mirror of
https://github.com/CJackHwang/ds2api.git
synced 2026-05-04 00:15:28 +08:00
234 lines
6.5 KiB
Go
234 lines
6.5 KiB
Go
package gemini
|
|
|
|
import (
|
|
"encoding/json"
|
|
"io"
|
|
"net/http"
|
|
"strings"
|
|
"time"
|
|
|
|
"ds2api/internal/assistantturn"
|
|
dsprotocol "ds2api/internal/deepseek/protocol"
|
|
"ds2api/internal/sse"
|
|
streamengine "ds2api/internal/stream"
|
|
)
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
func (h *Handler) handleStreamGenerateContent(w http.ResponseWriter, r *http.Request, resp *http.Response, model, finalPrompt string, thinkingEnabled, searchEnabled bool, toolNames []string, toolsRaw any) {
|
|
defer func() { _ = resp.Body.Close() }()
|
|
if resp.StatusCode != http.StatusOK {
|
|
body, _ := io.ReadAll(resp.Body)
|
|
writeGeminiError(w, resp.StatusCode, strings.TrimSpace(string(body)))
|
|
return
|
|
}
|
|
|
|
w.Header().Set("Content-Type", "text/event-stream")
|
|
w.Header().Set("Cache-Control", "no-cache, no-transform")
|
|
w.Header().Set("Connection", "keep-alive")
|
|
w.Header().Set("X-Accel-Buffering", "no")
|
|
|
|
rc := http.NewResponseController(w)
|
|
_, canFlush := w.(http.Flusher)
|
|
runtime := newGeminiStreamRuntime(w, rc, canFlush, model, finalPrompt, thinkingEnabled, searchEnabled, stripReferenceMarkersEnabled(), toolNames, toolsRaw)
|
|
|
|
initialType := "text"
|
|
if thinkingEnabled {
|
|
initialType = "thinking"
|
|
}
|
|
streamengine.ConsumeSSE(streamengine.ConsumeConfig{
|
|
Context: r.Context(),
|
|
Body: resp.Body,
|
|
ThinkingEnabled: thinkingEnabled,
|
|
InitialType: initialType,
|
|
KeepAliveInterval: time.Duration(dsprotocol.KeepAliveTimeout) * time.Second,
|
|
IdleTimeout: time.Duration(dsprotocol.StreamIdleTimeout) * time.Second,
|
|
MaxKeepAliveNoInput: dsprotocol.MaxKeepaliveCount,
|
|
}, streamengine.ConsumeHooks{
|
|
OnParsed: runtime.onParsed,
|
|
OnFinalize: func(_ streamengine.StopReason, _ error) {
|
|
runtime.finalize()
|
|
},
|
|
})
|
|
}
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
type geminiStreamRuntime struct {
|
|
w http.ResponseWriter
|
|
rc *http.ResponseController
|
|
canFlush bool
|
|
|
|
model string
|
|
finalPrompt string
|
|
|
|
thinkingEnabled bool
|
|
searchEnabled bool
|
|
bufferContent bool
|
|
stripReferenceMarkers bool
|
|
toolNames []string
|
|
toolsRaw any
|
|
|
|
accumulator *assistantturn.Accumulator
|
|
contentFilter bool
|
|
responseMessageID int
|
|
}
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
func newGeminiStreamRuntime(
|
|
w http.ResponseWriter,
|
|
rc *http.ResponseController,
|
|
canFlush bool,
|
|
model string,
|
|
finalPrompt string,
|
|
thinkingEnabled bool,
|
|
searchEnabled bool,
|
|
stripReferenceMarkers bool,
|
|
toolNames []string,
|
|
toolsRaw any,
|
|
) *geminiStreamRuntime {
|
|
return &geminiStreamRuntime{
|
|
w: w,
|
|
rc: rc,
|
|
canFlush: canFlush,
|
|
model: model,
|
|
finalPrompt: finalPrompt,
|
|
thinkingEnabled: thinkingEnabled,
|
|
searchEnabled: searchEnabled,
|
|
bufferContent: len(toolNames) > 0,
|
|
stripReferenceMarkers: stripReferenceMarkers,
|
|
toolNames: toolNames,
|
|
toolsRaw: toolsRaw,
|
|
accumulator: assistantturn.NewAccumulator(assistantturn.AccumulatorOptions{
|
|
ThinkingEnabled: thinkingEnabled,
|
|
SearchEnabled: searchEnabled,
|
|
StripReferenceMarkers: stripReferenceMarkers,
|
|
}),
|
|
}
|
|
}
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
func (s *geminiStreamRuntime) sendChunk(payload map[string]any) {
|
|
b, _ := json.Marshal(payload)
|
|
_, _ = s.w.Write([]byte("data: "))
|
|
_, _ = s.w.Write(b)
|
|
_, _ = s.w.Write([]byte("\n\n"))
|
|
if s.canFlush {
|
|
_ = s.rc.Flush()
|
|
}
|
|
}
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
func (s *geminiStreamRuntime) onParsed(parsed sse.LineResult) streamengine.ParsedDecision {
|
|
if !parsed.Parsed {
|
|
return streamengine.ParsedDecision{}
|
|
}
|
|
if parsed.ResponseMessageID > 0 {
|
|
s.responseMessageID = parsed.ResponseMessageID
|
|
}
|
|
if parsed.ContentFilter || parsed.ErrorMessage != "" || parsed.Stop {
|
|
if parsed.ContentFilter {
|
|
s.contentFilter = true
|
|
}
|
|
return streamengine.ParsedDecision{Stop: true}
|
|
}
|
|
|
|
accumulated := s.accumulator.Apply(parsed)
|
|
for _, p := range accumulated.Parts {
|
|
if p.Type == "thinking" {
|
|
if p.VisibleText == "" || s.bufferContent {
|
|
continue
|
|
}
|
|
s.sendChunk(map[string]any{
|
|
"candidates": []map[string]any{
|
|
{
|
|
"index": 0,
|
|
"content": map[string]any{
|
|
"role": "model",
|
|
"parts": []map[string]any{{"text": p.VisibleText, "thought": true}},
|
|
},
|
|
},
|
|
},
|
|
"modelVersion": s.model,
|
|
})
|
|
continue
|
|
}
|
|
if p.RawText == "" || p.CitationOnly || p.VisibleText == "" {
|
|
continue
|
|
}
|
|
if s.bufferContent {
|
|
continue
|
|
}
|
|
s.sendChunk(map[string]any{
|
|
"candidates": []map[string]any{
|
|
{
|
|
"index": 0,
|
|
"content": map[string]any{
|
|
"role": "model",
|
|
"parts": []map[string]any{{"text": p.VisibleText}},
|
|
},
|
|
},
|
|
},
|
|
"modelVersion": s.model,
|
|
})
|
|
}
|
|
return streamengine.ParsedDecision{ContentSeen: accumulated.ContentSeen}
|
|
}
|
|
|
|
//nolint:unused // retained for native Gemini stream handling path.
|
|
func (s *geminiStreamRuntime) finalize() {
|
|
rawText, text, rawThinking, thinking, detectionThinking := s.accumulator.Snapshot()
|
|
turn := assistantturn.BuildTurnFromStreamSnapshot(assistantturn.StreamSnapshot{
|
|
RawText: rawText,
|
|
VisibleText: text,
|
|
RawThinking: rawThinking,
|
|
VisibleThinking: thinking,
|
|
DetectionThinking: detectionThinking,
|
|
ContentFilter: s.contentFilter,
|
|
ResponseMessageID: s.responseMessageID,
|
|
}, assistantturn.BuildOptions{
|
|
Model: s.model,
|
|
Prompt: s.finalPrompt,
|
|
SearchEnabled: s.searchEnabled,
|
|
StripReferenceMarkers: s.stripReferenceMarkers,
|
|
ToolNames: s.toolNames,
|
|
ToolsRaw: s.toolsRaw,
|
|
})
|
|
outcome := assistantturn.FinalizeTurn(turn, assistantturn.FinalizeOptions{})
|
|
|
|
if s.bufferContent {
|
|
parts := buildGeminiPartsFromTurn(turn)
|
|
s.sendChunk(map[string]any{
|
|
"candidates": []map[string]any{
|
|
{
|
|
"index": 0,
|
|
"content": map[string]any{
|
|
"role": "model",
|
|
"parts": parts,
|
|
},
|
|
},
|
|
},
|
|
"modelVersion": s.model,
|
|
})
|
|
}
|
|
|
|
s.sendChunk(map[string]any{
|
|
"candidates": []map[string]any{
|
|
{
|
|
"index": 0,
|
|
"content": map[string]any{
|
|
"role": "model",
|
|
"parts": []map[string]any{
|
|
{"text": ""},
|
|
},
|
|
},
|
|
"finishReason": "STOP",
|
|
},
|
|
},
|
|
"modelVersion": s.model,
|
|
"usageMetadata": map[string]any{
|
|
"promptTokenCount": outcome.Usage.InputTokens,
|
|
"candidatesTokenCount": outcome.Usage.OutputTokens,
|
|
"totalTokenCount": outcome.Usage.TotalTokens,
|
|
},
|
|
})
|
|
}
|