable|opus-5|sonnet-5)/.test(value)) {
} else if (/^claude-(sonnet-4-5|sonnet-4-6|sonnet-5|opus-4-6|opus-4-7|opus-4-8|opus-5|fable)/.test(value)) {
$('#openai_max_context').attr('max', max_1mil);
} else if (/^claude-(3|opus|haiku|sonnet)/.test(value)) {
$('#openai_max_context').attr('max', max_200k);
@@ -6135,10 +6193,8 @@ function reconnectOpenAi() {
}
}
function onProxyPasswordShowClick() {
const $input = $('#openai_proxy_password');
const type = $input.attr('type') === 'password' ? 'text' : 'password';
$input.attr('type', type);
function onProxyAccessKeyShowClick() {
$('#openai_proxy_access_key').toggleClass('masked-secret');
$(this).toggleClass('fa-eye-slash fa-eye');
}
@@ -6185,6 +6241,7 @@ export function isImageInliningSupported() {
'gpt-4.1',
'gpt-4.5-preview',
'gpt-4o',
'gpt-6-astra',
'gpt-5',
'o1',
'o3',
@@ -6192,10 +6249,10 @@ export function isImageInliningSupported() {
// Claude
'claude-3',
'claude-fable',
'claude-opus-5',
'claude-sonnet-5',
'claude-opus-4',
'claude-opus-5',
'claude-sonnet-4',
'claude-sonnet-5',
'claude-haiku-4',
// Cohere
'c4ai-aya-vision',
@@ -6228,6 +6285,8 @@ export function isImageInliningSupported() {
'moonshot-v1-128k-vision-preview',
'kimi-k2.5',
'kimi-latest',
// DeepSeek
'deepseek-v4-flash-vision-exp',
// Z.AI (GLM)
'glm-4.5v',
'glm-4.6v',
@@ -6287,10 +6346,14 @@ export function isImageInliningSupported() {
return visionSupportedModels.some(model => oai_settings.zai_model.includes(model));
case chat_completion_sources.SILICONFLOW:
return visionSupportedModels.some(model => oai_settings.siliconflow_model.includes(model));
case chat_completion_sources.DEEPSEEK:
return visionSupportedModels.some(model => oai_settings.deepseek_model.includes(model));
case chat_completion_sources.WORKERS_AI: {
const waiModel = Array.isArray(model_list) && model_list.find(m => m.id === oai_settings.workers_ai_model);
return Boolean(waiModel && Array.isArray(waiModel.properties) && waiModel.properties.some(p => p.property_id === 'vision' && p.value === 'true'));
}
case chat_completion_sources.FIREWORKS:
return (Array.isArray(model_list) && model_list.find(m => m.id === oai_settings.fireworks_model)?.supports_image_input);
default:
return false;
}
@@ -6459,7 +6522,7 @@ function setProxyPreset(name, url, password) {
oai_settings.reverse_proxy = url;
$('#openai_reverse_proxy').val(oai_settings.reverse_proxy);
oai_settings.proxy_password = password;
$('#openai_proxy_password').val(oai_settings.proxy_password);
$('#openai_proxy_access_key').val(oai_settings.proxy_password);
reconnectOpenAi();
}
@@ -6478,7 +6541,7 @@ function onProxyPresetChange() {
$('#save_proxy').on('click', async function () {
const presetName = $('#openai_reverse_proxy_name').val();
const reverseProxy = $('#openai_reverse_proxy').val();
const proxyPassword = $('#openai_proxy_password').val();
const proxyPassword = $('#openai_proxy_access_key').val();
setProxyPreset(presetName, reverseProxy, proxyPassword);
saveSettingsDebounced();
@@ -6512,7 +6575,7 @@ $('#delete_proxy').on('click', async function () {
oai_settings.reverse_proxy = selected_proxy.url;
$('#openai_reverse_proxy').val(selected_proxy.url);
oai_settings.proxy_password = selected_proxy.password;
$('#openai_proxy_password').val(selected_proxy.password);
$('#openai_proxy_access_key').val(selected_proxy.password);
saveSettingsDebounced();
$('#openai_proxy_preset').val(selected_proxy.name);
@@ -6927,11 +6990,17 @@ export function initOpenAI() {
saveSettingsDebounced();
});
$('#openai_proxy_password').on('input', function () {
$('#openai_proxy_access_key').on('input', function () {
oai_settings.proxy_password = String($(this).val());
saveSettingsDebounced();
});
$('#openai_proxy_access_key').on('copy cut', function (event) {
if ($(this).hasClass('masked-secret')) {
event.preventDefault();
}
});
$('#claude_assistant_prefill').on('input', function () {
oai_settings.assistant_prefill = String($(this).val());
saveSettingsDebounced();
@@ -7321,7 +7390,7 @@ export function initOpenAI() {
$('#openai_logit_bias_export_preset').on('click', onLogitBiasPresetExportClick);
$('#openai_logit_bias_delete_preset').on('click', onLogitBiasPresetDeleteClick);
$('#import_oai_preset').on('click', onImportPresetClick);
$('#openai_proxy_password_show').on('click', onProxyPasswordShowClick);
$('#openai_proxy_access_key_show').on('click', onProxyAccessKeyShowClick);
$('#customize_additional_parameters').on('click', onCustomizeParametersClick);
$('#openai_proxy_preset').on('change', onProxyPresetChange);
}
+33 -9
View File
@@ -2818,9 +2818,10 @@ async function loadUntilMesId(mesId) {
return target;
}
async function doMesCut(_, text) {
async function doMesCut(args, text) {
console.debug(`was asked to cut message id #${text}`);
const range = stringToRange(text, 0, chat.length - 1);
const deleteToolCalls = args?.toolcalls === undefined || isTrueBoolean(args.toolcalls);
//reject invalid args or no args
if (!range) {
@@ -2828,12 +2829,16 @@ async function doMesCut(_, text) {
return;
}
let totalMesToCut = (range.end - range.start) + 1;
let mesIDToCut = range.start;
const messagesToCut = chat.slice(range.start, range.end + 1);
let cutText = '';
for (let i = 0; i < totalMesToCut; i++) {
cutText += (chat[mesIDToCut]?.mes || '') + '\n';
for (const message of messagesToCut) {
const mesIDToCut = chat.indexOf(message);
if (mesIDToCut === -1) {
continue;
}
cutText += (message?.mes || '') + '\n';
let mesToCut = $('#chat').find(`.mes[mesid=${mesIDToCut}]`);
if (!mesToCut.length) {
@@ -2845,7 +2850,7 @@ async function doMesCut(_, text) {
}
setEditedMessageId(mesIDToCut);
await deleteMessage(mesIDToCut, null, false);
await deleteMessage(mesIDToCut, null, false, deleteToolCalls);
}
await saveChatConditional();
@@ -2853,7 +2858,7 @@ async function doMesCut(_, text) {
return cutText;
}
async function doDelMode(_, text) {
async function doDelMode(args, text) {
//reject invalid args
if (text && isNaN(text)) {
toastr.warning('Must enter a number or nothing.');
@@ -2862,7 +2867,8 @@ async function doDelMode(_, text) {
// Just enter the delete mode.
if (!text) {
$('#option_delete_mes').trigger('click', { fromSlashCommand: true });
const deleteToolCalls = args?.toolcalls === undefined || isTrueBoolean(args.toolcalls);
$('#option_delete_mes').trigger('click', { fromSlashCommand: true, deleteToolCalls });
return '';
}
@@ -2879,7 +2885,7 @@ async function doDelMode(_, text) {
}
const range = `${chat.length - count}-${chat.length - 1}`;
return doMesCut(_, range);
return doMesCut(args, range);
}
function doResetPanels() {
@@ -4145,6 +4151,15 @@ jQuery(() => {
name: 'del',
callback: doDelMode,
aliases: ['delete', 'delmode'],
namedArgumentList: [
SlashCommandNamedArgument.fromProps({
name: 'toolcalls',
description: 'also delete associated tool-call messages',
typeList: [ARGUMENT_TYPE.BOOLEAN],
defaultValue: 'true',
enumList: commonEnumProviders.boolean('trueFalse')(),
}),
],
unnamedArgumentList: [
new SlashCommandArgument(
'optional number', [ARGUMENT_TYPE.NUMBER], false,
@@ -4157,6 +4172,15 @@ jQuery(() => {
name: 'cut',
callback: doMesCut,
returns: 'the text of cut messages separated by a newline',
namedArgumentList: [
SlashCommandNamedArgument.fromProps({
name: 'toolcalls',
description: 'also delete associated tool-call messages',
typeList: [ARGUMENT_TYPE.BOOLEAN],
defaultValue: 'true',
enumList: commonEnumProviders.boolean('trueFalse')(),
}),
],
unnamedArgumentList: [
SlashCommandArgument.fromProps({
description: 'number or range',
+1
View File
@@ -154,6 +154,7 @@ export function extractReasoningFromData(data, {
case chat_completion_sources.SILICONFLOW:
case chat_completion_sources.ZAI:
case chat_completion_sources.WORKERS_AI:
case chat_completion_sources.FIREWORKS:
case chat_completion_sources.CUSTOM: {
return data?.choices?.[0]?.message?.reasoning_content
?? data?.choices?.[0]?.message?.reasoning
+18 -14
View File
@@ -2289,28 +2289,32 @@ function printViewTagList(tagContainer, empty = true) {
function removeMissingTagFilters() {
const tagIds = new Set(tags.map(tag => tag.id));
const assignedTagIds = new Set(Object.values(tag_map).flat());
const openBogusFolderIds = new Set(getOpenBogusFolders().map(tag => tag.id));
const isEmptyOpenBogusFolder = (tagId) => openBogusFolderIds.has(tagId) && !assignedTagIds.has(tagId);
// Filter lists only print tags that are assigned to at least one entity. A filter on an unassigned
// tag therefore has no element to toggle, and "Clear all filters" can't reset it either, because it
// works by clicking the printed elements. Drop those filters instead of leaving them stuck.
const isUnclearable = (tagId) => !tagIds.has(tagId) || !assignedTagIds.has(tagId);
for (const helper of [groupCandidatesFilter, groupMembersFilter, entitiesFilter]) {
const { selected, excluded } = helper.getFilterData(FILTER_TYPES.TAG);
const storagePrefix = getFilterStorageKey(helper);
let anyRemoved = false;
if (Array.isArray(selected)) {
for (let i = selected.length - 1; i >= 0; i--) {
if (!tagIds.has(selected[i]) || isEmptyOpenBogusFolder(selected[i])) {
selected.splice(i, 1);
anyRemoved = true;
}
for (const tagIdList of [selected, excluded]) {
if (!Array.isArray(tagIdList)) {
continue;
}
}
if (Array.isArray(excluded)) {
for (let i = excluded.length - 1; i >= 0; i--) {
if (!tagIds.has(excluded[i]) || isEmptyOpenBogusFolder(excluded[i])) {
excluded.splice(i, 1);
anyRemoved = true;
for (let i = tagIdList.length - 1; i >= 0; i--) {
if (!isUnclearable(tagIdList[i])) {
continue;
}
if (storagePrefix) {
accountStorage.removeItem(`${storagePrefix}_tag_${tagIdList[i]}`);
}
tagIdList.splice(i, 1);
anyRemoved = true;
}
}
+39 -2
View File
@@ -610,15 +610,37 @@ export function loadGenericModels(data) {
return;
}
data.sort((a, b) => a.id.localeCompare(b.id));
const models = data
.filter(model => model && typeof model.id === 'string' && model.id.length > 0)
.sort((a, b) => a.id.localeCompare(b.id));
const dataList = $('#generic_model_fill');
dataList.empty();
for (const model of data) {
const modelSelect = $('#generic_model_select');
modelSelect.empty();
modelSelect.append($('<option>', { value: '' }));
for (const model of models) {
const option = document.createElement('option');
option.value = model.id;
option.text = model.id;
dataList.append(option);
const selectOption = document.createElement('option');
selectOption.value = model.id;
selectOption.text = model.id;
selectOption.selected = model.id === textgen_settings.generic_model;
modelSelect.append(selectOption);
}
// Keep free-text entry for IDs that are not in the /v1/models list
if (textgen_settings.generic_model && !models.find(x => x.id === textgen_settings.generic_model)) {
modelSelect.append($('<option>', {
value: textgen_settings.generic_model,
text: textgen_settings.generic_model,
selected: true,
}));
}
}
@@ -1051,6 +1073,13 @@ function onLlamaCppModelSelect() {
$('#api_button_textgenerationwebui').trigger('click');
}
function onGenericModelSelect() {
const modelId = String($('#generic_model_select').val() ?? '');
textgen_settings.generic_model = modelId;
$('#generic_model_textgenerationwebui').val(modelId);
$('#api_button_textgenerationwebui').trigger('click');
}
function onOpenRouterModelSelect() {
const modelId = String($('#openrouter_model').val());
textgen_settings.openrouter_model = modelId;
@@ -1368,6 +1397,7 @@ export function initTextGenModels() {
$('#tabby_download_model').on('click', downloadTabbyModel);
$('#tabby_model').on('change', onTabbyModelSelect);
$('#llamacpp_model').on('change', onLlamaCppModelSelect);
$('#generic_model_select').on('change', onGenericModelSelect);
$('#featherless_model').on('change', () => onFeatherlessModelSelect(String($('#featherless_model').val())));
const providersSelect = $('.openrouter_providers');
@@ -1421,6 +1451,13 @@ export function initTextGenModels() {
width: '100%',
allowClear: true,
});
$('#generic_model_select').select2({
placeholder: t`Select a model`,
searchInputPlaceholder: t`Search models...`,
searchInputCssClass: 'text_pole',
width: '100%',
allowClear: true,
});
$('#model_infermaticai_select').select2({
placeholder: t`Select a model`,
searchInputPlaceholder: t`Search models...`,
+1 -1
View File
@@ -639,7 +639,7 @@ export function getTokenizerModel() {
}
if (oai_settings.chat_completion_source == chat_completion_sources.ELECTRONHUB && oai_settings.electronhub_model) {
if (oai_settings.electronhub_model.includes('gpt-4o') || oai_settings.electronhub_model.includes('gpt-5')) {
if (oai_settings.electronhub_model.includes('gpt-4o') || oai_settings.electronhub_model.includes('gpt-5') || oai_settings.electronhub_model.includes('gpt-6-astra')) {
return gpt4oTokenizer;
} else if (oai_settings.electronhub_model.includes('gpt-4.1') || oai_settings.electronhub_model.includes('gpt-4.5')) {
return gpt4oTokenizer;
+13 -1
View File
@@ -580,7 +580,13 @@ export class ToolManager {
}
if (typeof deltaValue === 'string') {
if (typeof targetValue === 'string') {
// `id`, `name`, `type` are sent in full by some providers on every
// streaming chunk; concatenating them would duplicate the value.
if (key === 'id' || key === 'name' || key === 'type') {
if (!targetValue) {
target[key] = deltaValue;
}
} else if (typeof targetValue === 'string') {
// Concatenate strings
target[key] = targetValue + deltaValue;
} else {
@@ -613,6 +619,12 @@ export class ToolManager {
return false;
}
// GPT-6 Astra supports tool calling only through the Responses API.
if ([chat_completion_sources.OPENAI, chat_completion_sources.AZURE_OPENAI].includes(settings.chat_completion_source)
&& /^gpt-6-astra/.test(model)) {
return false;
}
// Post-processing will forcefully remove past tool calls from the prompt, making them useless
const { NONE, MERGE_TOOLS, SEMI_TOOLS, STRICT_TOOLS } = custom_prompt_post_processing_types;
const allowedPromptPostProcessing = [NONE, MERGE_TOOLS, SEMI_TOOLS, STRICT_TOOLS];
+5
View File
@@ -2681,6 +2681,11 @@ textarea::placeholder {
height: fit-content;
}
/* Keep non-login secrets visually masked without using password input semantics. */
.masked-secret {
-webkit-text-security: disc;
}
select.text_pole {
padding-right: 20px;
}
+2 -2
View File
@@ -4,7 +4,7 @@ import yaml from 'yaml';
import color from 'chalk';
import _ from 'lodash';
import { serverDirectory } from './server-directory.js';
import { keyToEnv, setConfigFilePath } from './util.js';
import { keyToEnv, setConfigFilePath, stringToBool } from './util.js';
const keyMigrationMap = [
{
@@ -181,7 +181,7 @@ export function addMissingConfigValues(configPath) {
const newEnvKey = keyToEnv(newKey);
if (process.env[oldEnvKey] && !process.env[newEnvKey]) {
const oldValue = process.env[oldEnvKey];
const newValue = migrate(oldValue);
const newValue = migrate(stringToBool(oldValue));
process.env[newEnvKey] = newValue;
delete process.env[oldEnvKey];
console.warn(color.yellow(`Warning: Using a deprecated environment variable: ${oldEnvKey}. Please use ${newEnvKey} instead.`));
+2 -1
View File
@@ -456,7 +456,7 @@ export const AZURE_OPENAI_KEYS = [
'reasoning_effort',
];
export const OPENAI_VERBOSITY_MODELS = /^gpt-5/;
export const OPENAI_VERBOSITY_MODELS = /^(?:gpt-5|gpt-6-astra)/;
export const OPENAI_REASONING_EFFORT_MODELS = [
'o1',
@@ -491,6 +491,7 @@ export const OPENAI_REASONING_EFFORT_MODELS = [
'gpt-5.6-sol',
'gpt-5.6-terra',
'gpt-5.6-luna',
'gpt-6-astra',
];
export const OPENAI_REASONING_EFFORT_MAP = {
+112 -32
View File
@@ -1,4 +1,5 @@
/* eslint-disable dot-notation */
import { createHmac } from 'node:crypto';
import process from 'node:process';
import util from 'node:util';
import express from 'express';
@@ -66,6 +67,8 @@ import {
getWebTokenizer,
} from '../tokenizers.js';
import { getVertexAIAuth, getProjectIdFromServiceAccount } from '../google.js';
import { getCookieSecret } from '../../users.js';
import { fetchGoogleModels, GoogleModelsHttpError } from './google-models.js';
const API_OPENAI = 'https://api.openai.com/v1';
const API_CLAUDE = 'https://api.anthropic.com/v1';
@@ -108,6 +111,18 @@ const cachingAtDepth = (() => {
})();
const enableAdaptiveThinking = getConfigValue('claude.enableAdaptiveThinking', true, 'boolean');
/**
* Lazily-cached HMAC key (instance cookie secret) for session-affinity hashing.
* @type {string|undefined}
*/
let affinityKey;
function getAffinityKey() {
if (affinityKey === undefined) {
affinityKey = getCookieSecret(globalThis.DATA_ROOT);
}
return affinityKey;
}
/**
* Cache for cacheable (writing) OpenRouter model IDs.
* @type {string[]}
@@ -235,8 +250,8 @@ async function sendClaudeRequest(request, response) {
const convertedPrompt = convertClaudeMessages(request.body.messages, request.body.assistant_prefill, useSystemPrompt, useTools, getPromptNames(request));
// Unanchored to also match prefixed ids passed through proxies, e.g. 'anthropic/claude-fable-5'
const isFableModel = /claude-fable/.test(request.body.model);
// Claude 5 generation (opus-5, sonnet-5): adaptive-only thinking, sampling parameters removed, no prefill
const isClaude5Model = /^claude-(opus-5|sonnet-5)/.test(request.body.model);
const isFable51Model = /claude-fable-5-1/.test(request.body.model);
const isClaude5Model = /claude-(opus-5|sonnet-5)/.test(request.body.model);
const useThinking = /^claude-(3-7|opus-4|sonnet-4|haiku-4-5|opus-4-5|opus-4-6|sonnet-4-6|opus-4-7)/.test(request.body.model) || isFableModel || isClaude5Model;
const useWebSearch = (/^claude-(3-5|3-7|opus-4|sonnet-4|haiku-4-5|opus-4-5|opus-4-6|sonnet-4-6|opus-4-7)/.test(request.body.model) || isFableModel || isClaude5Model) && Boolean(request.body.enable_web_search);
const isLimitedSampling = /^claude-(opus-4-1|sonnet-4-5|haiku-4-5|opus-4-5|opus-4-6|sonnet-4-6)/.test(request.body.model);
@@ -284,15 +299,24 @@ async function sendClaudeRequest(request, response) {
}
}
// Structured output is a forced tool
// Fable 5.1 rejects forced tools, but supports native JSON outputs.
if (request.body.json_schema) {
const jsonTool = {
name: request.body.json_schema.name,
description: request.body.json_schema.description || 'Well-formed JSON object',
input_schema: request.body.json_schema.value,
};
requestBody.tools = [...(requestBody.tools || []), jsonTool];
requestBody.tool_choice = { type: 'tool', name: request.body.json_schema.name };
if (isFable51Model) {
requestBody.output_config = {
format: {
type: 'json_schema',
schema: request.body.json_schema.value,
},
};
} else {
const jsonTool = {
name: request.body.json_schema.name,
description: request.body.json_schema.description || 'Well-formed JSON object',
input_schema: request.body.json_schema.value,
};
requestBody.tools = [...(requestBody.tools || []), jsonTool];
requestBody.tool_choice = { type: 'tool', name: request.body.json_schema.name };
}
}
if (useWebSearch) {
@@ -341,8 +365,8 @@ async function sendClaudeRequest(request, response) {
requestBody.output_config.effort = budgetTokens;
// top_k is not allowed in adaptive mode
delete requestBody.top_k;
} else if (useThinking && isFableModel && reasoningEffort === 'auto' && includeReasoning) {
// Fable auto thinking is already enabled, but readable summaries require an explicit display request.
} else if (useThinking && (isFableModel || isClaude5Model) && reasoningEffort === 'auto' && includeReasoning) {
// Fable/Claude 5 auto thinking is already enabled, but readable summaries require an explicit display request.
fixThinkingPrefill = true;
requestBody.thinking = { type: 'adaptive', display: 'summarized' };
} else if (useThinking && Number.isInteger(budgetTokens)) {
@@ -501,6 +525,8 @@ async function sendMakerSuiteRequest(request, response) {
const isThinkingConfigModel = m => (/^gemini-2.5-(flash|pro)/.test(m) && !/-image(-preview)?$/.test(m)) || (/^gemini-3[.\d]*-(flash|pro)/.test(m));
const isImageSizeModel = m => /^gemini-3/.test(m);
// https://ai.google.dev/gemini-api/docs/latest-model#api-changes-and-parameter-updates
const noSamplingModel = /gemini-3\.[67]-flash|gemini-3\.5-flash-lite/.test(model);
const noSearchModels = [
'gemini-2.0-flash-lite',
@@ -514,6 +540,13 @@ async function sendMakerSuiteRequest(request, response) {
delete generationConfig.stopSequences;
}
if (noSamplingModel) {
delete generationConfig.temperature;
delete generationConfig.topP;
delete generationConfig.topK;
delete generationConfig.candidateCount;
}
const enableImageModality = requestImages && imageGenerationModels.includes(model);
const enableImageConfig = enableImageModality && (aspectRatio || imageSize);
if (enableImageModality) {
@@ -1817,9 +1850,61 @@ router.post('/status', async function (request, statusResponse) {
apiKey = request.body.reverse_proxy ? request.body.proxy_password : readSecret(request.user.directories, SECRET_KEYS.MOONSHOT, request.body.secret_id);
headers = {};
} else if (request.body.chat_completion_source === CHAT_COMPLETION_SOURCES.FIREWORKS) {
apiUrl = API_FIREWORKS;
apiKey = readSecret(request.user.directories, SECRET_KEYS.FIREWORKS, request.body.secret_id);
headers = {};
const modelsUrl = 'https://api.fireworks.ai/v1/accounts/fireworks/models?filter=supports_serverless%3Dtrue&pageSize=200';
try {
const response = await fetch(modelsUrl, {
method: 'GET',
headers: {
'Authorization': 'Bearer ' + apiKey,
...headers,
},
});
if (response.ok) {
/** @type {any} */
const data = await response.json();
const models = Array.isArray(data?.models)
? data.models
.filter(m => m?.contextLength > 0 && m?.kind !== 'EMBEDDING_MODEL')
.map(m => ({
id: m.name,
name: m.displayName,
context_length: m.contextLength,
supports_tools: m.supportsTools,
supports_image_input: m.supportsImageInput,
}))
: [];
// Add fast router versions for models that have them
const fastRouters = {
'accounts/fireworks/models/glm-5p2': 'accounts/fireworks/routers/glm-5p2-fast',
'accounts/fireworks/models/kimi-k2p6': 'accounts/fireworks/routers/kimi-k2p6-fast',
'accounts/fireworks/models/kimi-k2p7-code': 'accounts/fireworks/routers/kimi-k2p7-code-fast',
'accounts/fireworks/models/kimi-k3': 'accounts/fireworks/routers/kimi-k3-fast',
};
for (const [standardId, fastId] of Object.entries(fastRouters)) {
const standard = models.find(m => m.id === standardId);
if (standard) {
models.push({
...standard,
id: fastId,
name: standard.name + ' (fast)',
});
}
}
console.debug('Available Fireworks models:', models.map(m => m.id));
return statusResponse.send({ data: models });
} else {
console.warn('Fireworks models endpoint failed:', response.status, response.statusText);
return statusResponse.send({ error: true, data: { data: [] } });
}
} catch (error) {
console.error('Error fetching Fireworks models:', error);
return statusResponse.send({ error: true, data: { data: [] } });
}
} else if (request.body.chat_completion_source === CHAT_COMPLETION_SOURCES.MAKERSUITE) {
apiKey = request.body.reverse_proxy ? request.body.proxy_password : readSecret(request.user.directories, SECRET_KEYS.MAKERSUITE, request.body.secret_id);
apiUrl = trimTrailingSlash(request.body.reverse_proxy || API_MAKERSUITE);
@@ -1834,26 +1919,15 @@ router.post('/status', async function (request, statusResponse) {
}
try {
const response = await fetch(modelsUrl);
if (response.ok) {
/** @type {any} */
const data = await response.json();
// Transform Google AI Studio models to OpenAI format
const models = data.models
?.filter(model => model.supportedGenerationMethods?.includes('generateContent'))
?.map(model => ({
...model,
id: model.name.replace('models/', ''),
})) || [];
console.info('Available Google AI Studio models:', models.map(m => m.id));
return statusResponse.send({ data: models });
} else {
console.warn('Google AI Studio models endpoint failed:', response.status, response.statusText);
const models = await fetchGoogleModels(modelsUrl);
console.info('Available Google AI Studio models:', models.map(m => m.id));
return statusResponse.send({ data: models });
} catch (error) {
if (error instanceof GoogleModelsHttpError) {
console.warn('Google AI Studio models endpoint failed:', error.status, error.statusText);
return statusResponse.send({ error: true, bypass: true, data: { data: [] } });
}
} catch (error) {
console.error('Error fetching Google AI Studio models:', error);
return statusResponse.send({ error: true, bypass: true, data: { data: [] } });
}
@@ -2382,6 +2456,9 @@ router.post('/generate', async function (request, response) {
apiKey = readSecret(request.user.directories, SECRET_KEYS.FIREWORKS, request.body.secret_id);
headers = {};
bodyParams = {};
if (request.body.reasoning_effort) {
bodyParams['reasoning_effort'] = request.body.reasoning_effort;
}
if (request.body.json_schema) {
bodyParams['response_format'] = {
type: 'json_schema',
@@ -2393,6 +2470,9 @@ router.post('/generate', async function (request, response) {
},
};
}
if (request.body.chat_id) {
headers['x-session-affinity'] = createHmac('sha256', getAffinityKey()).update(request.body.chat_id).digest('hex').slice(0, 16);
}
} else if (request.body.chat_completion_source === CHAT_COMPLETION_SOURCES.NANOGPT) {
apiUrl = API_NANOGPT;
apiKey = readSecret(request.user.directories, SECRET_KEYS.NANOGPT, request.body.secret_id);
+99
View File
@@ -0,0 +1,99 @@
import fetch from 'node-fetch';
const GOOGLE_MODELS_PAGE_SIZE = 1000;
export class GoogleModelsHttpError extends Error {
/**
* @param {number} status HTTP status code
* @param {string} statusText HTTP status text
*/
constructor(status, statusText) {
super(`Google AI Studio models endpoint returned ${status} ${statusText}`.trim());
this.name = 'GoogleModelsHttpError';
this.status = status;
this.statusText = statusText;
}
}
export class GoogleModelsResponseError extends Error {
/**
* @param {string} message Error message
* @param {ErrorOptions} [options] Error options
*/
constructor(message, options) {
super(message, options);
this.name = 'GoogleModelsResponseError';
}
}
/**
* Fetches every page of Google AI Studio models and converts supported models.
*
* @param {string | URL} modelsUrl Google AI Studio models endpoint
* @param {typeof fetch} [fetchImpl] Fetch implementation
* @returns {Promise<object[]>} Models that support content generation
*/
export async function fetchGoogleModels(modelsUrl, fetchImpl = fetch) {
const requestUrl = new URL(modelsUrl);
requestUrl.searchParams.set('pageSize', String(GOOGLE_MODELS_PAGE_SIZE));
requestUrl.searchParams.delete('pageToken');
const models = [];
const pageTokens = new Set();
let pageNumber = 1;
while (true) {
let response;
try {
response = await fetchImpl(requestUrl.toString());
} catch (cause) {
throw new GoogleModelsResponseError(`Failed to fetch Google AI Studio models page ${pageNumber}.`, { cause });
}
if (!response.ok) {
throw new GoogleModelsHttpError(response.status, response.statusText);
}
let data;
try {
data = await response.json();
} catch (cause) {
throw new GoogleModelsResponseError(`Failed to parse Google AI Studio models page ${pageNumber}.`, { cause });
}
if (!data || !Array.isArray(data.models)) {
throw new GoogleModelsResponseError(`Google AI Studio models page ${pageNumber} has an invalid models field.`);
}
models.push(...data.models);
const nextPageToken = data.nextPageToken;
if (nextPageToken === undefined || nextPageToken === '') {
break;
}
if (typeof nextPageToken !== 'string') {
throw new GoogleModelsResponseError(`Google AI Studio models page ${pageNumber} has an invalid nextPageToken.`);
}
if (pageTokens.has(nextPageToken)) {
throw new GoogleModelsResponseError(`Google AI Studio models pagination repeated token on page ${pageNumber}.`);
}
pageTokens.add(nextPageToken);
requestUrl.searchParams.set('pageToken', nextPageToken);
pageNumber++;
}
return models
.filter(model => model?.supportedGenerationMethods?.includes('generateContent'))
.map((model, index) => {
if (typeof model.name !== 'string') {
throw new GoogleModelsResponseError(`Google AI Studio model at index ${index} has an invalid name.`);
}
return {
...model,
id: model.name.replace('models/', ''),
};
});
}
+6 -4
View File
@@ -14,7 +14,7 @@ import storage from 'node-persist';
import { AVATAR_WIDTH, AVATAR_HEIGHT, DEFAULT_AVATAR_PATH } from '../constants.js';
import { default as validateAvatarUrlMiddleware, getFileNameValidationFunction, forbiddenRegExp } from '../middleware/validateFileName.js';
import { deepMerge, humanizedDateTime, tryParse, MemoryLimitedMap, getConfigValue, mutateJsonString, clientRelativePath, getUniqueName, sanitizeSafeCharacterReplacements } from '../util.js';
import { deepMerge, humanizedDateTime, tryParse, MemoryLimitedMap, getConfigValue, mutateJsonString, clientRelativePath, getUniqueName, sanitizeSafeCharacterReplacements, getArrayBufferSlice } from '../util.js';
import { TavernCardValidator } from '../validator/TavernCardValidator.js';
import { parse, read, write } from '../character-card-parser.js';
import { readWorldInfoFile } from './worldinfo.js';
@@ -765,7 +765,7 @@ async function importFromYaml(uploadPath, context, preservedFileName) {
async function importFromCharX(uploadPath, { request }, preservedFileName) {
const fileBuffer = fs.readFileSync(uploadPath);
// Create a properly-sized ArrayBuffer (Node's buffer pool can cause oversized .buffer)
const data = fileBuffer.buffer.slice(fileBuffer.byteOffset, fileBuffer.byteOffset + fileBuffer.byteLength);
const data = getArrayBufferSlice(fileBuffer);
fs.unlinkSync(uploadPath);
const parser = new CharXParser(data);
@@ -801,7 +801,7 @@ async function importFromCharX(uploadPath, { request }, preservedFileName) {
}
async function importFromByaf(uploadPath, { request }, preservedFileName) {
const data = (await fsPromises.readFile(uploadPath)).buffer;
const data = getArrayBufferSlice(await fsPromises.readFile(uploadPath));
await fsPromises.unlink(uploadPath);
console.info('Importing from BYAF');
@@ -1081,7 +1081,9 @@ router.post('/rename', validateAvatarUrlMiddleware, async function (request, res
// Rename chats folder
if (fs.existsSync(oldChatsPath) && !fs.existsSync(newChatsPath)) {
fs.cpSync(oldChatsPath, newChatsPath, { recursive: true });
// Supplying a filter avoids a Node.js Windows copyDir crash while preserving all entries.
// https://github.com/nodejs/node/issues/63970
fs.cpSync(oldChatsPath, newChatsPath, { recursive: true, filter: () => true });
fs.rmSync(oldChatsPath, { recursive: true, force: true });
}
+40 -12
View File
@@ -1,3 +1,4 @@
import crypto from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import readline from 'node:readline';
@@ -30,6 +31,23 @@ const checkIntegrity = !!getConfigValue('backups.chat.checkIntegrity', true, 'bo
export const CHAT_BACKUPS_PREFIX = 'chat_';
/**
* Builds a stable filename key for a chat's backups.
* Non-ASCII characters are replaced with underscores, so names such as CJK ones
* would all collapse to the same key and share one backup quota. A short hash of
* the raw name keeps those keys distinct while ASCII names stay unchanged (#5780).
* @param {string} name The name of the chat.
* @returns {string} Sanitized filename key for the backup files.
*/
export function getBackupKey(name) {
const sanitized = sanitize(name).replace(/[^a-z0-9]/gi, '_').toLowerCase();
if (/[^\x20-\x7E]/.test(name)) {
const hash = crypto.createHash('sha256').update(name).digest('hex').slice(0, 8);
return `${sanitized}_${hash}`;
}
return sanitized;
}
/**
* Saves a chat to the backups directory.
* @param {string} directory The user's backup directory.
@@ -44,8 +62,7 @@ function backupChat(directory, name, data, backupPrefix = CHAT_BACKUPS_PREFIX) {
if (!fs.existsSync(directory)) {
console.error(`The chat couldn't be backed up because no directory exists at ${directory}!`);
}
// replace non-alphanumeric characters with underscores
name = sanitize(name).replace(/[^a-z0-9]/gi, '_').toLowerCase();
name = getBackupKey(name);
const backupFile = path.join(directory, `${backupPrefix}${name}_${generateTimestamp()}.jsonl`);
@@ -66,15 +83,19 @@ function backupChat(directory, name, data, backupPrefix = CHAT_BACKUPS_PREFIX) {
const backupFunctions = new Map();
/**
* Gets a backup function for a user.
* Gets a backup function for a user and chat.
* Throttling is keyed per user and chat so that rapid saves in one chat cannot
* swallow the throttled backup of another chat saved in the same window.
* @param {string} handle User handle
* @param {string} name The name of the chat, as passed to backupChat
* @returns {typeof backupChat} Backup function
*/
function getBackupFunction(handle) {
if (!backupFunctions.has(handle)) {
backupFunctions.set(handle, _.throttle(backupChat, throttleInterval, { leading: true, trailing: true }));
function getBackupFunction(handle, name) {
const key = `${handle} ${name}`;
if (!backupFunctions.has(key)) {
backupFunctions.set(key, _.throttle(backupChat, throttleInterval, { leading: true, trailing: true }));
}
return backupFunctions.get(handle) || (() => { });
return backupFunctions.get(key) || (() => { });
}
/**
@@ -466,12 +487,19 @@ export async function getChatInfo(pathToFile, additionalData = {}, withMetadata
res(chatData);
} else {
console.warn('Found an invalid or corrupted chat file:', pathToFile);
res({});
// The last line is unparseable or lacks known fields (e.g. a truncated write or an external edit).
// Resolve a degraded preview from the stat data instead of hiding an otherwise intact chat
// from the chat list, search and recents.
console.warn('Found an invalid or corrupted last line in a chat file:', pathToFile);
// Exclude both the metadata line and the unreadable trailing line.
chatData.chat_items = Math.max(itemCounter - 2, 0);
chatData.mes = '[The message is empty]';
chatData.match = hasMatcher ? hasAnyMatch : true;
res(chatData);
}
} else {
// The file was truncated after the stat reported a non-zero size
res({});
// The file was truncated after the stat reported a non-zero size; treat it like an empty chat
res(chatData);
}
});
});
@@ -511,7 +539,7 @@ export async function trySaveChat(chatData, filePath, skipIntegrityCheck = false
throw new IntegrityMismatchError(`Chat integrity check failed for "${filePath}". The expected integrity slug was "${chatIntegritySlug}".`);
}
tryWriteFileSync(filePath, jsonlData);
getBackupFunction(handle)(backupDirectory, cardName, jsonlData);
getBackupFunction(handle, cardName)(backupDirectory, cardName, jsonlData);
}
router.post('/save', validateAvatarUrlMiddleware, async function (request, response) {
+20 -1
View File
@@ -6,7 +6,7 @@ import sanitize from 'sanitize-filename';
import { CheckRepoActions, default as simpleGit } from 'simple-git';
import { PUBLIC_DIRECTORIES } from '../constants.js';
import { getConfigValue, isValidUrl } from '../util.js';
import { color, getConfigValue, isValidUrl } from '../util.js';
import { createGitClient } from '../git/client.js';
const gitBackend = getConfigValue('git.backend', 'auto');
@@ -486,17 +486,35 @@ router.get('/discover', function (request, response) {
fs.mkdirSync(PUBLIC_DIRECTORIES.globalExtensions);
}
/**
* A directory without a manifest.json is not a usable extension: typically a leftover
* from a manual deletion (e.g. a locked .git directory on Windows). Listing it would
* make the client report a ghost extension and request its manifest in vain on every load.
* @param {string} parentDirectory Directory containing extension folders
* @param {string} folder Extension folder name
* @returns {boolean} Whether the folder contains a manifest
*/
const hasManifest = (parentDirectory, folder) => {
const manifestExists = fs.existsSync(path.join(parentDirectory, folder, 'manifest.json'));
if (!manifestExists) {
console.warn(color.yellow(`Extension folder "${path.join(parentDirectory, folder)}" has no manifest.json and was skipped. Remove the folder to get rid of this warning.`));
}
return manifestExists;
};
// Get all folders in system extensions folder, excluding third-party
const builtInExtensions = fs
.readdirSync(PUBLIC_DIRECTORIES.extensions)
.filter(f => fs.statSync(path.join(PUBLIC_DIRECTORIES.extensions, f)).isDirectory())
.filter(f => f !== 'third-party')
.filter(f => hasManifest(PUBLIC_DIRECTORIES.extensions, f))
.map(f => ({ type: 'system', name: f }));
// Get all folders in local extensions folder
const userExtensions = fs
.readdirSync(path.join(request.user.directories.extensions))
.filter(f => fs.statSync(path.join(request.user.directories.extensions, f)).isDirectory())
.filter(f => hasManifest(request.user.directories.extensions, f))
.map(f => ({ type: 'local', name: `third-party/${f}` }));
// Get all folders in global extensions folder
@@ -504,6 +522,7 @@ router.get('/discover', function (request, response) {
const globalExtensions = fs
.readdirSync(PUBLIC_DIRECTORIES.globalExtensions)
.filter(f => fs.statSync(path.join(PUBLIC_DIRECTORIES.globalExtensions, f)).isDirectory())
.filter(f => hasManifest(PUBLIC_DIRECTORIES.globalExtensions, f))
.map(f => ({ type: 'global', name: `third-party/${f}` }))
.filter(f => !userExtensions.some(e => e.name === f.name));
+3 -4
View File
@@ -33,12 +33,11 @@ router.post('/generate-voice', async (request, response) => {
} = request.body;
const apiKey = readSecret(request.user.directories, SECRET_KEYS.MINIMAX);
const groupId = readSecret(request.user.directories, SECRET_KEYS.MINIMAX_GROUP_ID);
// Validate required parameters
if (!text || !voiceId || !apiKey || !groupId) {
if (!text || !voiceId || !apiKey) {
console.warn('MiniMax TTS: Missing required parameters');
return response.status(400).json({ error: 'Missing required parameters: text, voiceId, apiKey, and groupId are required' });
return response.status(400).json({ error: 'Missing required parameters: text, voiceId, and apiKey are required' });
}
const requestBody = {
@@ -64,7 +63,7 @@ router.post('/generate-voice', async (request, response) => {
requestBody.lang = language;
}
const apiUrl = `${apiHost}/v1/t2a_v2?GroupId=${groupId}`;
const apiUrl = `${apiHost}/v1/t2a_v2`;
console.debug('MiniMax TTS Request:', {
url: apiUrl,
-13
View File
@@ -235,19 +235,6 @@ router.post('/caption-image', async (request, response) => {
apiUrl = `${trimV1(request.body.server_url)}/v1/chat/completions`;
}
if (request.body.api === 'ooba') {
const imgMessage = body.messages.pop();
body.messages.push({
role: 'user',
content: imgMessage?.content?.[0]?.text,
});
body.messages.push({
role: 'user',
content: [],
image_url: imgMessage?.content?.[1]?.image_url?.url,
});
}
setAdditionalHeaders(request, { headers }, apiUrl);
console.debug('Multimodal captioning request', body);
+1 -1
View File
@@ -101,7 +101,7 @@ pollinations.post('/voices', async (req, res) => {
throw new Error('Invalid data format received from Pollinations');
}
const audioModelData = data.find(m => m.name === model);
const audioModelData = data.find(m => m.name === model || m.aliases?.includes(model));
if (!audioModelData || !Array.isArray(audioModelData.voices)) {
throw new Error('No voices found for the specified model');
}
+2 -1
View File
@@ -611,7 +611,8 @@ comfy.post('/generate', async (request, response) => {
}
const outputs = Object.keys(item.outputs).map(it => item.outputs[it]);
console.debug('ComfyUI outputs:', outputs);
const imgInfo = outputs.map(it => it.images).flat()[0] ?? outputs.map(it => it.gifs).flat()[0];
const imgInfo = outputs.map(it => it.images).filter(Array.isArray).flat()[0]
?? outputs.map(it => it.gifs).filter(Array.isArray).flat()[0];
if (!imgInfo) {
throw new Error('ComfyUI did not return any recognizable outputs.');
}
+1 -1
View File
@@ -442,7 +442,7 @@ export function getTokenizerModel(requestModel) {
return 'o1';
}
if (requestModel.includes('gpt-5') || requestModel.includes('o3') || requestModel.includes('o4-mini')) {
if (requestModel.includes('gpt-5') || requestModel.includes('gpt-6-astra') || requestModel.includes('o3') || requestModel.includes('o4-mini')) {
return 'o1';
}
+6
View File
@@ -26,6 +26,12 @@ export default async function corsProxyMiddleware(req, res) {
headersToRemove.forEach(header => delete headers[header]);
// Force a set of encodings that node-fetch/undici can transparently decompress.
// Some origins (e.g. Cloudflare-fronted sites) will respond with zstd when a browser's
// full Accept-Encoding list is forwarded as-is; undici does not decode zstd, so the raw
// compressed bytes would otherwise be piped straight through to the client as garbage.
headers['accept-encoding'] = 'gzip, deflate, br';
const bodyMethods = ['POST', 'PUT', 'PATCH'];
const response = await fetch(url, {
+8 -2
View File
@@ -453,13 +453,17 @@ export function convertGooglePrompt(messages, model, useSysPrompt, names) {
const system_instruction = { parts: sysPrompt.map(text => ({ text })) };
const toolNameMap = {};
// https://ai.google.dev/gemini-api/docs/latest-model#prefilled-model-turn-validation
const noPrefillModel = /gemini-3\.[67]-flash|gemini-3\.5-flash-lite/.test(model);
const contents = [];
messages.forEach((message, index) => {
// fix the roles
if (message.role === 'system' || message.role === 'tool') {
message.role = 'user';
} else if (message.role === 'assistant') {
message.role = 'model';
// A trailing model turn is a prefill, which is rejected by the newest models
message.role = noPrefillModel && index === messages.length - 1 ? 'user' : 'model';
}
// Convert the content to an array of parts
@@ -1259,11 +1263,13 @@ export function calculateGoogleBudgetTokens(maxTokens, reasoningEffort, model) {
}
function getGemini3FlashBudget() {
// https://ai.google.dev/gemini-api/docs/models/gemini-3.7-flash
const noMinimalThinking = /gemini-3\.7-flash/.test(model);
switch (reasoningEffort) {
case REASONING_EFFORT.auto:
return null;
case REASONING_EFFORT.min:
return 'minimal';
return noMinimalThinking ? 'low' : 'minimal';
case REASONING_EFFORT.low:
return 'low';
case REASONING_EFFORT.medium:
+14
View File
@@ -486,6 +486,20 @@ export async function readAllChunks(readableStream) {
});
}
/**
* Creates a precisely-sized ArrayBuffer from a Uint8Array view such as a Node Buffer.
* This avoids leaking unrelated bytes from the underlying backing store.
* @param {Uint8Array} view Source byte view
* @returns {ArrayBuffer} Exact ArrayBuffer slice for the provided view
*/
export function getArrayBufferSlice(view) {
if (!(view instanceof Uint8Array)) {
throw new TypeError('Expected Uint8Array');
}
return view.buffer.slice(view.byteOffset, view.byteOffset + view.byteLength);
}
function isObject(item) {
return (item && typeof item === 'object' && !Array.isArray(item));
}
+73
View File
@@ -0,0 +1,73 @@
import path from 'node:path';
import process from 'node:process';
import { afterAll, beforeAll, describe, expect, test } from '@jest/globals';
import archiver from 'archiver';
import { ByafParser } from '../src/byaf.js';
import { getArrayBufferSlice } from '../src/util.js';
// The parser reads the default avatar from a repo-root-relative path
const originalCwd = process.cwd();
beforeAll(() => process.chdir(path.resolve(originalCwd, '..')));
afterAll(() => process.chdir(originalCwd));
/**
* Builds a minimal valid BYAF archive in memory.
* @returns {Promise<Buffer>} Zipped BYAF bytes
*/
function buildByafArchive() {
return new Promise((resolve, reject) => {
const archive = archiver('zip', { zlib: { level: 9 } });
const chunks = [];
archive.on('data', chunk => chunks.push(chunk));
archive.on('error', reject);
archive.on('end', () => resolve(Buffer.concat(chunks)));
archive.append(JSON.stringify({
characters: ['characters/character.json'],
scenarios: ['scenarios/scenario.json'],
}), { name: 'manifest.json' });
archive.append(JSON.stringify({
name: 'Byaf Test Character',
persona: 'A test persona.',
images: [{ path: 'images/icon.png', label: '' }],
}), { name: 'characters/character.json' });
const pngPixel = Buffer.from('iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mP8z8BQDwAEhQGAhKmMIQAAAABJRU5ErkJggg==', 'base64');
archive.append(pngPixel, { name: 'characters/images/icon.png' });
archive.append(JSON.stringify({
name: 'Test Scenario',
messages: [],
}), { name: 'scenarios/scenario.json' });
archive.finalize();
});
}
describe('BYAF import buffer handling', () => {
/** @type {Buffer} */
let zipBuffer;
/** @type {Buffer} */
let pooledView;
beforeAll(async () => {
zipBuffer = await buildByafArchive();
// Simulate Node's buffer pooling: the archive bytes sit at an offset inside
// a larger backing buffer surrounded by unrelated bytes.
const backing = Buffer.alloc(zipBuffer.length + 8192, 0xFF);
zipBuffer.copy(backing, 4096);
pooledView = backing.subarray(4096, 4096 + zipBuffer.length);
});
test('parses a real BYAF archive from an exact ArrayBuffer slice of a pooled view', async () => {
const exact = getArrayBufferSlice(pooledView);
expect(exact.byteLength).toBe(zipBuffer.length);
const parser = new ByafParser(exact);
const result = await parser.parse();
expect(result.card?.data?.name).toBe('Byaf Test Character');
});
test('the raw oversized backing buffer is not parseable, which is why callers must slice', async () => {
const parser = new ByafParser(pooledView.buffer);
await expect(parser.parse()).rejects.toThrow();
});
});
+78
View File
@@ -0,0 +1,78 @@
import { beforeAll, afterAll, describe, test, expect } from '@jest/globals';
import fs from 'node:fs';
import os from 'node:os';
import path from 'node:path';
// chats.js reads these config values at module load. Environment variables take
// precedence over config.yaml, so setting them here lets the module import
// without a config file present. The short throttle interval keeps lodash
// timers from delaying the Jest process while still being long enough that a
// per-user (rather than per-chat) throttle would defer the second chat's backup.
process.env.SILLYTAVERN_BACKUPS_CHAT_ENABLED = 'true';
process.env.SILLYTAVERN_BACKUPS_CHAT_MAXTOTALBACKUPS = '-1';
process.env.SILLYTAVERN_BACKUPS_CHAT_THROTTLEINTERVAL = '200';
process.env.SILLYTAVERN_BACKUPS_CHAT_CHECKINTEGRITY = 'false';
process.env.SILLYTAVERN_BACKUPS_COMMON_NUMBEROFBACKUPS = '1';
/** @type {import('../src/endpoints/chats.js')} */
let chats;
let workDir;
beforeAll(async () => {
chats = await import('../src/endpoints/chats.js');
workDir = fs.mkdtempSync(path.join(os.tmpdir(), 'st-backups-'));
});
afterAll(() => {
fs.rmSync(workDir, { recursive: true, force: true });
});
/**
* Builds a minimal chat array like the client sends it.
* @param {string} message Message text
* @returns {object[]} Chat array
*/
function makeChat(message) {
return [
{ user_name: 'User', character_name: 'Char', chat_metadata: {} },
{ name: 'Char', is_user: false, mes: message },
];
}
describe('getBackupKey', () => {
test('keeps plain ASCII names unchanged', () => {
expect(chats.getBackupKey('Seraphina')).toBe('seraphina');
expect(chats.getBackupKey('Some Test')).toBe('some_test');
});
test('produces distinct stable keys for names that sanitize identically', () => {
const keyA = chats.getBackupKey('雷电将军');
const keyB = chats.getBackupKey('测试角色');
expect(keyA).not.toBe(keyB);
expect(chats.getBackupKey('雷电将军')).toBe(keyA);
});
});
describe('chat backups', () => {
test('chats with equal-length CJK names get independent backups and quotas', async () => {
const backupsDir = path.join(workDir, 'backups');
const chatsDir = path.join(workDir, 'chats');
fs.mkdirSync(backupsDir, { recursive: true });
fs.mkdirSync(chatsDir, { recursive: true });
// numberOfBackups is 1: with a shared backup key, the second chat's
// backup would evict the first; with a shared throttle, it would not be
// written at the leading edge at all
await chats.trySaveChat(makeChat('first'), path.join(chatsDir, 'a.jsonl'), true, 'test-user', '雷电将军', backupsDir);
await chats.trySaveChat(makeChat('second'), path.join(chatsDir, 'b.jsonl'), true, 'test-user', '测试角色', backupsDir);
const backupFiles = fs.readdirSync(backupsDir);
expect(backupFiles).toHaveLength(2);
const keyA = chats.getBackupKey('雷电将军');
const keyB = chats.getBackupKey('测试角色');
expect(backupFiles.some(f => f.startsWith(`chat_${keyA}_`))).toBe(true);
expect(backupFiles.some(f => f.startsWith(`chat_${keyB}_`))).toBe(true);
});
});
+31 -3
View File
@@ -85,7 +85,10 @@ describe('getChatInfo', () => {
fs.writeFileSync(chatFile, '');
jest.spyOn(fs.promises, 'stat').mockResolvedValue(/** @type {any} */({ size: 100, mtimeMs: 1_000_000 }));
const info = await chats.getChatInfo(chatFile);
expect(info).toEqual({});
expect(info.file_name).toBe('chat.jsonl');
expect(info.chat_items).toBe(0);
expect(info.mes).toBe('[The chat is empty]');
expect(info.match).toBe(false);
});
test('rejects with the original error for non-ENOENT stat failures', async () => {
@@ -94,9 +97,34 @@ describe('getChatInfo', () => {
await expect(chats.getChatInfo(chatFile)).rejects.toThrow('EACCES');
});
test('resolves an empty object for a corrupted chat file', async () => {
test('resolves a degraded preview for a fully corrupted chat file', async () => {
fs.writeFileSync(chatFile, 'not json at all\nstill not json');
const info = await chats.getChatInfo(chatFile);
expect(info).toEqual({});
expect(info.file_id).toBe('chat');
expect(info.file_name).toBe('chat.jsonl');
expect(info.chat_items).toBe(0);
expect(info.mes).toBe('[The message is empty]');
expect(info.match).toBe(true);
});
test('resolves a degraded preview when only the last line is unparseable', async () => {
// e.g. a partially flushed write or an external edit truncated the final line
fs.writeFileSync(chatFile, makeChatJsonl() + '\n' + '{"name": "User", "is_user": true, "mes": "Trunc');
const info = await chats.getChatInfo(chatFile);
expect(info.file_id).toBe('chat');
expect(info.file_name).toBe('chat.jsonl');
expect(info.chat_items).toBe(2);
expect(info.mes).toBe('[The message is empty]');
expect(info.match).toBe(true);
});
test('matcher still applies to intact messages when the last line is unparseable', async () => {
fs.writeFileSync(chatFile, makeChatJsonl() + '\n' + '{"broken');
const matching = await chats.getChatInfo(chatFile, {}, false, lines => lines.some(l => l.includes('Second')));
expect(matching.match).toBe(true);
expect(matching.file_name).toBe('chat.jsonl');
const nonMatching = await chats.getChatInfo(chatFile, {}, false, lines => lines.some(l => l.includes('nope')));
expect(nonMatching.match).toBe(false);
expect(nonMatching.file_name).toBe('chat.jsonl');
});
});
+42
View File
@@ -0,0 +1,42 @@
import fs from 'node:fs';
import os from 'node:os';
import path from 'node:path';
import { afterEach, beforeEach, describe, expect, jest, test } from '@jest/globals';
import { addMissingConfigValues } from '../src/config-init.js';
const oldEnvKey = 'SILLYTAVERN_DISABLETHUMBNAILS';
const newEnvKey = 'SILLYTAVERN_THUMBNAILS_ENABLED';
describe('addMissingConfigValues', () => {
let tempDirectory;
let configPath;
beforeEach(() => {
tempDirectory = fs.mkdtempSync(path.join(os.tmpdir(), 'sillytavern-config-'));
configPath = path.join(tempDirectory, 'config.yaml');
fs.writeFileSync(configPath, '{}\n');
delete process.env[oldEnvKey];
delete process.env[newEnvKey];
jest.spyOn(console, 'log').mockImplementation(() => {});
jest.spyOn(console, 'warn').mockImplementation(() => {});
});
afterEach(() => {
delete process.env[oldEnvKey];
delete process.env[newEnvKey];
fs.rmSync(tempDirectory, { recursive: true, force: true });
jest.restoreAllMocks();
});
test.each([
['false', 'true'],
['true', 'false'],
])('migrates the boolean environment value %s to %s', (oldValue, newValue) => {
process.env[oldEnvKey] = oldValue;
addMissingConfigValues(configPath);
expect(process.env[newEnvKey]).toBe(newValue);
expect(process.env[oldEnvKey]).toBeUndefined();
});
});
+63
View File
@@ -0,0 +1,63 @@
import fs from 'node:fs';
import os from 'node:os';
import path from 'node:path';
import process from 'node:process';
import { afterAll, beforeAll, describe, expect, test } from '@jest/globals';
// extensions.js reads these config values at module load or per request; the env var
// overrides avoid needing a config.yaml in the test environment
process.env.SILLYTAVERN_GIT_BACKEND = 'auto';
process.env.SILLYTAVERN_EXTENSIONS_ENABLED = 'true';
// The extensions endpoint resolves the built-in extensions folder relative to the repo root
const originalCwd = process.cwd();
beforeAll(() => process.chdir(path.resolve(originalCwd, '..')));
afterAll(() => process.chdir(originalCwd));
describe('extensions discover', () => {
/** @type {import('node:http').Server} */
let server;
let baseUrl;
let userExtensionsDir;
beforeAll(async () => {
userExtensionsDir = fs.mkdtempSync(path.join(os.tmpdir(), 'st-extensions-'));
// A valid extension with a manifest
fs.mkdirSync(path.join(userExtensionsDir, 'Valid-Extension'));
fs.writeFileSync(path.join(userExtensionsDir, 'Valid-Extension', 'manifest.json'), JSON.stringify({ display_name: 'Valid' }));
// A leftover folder without a manifest (e.g. locked .git remnant after manual deletion)
fs.mkdirSync(path.join(userExtensionsDir, 'Deleted-Husk'));
fs.mkdirSync(path.join(userExtensionsDir, 'Deleted-Husk', '.git'));
const { default: express } = await import('express');
const { router } = await import('../src/endpoints/extensions.js');
const app = express();
app.use((req, _res, next) => {
req.user = {
profile: { handle: 'test-user' },
directories: { extensions: userExtensionsDir },
};
next();
});
app.use(router);
server = app.listen(0, '127.0.0.1');
await new Promise(resolve => server.once('listening', resolve));
baseUrl = `http://127.0.0.1:${server.address().port}`;
});
afterAll(async () => {
await new Promise((resolve, reject) => server.close(error => error ? reject(error) : resolve()));
fs.rmSync(userExtensionsDir, { recursive: true, force: true });
});
test('lists folders with a manifest and skips folders without one', async () => {
const response = await fetch(`${baseUrl}/discover`);
expect(response.status).toBe(200);
const extensions = await response.json();
const names = extensions.map(x => x.name);
expect(names).toContain('third-party/Valid-Extension');
expect(names).not.toContain('third-party/Deleted-Husk');
});
});
+76
View File
@@ -0,0 +1,76 @@
import { test, expect } from '@playwright/test';
import { testSetup } from './frontent-test-utils.js';
test.describe('deleting assistant messages with tool calls', () => {
test.beforeEach(testSetup.awaitST);
test('deletes preceding tool calls unless explicitly disabled', async ({ page }) => {
const result = await page.evaluate(async () => {
const { addOneMessage, chat, deleteMessage, newAssistantChat } = await import('./script.js');
const commands = window.SillyTavern.getContext().SlashCommandParser.commands;
const testMessages = new Set(['Run a tool', 'Tool result', 'Done']);
const snapshot = () => chat
.filter(message => testMessages.has(message.mes))
.map(message => ({
mes: message.mes,
toolCall: Array.isArray(message.extra?.tool_invocations),
}));
const addMessages = async () => {
const messages = [
{ name: 'User', is_user: true, is_system: false, mes: 'Run a tool', extra: {} },
{
name: 'System',
is_user: false,
is_system: true,
mes: 'Tool result',
extra: { isSmallSys: true, tool_invocations: [{ id: 'call-1', name: 'lookup' }] },
},
{ name: 'Assistant', is_user: false, is_system: false, mes: 'Done', extra: {} },
];
for (const message of messages) {
chat.push(message);
await addOneMessage(message);
}
};
await newAssistantChat({ temporary: true });
await addMessages();
await deleteMessage(chat.length - 1, undefined, false);
const automatic = snapshot();
await newAssistantChat({ temporary: true });
await addMessages();
await commands.cut.callback({ toolcalls: 'false' }, String(chat.length - 1));
const disabled = snapshot();
await newAssistantChat({ temporary: true });
await addMessages();
chat.push({ name: 'Assistant', is_user: false, is_system: false, mes: 'Later', extra: {} });
await addOneMessage(chat.at(-1));
await commands.del.callback({ toolcalls: 'true' }, '2');
const range = chat
.filter(message => new Set([...testMessages, 'Later']).has(message.mes))
.map(message => message.mes);
await newAssistantChat({ temporary: true });
await addMessages();
document.querySelector('#option_delete_mes').click();
await new Promise(resolve => setTimeout(resolve, 200));
document.querySelector(`.mes[mesid="${chat.length - 1}"]`).click();
document.querySelector('#dialogue_del_mes_ok').click();
await new Promise(resolve => setTimeout(resolve, 50));
const deleteMode = snapshot();
return { automatic, deleteMode, disabled, range };
});
expect(result.automatic).toEqual([{ mes: 'Run a tool', toolCall: false }]);
expect(result.disabled).toEqual([
{ mes: 'Run a tool', toolCall: false },
{ mes: 'Tool result', toolCall: true },
]);
expect(result.range).toEqual(['Run a tool']);
expect(result.deleteMode).toEqual([{ mes: 'Run a tool', toolCall: false }]);
});
});
@@ -0,0 +1,78 @@
import { test, expect } from '@playwright/test';
import { testSetup } from './frontent-test-utils.js';
test.describe('Quick Reply overwrite cancellation', () => {
test.beforeEach(testSetup.awaitST);
for (const operation of ['create', 'import']) {
test(`should keep the existing set when ${operation} overwrite is cancelled`, async ({ page }) => {
const result = await page.evaluate(async (operation) => {
const { Popup, POPUP_RESULT } = await import('./scripts/popup.js');
const { QuickReplySet } = await import('./scripts/extensions/quick-reply/src/QuickReplySet.js');
const { SettingsUi } = await import('./scripts/extensions/quick-reply/src/ui/SettingsUi.js');
const setName = 'STAGE_REPRO_OVERWRITE_CANCEL';
const existingSet = { name: setName };
const originalSets = QuickReplySet.list;
const originalInput = Popup.show.input;
const originalConfirm = Popup.show.confirm;
const originalFrom = QuickReplySet.from;
const originalAddQuickReply = QuickReplySet.prototype.addQuickReply;
let deleteCalls = 0;
let confirmMessage = '';
try {
QuickReplySet.list = [existingSet];
Popup.show.input = async () => setName;
Popup.show.confirm = async (_title, message) => {
confirmMessage = message;
return POPUP_RESULT.NEGATIVE;
};
QuickReplySet.prototype.addQuickReply = () => {};
QuickReplySet.from = props => ({
...props,
qrList: [],
init: () => {},
save: async () => {},
});
const settingsUi = new SettingsUi({});
settingsUi.currentSet = document.createElement('select');
settingsUi.doDeleteQrSet = async () => deleteCalls++;
settingsUi.rerender = () => {};
settingsUi.onQrSetChange = () => {};
settingsUi.prepareGlobalSetList = () => {};
settingsUi.prepareChatSetList = () => {};
settingsUi.prepareCharacterSetList = () => {};
if (operation === 'create') {
await settingsUi.addQrSet();
} else {
const file = {
name: 'duplicate.json',
text: async () => JSON.stringify({ version: 2, name: setName, qrList: [] }),
};
await settingsUi.importSingleQrSet(file);
}
return {
confirmMessage,
deleteCalls,
setNames: QuickReplySet.list.map(set => set.name),
setName,
};
} finally {
QuickReplySet.list = originalSets;
Popup.show.input = originalInput;
Popup.show.confirm = originalConfirm;
QuickReplySet.from = originalFrom;
QuickReplySet.prototype.addQuickReply = originalAddQuickReply;
}
}, operation);
expect(result.confirmMessage).toContain(`"${result.setName}"`);
expect(result.deleteCalls).toBe(0);
expect(result.setNames).toEqual([result.setName]);
});
}
});
+123
View File
@@ -0,0 +1,123 @@
import { describe, expect, jest, test } from '@jest/globals';
import {
fetchGoogleModels,
GoogleModelsHttpError,
GoogleModelsResponseError,
} from '../src/endpoints/backends/google-models.js';
function createResponse(data, overrides = {}) {
return {
ok: true,
status: 200,
statusText: 'OK',
json: async () => data,
...overrides,
};
}
describe('Google AI Studio model pagination', () => {
test('fetches every page and returns generateContent models', async () => {
const fetchMock = jest.fn()
.mockResolvedValueOnce(createResponse({
models: [
{
name: 'models/gemini-2.5-flash',
supportedGenerationMethods: ['generateContent', 'countTokens'],
},
{
name: 'models/gemini-embedding-001',
supportedGenerationMethods: ['embedContent'],
},
],
nextPageToken: 'second-page',
}))
.mockResolvedValueOnce(createResponse({
models: [{
name: 'models/gemini-2.5-pro',
supportedGenerationMethods: ['generateContent'],
}],
}));
const models = await fetchGoogleModels('https://example.com/v1beta/models?key=secret', fetchMock);
expect(models).toEqual([
{
id: 'gemini-2.5-flash',
name: 'models/gemini-2.5-flash',
supportedGenerationMethods: ['generateContent', 'countTokens'],
},
{
id: 'gemini-2.5-pro',
name: 'models/gemini-2.5-pro',
supportedGenerationMethods: ['generateContent'],
},
]);
expect(fetchMock).toHaveBeenCalledTimes(2);
const firstUrl = new URL(fetchMock.mock.calls[0][0]);
expect(firstUrl.searchParams.get('key')).toBe('secret');
expect(firstUrl.searchParams.get('pageSize')).toBe('1000');
expect(firstUrl.searchParams.has('pageToken')).toBe(false);
const secondUrl = new URL(fetchMock.mock.calls[1][0]);
expect(secondUrl.searchParams.get('key')).toBe('secret');
expect(secondUrl.searchParams.get('pageSize')).toBe('1000');
expect(secondUrl.searchParams.get('pageToken')).toBe('second-page');
});
test('preserves reverse proxy query parameters', async () => {
const fetchMock = jest.fn().mockResolvedValue(createResponse({ models: [] }));
await fetchGoogleModels('https://proxy.example/v1beta/models?tenant=test', fetchMock);
const requestUrl = new URL(fetchMock.mock.calls[0][0]);
expect(requestUrl.searchParams.get('tenant')).toBe('test');
expect(requestUrl.searchParams.get('pageSize')).toBe('1000');
});
test('rejects an HTTP failure without returning partial models', async () => {
const fetchMock = jest.fn()
.mockResolvedValueOnce(createResponse({
models: [],
nextPageToken: 'second-page',
}))
.mockResolvedValueOnce(createResponse(null, {
ok: false,
status: 503,
statusText: 'Service Unavailable',
}));
await expect(fetchGoogleModels('https://example.com/v1beta/models', fetchMock))
.rejects.toEqual(expect.objectContaining({
name: GoogleModelsHttpError.name,
status: 503,
statusText: 'Service Unavailable',
}));
});
test('rejects a malformed models field', async () => {
const fetchMock = jest.fn().mockResolvedValue(createResponse({ models: null }));
await expect(fetchGoogleModels('https://example.com/v1beta/models', fetchMock))
.rejects.toThrow('invalid models field');
});
test('rejects a repeated page token', async () => {
const fetchMock = jest.fn()
.mockResolvedValueOnce(createResponse({ models: [], nextPageToken: 'repeated' }))
.mockResolvedValueOnce(createResponse({ models: [], nextPageToken: 'repeated' }));
await expect(fetchGoogleModels('https://example.com/v1beta/models', fetchMock))
.rejects.toBeInstanceOf(GoogleModelsResponseError);
expect(fetchMock).toHaveBeenCalledTimes(2);
});
test('rejects a supported model without a valid name', async () => {
const fetchMock = jest.fn().mockResolvedValue(createResponse({
models: [{ supportedGenerationMethods: ['generateContent'] }],
}));
await expect(fetchGoogleModels('https://example.com/v1beta/models', fetchMock))
.rejects.toThrow('invalid name');
});
});
+80
View File
@@ -0,0 +1,80 @@
import { beforeAll, describe, expect, jest, test } from '@jest/globals';
const jqueryElement = {
append: jest.fn(function () { return this; }),
children: jest.fn(() => ({ each: jest.fn() })),
detach: jest.fn(function () { return this; }),
empty: jest.fn(function () { return this; }),
find: jest.fn(function () { return this; }),
on: jest.fn(function () { return this; }),
prop: jest.fn(function () { return this; }),
trigger: jest.fn(function () { return this; }),
val: jest.fn(function () { return this; }),
};
global.$ = jest.fn(() => jqueryElement);
jest.unstable_mockModule('../public/script.js', () => ({
abortStatusCheck: { signal: undefined },
event_types: {},
eventSource: { emit: jest.fn() },
getRequestHeaders: jest.fn(),
getStoppingStrings: jest.fn(),
resultCheckStatus: jest.fn(),
saveSettingsDebounced: jest.fn(),
setGenerationParamsFromPreset: jest.fn(),
setOnlineStatus: jest.fn(),
startStatusLoading: jest.fn(),
}));
jest.unstable_mockModule('../public/scripts/power-user.js', () => ({
MAX_CONTEXT_DEFAULT: 8192,
MAX_RESPONSE_DEFAULT: 150,
power_user: {},
}));
jest.unstable_mockModule('../public/scripts/tokenizers.js', () => ({
getTextTokens: jest.fn(),
tokenizers: { NONE: 0, NERD: 1, NERD2: 2, LLAMA3: 3 },
}));
jest.unstable_mockModule('../public/scripts/sse-stream.js', () => ({ getEventSourceStream: jest.fn() }));
jest.unstable_mockModule('../public/scripts/utils.js', () => ({
getSortableDelay: jest.fn(),
getStringHash: jest.fn(),
onlyUnique: jest.fn(),
}));
jest.unstable_mockModule('../public/scripts/logit-bias.js', () => ({
BIAS_CACHE: new Map(),
createNewLogitBiasEntry: jest.fn(),
displayLogitBias: jest.fn(),
getLogitBiasListResult: jest.fn(),
}));
jest.unstable_mockModule('../public/scripts/secrets.js', () => ({
SECRET_KEYS: {},
secret_state: {},
writeSecret: jest.fn(),
}));
let loadNovelPreset;
let loadNovelSettings;
let naiSettings;
beforeAll(async () => {
({
loadNovelPreset,
loadNovelSettings,
nai_settings: naiSettings,
} = await import('../public/scripts/nai-settings.js'));
});
describe('NovelAI settings', () => {
test('preserves a zero Math1 temperature when loading settings', () => {
loadNovelSettings(
{ novelai_setting_names: [], novelai_settings: [] },
{ ...naiSettings, math1_temp: 0 },
);
expect(naiSettings.math1_temp).toBe(0);
naiSettings.math1_temp = 1;
loadNovelPreset({ ...naiSettings, genamt: 150, math1_temp: 0 });
expect(naiSettings.math1_temp).toBe(0);
});
});
+67
View File
@@ -0,0 +1,67 @@
import { afterAll, beforeAll, beforeEach, describe, expect, jest, test } from '@jest/globals';
const fetchMock = jest.fn();
jest.unstable_mockModule('node-fetch', () => ({ default: fetchMock }));
jest.unstable_mockModule('../src/transformers.js', () => ({ getPipeline: jest.fn() }));
jest.unstable_mockModule('../src/endpoints/secrets.js', () => ({
readSecret: jest.fn(),
SECRET_KEYS: {},
}));
describe('Pollinations TTS model aliases', () => {
/** @type {import('node:http').Server} */
let server;
let baseUrl;
beforeAll(async () => {
const { default: express } = await import('express');
const { router: speechRouter } = await import('../src/endpoints/speech.js');
const app = express();
app.use(express.json());
app.use('/speech', speechRouter);
server = app.listen(0, '127.0.0.1');
await new Promise(resolve => server.once('listening', resolve));
const address = server.address();
baseUrl = `http://127.0.0.1:${address.port}`;
});
afterAll(async () => {
await new Promise((resolve, reject) => server.close(error => error ? reject(error) : resolve()));
});
beforeEach(() => {
fetchMock.mockReset();
});
test.each([undefined, 'openai-audio', 'openai/gpt-audio-mini'])('looks up voices for %s', async (model) => {
fetchMock.mockResolvedValue({
ok: true,
json: async () => [{ name: 'openai/gpt-audio-mini', aliases: ['openai-audio'], voices: ['alloy', 'nova'] }],
});
const response = await fetch(`${baseUrl}/speech/pollinations/voices`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ model }),
});
expect(response.status).toBe(200);
expect(await response.json()).toEqual(['alloy', 'nova']);
});
test('still accepts catalogs without aliases', async () => {
fetchMock.mockResolvedValue({
ok: true,
json: async () => [{ name: 'openai-audio', voices: ['alloy'] }],
});
const response = await fetch(`${baseUrl}/speech/pollinations/voices`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: '{}',
});
expect(response.status).toBe(200);
expect(await response.json()).toEqual(['alloy']);
});
});
+44
View File
@@ -166,6 +166,27 @@ describe('calculateGoogleBudgetTokens', () => {
test('max returns high', () => expect(mod.calculateGoogleBudgetTokens(8192, 'max', 'gemini-3.5-flash')).toBe('high'));
});
test('Gemini 3.7 Flash uses Gemini 3 thinking levels without minimal', () => {
expect(mod.calculateGoogleBudgetTokens(8192, 'auto', 'gemini-3.7-flash')).toBeNull();
expect(mod.calculateGoogleBudgetTokens(8192, 'min', 'gemini-3.7-flash')).toBe('low');
expect(mod.calculateGoogleBudgetTokens(8192, 'medium', 'gemini-3.7-flash')).toBe('medium');
expect(mod.calculateGoogleBudgetTokens(8192, 'max', 'gemini-3.7-flash')).toBe('high');
});
test('Gemini 3.6 Flash uses Gemini 3 thinking levels', () => {
expect(mod.calculateGoogleBudgetTokens(8192, 'auto', 'gemini-3.6-flash')).toBeNull();
expect(mod.calculateGoogleBudgetTokens(8192, 'min', 'gemini-3.6-flash')).toBe('minimal');
expect(mod.calculateGoogleBudgetTokens(8192, 'medium', 'gemini-3.6-flash')).toBe('medium');
expect(mod.calculateGoogleBudgetTokens(8192, 'max', 'gemini-3.6-flash')).toBe('high');
});
test('Gemini 3.5 Flash-Lite uses Gemini 3 thinking levels', () => {
expect(mod.calculateGoogleBudgetTokens(8192, 'auto', 'gemini-3.5-flash-lite')).toBeNull();
expect(mod.calculateGoogleBudgetTokens(8192, 'min', 'gemini-3.5-flash-lite')).toBe('minimal');
expect(mod.calculateGoogleBudgetTokens(8192, 'medium', 'gemini-3.5-flash-lite')).toBe('medium');
expect(mod.calculateGoogleBudgetTokens(8192, 'max', 'gemini-3.5-flash-lite')).toBe('high');
});
test('stable Gemini 3.1 Flash-Lite uses Gemini 3 thinking levels', () => {
expect(mod.calculateGoogleBudgetTokens(8192, 'min', 'gemini-3.1-flash-lite')).toBe('minimal');
expect(mod.calculateGoogleBudgetTokens(8192, 'medium', 'gemini-3.1-flash-lite')).toBe('medium');
@@ -1041,6 +1062,29 @@ describe('convertGooglePrompt', () => {
expect(result.contents.filter(c => c.role === 'user')).toHaveLength(1);
});
for (const model of ['gemini-3.7-flash', 'gemini-3.6-flash', 'gemini-3.5-flash-lite']) {
test(`merges a trailing prefill into the user turn on ${model}`, () => {
const messages = [
{ role: 'user', content: 'Hi' },
{ role: 'assistant', content: 'Prefill' },
];
const result = mod.convertGooglePrompt(messages, model, false, names);
expect(result.contents).toHaveLength(1);
expect(result.contents[0].role).toBe('user');
expect(result.contents[0].parts[0].text).toBe('Hi\n\nPrefill');
});
}
test('keeps non-trailing model turns on models without prefill support', () => {
const messages = [
{ role: 'user', content: 'Hi' },
{ role: 'assistant', content: 'Hello' },
{ role: 'user', content: 'How are you?' },
];
const result = mod.convertGooglePrompt(messages, 'gemini-3.6-flash', false, names);
expect(result.contents.map(c => c.role)).toEqual(['user', 'model', 'user']);
});
test('converts image_url to inlineData', () => {
const messages = [
{ role: 'user', content: [
+35
View File
@@ -0,0 +1,35 @@
import { describe, test, expect } from '@jest/globals';
import fs from 'node:fs';
const indexHtml = fs.readFileSync(new URL('../public/index.html', import.meta.url), 'utf8');
const openaiJs = fs.readFileSync(new URL('../public/scripts/openai.js', import.meta.url), 'utf8');
const styleCss = fs.readFileSync(new URL('../public/style.css', import.meta.url), 'utf8');
const inputTags = indexHtml.match(/<input\b[^>]*>/gi) ?? [];
function getInputById(id) {
return inputTags.find(tag => tag.includes(`id="${id}"`));
}
describe('secret inputs', () => {
test('keeps the proxy access key out of password manager form detection', () => {
const input = getInputById('openai_proxy_access_key');
expect(input).toBeDefined();
expect(input).toContain('type="text"');
expect(input).toContain('masked-secret');
expect(input).not.toMatch(/\sform=/i);
expect(indexHtml).not.toContain('id="openai_proxy_password"');
expect(openaiJs).not.toContain('#openai_proxy_password');
expect(openaiJs).toMatch(/proxy_password:\s*\['#openai_proxy_access_key', 'proxy_password'/);
expect(openaiJs).toMatch(/\$\('#openai_proxy_access_key'\)\.toggleClass\('masked-secret'\)/);
expect(openaiJs).toMatch(/\$\('#openai_proxy_access_key'\)\.on\('copy cut', function \(event\) \{\s*if \(\$\(this\)\.hasClass\('masked-secret'\)\) \{\s*event\.preventDefault\(\);/s);
expect(styleCss).toMatch(/\.masked-secret\s*{[^}]*-webkit-text-security:\s*disc;/s);
});
test('uses a regular text input for the Azure API key', () => {
const input = getInputById('api_key_azure_openai');
expect(input).toBeDefined();
expect(input).toContain('type="text"');
});
});
+113
View File
@@ -0,0 +1,113 @@
import { afterAll, beforeAll, beforeEach, describe, expect, jest, test } from '@jest/globals';
const fetchMock = jest.fn();
jest.unstable_mockModule('node-fetch', () => ({ default: fetchMock }));
jest.unstable_mockModule('../src/endpoints/secrets.js', () => ({
readSecret: jest.fn(),
SECRET_KEYS: {},
}));
describe('ComfyUI generation', () => {
/** @type {import('node:http').Server} */
let server;
let baseUrl;
beforeAll(async () => {
const { default: express } = await import('express');
const { router } = await import('../src/endpoints/stable-diffusion.js');
const app = express();
app.use(express.json());
app.use(router);
server = app.listen(0, '127.0.0.1');
await new Promise(resolve => server.once('listening', resolve));
const address = server.address();
baseUrl = `http://127.0.0.1:${address.port}`;
});
afterAll(async () => {
await new Promise((resolve, reject) => server.close(error => error ? reject(error) : resolve()));
});
beforeEach(() => {
fetchMock.mockReset();
});
test('skips output nodes without images', async () => {
fetchMock
.mockResolvedValueOnce({ ok: true, json: async () => ({ prompt_id: 'prompt-1' }) })
.mockResolvedValueOnce({
ok: true,
json: async () => ({
'prompt-1': {
status: { status_str: 'success' },
outputs: {
117: { a_images: [{ filename: 'comparison.png' }] },
197: { images: [{ filename: 'result.png', subfolder: '', type: 'temp' }] },
},
},
}),
})
.mockResolvedValueOnce({ ok: true, arrayBuffer: async () => Uint8Array.from([1, 2, 3]).buffer });
const response = await fetch(`${baseUrl}/comfy/generate`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ url: 'http://127.0.0.1:8188', prompt: '{}' }),
});
expect(response.status).toBe(200);
expect(await response.json()).toEqual({ format: 'png', data: 'AQID' });
});
test('falls back to gifs when no output node has images', async () => {
fetchMock
.mockResolvedValueOnce({ ok: true, json: async () => ({ prompt_id: 'prompt-2' }) })
.mockResolvedValueOnce({
ok: true,
json: async () => ({
'prompt-2': {
status: { status_str: 'success' },
outputs: {
10: { text: ['some non-media output'] },
42: { gifs: [{ filename: 'animation.webp', subfolder: '', type: 'output' }] },
},
},
}),
})
.mockResolvedValueOnce({ ok: true, arrayBuffer: async () => Uint8Array.from([4, 5, 6]).buffer });
const response = await fetch(`${baseUrl}/comfy/generate`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ url: 'http://127.0.0.1:8188', prompt: '{}' }),
});
expect(response.status).toBe(200);
expect(await response.json()).toEqual({ format: 'webp', data: 'BAUG' });
});
test('reports an error when no output has images or gifs', async () => {
fetchMock
.mockResolvedValueOnce({ ok: true, json: async () => ({ prompt_id: 'prompt-3' }) })
.mockResolvedValueOnce({
ok: true,
json: async () => ({
'prompt-3': {
status: { status_str: 'success' },
outputs: {
10: { text: ['no media at all'] },
},
},
}),
});
const response = await fetch(`${baseUrl}/comfy/generate`, {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ url: 'http://127.0.0.1:8188', prompt: '{}' }),
});
expect(response.status).toBe(500);
expect(await response.text()).toContain('did not return any recognizable outputs');
});
});
+20
View File
@@ -29,6 +29,7 @@ import {
generateTimestamp,
mergeObjectWithYaml,
excludeKeysByYaml,
getArrayBufferSlice,
Cache,
MemoryLimitedMap,
} from '../src/util';
@@ -519,6 +520,25 @@ describe('mergeObjectWithYaml', () => {
});
});
describe('getArrayBufferSlice', () => {
test('returns an exact ArrayBuffer for Buffer views', () => {
const source = Buffer.alloc(16, 0);
source.write('abc', 4, 'utf8');
const view = source.subarray(4, 7);
expect(view.byteLength).toBe(3);
expect(view.buffer.byteLength).toBe(16);
const exact = getArrayBufferSlice(view);
expect(exact.byteLength).toBe(3);
expect(Buffer.from(exact).toString('utf8')).toBe('abc');
});
test('throws for non-Uint8Array input', () => {
expect(() => getArrayBufferSlice('abc')).toThrow(TypeError);
});
});
describe('excludeKeysByYaml', () => {
test('should delete keys listed in a YAML array', () => {
const obj = { a: 1, b: 2, c: 3 };