Commit Graph
74 Commits
Author SHA1 Message Date
Seefs 8ad159a3bb fix(ollama): preserve reasoning and tool-call context (#6605) 2026-08-10 12:49:26 +08:00
Calcium-Ion 86ac0f7745 refactor: extract protocol conversion layer into standalone relaykit module (#6369)
* test(relayconvert): add golden snapshot matrix and relaykit boundary guard

Phase 0 of the relaykit extraction plan: pin byte-level output of every
registered (from,to) request/response/stream conversion route, and
forbid kit-bound packages from growing host-only imports.

* wip(relayconvert): drop gin.Context from converter signatures; add convmeta draft

Phase 1 in progress: relayconvert now takes context.Context; host media
resolver adapts gin.Context back at the service boundary.

* refactor(relayconvert): decouple converters from RelayInfo, gin, and settings

Phase 1 of the relaykit extraction plan:
- converters now depend on convmeta.Meta (implemented by RelayInfo) instead
  of *relaycommon.RelayInfo; ClaudeConvertInfo and the format guesser move
  to convmeta with aliases left behind
- host settings reach converters via a convmeta.Options snapshot built in
  RelayInfo.ConvOptions; no more model_setting/reasoning global reads inside
  the conversion layer
- effort-suffix helpers move to service/relayconvert/reasoning (old package
  forwards); chat-to-responses upgrade policy moves to service (host routing
  logic, not conversion)
- golden conversion matrix unchanged

* test(relayconvert): tighten boundary — kit packages now free of gin/setting imports

* refactor(dto): drop gin and logger dependencies

Phase 2 (part 1): dto.Request.IsStream now takes *http.Request instead of
*gin.Context (Gemini's impl reads query/path off the std request); dto's
three logger calls become common.SysError. Boundary test allowlist is now
empty — kit-bound packages import no gin/setting/logger/model.

* refactor(kit): extract dependency-free kitutil; dto/types/relayconvert stop importing common

Phase 2 of the relaykit extraction plan:
- new service/relayconvert/kitutil holds the pure helpers the kit needs
  (JSON wrappers, pointer/string/uuid/timestamp utils, MaskSensitiveInfo,
  pluggable LogInfo/LogError hooks, Debug flag)
- dto, types, and all relayconvert packages now use kitutil; their only
  remaining internal deps are dto/types/constant
- common keeps every original symbol (MaskSensitiveInfo delegates to
  kitutil) so host code is untouched; main.go routes kit logging into
  common.SysLog/SysError and mirrors DebugEnabled
- golden conversion matrix unchanged

* refactor(kit): move EndpointType/FinishReason to types; OpenRouter dialect via Options

Kit packages (dto/types/relayconvert/reasonmap) no longer import constant:
- EndpointType and finish-reason values live in types; constant re-exports
- the OpenRouter special-case in claude->openai request conversion reads
  Options.OpenRouterDialect, set by the host from the channel type;
  InitChannelMeta invalidates the cached snapshot on channel switch

* refactor: extract relaykit submodule (dto/types/relayconvert/reasonmap)

Phase 3 of the relaykit extraction plan:
- new go module github.com/QuantumNous/new-api/relaykit containing dto
  (minus task family), types, relayconvert (with convmeta/kitutil/reasoning),
  and reasonmap; host consumes it via require + replace, go.work for dev
- task-family dto (task/suno/midjourney/video) stays in the host dto
  package; dual-consumer host files alias it as taskdto
- relaykit builds and tests standalone (GOWORK=off): no host imports,
  no gin, no DB, no settings
- golden conversion matrix unchanged

* build(docker): copy relaykit/go.mod before go mod download

The local-replace submodule's go.mod must exist inside the build context
for the main module graph to resolve.

* fix: address relaykit extraction regressions

* fix: address relaykit review regressions

* docs: document Meta nil receiver contract

* fix(relaykit): fail OpenAI→Claude conversion without max_tokens; reject negative default_max_tokens

The Claude Messages API requires max_tokens (omitting it is a 400
"Field required"), but with a nil Options.Claude.DefaultMaxTokens hook
the converters silently emitted a request the upstream is guaranteed to
reject. Both OpenAI Chat and Responses → Claude conversions now return
sharedclaude.ErrMissingMaxTokens when no path (client value, default
hook, thinking-adapter floor) supplied one. Unreachable in the host,
which always configures the hook.

Host side, claude.default_max_tokens now rejects negative values at the
option API before persisting — they would wrap into huge unsigned values
during conversion. Zero stays allowed: the current API treats
max_tokens: 0 as cache pre-warming.

* fix: make Gemini safety settings read path race-free
2026-07-27 15:56:21 +08:00
MimiandCaIon 0977965d93 fix: handle ollama non-stream tool calls (#5865)
* fix: handle ollama non-stream tool calls

* test: cover ollama non-stream tool call paths

---------

Co-authored-by: CaIon <i@caion.me>
2026-07-03 15:35:34 +08:00
xujiantop-crypto 32805849d6 fix: reuse stream scanner buffer in channel handlers (#5225) 2026-06-05 12:18:57 +08:00
heimoshuiyu 8ca103342d fix: Message.ReasoningContent/Reasoning 改为 *string,修复空思考内容在请求转发时被静默丢弃的问题
问题:
在非 passThrough 模式下,客户端发送的 reasoning_content: "" 经过
Go struct 反序列化再序列化后,因 string + omitempty 无法区分空串和
字段缺失,导致空的思考内容被静默丢弃。

根因:
dto.Message.ReasoningContent 和 Message.Reasoning 使用 string(非指针)
加 omitempty,违反 AGENTS.md Rule 6(可选标量字段必须用指针类型)。

修复:
1. Message.ReasoningContent/Reasoning 类型从 string 改为 *string
   - nil = 字段缺失 → JSON 省略
   - &"" = 显式空串 → JSON 保留 reasoning_content: ""
2. 新增 Message.GetReasoningContent() 辅助方法
3. 更新所有读写处:relay-openai, relay-claude, relay-gemini, ollama
4. 新增测试覆盖空串保留、字段省略、getter 回退逻辑
2026-04-29 13:43:26 +08:00
CaIon 03758a4a85 refactor(file-source): unify file source creation and enhance caching mechanisms 2026-04-06 15:54:55 +08:00
Seefs 2cf3c1836c fix: preserve explicit zero values in native relay requests 2026-03-01 15:47:03 +08:00
CaIon 9ef9e78821 feat(file): unify file handling with a new FileSource abstraction for URL and base64 data 2026-02-04 18:23:17 +08:00
Seefs b10f1f7b85 feat: ionet integrate (#2105)
* wip ionet integrate

* wip ionet integrate

* wip ionet integrate

* ollama wip

* wip

* feat: ionet integration & ollama manage

* fix merge conflict

* wip

* fix: test conn cors

* wip

* fix ionet

* fix ionet

* wip

* fix model select

* refactor: Remove `pkg/ionet` test files and update related Go source and web UI model deployment components.

* feat: Enhance model deployment UI with styling improvements, updated text, and a new description component.

* Revert "feat: Enhance model deployment UI with styling improvements, updated text, and a new description component."

This reverts commit 8b75cb5bf0d1a534b339df8c033be9a6c7df7964.
2025-12-28 15:55:35 +08:00
somnifex e4a2904c56 fix: handle JSON parsing for thinking content in ollama stream and chat handlers 2025-10-17 18:35:08 +08:00
Seefs 5010f2d004 format: package name -> github.com/QuantumNous/new-api (#2017) 2025-10-11 15:30:09 +08:00
t0ng7u 8294a76bc2 💱 feat(settings): introduce site-wide quota display type (USD/CNY/TOKENS/CUSTOM)
Replace the legacy boolean “DisplayInCurrencyEnabled” with an injected, type-safe
configuration `general_setting.quota_display_type`, and wire it through the
backend and frontend.

Backend
- Add `QuotaDisplayType` to `operation_setting.GeneralSetting` with injected
  registration via `config.GlobalConfig.Register("general_setting", ...)`.
  Helpers: `IsCurrencyDisplay()`, `IsCNYDisplay()`, `GetQuotaDisplayType()`.
- Expose `quota_display_type` in `/api/status` and keep legacy
  `display_in_currency` for backward compatibility.
- Logger: update `LogQuota` and `FormatQuota` to support USD/CNY/TOKENS. When
  CNY is selected, convert using `operation_setting.USDExchangeRate`.
- Controllers:
  - `billing`: compute subscription/usage amounts based on the selected type
    (USD: divide by `QuotaPerUnit`; CNY: USD→CNY; TOKENS: keep raw tokens).
  - `topup` / `topup_stripe`: treat inputs as “amount” for USD/CNY and as
    token-count for TOKENS; adjust min topup and pay money accordingly.
  - `misc`: include `quota_display_type` in status payload.
- Compatibility: in `model/option.UpdateOption`, map updates to
  `DisplayInCurrencyEnabled` → `general_setting.quota_display_type`
  (true→USD, false→TOKENS). Keep exporting the legacy key in `OptionMap`.

Frontend
- Settings: replace the “display in currency” switch with a Select
  (`general_setting.quota_display_type`) offering USD / CNY / Tokens.
  Provide fallback mapping from legacy `DisplayInCurrencyEnabled`.
- Persist `quota_display_type` to localStorage (keep `display_in_currency`
  for legacy components).
- Rendering helpers: base all quota/price rendering on `quota_display_type`;
  use `usd_exchange_rate` for CNY symbol/values.
- Pricing page: default view currency follows site display type (USD/CNY),
  while TOKENS mode still allows per-view currency toggling when needed.

Notes
- No database migrations required.
- Legacy clients remain functional via compatibility fields.
2025-09-29 23:23:31 +08:00
somnifex 05241aea92 fix: 修复ollamaChatHandler中ReasoningContent字段的赋值逻辑 2025-09-16 08:54:34 +08:00
somnifex 414b959e65 fix: 添加对Thinking字段的处理逻辑,确保推理内容正确传递 2025-09-16 08:51:29 +08:00
somnifex b80811a412 fix: 修复图像URL处理逻辑,确保正确生成base64数据 2025-09-16 08:13:28 +08:00
somnifex f57c539a1f fix: 修复ollamaChatHandler中的FinishReason字段赋值逻辑 2025-09-15 23:43:39 +08:00
somnifex a1f79761cf refactor: 简化请求转换函数和流处理逻辑 2025-09-15 23:41:09 +08:00
somnifex a9354df549 fix: 优化ollamaStreamHandler中的停止和最终使用响应逻辑 2025-09-15 23:23:53 +08:00
somnifex 503a2669d5 refactor: 更新请求转换逻辑,优化工具调用解析 2025-09-15 23:15:46 +08:00
somnifex cdf0c9dae4 fix: 优化ollamaChatStreamChunk结构体字段格式 2025-09-15 23:09:10 +08:00
somnifex 767d46e829 feat: 重构ollama渠道请求 2025-09-15 23:01:14 +08:00
CaIon 781a708173 feat: support qwen-image-edit 2025-08-25 14:33:12 +08:00
CaIon 77b100ba2b refactor: update function signatures to include context and improve file handling #1599 2025-08-15 18:40:54 +08:00
CaIon c9e9a24e99 refactor: centralize logging and update resource initialization
This commit refactors the logging mechanism across the application by replacing direct logger calls with a centralized logging approach using the `common` package. Key changes include:

- Replaced instances of `logger.SysLog` and `logger.FatalLog` with `common.SysLog` and `common.FatalLog` for consistent logging practices.
- Updated resource initialization error handling to utilize the new logging structure, enhancing maintainability and readability.
- Minor adjustments to improve code clarity and organization throughout various modules.

This change aims to streamline logging and improve the overall architecture of the codebase.
2025-08-14 21:10:04 +08:00
CaIon 97ea8b6560 refactor: Introduce pre-consume quota and unify relay handlers
This commit introduces a major architectural refactoring to improve quota management, centralize logging, and streamline the relay handling logic.

Key changes:
- **Pre-consume Quota:** Implements a new mechanism to check and reserve user quota *before* making the request to the upstream provider. This ensures more accurate quota deduction and prevents users from exceeding their limits due to concurrent requests.

- **Unified Relay Handlers:** Refactors the relay logic to use generic handlers (e.g., `ChatHandler`, `ImageHandler`) instead of provider-specific implementations. This significantly reduces code duplication and simplifies adding new channels.

- **Centralized Logger:** A new dedicated `logger` package is introduced, and all system logging calls are migrated to use it, moving this responsibility out of the `common` package.

- **Code Reorganization:** DTOs are generalized (e.g., `dalle.go` -> `openai_image.go`) and utility code is moved to more appropriate packages (e.g., `common/http.go` -> `service/http.go`) for better code structure.
2025-08-14 20:05:06 +08:00
CaIon a4b02107dd feat: update MaxTokens handling 2025-08-07 16:15:59 +08:00
creamlike1024 689dbfe71a feat: convert gemini format to openai chat completions 2025-08-01 22:23:35 +08:00
CaIon 6ddeab2f2f fix(adaptor): enhance response handling and error logging for Claude format 2025-07-23 20:59:56 +08:00
CaIon c73b5886b9 feat: support ollama claude format 2025-07-23 20:01:03 +08:00
CaIon c5ec332ab3 fix: add Think field to OllamaRequest and support extra parameters in GeneralOpenAIRequest. (close #1125
)
2025-07-22 12:06:21 +08:00
CaIon 5bf5d3d16a feat(adaptor): refactor response handlers to return usage first and improve error handling 2025-07-12 21:12:46 +08:00
CaIon 4a9ca4209a refactor: Introduce standardized API error
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`).
2025-07-10 15:02:40 +08:00
CaIon aae9e81c20 🐛 fix: refactor response body handling in multiple relay handlers to utilize IOCopyBytesGracefully 2025-06-27 23:35:56 +08:00
CaIon ae586e1be9 feat: add CloseResponseBodyGracefully function to handle HTTP response body closure 2025-06-27 21:37:13 +08:00
creamlike1024 8ea4c76f2b feat: support /v1/responses API 2025-05-02 13:59:46 +08:00
1808837298@qq.com 921ad4530e refactor: Replace direct access to ImageUrl with GetImageMedia method across multiple relay channels 2025-03-15 19:43:37 +08:00
CalciumIon 5021000c5d refactor: Simplify OpenAI handler function signature and remove unused TextResponseWithError struct; introduce common_handler for rerank functionality 2025-03-14 17:31:05 +08:00
1808837298@qq.com 47c297794d feat: 初步兼容流模式下openai渠道类型转为claude格式访问 #862 2025-03-13 19:32:08 +08:00
1808837298@qq.com 78fc3a191c feat: claude relay 2025-03-12 21:31:46 +08:00
1808837298@qq.com 0646fa1892 fix: gemini&claude tool call format #795 #766 2025-02-26 23:56:10 +08:00
1808837298@qq.com 8f3c7280cf feat: Support max_tokens parameter for Ollama channel #782 2025-02-24 17:35:49 +08:00
1808837298@qq.com 8dbac87e92 fix: Correct Ollama channel authentication header setting 2025-02-20 01:28:15 +08:00
Coming 368b5fbaa1 fix: Fix Ollama channel authentication 2025-02-20 00:52:30 +08:00
1808837298@qq.com 5ef44f6898 feat: Improve image handling for Ollama channels 2025-02-19 20:45:42 +08:00
1808837298@qq.com 604a5b1ad5 feat: Enhance Ollama channel support with additional request parameters #771 2025-02-19 19:58:34 +08:00
1808837298@qq.com 61ad1adbda feat: Improve embedding request handling and support across channels
- Update EmbeddingRequest DTO to support more flexible input types
- Add input parsing method to handle various input formats
- Implement ConvertEmbeddingRequest for multiple channel adaptors
- Remove relayMode parameter from EmbeddingHelper
- Add input validation for embedding requests
- Simplify embedding request conversion for different channels
2025-02-12 14:39:36 +08:00
kuwork fdebb6e6e8 Merge branch 'main' into main 2025-02-04 22:52:37 +08:00
Jerry fcc32ffbc9 Fix M3E not working 2025-01-23 05:54:39 +08:00
Butui Hu 6e710f3210 Fix temperature not being set to 0 due to json omitempty
The issue was caused by the `omitempty` tag in the Go struct, which prevented the `temperature` field from being included in the JSON output when it was set to 0.

Signed-off-by: Butui Hu <hot123tea123@gmail.com>
2025-01-21 12:54:09 +08:00
HowieWood 0788a6c25f Continue fixing Ollama embedding return issue 2024-11-06 01:21:02 +00:00