Skip to content

phsl / new-api

Go to a project
Commit 414b959e by somnifex
Options

fix: 添加对Thinking字段的处理逻辑，确保推理内容正确传递

parent b80811a4
Showing with 21 additions and 10 deletions
relay/channel/ollama/stream.go
...@@ -26,6 +26,7 @@ type ollamaChatStreamChunk struct { ...@@ -26,6 +26,7 @@ type ollamaChatStreamChunk struct {
Message *struct { Message *struct {
Role string `json:"role"` Role string `json:"role"`
Content string `json:"content"` Content string `json:"content"`
Thinking json.RawMessage `json:"thinking"`
ToolCalls []struct { ToolCalls []struct {
Function struct { Function struct {
Name string `json:"name"` Name string `json:"name"`
...@@ -41,7 +42,6 @@ type ollamaChatStreamChunk struct { ...@@ -41,7 +42,6 @@ type ollamaChatStreamChunk struct {
LoadDuration int64 `json:"load_duration"` LoadDuration int64 `json:"load_duration"`
PromptEvalCount int `json:"prompt_eval_count"` PromptEvalCount int `json:"prompt_eval_count"`
EvalCount int `json:"eval_count"` EvalCount int `json:"eval_count"`
// generate mode may use these
PromptEvalDuration int64 `json:"prompt_eval_duration"` PromptEvalDuration int64 `json:"prompt_eval_duration"`
EvalDuration int64 `json:"eval_duration"` EvalDuration int64 `json:"eval_duration"`
} }
...@@ -95,13 +95,18 @@ func ollamaStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http ...@@ -95,13 +95,18 @@ func ollamaStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
} }, } },
} }
if content != "" { delta.Choices[0].Delta.SetContentString(content) } if content != "" { delta.Choices[0].Delta.SetContentString(content) }
if chunk.Message != nil && len(chunk.Message.Thinking) > 0 {
raw := strings.TrimSpace(string(chunk.Message.Thinking))
if raw != "" && raw != "null" { delta.Choices[0].Delta.SetReasoningContent(raw) }
}
// tool calls // tool calls
if chunk.Message != nil && len(chunk.Message.ToolCalls) > 0 { if chunk.Message != nil && len(chunk.Message.ToolCalls) > 0 {
delta.Choices[0].Delta.ToolCalls = make([]dto.ToolCallResponse,0,len(chunk.Message.ToolCalls)) delta.Choices[0].Delta.ToolCalls = make([]dto.ToolCallResponse,0,len(chunk.Message.ToolCalls))
for _, tc := range chunk.Message.ToolCalls { for _, tc := range chunk.Message.ToolCalls {
// arguments -> string // arguments -> string
argBytes, _ := json.Marshal(tc.Function.Arguments) argBytes, _ := json.Marshal(tc.Function.Arguments)
tr := dto.ToolCallResponse{ID:"", Type:nil, Function: dto.FunctionResponse{Name: tc.Function.Name, Arguments: string(argBytes)}} toolId := fmt.Sprintf("call_%d", toolCallIndex)
tr := dto.ToolCallResponse{ID:toolId, Type:"function", Function: dto.FunctionResponse{Name: tc.Function.Name, Arguments: string(argBytes)}}
tr.SetIndex(toolCallIndex) tr.SetIndex(toolCallIndex)
toolCallIndex++ toolCallIndex++
delta.Choices[0].Delta.ToolCalls = append(delta.Choices[0].Delta.ToolCalls, tr) delta.Choices[0].Delta.ToolCalls = append(delta.Choices[0].Delta.ToolCalls, tr)
...@@ -115,8 +120,8 @@ func ollamaStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http ...@@ -115,8 +120,8 @@ func ollamaStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
usage.PromptTokens = chunk.PromptEvalCount usage.PromptTokens = chunk.PromptEvalCount
usage.CompletionTokens = chunk.EvalCount usage.CompletionTokens = chunk.EvalCount
usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
finishReason := chunk.DoneReason finishReason := chunk.DoneReason
if finishReason == "" { finishReason = "stop" } if finishReason == "" { finishReason = "stop" }
// emit stop delta // emit stop delta
if stop := helper.GenerateStopResponse(responseId, created, model, finishReason); stop != nil { if stop := helper.GenerateStopResponse(responseId, created, model, finishReason); stop != nil {
if data, err := common.Marshal(stop); err == nil { _ = helper.StringData(c, string(data)) } if data, err := common.Marshal(stop); err == nil { _ = helper.StringData(c, string(data)) }
...@@ -144,6 +149,7 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R ...@@ -144,6 +149,7 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R
lines := strings.Split(raw, "\n") lines := strings.Split(raw, "\n")
var ( var (
aggContent strings.Builder aggContent strings.Builder
reasoningBuilder strings.Builder
lastChunk ollamaChatStreamChunk lastChunk ollamaChatStreamChunk
parsedAny bool parsedAny bool
) )
...@@ -157,18 +163,21 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R ...@@ -157,18 +163,21 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R
} }
parsedAny = true parsedAny = true
lastChunk = ck lastChunk = ck
if !ck.Done { if ck.Message != nil && len(ck.Message.Thinking) > 0 {
if ck.Message != nil && ck.Message.Content != "" { aggContent.WriteString(ck.Message.Content) } else if ck.Response != "" { aggContent.WriteString(ck.Response) } raw := strings.TrimSpace(string(ck.Message.Thinking))
} else { if raw != "" && raw != "null" { reasoningBuilder.WriteString(raw) }
if ck.Message != nil && ck.Message.Content != "" { aggContent.WriteString(ck.Message.Content) } else if ck.Response != "" { aggContent.WriteString(ck.Response) }
} }
if ck.Message != nil && ck.Message.Content != "" { aggContent.WriteString(ck.Message.Content) } else if ck.Response != "" { aggContent.WriteString(ck.Response) }
} }
if !parsedAny { if !parsedAny {
var single ollamaChatStreamChunk var single ollamaChatStreamChunk
if err := json.Unmarshal(body, &single); err != nil { return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError) } if err := json.Unmarshal(body, &single); err != nil { return nil, types.NewOpenAIError(err, types.ErrorCodeBadResponseBody, http.StatusInternalServerError) }
lastChunk = single lastChunk = single
if single.Message != nil { aggContent.WriteString(single.Message.Content) } else { aggContent.WriteString(single.Response) } if single.Message != nil {
if len(single.Message.Thinking) > 0 { raw := strings.TrimSpace(string(single.Message.Thinking)); if raw != "" && raw != "null" { reasoningBuilder.WriteString(raw) } }
aggContent.WriteString(single.Message.Content)
} else { aggContent.WriteString(single.Response) }
} }
model := lastChunk.Model model := lastChunk.Model
...@@ -179,6 +188,8 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R ...@@ -179,6 +188,8 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R
finishReason := lastChunk.DoneReason finishReason := lastChunk.DoneReason
if finishReason == "" { finishReason = "stop" } if finishReason == "" { finishReason = "stop" }
msg := dto.Message{Role: "assistant", Content: contentPtr(content)}
if rc := reasoningBuilder.String(); rc != "" { msg.ReasoningContent = &rc }
full := dto.OpenAITextResponse{ full := dto.OpenAITextResponse{
Id: common.GetUUID(), Id: common.GetUUID(),
Model: model, Model: model,
...@@ -186,7 +197,7 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R ...@@ -186,7 +197,7 @@ func ollamaChatHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R
Created: created, Created: created,
Choices: []dto.OpenAITextResponseChoice{ { Choices: []dto.OpenAITextResponseChoice{ {
Index: 0, Index: 0,
Message: dto.Message{Role: "assistant", Content: contentPtr(content)}, Message: msg,
FinishReason: finishReason, FinishReason: finishReason,
} }, } },
Usage: *usage, Usage: *usage,
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or sign in to comment