phsl / new-api

# Conflicts:
#	controller/channel.go
#	middleware/distributor.go
#	model/channel.go
#	model/user.go
#	model/user_cache.go
#	relay/common/relay_info.go

This commit refactors the application's error handling mechanism by introducing a new standardized error type, `types.NewAPIError`. It also renames common JSON utility functions for better clarity.

Previously, internal error handling was tightly coupled to the `dto.OpenAIError` format. This change decouples the internal logic from the external API representation.

Key changes:
- A new `types.NewAPIError` struct is introduced to serve as a canonical internal representation for all API errors.
- All relay adapters (OpenAI, Claude, Gemini, etc.) are updated to return `*types.NewAPIError`.
- Controllers now convert the internal `NewAPIError` to the client-facing `OpenAIError` format at the API boundary, ensuring backward compatibility.
- Channel auto-disable/enable logic is updated to use the new standardized error type.
- JSON utility functions are renamed to align with Go's standard library conventions (e.g., `UnmarshalJson` -> `Unmarshal`, `EncodeJson` -> `Marshal`).

✨ feat(ability): enhance FixAbility function

修复了百度请求时候需要传appid的bug

refactor: log params and channel params

refactor: log params and channel params

Summary
1. Load `channel_info` when editing:
   • Detect if the channel is in multi-key mode (`is_multi_key`).
   • Auto-initialize `batch`, `multiToSingle`, and `multiKeyMode` from backend data.

2. Visibility logic
   • Creation page: “Batch create / Multi-key mode” always available.
   • Edit page: show these controls **only when** the channel itself is multi-key.

3. State consistency
   • `multi_key_mode` added to `inputs`; `setValues(inputs)` now preserves the user’s selection.

Result
Single-key channels no longer display irrelevant “key aggregation” options, while multi-key channels open with the correct defaults, providing a cleaner and more accurate editing experience.

🔧 **fix(model, controller): robust serialization for `ChannelInfo` & correct Vertex-AI key storage**

Summary
1. **model/channel.go**
   • Replaced the pointer-only `Value()` with a value-receiver implementation
   • GORM can now marshal both `ChannelInfo` and `*ChannelInfo`, eliminating
     `unsupported type model.ChannelInfo` runtime error.

2. **controller/channel.go**
   • Refactored `getVertexArrayKeys()` – every element is now
     - passed through `json.Marshal` when not already a string
     - trimmed & validated before insertion
   • Guarantees each service-account key is persisted as a **pure JSON string**
     instead of the previous `map[...]` dump.

Result
• Channel creation / update succeeds without SQL driver errors.
• Vertex-AI batch & multi-key uploads are stored in canonical JSON, ready for
  downstream SDKs to consume.

Summary
-------
1. Vertex AI JSON key upload
   • Accept multiple `.json` files (drag & drop / click).
   • Parse each `fileInstance`; build valid key array.
   • Malformed files are skipped and collected into **one** toast message.
   • Upload list now仅displays valid files; form state (`vertex_files`) 同步保持.
   • On *Submit* keys are re-parsed to prevent async timing loss.

2. Multi-key mode stability
   • Added `multi_key_mode: "random"` to initial form values.
   • `Form.Select` becomes fully controlled (`value`/`onChange`).
   • Toggling the “multi-key mode” checkbox writes default / removes field, so
     `setValues(inputs)` no longer clears the user’s choice.

3. UX / compatibility tweaks
   • Preserve uploaded files when editing any other field.
   • Use `fileInstance` only – compatible with legacy Semi Upload API.
   • Removed redundant `limit` prop on `Form.Upload`.
   • Aggregated error handling avoids toast spam.

Result
------
Channel creation/update now supports:
• Reliable batch import of Vertex AI service-account keys.
• Consistent retention of multi-key strategy (`random` / `polling`).
• Cleaner, user-friendly error feedback.

# Conflicts:
#	controller/channel.go
#	docker-compose.yml
#	web/src/components/table/ChannelsTable.js
#	web/src/pages/Channel/EditChannel.js

支持Midjourney视频任务和图片编辑

Introduce two new visualizations to the “Model Data Analysis” panel:

1. Model Consumption Trend (line chart)
   • Added `spec_model_line` state and legend support.
   • Calculates per-model counts over time and updates via `updateChartData`.
2. Model Call Ranking (bar chart)
   • Added `spec_rank_bar` state with `seriesField` and legend enabled.
   • Ranks models by total call count.

Additional changes:
• Extended tab navigation with two new `TabPane`s and adjusted chart rendering logic.
• Swapped icons/texts to match new chart purposes.
• Reused existing color mapping to ensure consistent palette.

No breaking changes; UI now offers richer insights into model usage patterns.

Summary:
• Imported InputNumber from @douyinfe/semi-ui.
• Swapped plain Input for InputNumber in “Add Quota” modal.
• Added UX tweaks: full-width styling, showClear, step = 500 000.
• Initialized addQuotaLocal to an empty string so the field starts blank.
• Adjusted state handling and kept quota calculation logic unchanged.

This improves numeric input accuracy and overall user experience without breaking existing functionality.

Motivation
• Remove unused UI components to keep the bundle lean and silence linter warnings.
• Ensure every time the side-sheet opens it reflects the latest tag data, avoiding stale form values (e.g., model / group mismatches).

Key Changes
1. UI Imports
   – Dropped `Input`, `Select`, `TextArea` from `@douyinfe/semi-ui` (unused in Form-based version).
2. State Reset & Form Sync
   – On `visible` or `tag` change:
     • Refresh model & group options.
     • Reset `inputs` to clean defaults (`originInputs`) carrying the current `tag`.
     • Pre-fill Form through `formApiRef` to keep controlled fields aligned.
3. Minor Cleanup
   – Added inline comment clarifying local state reset purpose.

Result
Opening the “Edit Tag” side-sheet now always displays accurate data without residual selections, and build output is cleaner due to removed dead imports.

✨ refactor(EditChannel&EditToken): refactor Channel & Token edit pages with Semi Form and UX enhancements

Overview
• Migrated both `EditChannel.js` and `EditToken.js` to fully leverage Semi UI `Form.*` components, removing legacy `Input/Select/TextArea` + manual labels.
• Unified data-loading strategy: when the drawer becomes visible we load (or reset) data via `props.visible + id` effect and `formApi.setValues()`, guaranteeing fields are always populated; form resets on close.
• Fixed blank-form bug when opening the same record twice.

Key improvements
1. Validation
   • `type`, `models` always required.
   • `key` required only while creating (not on edit).
2. Batch key creation
   • Checkbox moved into `extraText`; hidden when editing or when channel type = 41.
3. Layout & UI
   • `Row / Col` (12 + 12) for “Priority” and “Weight”.
   • Placeholders revised; model selector now shows creation hint; removed obsolete banner.
   • Help / extraText used for long hints, template buttons (`model_mapping`, `status_code_mapping`, `param_override`, etc.), and API address notice.
   • Added `showClear`, `min`, rounded card class names for consistency.
4. Reusable helpers
   • `batchAllowed`, `batchExtra` utilities.
   • `getInitValues()` + centralized `inputs`→form synchronization.
5. Token editor aligned to the same pattern (`props.visiable` watcher).

Result
Cleaner code, consistent UX, instant field population on every open, and clearer validation/error feedback across both editors.

* backend
  - constant/endpoint_type.go
    • Add EndpointTypeMidjourney, EndpointTypeSuno, EndpointTypeKling, EndpointTypeJimeng.
  - common/endpoint_type.go
    • Map Midjourney / MidjourneyPlus, SunoAPI, Kling, Jimeng channel types to the new endpoint types.

* frontend
  - ModelPricing.js
    • Add “Supported Endpoint Type” column.
    • Implement renderSupportedEndpoints with `stringToColor` for consistent tag colors.

These changes allow `/api/pricing` and model lists to return accurate
`supported_endpoint_types` covering all non-OpenAI providers and display
them clearly in the UI.

No breaking changes.

修复使用gemini-balance作为上游时，测试gemini2.5pro模型时出现的错误问题

fix: ali embedding lose prompt_tokens

✨ feat: refactor environment variable initialization

✨ feat: refactor environment variable initialization and introduce new constant types for API and context keys

feat: video action to constant