LLM/CLIProxyAPIPlus
1
0
Fork 0
mirror of https://github.com/router-for-me/CLIProxyAPIPlus.git synced 2026-03-30 01:06:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'router-for-me:main' into main

Browse Source
This commit is contained in:
Luis Pater
2026-02-07 06:43:17 +08:00
committed by GitHub
parent d182e893b6 f7d0019df7
commit c20507c15e
10 changed files with 421 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
internal/auth/kimi/kimi.go
View File

@@ -30,7 +30,7 @@ const (
// kimiTokenURL is the endpoint for exchanging device codes for tokens.
kimiTokenURL = kimiOAuthHost + "/api/oauth/token"
// KimiAPIBaseURL is the base URL for Kimi API requests.
KimiAPIBaseURL = "https://api.kimi.com/coding/v1"
KimiAPIBaseURL = "https://api.kimi.com/coding"
// defaultPollInterval is the default interval for polling token endpoint.
defaultPollInterval = 5 * time.Second
// maxPollDuration is the maximum time to wait for user authorization.

42
internal/runtime/executor/kimi_executor.go
View File

@@ -25,6 +25,7 @@ import (
// KimiExecutor is a stateless executor for Kimi API using OpenAI-compatible chat completions.
type KimiExecutor struct {
ClaudeExecutor
cfg *config.Config
}
@@ -64,6 +65,12 @@ func (e *KimiExecutor) HttpRequest(ctx context.Context, auth *cliproxyauth.Auth,
// Execute performs a non-streaming chat completion request to Kimi.
func (e *KimiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
from := opts.SourceFormat
if from.String() == "claude" {
auth.Attributes["base_url"] = kimiauth.KimiAPIBaseURL
return e.ClaudeExecutor.Execute(ctx, auth, req, opts)
}
baseModel := thinking.ParseSuffix(req.Model).ModelName
token := kimiCreds(auth)
@@ -71,7 +78,6 @@ func (e *KimiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req
reporter := newUsageReporter(ctx, e.Identifier(), baseModel, auth)
defer reporter.trackFailure(ctx, &err)
from := opts.SourceFormat
to := sdktranslator.FromString("openai")
originalPayload := bytes.Clone(req.Payload)
if len(opts.OriginalRequest) > 0 {
@@ -95,7 +101,7 @@ func (e *KimiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req
requestedModel := payloadRequestedModel(opts, req.Model)
body = applyPayloadConfigWithRoot(e.cfg, baseModel, to.String(), "", body, originalTranslated, requestedModel)
url := kimiauth.KimiAPIBaseURL + "/chat/completions"
url := kimiauth.KimiAPIBaseURL + "/v1/chat/completions"
httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
if err != nil {
return resp, err
@@ -155,14 +161,18 @@ func (e *KimiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req
// ExecuteStream performs a streaming chat completion request to Kimi.
func (e *KimiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
baseModel := thinking.ParseSuffix(req.Model).ModelName
from := opts.SourceFormat
if from.String() == "claude" {
auth.Attributes["base_url"] = kimiauth.KimiAPIBaseURL
return e.ClaudeExecutor.ExecuteStream(ctx, auth, req, opts)
}
baseModel := thinking.ParseSuffix(req.Model).ModelName
token := kimiCreds(auth)
reporter := newUsageReporter(ctx, e.Identifier(), baseModel, auth)
defer reporter.trackFailure(ctx, &err)
from := opts.SourceFormat
to := sdktranslator.FromString("openai")
originalPayload := bytes.Clone(req.Payload)
if len(opts.OriginalRequest) > 0 {
@@ -190,7 +200,7 @@ func (e *KimiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
requestedModel := payloadRequestedModel(opts, req.Model)
body = applyPayloadConfigWithRoot(e.cfg, baseModel, to.String(), "", body, originalTranslated, requestedModel)
url := kimiauth.KimiAPIBaseURL + "/chat/completions"
url := kimiauth.KimiAPIBaseURL + "/v1/chat/completions"
httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
if err != nil {
return nil, err
@@ -269,26 +279,8 @@ func (e *KimiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
// CountTokens estimates token count for Kimi requests.
func (e *KimiExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
baseModel := thinking.ParseSuffix(req.Model).ModelName
from := opts.SourceFormat
to := sdktranslator.FromString("openai")
body := sdktranslator.TranslateRequest(from, to, baseModel, bytes.Clone(req.Payload), false)
// Use a generic tokenizer for estimation
enc, err := tokenizerForModel("gpt-4")
if err != nil {
return cliproxyexecutor.Response{}, fmt.Errorf("kimi executor: tokenizer init failed: %w", err)
}
count, err := countOpenAIChatTokens(enc, body)
if err != nil {
return cliproxyexecutor.Response{}, fmt.Errorf("kimi executor: token counting failed: %w", err)
}
usageJSON := buildOpenAIUsageJSON(count)
translated := sdktranslator.TranslateTokenCount(ctx, to, from, count, usageJSON)
return cliproxyexecutor.Response{Payload: []byte(translated)}, nil
auth.Attributes["base_url"] = kimiauth.KimiAPIBaseURL
return e.ClaudeExecutor.CountTokens(ctx, auth, req, opts)
}
// Refresh refreshes the Kimi token using the refresh token.

22
internal/store/gitstore.go
View File

@@ -21,6 +21,9 @@ import (
cliproxyauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
)
// gcInterval defines minimum time between garbage collection runs.
const gcInterval = 5 * time.Minute
// GitTokenStore persists token records and auth metadata using git as the backing storage.
type GitTokenStore struct {
mu sync.Mutex
@@ -31,6 +34,7 @@ type GitTokenStore struct {
remote string
username string
password string
lastGC time.Time
}
// NewGitTokenStore creates a token store that saves credentials to disk through the
@@ -613,6 +617,7 @@ func (s *GitTokenStore) commitAndPushLocked(message string, relPaths ...string)
} else if errRewrite := s.rewriteHeadAsSingleCommit(repo, headRef.Name(), commitHash, message, signature); errRewrite != nil {
return errRewrite
}
s.maybeRunGC(repo)
if err = repo.Push(&git.PushOptions{Auth: s.gitAuth(), Force: true}); err != nil {
if errors.Is(err, git.NoErrAlreadyUpToDate) {
return nil
@@ -652,6 +657,23 @@ func (s *GitTokenStore) rewriteHeadAsSingleCommit(repo *git.Repository, branch p
return nil
}
func (s *GitTokenStore) maybeRunGC(repo *git.Repository) {
now := time.Now()
if now.Sub(s.lastGC) < gcInterval {
return
}
s.lastGC = now
pruneOpts := git.PruneOptions{
OnlyObjectsOlderThan: now,
Handler: repo.DeleteObject,
}
if err := repo.Prune(pruneOpts); err != nil && !errors.Is(err, git.ErrLooseObjectsNotSupported) {
return
}
_ = repo.RepackObjects(&git.RepackConfig{})
}
// PersistConfig commits and pushes configuration changes to git.
func (s *GitTokenStore) PersistConfig(_ context.Context) error {
if err := s.EnsureRepository(); err != nil {

6
internal/translator/codex/claude/codex_claude_response.go
View File

@@ -113,10 +113,10 @@ func ConvertCodexResponseToClaude(_ context.Context, _ string, originalRequestRa
template = `{"type":"message_delta","delta":{"stop_reason":"tool_use","stop_sequence":null},"usage":{"input_tokens":0,"output_tokens":0}}`
p := (*param).(*ConvertCodexResponseToClaudeParams).HasToolCall
stopReason := rootResult.Get("response.stop_reason").String()
if stopReason != "" {
template, _ = sjson.Set(template, "delta.stop_reason", stopReason)
} else if p {
if p {
template, _ = sjson.Set(template, "delta.stop_reason", "tool_use")
} else if stopReason == "max_tokens" || stopReason == "stop" {
template, _ = sjson.Set(template, "delta.stop_reason", stopReason)
} else {
template, _ = sjson.Set(template, "delta.stop_reason", "end_turn")
}

19
internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go
View File

@@ -78,11 +78,16 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
template, _ = sjson.Set(template, "id", responseIDResult.String())
}
// Extract and set the finish reason.
if finishReasonResult := gjson.GetBytes(rawJSON, "response.candidates.0.finishReason"); finishReasonResult.Exists() {
template, _ = sjson.Set(template, "choices.0.finish_reason", strings.ToLower(finishReasonResult.String()))
template, _ = sjson.Set(template, "choices.0.native_finish_reason", strings.ToLower(finishReasonResult.String()))
finishReason := ""
if stopReasonResult := gjson.GetBytes(rawJSON, "response.stop_reason"); stopReasonResult.Exists() {
finishReason = stopReasonResult.String()
}
if finishReason == "" {
if finishReasonResult := gjson.GetBytes(rawJSON, "response.candidates.0.finishReason"); finishReasonResult.Exists() {
finishReason = finishReasonResult.String()
}
}
finishReason = strings.ToLower(finishReason)
// Extract and set usage metadata (token counts).
if usageResult := gjson.GetBytes(rawJSON, "response.usageMetadata"); usageResult.Exists() {
@@ -197,6 +202,12 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
if hasFunctionCall {
template, _ = sjson.Set(template, "choices.0.finish_reason", "tool_calls")
template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
} else if finishReason != "" && (*param).(*convertCliResponseToOpenAIChatParams).FunctionIndex == 0 {
// Only pass through specific finish reasons
if finishReason == "max_tokens" || finishReason == "stop" {
template, _ = sjson.Set(template, "choices.0.finish_reason", finishReason)
template, _ = sjson.Set(template, "choices.0.native_finish_reason", finishReason)
}
}
return []string{template}

19
internal/translator/gemini/openai/chat-completions/gemini_openai_response.go
View File

@@ -129,11 +129,16 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
candidateIndex := int(candidate.Get("index").Int())
template, _ = sjson.Set(template, "choices.0.index", candidateIndex)
// Extract and set the finish reason.
if finishReasonResult := candidate.Get("finishReason"); finishReasonResult.Exists() {
template, _ = sjson.Set(template, "choices.0.finish_reason", strings.ToLower(finishReasonResult.String()))
template, _ = sjson.Set(template, "choices.0.native_finish_reason", strings.ToLower(finishReasonResult.String()))
finishReason := ""
if stopReasonResult := gjson.GetBytes(rawJSON, "stop_reason"); stopReasonResult.Exists() {
finishReason = stopReasonResult.String()
}
if finishReason == "" {
if finishReasonResult := gjson.GetBytes(rawJSON, "candidates.0.finishReason"); finishReasonResult.Exists() {
finishReason = finishReasonResult.String()
}
}
finishReason = strings.ToLower(finishReason)
partsResult := candidate.Get("content.parts")
hasFunctionCall := false
@@ -225,6 +230,12 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
if hasFunctionCall {
template, _ = sjson.Set(template, "choices.0.finish_reason", "tool_calls")
template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
} else if finishReason != "" {
// Only pass through specific finish reasons
if finishReason == "max_tokens" || finishReason == "stop" {
template, _ = sjson.Set(template, "choices.0.finish_reason", finishReason)
template, _ = sjson.Set(template, "choices.0.native_finish_reason", finishReason)
}
}
responseStrings = append(responseStrings, template)