phsl / new-api

Merge pull request #1534 from HynoR/feat/je

Ensure models and vendors can be re-created after soft deletion by switching to composite unique indexes on (name, deleted_at) and cleaning up legacy single-column unique indexes on MySQL.

Why
- MySQL raised 1062 duplicate key errors when re-adding a soft-deleted model/vendor because the legacy unique index enforced uniqueness on the name column alone (uk_model_name / uk_vendor_name), despite soft deletes.
- Users encountered errors such as:
  - Error 1062 (23000): Duplicate entry 'deepseek-chat' for key 'models.uk_model_name'
  - Error 1062 (23000): Duplicate entry 'DeepSeek' for key 'vendors.uk_vendor_name'

How
- Model indices:
  - model/model_meta.go:
    - Model.ModelName → gorm: uniqueIndex:uk_model_name,priority:1
    - Model.DeletedAt → gorm: index; uniqueIndex:uk_model_name,priority:2
- Vendor indices:
  - model/vendor_meta.go:
    - Vendor.Name → gorm: uniqueIndex:uk_vendor_name,priority:1
    - Vendor.DeletedAt → gorm: index; uniqueIndex:uk_vendor_name,priority:2
- Migration (automatic, idempotent):
  - model/main.go (migrateDB, migrateDBFast):
    - On MySQL, drop legacy single-column unique indexes if present:
      - ALTER TABLE models DROP INDEX uk_model_name;
      - ALTER TABLE vendors DROP INDEX uk_vendor_name;
    - Then run AutoMigrate to create composite unique indexes.
  - Missing-index errors are ignored to keep the migration safe to run multiple times.

Result
- Users can delete and re-add the same model/vendor name without manual SQL.
- Migration runs automatically at startup; no user action required.
- PostgreSQL and SQLite remain unaffected.

Files changed
- model/model_meta.go
- model/vendor_meta.go
- model/main.go (migrateDB, migrateDBFast)

Testing
- Create model "deepseek-chat" → delete (soft) → re-create → succeeds.
- Create vendor "DeepSeek" → delete (soft) → re-create → succeeds.

Backward compatibility
- Data remains intact; only index definitions are updated.
- Behavior is unchanged except for fixing the uniqueness constraint with soft deletes.

fix: Gemini embedding model only embeds the first text in a batch

feat: add video preview modal

fix(midjourney): 为 Midjourney 任务添加视频 URL 字段

fix: 当OIDC的AuthUrl带有param时，跳转参数拼接错误

feats:replace GLM-4v authentication headers to support customize api key

* Integrate `useIsMobile` hook to detect mobile devices.
* Pagination now automatically:
  * sets `size="small"` on mobile screens
  * enables `showQuickJumper` for quicker navigation on small screens
* Desktop behaviour remains unchanged.

This commit updates the quick-fill endpoint templates used in the model and pre-fill group editors:

• `EditModelModal.jsx`
• `EditPrefillGroupModal.jsx`

Key changes
1. Added missing default endpoints defined in `common/endpoint_defaults.go`.
   - `openai-response`
   - `gemini`
   - `jina-rerank`
2. Ensured each template entry includes both `path` and `method` for clarity.

Benefits
• Provides one-click access to every built-in upstream endpoint, reducing manual input.
• Keeps the UI definitions in sync with backend defaults, preventing mismatch errors.

Problem
Choosing a different token-group in the pricing sidebar only updated the filter but did **not** refresh the displayed group ratio in both the Table (`@table/`) and Card (`@card/`) views. The callback used by the sidebar changed `filterGroup` yet left `selectedGroup` untouched, so ratio columns/cards kept showing the previous value.

Solution
• `PricingSidebar.jsx`
  – Accept new prop `handleGroupClick` (from `useModelPricingData`).
  – Forward this callback to `PricingGroups` (`setFilterGroup={handleGroupClick}`) while retaining `setFilterGroup` for reset logic.
  – Keeps both `filterGroup` filtering and `selectedGroup` state in sync via the single unified handler.

Result
Switching groups in the sidebar now simultaneously updates:
1. the model list filtering, and
2. the ratio information shown in both pricing Table and Card views.

No UI/UX regression; linter passes.

The sidebar’s admin section now displays “Channel Management” before “Model Management” to better reflect common user workflows and improve navigation clarity.

Details:
• Updated `web/src/components/layout/SiderBar.js`
  – Re-ordered items in `adminItems` array so `channel` precedes `models`.
• No logic or route changes; this is purely a UI ordering adjustment.

This change enhances usability for administrators by presenting frequently accessed channel settings first.

feat: 完善格式转换，修复gemini渠道和openai渠道在claude code中使用的问题

Changes
1. ModelPricingTable.jsx
   • Compute `autoChain` as the intersection of `autoGroups` and the model’s `enable_groups` (order preserved).
   • Display the chain banner only when `autoChain.length > 0`; banner shows the reduced path (e.g. `a → c → e`).
   • Dropped obsolete `selectedGroup` prop; all callers updated.

2. ModelDetailSideSheet.jsx / PricingPage.jsx
   • Removed forwarding of deleted `selectedGroup` prop.

Outcome
– “Auto group routing” appears only for models that actually participate in the chain, avoiding empty or irrelevant banners.
– Codebase simplified by eliminating an unused prop.

fix: ensure last message is sent successfully

Detailed changes
Backend
• `controller/pricing.go` now includes `auto_groups` in `/api/pricing` response, sourced from `setting.AutoGroups`.

Frontend
• `useModelPricingData.js`
  – Parses `auto_groups` and exposes `autoGroups` state.
• `PricingPage.jsx` → `ModelDetailSideSheet.jsx` → `ModelPricingTable.jsx`
  – Thread `autoGroups` through component tree.
• `ModelPricingTable.jsx`
  – Removes deprecated `getGroupDescription` / `Tooltip`.
  – Filters out `auto` when building price table rows.
  – Renders a descriptive banner: “auto 分组调用链路 → auto → group1 → …”, clarifying fallback order without showing prices.
• Minor i18n tweak: adds `auto分组调用链路` key for the banner text.

Why
Users were confused by the “auto” tag appearing alongside regular groups with no price.
This change:
1. Makes the routing chain explicit.
2. Keeps the pricing table focused on billable groups.

No breaking API changes; existing clients can ignore the new `auto_groups` field.

- Updated the model list to include various gpt-5 variants.
- Enhanced the ConvertOpenAIRequest function to handle gpt-5 model temperature settings based on specific model prefixes.
- Adjusted default cache and model ratios for new gpt-5 models.

feat: sync gpt-5 model ratio and support new reasoning effort

fix: 修复通配符处理对判断可用渠道的破坏

- Added support to fetch and render “model prefill groups” in `EditChannelModal.jsx`
- Users can now click a group button to instantly merge that group’s models into the models Select
- Mirrors the prefill-group UX used for tags/endpoints in `EditModelModal.jsx`

Details
- UI/UX:
  - Renders one button per model group inside the models field’s extra actions
  - Clicking a button merges its items into the selected models (trimmed, deduplicated), updating immediately
  - Non-destructive and works alongside existing actions (fill related/all models, fetch upstream, clear, copy)
- API:
  - GET `/api/prefill_group?type=model`
  - Handles `items` as either an array or a JSON string array for robustness
  - If request fails or returns no groups, buttons are simply not shown
- i18n:
  - Reuses existing i18n; group names come from backend and are displayed as-is
- Performance:
  - Simple set merge; negligible overhead
- Backward compatibility:
  - No changes required on the backend or elsewhere; feature is additive
- Testing (manual):
  1) Open channel modal (new or edit) and navigate to the Models section
  2) Confirm model group buttons render when groups are configured
  3) Click a group button → models Select updates with merged models (no duplicates)
  4) Verify other actions (fill related/all, fetch upstream, clear, copy) still work
  5) Close/reopen modal → state resets as expected

Implementation
- `web/src/components/table/channels/modals/EditChannelModal.jsx`
  - Added `modelGroups` state and `fetchModelGroups()` (GET `/api/prefill_group?type=model`)
  - Invoked `fetchModelGroups()` when the modal opens
  - Rendered group buttons in the models Select `extraText`, merging group items into current selection

Chore
- Verified no new linter errors were introduced.

- Move model price column to fixed right position
- Convert endpoint types column from fixed to regular column
- Reorder columns: endpoint types now appears before ratio column
- Improve table layout and user experience for pricing data viewing

Changes made to web/src/components/table/model-pricing/view/table/PricingTableColumns.js:
* Removed `fixed: 'right'` from endpointColumn
* Added `fixed: 'right'` to priceColumn
* Updated column order in the columns array

Merge pull request #1452 from QuantumNous/refactor/model-pricing

- Why:
  - Eliminate `gorm.io/datatypes` for a single field and fix scan errors when drivers return JSON as string.
  - Prevent JSONEditor manual mode from locking on invalid JSON and from appending stray characters after “Fill Template”.

- What:
  - Backend (`model/prefill_group.go`):
    - Replaced `datatypes.JSON` with `JSONValue` (based on `json.RawMessage`) for `PrefillGroup.Items`.
    - Implemented `sql.Scanner` and `driver.Valuer` to accept both `[]byte` and `string`.
    - Implemented `MarshalJSON`/`UnmarshalJSON` to preserve raw JSON in API without base64.
    - Converted comments to Chinese.
  - Frontend (`web/src/components/common/ui/JSONEditor.js`):
    - Added `manualText` buffer for manual mode to avoid input being overridden by external value.
    - Only propagate `onChange` when manual text is valid JSON; otherwise show error but do not block typing.
    - Safe manual-mode rendering: derive rows from `manualText` and avoid calling `split` on non-strings.
    - Improved mode toggle: populate `manualText` from visual data; validate before switching back to visual.
    - Fixed “Fill Template” to sync `manualText`, `jsonData`, and `onChange` to avoid stray trailing characters.

- Impact:
  - Resolves: “unsupported Scan, storing driver.Value type string into type *json.RawMessage”.
  - Resolves: `value.split is not a function` in manual mode.
  - Resolves: extra `s` appended after inserting template.
  - API shape and DB column type remain the same (`gorm:"type:json"`); no `go.mod` changes.
  - Lints pass for modified files.

Files changed:
- model/prefill_group.go
- web/src/components/common/ui/JSONEditor.js

♻️ refactor(model): replace gorm.io/datatypes with JSONValue for PrefillGroup.Items; fix JSON scan across drivers

- Why:
  - Avoid introducing `gorm.io/datatypes` for a single field.
  - Align with existing pattern (`ChannelInfo`, `Properties`) using `Scanner`/`Valuer`.
  - Fix runtime error when drivers return JSON as string.

- What:
  - Introduced `JSONValue` (based on `json.RawMessage`) implementing `sql.Scanner` and `driver.Valuer`, with `MarshalJSON`/`UnmarshalJSON` to preserve raw JSON in API.
  - Updated `PrefillGroup.Items` to use `JSONValue` with `gorm:"type:json"`.
  - Localized comments in `model/prefill_group.go` to Chinese.

- Impact:
  - Resolves “unsupported Scan, storing driver.Value type string into type *json.RawMessage”.
  - Works with MySQL/Postgres/SQLite whether JSON is returned as `[]byte` or `string`.
  - API and DB schema remain unchanged; no `go.mod` changes; lints pass.

Files changed:
- model/prefill_group.go

- Why: Avoid adding `gorm.io/datatypes` for a single field; the rest of the codebase does not use it, and using the standard library keeps dependencies lean.
- What:
  - Switched `PrefillGroup.Items` from `datatypes.JSON` to `json.RawMessage`.
  - Updated imports in `model/prefill_group.go` to use `encoding/json` and removed the unused `gorm.io/datatypes`.
  - Preserved `gorm:"type:json"` so DB column behavior remains the same.
- Impact:
  - API response/request shape for `items` remains unchanged (still JSON).
  - DB schema behavior is unchanged; GORM migration continues to handle the field as JSON.
  - No other references to `datatypes` exist; no `go.mod` changes needed.
  - Lints pass for the modified file.

Files changed:
- model/prefill_group.go

No breaking changes.