jiazhizhong/higress
1
0
Fork 0
mirror of https://github.com/alibaba/higress.git synced 2026-06-09 20:57:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: resolve blocking issue with minimax responses in ai-proxy (#1663)

Browse Source
This commit is contained in:
韩贤涛
2025-01-14 09:43:19 +08:00
committed by GitHub
parent b3d9123d59
commit a1bf315b13
1 changed files with 39 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
plugins/wasm-go/extensions/ai-proxy/provider/minimax.go
View File

@@ -93,10 +93,6 @@ func (m *minimaxProvider) OnRequestBody(ctx wrapper.HttpContext, apiName ApiName
} }
} }
func (m *minimaxProvider) TransformRequestBodyHeaders(ctx wrapper.HttpContext, apiName ApiName, body []byte, headers http.Header, log wrapper.Log) ([]byte, error) {
return m.handleRequestBodyByChatCompletionV2(body, headers, log)
}
// handleRequestBodyByChatCompletionPro processes the request body using the chat completion Pro API. // handleRequestBodyByChatCompletionPro processes the request body using the chat completion Pro API.
func (m *minimaxProvider) handleRequestBodyByChatCompletionPro(body []byte, log wrapper.Log) (types.Action, error) { func (m *minimaxProvider) handleRequestBodyByChatCompletionPro(body []byte, log wrapper.Log) (types.Action, error) {
request := &chatCompletionRequest{} request := &chatCompletionRequest{}
@@ -109,7 +105,7 @@ func (m *minimaxProvider) handleRequestBodyByChatCompletionPro(body []byte, log
_ = util.OverwriteRequestPath(fmt.Sprintf("%s?GroupId=%s", minimaxChatCompletionProPath, m.config.minimaxGroupId)) _ = util.OverwriteRequestPath(fmt.Sprintf("%s?GroupId=%s", minimaxChatCompletionProPath, m.config.minimaxGroupId))
if m.config.context == nil { if m.config.context == nil {
minimaxRequest := m.buildMinimaxChatCompletionV2Request(request, "") minimaxRequest := m.buildMinimaxChatCompletionProRequest(request, "")
return types.ActionContinue, replaceJsonRequestBody(minimaxRequest, log) return types.ActionContinue, replaceJsonRequestBody(minimaxRequest, log)
} }
@@ -124,7 +120,7 @@ func (m *minimaxProvider) handleRequestBodyByChatCompletionPro(body []byte, log
// Since minimaxChatCompletionV2 (format consistent with OpenAI) and minimaxChatCompletionPro (different format from OpenAI) have different logic for insertHttpContextMessage, we cannot unify them within one provider. // Since minimaxChatCompletionV2 (format consistent with OpenAI) and minimaxChatCompletionPro (different format from OpenAI) have different logic for insertHttpContextMessage, we cannot unify them within one provider.
// For minimaxChatCompletionPro, we need to manually handle context messages. // For minimaxChatCompletionPro, we need to manually handle context messages.
// minimaxChatCompletionV2 uses the default defaultInsertHttpContextMessage method to insert context messages. // minimaxChatCompletionV2 uses the default defaultInsertHttpContextMessage method to insert context messages.
minimaxRequest := m.buildMinimaxChatCompletionV2Request(request, content) minimaxRequest := m.buildMinimaxChatCompletionProRequest(request, content)
if err := replaceJsonRequestBody(minimaxRequest, log); err != nil { if err := replaceJsonRequestBody(minimaxRequest, log); err != nil {
util.ErrorHandler("ai-proxy.minimax.insert_ctx_failed", fmt.Errorf("failed to replace Request body: %v", err)) util.ErrorHandler("ai-proxy.minimax.insert_ctx_failed", fmt.Errorf("failed to replace Request body: %v", err))
} }
@@ -135,6 +131,10 @@ func (m *minimaxProvider) handleRequestBodyByChatCompletionPro(body []byte, log
return types.ActionContinue, err return types.ActionContinue, err
} }
func (m *minimaxProvider) TransformRequestBodyHeaders(ctx wrapper.HttpContext, apiName ApiName, body []byte, headers http.Header, log wrapper.Log) ([]byte, error) {
return m.handleRequestBodyByChatCompletionV2(body, headers, log)
}
// handleRequestBodyByChatCompletionV2 processes the request body using the chat completion V2 API. // handleRequestBodyByChatCompletionV2 processes the request body using the chat completion V2 API.
func (m *minimaxProvider) handleRequestBodyByChatCompletionV2(body []byte, headers http.Header, log wrapper.Log) ([]byte, error) { func (m *minimaxProvider) handleRequestBodyByChatCompletionV2(body []byte, headers http.Header, log wrapper.Log) ([]byte, error) {
util.OverwriteRequestPathHeader(headers, minimaxChatCompletionV2Path) util.OverwriteRequestPathHeader(headers, minimaxChatCompletionV2Path)
@@ -144,15 +144,13 @@ func (m *minimaxProvider) handleRequestBodyByChatCompletionV2(body []byte, heade
return sjson.SetBytes(body, "model", mappedModel) return sjson.SetBytes(body, "model", mappedModel)
} }
// Skip OnStreamingResponseBody() and OnResponseBody() when using original protocol.
func (m *minimaxProvider) TransformResponseHeaders(ctx wrapper.HttpContext, apiName ApiName, headers http.Header, log wrapper.Log) { func (m *minimaxProvider) TransformResponseHeaders(ctx wrapper.HttpContext, apiName ApiName, headers http.Header, log wrapper.Log) {
if m.config.protocol == protocolOriginal { // Skip OnStreamingResponseBody() and OnResponseBody() when using the original protocol
ctx.DontReadResponseBody() // or when the model corresponds to the chat completion V2 interface.
} if m.config.protocol == protocolOriginal || minimaxApiTypePro != m.config.minimaxApiType {
// Skip OnStreamingResponseBody() and OnResponseBody() when the model corresponds to the chat completion V2 interface.
if minimaxApiTypePro != m.config.minimaxApiType {
ctx.DontReadResponseBody() ctx.DontReadResponseBody()
} else {
headers.Del("Content-Length")
} }
} }
@@ -174,12 +172,12 @@ func (m *minimaxProvider) OnStreamingResponseBody(ctx wrapper.HttpContext, name
continue continue
} }
data = data[6:] data = data[6:]
var minimaxResp minimaxChatCompletionV2Resp var minimaxResp minimaxChatCompletionProResp
if err := json.Unmarshal([]byte(data), &minimaxResp); err != nil { if err := json.Unmarshal([]byte(data), &minimaxResp); err != nil {
log.Errorf("unable to unmarshal minimax response: %v", err) log.Errorf("unable to unmarshal minimax response: %v", err)
continue continue
} }
response := m.responseV2ToOpenAI(&minimaxResp) response := m.responseProToOpenAI(&minimaxResp)
responseBody, err := json.Marshal(response) responseBody, err := json.Marshal(response)
if err != nil { if err != nil {
log.Errorf("unable to marshal response: %v", err) log.Errorf("unable to marshal response: %v", err)
@@ -192,21 +190,21 @@ func (m *minimaxProvider) OnStreamingResponseBody(ctx wrapper.HttpContext, name
return []byte(modifiedResponseChunk), nil return []byte(modifiedResponseChunk), nil
} }
// OnResponseBody handles the final response body from the Minimax service only for requests using the OpenAI protocol and corresponding to the chat completion Pro API. // TransformResponseBody handles the final response body from the Minimax service only for requests using the OpenAI protocol and corresponding to the chat completion Pro API.
func (m *minimaxProvider) TransformResponseBody(ctx wrapper.HttpContext, apiName ApiName, body []byte, log wrapper.Log) ([]byte, error) { func (m *minimaxProvider) TransformResponseBody(ctx wrapper.HttpContext, apiName ApiName, body []byte, log wrapper.Log) ([]byte, error) {
minimaxResp := &minimaxChatCompletionV2Resp{} minimaxResp := &minimaxChatCompletionProResp{}
if err := json.Unmarshal(body, minimaxResp); err != nil { if err := json.Unmarshal(body, minimaxResp); err != nil {
return nil, fmt.Errorf("unable to unmarshal minimax response: %v", err) return nil, fmt.Errorf("unable to unmarshal minimax response: %v", err)
} }
if minimaxResp.BaseResp.StatusCode != 0 { if minimaxResp.BaseResp.StatusCode != 0 {
return nil, fmt.Errorf("minimax response error, error_code: %d, error_message: %s", minimaxResp.BaseResp.StatusCode, minimaxResp.BaseResp.StatusMsg) return nil, fmt.Errorf("minimax response error, error_code: %d, error_message: %s", minimaxResp.BaseResp.StatusCode, minimaxResp.BaseResp.StatusMsg)
} }
response := m.responseV2ToOpenAI(minimaxResp) response := m.responseProToOpenAI(minimaxResp)
return json.Marshal(response) return json.Marshal(response)
} }
// minimaxChatCompletionV2Request represents the structure of a chat completion V2 request. // minimaxChatCompletionProRequest represents the structure of a chat completion Pro request.
type minimaxChatCompletionV2Request struct { type minimaxChatCompletionProRequest struct {
Model string `json:"model"` Model string `json:"model"`
Stream bool `json:"stream,omitempty"` Stream bool `json:"stream,omitempty"`
TokensToGenerate int64 `json:"tokens_to_generate,omitempty"` TokensToGenerate int64 `json:"tokens_to_generate,omitempty"`
@@ -237,15 +235,13 @@ type minimaxReplyConstraints struct {
SenderName string `json:"sender_name"` SenderName string `json:"sender_name"`
} }
// minimaxChatCompletionV2Resp represents the structure of a Minimax Chat Completion V2 response. // minimaxChatCompletionProResp represents the structure of a Minimax Chat Completion Pro response.
type minimaxChatCompletionV2Resp struct { type minimaxChatCompletionProResp struct {
Created int64 `json:"created"` Created int64 `json:"created"`
Model string `json:"model"` Model string `json:"model"`
Reply string `json:"reply"` Reply string `json:"reply"`
InputSensitive bool `json:"input_sensitive,omitempty"` InputSensitive bool `json:"input_sensitive,omitempty"`
InputSensitiveType int64 `json:"input_sensitive_type,omitempty"`
OutputSensitive bool `json:"output_sensitive,omitempty"` OutputSensitive bool `json:"output_sensitive,omitempty"`
OutputSensitiveType int64 `json:"output_sensitive_type,omitempty"`
Choices []minimaxChoice `json:"choices,omitempty"` Choices []minimaxChoice `json:"choices,omitempty"`
Usage minimaxUsage `json:"usage,omitempty"` Usage minimaxUsage `json:"usage,omitempty"`
Id string `json:"id"` Id string `json:"id"`
@@ -268,6 +264,8 @@ type minimaxChoice struct {
// minimaxUsage represents token usage statistics. // minimaxUsage represents token usage statistics.
type minimaxUsage struct { type minimaxUsage struct {
TotalTokens int64 `json:"total_tokens"` TotalTokens int64 `json:"total_tokens"`
PromptTokens int64 `json:"prompt_tokens"`
CompletionTokens int64 `json:"completion_tokens"`
} }
func (m *minimaxProvider) parseModel(body []byte) (string, error) { func (m *minimaxProvider) parseModel(body []byte) (string, error) {
@@ -282,7 +280,7 @@ func (m *minimaxProvider) parseModel(body []byte) (string, error) {
return model, nil return model, nil
} }
func (m *minimaxProvider) setBotSettings(request *minimaxChatCompletionV2Request, botSettingContent string) { func (m *minimaxProvider) setBotSettings(request *minimaxChatCompletionProRequest, botSettingContent string) {
if len(request.BotSettings) == 0 { if len(request.BotSettings) == 0 {
request.BotSettings = []minimaxBotSetting{ request.BotSettings = []minimaxBotSetting{
{ {
@@ -304,7 +302,7 @@ func (m *minimaxProvider) setBotSettings(request *minimaxChatCompletionV2Request
} }
} }
func (m *minimaxProvider) buildMinimaxChatCompletionV2Request(request *chatCompletionRequest, botSettingContent string) *minimaxChatCompletionV2Request { func (m *minimaxProvider) buildMinimaxChatCompletionProRequest(request *chatCompletionRequest, botSettingContent string) *minimaxChatCompletionProRequest {
var messages []minimaxMessage var messages []minimaxMessage
var botSetting []minimaxBotSetting var botSetting []minimaxBotSetting
var botName string var botName string
@@ -343,7 +341,7 @@ func (m *minimaxProvider) buildMinimaxChatCompletionV2Request(request *chatCompl
SenderType: senderTypeBot, SenderType: senderTypeBot,
SenderName: determineName(botName, defaultBotName), SenderName: determineName(botName, defaultBotName),
} }
result := &minimaxChatCompletionV2Request{ result := &minimaxChatCompletionProRequest{
Model: request.Model, Model: request.Model,
Stream: request.Stream, Stream: request.Stream,
TokensToGenerate: int64(request.MaxTokens), TokensToGenerate: int64(request.MaxTokens),
@@ -359,7 +357,7 @@ func (m *minimaxProvider) buildMinimaxChatCompletionV2Request(request *chatCompl
return result return result
} }
func (m *minimaxProvider) responseV2ToOpenAI(response *minimaxChatCompletionV2Resp) *chatCompletionResponse { func (m *minimaxProvider) responseProToOpenAI(response *minimaxChatCompletionProResp) *chatCompletionResponse {
var choices []chatCompletionChoice var choices []chatCompletionChoice
messageIndex := 0 messageIndex := 0
for _, choice := range response.Choices { for _, choice := range response.Choices {
@@ -385,6 +383,8 @@ func (m *minimaxProvider) responseV2ToOpenAI(response *minimaxChatCompletionV2Re
Choices: choices, Choices: choices,
Usage: usage{ Usage: usage{
TotalTokens: int(response.Usage.TotalTokens), TotalTokens: int(response.Usage.TotalTokens),
PromptTokens: int(response.Usage.PromptTokens),
CompletionTokens: int(response.Usage.CompletionTokens),
}, },
} }
} }