lian/mchat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

faiss server

Browse Source
This commit is contained in:
1iaan
2026-04-10 11:55:00 +00:00
parent bc82e3e708
commit 8e39e609cc
30 changed files with 1271 additions and 1048 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
ai-chat-service/chat-server/server/app.go
View File

@@ -149,6 +149,30 @@ func (a *app) buildChatCompletionResponse(msg string) *proto.ChatCompletionRespo
}
}
func (a *app) countMessageTokens(role, content string) (int, error) {
message := openai.ChatCompletionMessage{
Role: role,
Content: content,
}
return tokenizer.GetTokens(&message, a.openaiConf.Model)
}
func (a *app) buildUsage(promptTokens int, answer string) (*proto.Usage, error) {
completionTokens := 0
if answer != "" {
tokens, err := a.countMessageTokens(openai.ChatMessageRoleAssistant, answer)
if err != nil {
return nil, err
}
completionTokens = tokens
}
return &proto.Usage{
PromptTokens: int32(promptTokens),
CompletionTokens: int32(completionTokens),
TotalTokens: int32(promptTokens + completionTokens),
}, nil
}
func (a *app) buildChatCompletionStreamResponse(id, delta, finishReason string) *proto.ChatCompletionStreamResponse {
return &proto.ChatCompletionStreamResponse{
Id: id,

104
ai-chat-service/chat-server/server/server.go
View File

@@ -8,7 +8,6 @@ import (
"ai-chat-service/proto"
"ai-chat-service/services/embedding"
"ai-chat-service/services/faiss"
"ai-chat-service/services/tokenizer"
"context"
"encoding/json"
"io"
@@ -19,6 +18,11 @@ import (
"github.com/sashabaranov/go-openai"
)
const (
replySourceSemanticMatch = "semantic_match"
replySourceLLM = "llm"
)
type chatService struct {
proto.UnimplementedChatServer
config *config.Config
@@ -50,7 +54,19 @@ func (s *chatService) ChatCompletion(ctx context.Context, in *proto.ChatCompleti
}
if !ok {
s.busMetrics.SensitiveQuestionsTotalCounter.Inc()
return app.buildChatCompletionResponse(msg), nil
res := app.buildChatCompletionResponse(msg)
promptTokens, tokenErr := app.countMessageTokens(openai.ChatMessageRoleUser, in.Message)
if tokenErr != nil {
s.log.Error(tokenErr)
return res, nil
}
usage, tokenErr := app.buildUsage(promptTokens, msg)
if tokenErr != nil {
s.log.Error(tokenErr)
return res, nil
}
res.Usage = usage
return res, nil
}
keywords := app.keywords(in)
@@ -58,7 +74,7 @@ func (s *chatService) ChatCompletion(ctx context.Context, in *proto.ChatCompleti
s.busMetrics.KeywordsQuestionsTotalCounter.Inc()
}
req, _, _, _, err := app.buildChatCompletionRequest(in, false)
req, _, currTokens, _, err := app.buildChatCompletionRequest(in, false)
if err != nil {
s.busMetrics.ErrQuestionsTotalCounter.Inc()
return nil, err
@@ -66,7 +82,15 @@ func (s *chatService) ChatCompletion(ctx context.Context, in *proto.ChatCompleti
questionEmbedding, cachedRecord := s.searchCachedAnswer(ctx, in.Message)
if cachedRecord != nil {
return app.buildChatCompletionResponse(cachedRecord.Answer), nil
res := app.buildChatCompletionResponse(cachedRecord.Answer)
usage, tokenErr := app.buildUsage(currTokens, cachedRecord.Answer)
if tokenErr != nil {
s.log.Error(tokenErr)
} else {
res.Usage = usage
}
res.Source = replySourceSemanticMatch
return res, nil
}
client := app.getOpenaiClient()
@@ -88,8 +112,20 @@ func (s *chatService) ChatCompletion(ctx context.Context, in *proto.ChatCompleti
return nil, err
}
answer := ""
if len(resp.Choices) > 0 {
if err = s.persistQA(ctx, questionEmbedding, in.Message, resp.Choices[0].Message.Content); err != nil {
answer = resp.Choices[0].Message.Content
}
usage, tokenErr := app.buildUsage(currTokens, answer)
if tokenErr != nil {
s.log.Error(tokenErr)
} else {
res.Usage = usage
}
res.Source = replySourceLLM
if len(resp.Choices) > 0 {
if err = s.persistQA(ctx, questionEmbedding, in.Message, answer); err != nil {
s.log.Error(err)
} else {
s.busMetrics.QuestionsTotalCounter.Inc()
@@ -109,7 +145,8 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
if !ok {
s.busMetrics.SensitiveQuestionsTotalCounter.Inc()
resID := uuid.New().String()
if err = stream.Send(app.buildChatCompletionStreamResponse(resID, "", "")); err != nil {
start := app.buildChatCompletionStreamResponse(resID, "", "")
if err = stream.Send(start); err != nil {
return err
}
for _, res := range app.buildChatCompletionStreamResponseList(resID, msg) {
@@ -117,7 +154,19 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
return err
}
}
return stream.Send(app.buildChatCompletionStreamResponse(resID, "", "stop"))
final := app.buildChatCompletionStreamResponse(resID, "", "stop")
promptTokens, tokenErr := app.countMessageTokens(openai.ChatMessageRoleUser, in.Message)
if tokenErr != nil {
s.log.Error(tokenErr)
} else {
usage, tokenErr := app.buildUsage(promptTokens, msg)
if tokenErr != nil {
s.log.Error(tokenErr)
} else {
final.Usage = usage
}
}
return stream.Send(final)
}
keywords := app.keywords(in)
@@ -125,7 +174,7 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
s.busMetrics.KeywordsQuestionsTotalCounter.Inc()
}
req, _, _, _, err := app.buildChatCompletionRequest(in, true)
req, _, currTokens, _, err := app.buildChatCompletionRequest(in, true)
if err != nil {
s.busMetrics.ErrQuestionsTotalCounter.Inc()
return err
@@ -133,7 +182,9 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
questionEmbedding, cachedRecord := s.searchCachedAnswer(stream.Context(), in.Message)
if cachedRecord != nil {
if err = stream.Send(app.buildChatCompletionStreamResponse(cachedRecord.ID, "", "")); err != nil {
start := app.buildChatCompletionStreamResponse(cachedRecord.ID, "", "")
start.Source = replySourceSemanticMatch
if err = stream.Send(start); err != nil {
return err
}
for _, res := range app.buildChatCompletionStreamResponseList(cachedRecord.ID, cachedRecord.Answer) {
@@ -141,7 +192,15 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
return err
}
}
return stream.Send(app.buildChatCompletionStreamResponse(cachedRecord.ID, "", "stop"))
final := app.buildChatCompletionStreamResponse(cachedRecord.ID, "", "stop")
final.Source = replySourceSemanticMatch
usage, tokenErr := app.buildUsage(currTokens, cachedRecord.Answer)
if tokenErr != nil {
s.log.Error(tokenErr)
} else {
final.Usage = usage
}
return stream.Send(final)
}
client := app.getOpenaiClient()
@@ -154,6 +213,7 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
defer chatStream.Close()
completionContent := ""
responseID := ""
for {
resp, err := chatStream.Recv()
if err != nil && err != io.EOF {
@@ -164,6 +224,9 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
if err == io.EOF {
break
}
if resp.ID != "" {
responseID = resp.ID
}
completionContent += resp.Choices[0].Delta.Content
res := &proto.ChatCompletionStreamResponse{}
@@ -176,21 +239,24 @@ func (s *chatService) ChatCompletionStream(in *proto.ChatCompletionRequest, stre
s.busMetrics.ErrQuestionsTotalCounter.Inc()
return err
}
res.Source = replySourceLLM
if err = stream.Send(res); err != nil {
return err
}
}
model := s.config.Chat.Model
if in.ChatParam != nil && in.ChatParam.Model != "" {
model = in.ChatParam.Model
}
resultMessage := openai.ChatCompletionMessage{
Role: openai.ChatMessageRoleAssistant,
Content: completionContent,
}
if _, err = tokenizer.GetTokens(&resultMessage, model); err != nil {
usage, tokenErr := app.buildUsage(currTokens, completionContent)
if tokenErr != nil {
s.busMetrics.ErrQuestionsTotalCounter.Inc()
return tokenErr
}
if responseID == "" {
responseID = uuid.New().String()
}
final := app.buildChatCompletionStreamResponse(responseID, "", "stop")
final.Usage = usage
final.Source = replySourceLLM
if err = stream.Send(final); err != nil {
return err
}