Files
new-api/relaykit/relayconvert/internal/oai_chat/to_claude_messages_req.go
T
Calcium-Ion 86ac0f7745 refactor: extract protocol conversion layer into standalone relaykit module (#6369)
* test(relayconvert): add golden snapshot matrix and relaykit boundary guard

Phase 0 of the relaykit extraction plan: pin byte-level output of every
registered (from,to) request/response/stream conversion route, and
forbid kit-bound packages from growing host-only imports.

* wip(relayconvert): drop gin.Context from converter signatures; add convmeta draft

Phase 1 in progress: relayconvert now takes context.Context; host media
resolver adapts gin.Context back at the service boundary.

* refactor(relayconvert): decouple converters from RelayInfo, gin, and settings

Phase 1 of the relaykit extraction plan:
- converters now depend on convmeta.Meta (implemented by RelayInfo) instead
  of *relaycommon.RelayInfo; ClaudeConvertInfo and the format guesser move
  to convmeta with aliases left behind
- host settings reach converters via a convmeta.Options snapshot built in
  RelayInfo.ConvOptions; no more model_setting/reasoning global reads inside
  the conversion layer
- effort-suffix helpers move to service/relayconvert/reasoning (old package
  forwards); chat-to-responses upgrade policy moves to service (host routing
  logic, not conversion)
- golden conversion matrix unchanged

* test(relayconvert): tighten boundary — kit packages now free of gin/setting imports

* refactor(dto): drop gin and logger dependencies

Phase 2 (part 1): dto.Request.IsStream now takes *http.Request instead of
*gin.Context (Gemini's impl reads query/path off the std request); dto's
three logger calls become common.SysError. Boundary test allowlist is now
empty — kit-bound packages import no gin/setting/logger/model.

* refactor(kit): extract dependency-free kitutil; dto/types/relayconvert stop importing common

Phase 2 of the relaykit extraction plan:
- new service/relayconvert/kitutil holds the pure helpers the kit needs
  (JSON wrappers, pointer/string/uuid/timestamp utils, MaskSensitiveInfo,
  pluggable LogInfo/LogError hooks, Debug flag)
- dto, types, and all relayconvert packages now use kitutil; their only
  remaining internal deps are dto/types/constant
- common keeps every original symbol (MaskSensitiveInfo delegates to
  kitutil) so host code is untouched; main.go routes kit logging into
  common.SysLog/SysError and mirrors DebugEnabled
- golden conversion matrix unchanged

* refactor(kit): move EndpointType/FinishReason to types; OpenRouter dialect via Options

Kit packages (dto/types/relayconvert/reasonmap) no longer import constant:
- EndpointType and finish-reason values live in types; constant re-exports
- the OpenRouter special-case in claude->openai request conversion reads
  Options.OpenRouterDialect, set by the host from the channel type;
  InitChannelMeta invalidates the cached snapshot on channel switch

* refactor: extract relaykit submodule (dto/types/relayconvert/reasonmap)

Phase 3 of the relaykit extraction plan:
- new go module github.com/QuantumNous/new-api/relaykit containing dto
  (minus task family), types, relayconvert (with convmeta/kitutil/reasoning),
  and reasonmap; host consumes it via require + replace, go.work for dev
- task-family dto (task/suno/midjourney/video) stays in the host dto
  package; dual-consumer host files alias it as taskdto
- relaykit builds and tests standalone (GOWORK=off): no host imports,
  no gin, no DB, no settings
- golden conversion matrix unchanged

* build(docker): copy relaykit/go.mod before go mod download

The local-replace submodule's go.mod must exist inside the build context
for the main module graph to resolve.

* fix: address relaykit extraction regressions

* fix: address relaykit review regressions

* docs: document Meta nil receiver contract

* fix(relaykit): fail OpenAI→Claude conversion without max_tokens; reject negative default_max_tokens

The Claude Messages API requires max_tokens (omitting it is a 400
"Field required"), but with a nil Options.Claude.DefaultMaxTokens hook
the converters silently emitted a request the upstream is guaranteed to
reject. Both OpenAI Chat and Responses → Claude conversions now return
sharedclaude.ErrMissingMaxTokens when no path (client value, default
hook, thinking-adapter floor) supplied one. Unreachable in the host,
which always configures the hook.

Host side, claude.default_max_tokens now rejects negative values at the
option API before persisting — they would wrap into huge unsigned values
during conversion. Zero stays allowed: the current API treats
max_tokens: 0 as cache pre-warming.

* fix: make Gemini safety settings read path race-free
2026-07-27 15:56:21 +08:00

410 lines
13 KiB
Go

package oaichat
import (
"encoding/json"
"fmt"
"strings"
"context"
"github.com/QuantumNous/new-api/relaykit/dto"
"github.com/QuantumNous/new-api/relaykit/relayconvert/convmeta"
relaymedia "github.com/QuantumNous/new-api/relaykit/relayconvert/internal/media"
sharedclaude "github.com/QuantumNous/new-api/relaykit/relayconvert/internal/shared/claude"
kitutil "github.com/QuantumNous/new-api/relaykit/relayconvert/kitutil"
"github.com/QuantumNous/new-api/relaykit/relayconvert/reasoning"
)
const (
webSearchMaxUsesLow = 1
webSearchMaxUsesMedium = 5
webSearchMaxUsesHigh = 10
)
type openRouterRequestReasoning struct {
Enabled bool `json:"enabled"`
Effort string `json:"effort,omitempty"`
MaxTokens int `json:"max_tokens,omitempty"`
Exclude bool `json:"exclude,omitempty"`
}
func OpenAIChatRequestToClaudeMessages(c context.Context, info convmeta.Meta, textRequest dto.GeneralOpenAIRequest) (*dto.ClaudeRequest, error) {
opts := convmeta.OptionsOf(info)
claudeTools := make([]any, 0, len(textRequest.Tools))
for _, tool := range textRequest.Tools {
if params, ok := tool.Function.Parameters.(map[string]any); ok {
claudeTool := dto.Tool{
Name: tool.Function.Name,
Description: tool.Function.Description,
}
claudeTool.InputSchema = make(map[string]interface{})
if params["type"] != nil {
claudeTool.InputSchema["type"] = params["type"].(string)
}
claudeTool.InputSchema["properties"] = params["properties"]
claudeTool.InputSchema["required"] = params["required"]
for key, value := range params {
if key == "type" || key == "properties" || key == "required" {
continue
}
claudeTool.InputSchema[key] = value
}
claudeTools = append(claudeTools, &claudeTool)
}
}
if textRequest.WebSearchOptions != nil {
webSearchTool := dto.ClaudeWebSearchTool{
Type: "web_search_20250305",
Name: "web_search",
}
if textRequest.WebSearchOptions.UserLocation != nil {
anthropicUserLocation := &dto.ClaudeWebSearchUserLocation{
Type: "approximate",
}
var userLocationMap map[string]interface{}
if err := kitutil.Unmarshal(textRequest.WebSearchOptions.UserLocation, &userLocationMap); err == nil {
if approximateData, ok := userLocationMap["approximate"].(map[string]interface{}); ok {
if timezone, ok := approximateData["timezone"].(string); ok && timezone != "" {
anthropicUserLocation.Timezone = timezone
}
if country, ok := approximateData["country"].(string); ok && country != "" {
anthropicUserLocation.Country = country
}
if region, ok := approximateData["region"].(string); ok && region != "" {
anthropicUserLocation.Region = region
}
if city, ok := approximateData["city"].(string); ok && city != "" {
anthropicUserLocation.City = city
}
}
}
webSearchTool.UserLocation = anthropicUserLocation
}
switch textRequest.WebSearchOptions.SearchContextSize {
case "low":
webSearchTool.MaxUses = webSearchMaxUsesLow
case "medium":
webSearchTool.MaxUses = webSearchMaxUsesMedium
case "high":
webSearchTool.MaxUses = webSearchMaxUsesHigh
}
claudeTools = append(claudeTools, &webSearchTool)
}
claudeRequest := dto.ClaudeRequest{
Model: textRequest.Model,
StopSequences: nil,
Temperature: textRequest.Temperature,
Tools: claudeTools,
}
if maxTokens := textRequest.GetMaxTokens(); maxTokens > 0 {
claudeRequest.MaxTokens = kitutil.GetPointer(maxTokens)
}
if textRequest.TopP != nil {
claudeRequest.TopP = kitutil.GetPointer(*textRequest.TopP)
}
if textRequest.TopK != nil {
claudeRequest.TopK = kitutil.GetPointer(*textRequest.TopK)
}
if textRequest.IsStream(nil) {
claudeRequest.Stream = kitutil.GetPointer(true)
}
if textRequest.ToolChoice != nil || textRequest.ParallelTooCalls != nil {
claudeToolChoice := sharedclaude.MapOpenAIToolChoice(textRequest.ToolChoice, textRequest.ParallelTooCalls)
if claudeToolChoice != nil {
claudeRequest.ToolChoice = claudeToolChoice
}
}
if claudeRequest.MaxTokens == nil || *claudeRequest.MaxTokens == 0 {
if defaultMaxTokens, configured := opts.Claude.DefaultMaxTokensFor(textRequest.Model); configured {
value := uint(defaultMaxTokens)
claudeRequest.MaxTokens = &value
}
}
if baseModel, effortLevel, ok := reasoning.TrimEffortSuffix(textRequest.Model); ok && effortLevel != "" &&
(strings.HasPrefix(textRequest.Model, "claude-opus-4-6") ||
strings.HasPrefix(textRequest.Model, "claude-opus-4-7") ||
strings.HasPrefix(textRequest.Model, "claude-opus-4-8")) {
claudeRequest.Model = baseModel
claudeRequest.Thinking = &dto.Thinking{
Type: "adaptive",
}
claudeRequest.OutputConfig = json.RawMessage(fmt.Sprintf(`{"effort":"%s"}`, effortLevel))
if strings.HasPrefix(baseModel, "claude-opus-4-7") ||
strings.HasPrefix(baseModel, "claude-opus-4-8") {
claudeRequest.Thinking.Display = "summarized"
claudeRequest.Temperature = nil
claudeRequest.TopP = nil
claudeRequest.TopK = nil
} else {
claudeRequest.TopP = nil
claudeRequest.Temperature = kitutil.GetPointer[float64](1.0)
}
} else if opts.Claude.ThinkingAdapterEnabled &&
strings.HasSuffix(textRequest.Model, "-thinking") {
trimmedModel := strings.TrimSuffix(textRequest.Model, "-thinking")
if strings.HasPrefix(trimmedModel, "claude-opus-4-7") ||
strings.HasPrefix(trimmedModel, "claude-opus-4-8") {
claudeRequest.Thinking = &dto.Thinking{Type: "adaptive", Display: "summarized"}
claudeRequest.OutputConfig = json.RawMessage(`{"effort":"high"}`)
claudeRequest.Temperature = nil
claudeRequest.TopP = nil
claudeRequest.TopK = nil
} else {
if claudeRequest.MaxTokens == nil || *claudeRequest.MaxTokens < 1280 {
claudeRequest.MaxTokens = kitutil.GetPointer[uint](1280)
}
claudeRequest.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: kitutil.GetPointer[int](int(float64(*claudeRequest.MaxTokens) * opts.Claude.ThinkingAdapterBudgetTokensPercentage)),
}
claudeRequest.TopP = nil
claudeRequest.Temperature = kitutil.GetPointer[float64](1.0)
}
if !opts.ShouldPreserveThinkingSuffix(textRequest.Model) {
claudeRequest.Model = trimmedModel
}
}
if textRequest.ReasoningEffort != "" {
switch textRequest.ReasoningEffort {
case "low":
claudeRequest.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: kitutil.GetPointer[int](1280),
}
case "medium":
claudeRequest.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: kitutil.GetPointer[int](2048),
}
case "high":
claudeRequest.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: kitutil.GetPointer[int](4096),
}
}
}
if textRequest.Reasoning != nil {
var reasoningConfig openRouterRequestReasoning
if err := kitutil.Unmarshal(textRequest.Reasoning, &reasoningConfig); err != nil {
return nil, err
}
budgetTokens := reasoningConfig.MaxTokens
if budgetTokens > 0 {
claudeRequest.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: &budgetTokens,
}
}
}
if textRequest.Stop != nil {
switch stop := textRequest.Stop.(type) {
case string:
claudeRequest.StopSequences = []string{stop}
case []interface{}:
stopSequences := make([]string, 0)
for _, item := range stop {
stopSequences = append(stopSequences, item.(string))
}
claudeRequest.StopSequences = stopSequences
}
}
formatMessages := make([]dto.Message, 0)
lastMessage := dto.Message{
Role: "tool",
}
for i, message := range textRequest.Messages {
if message.Role == "" {
textRequest.Messages[i].Role = "user"
}
fmtMessage := dto.Message{
Role: message.Role,
Content: message.Content,
}
if message.Role == "tool" {
fmtMessage.ToolCallId = message.ToolCallId
}
if message.Role == "assistant" && message.ToolCalls != nil {
fmtMessage.ToolCalls = message.ToolCalls
}
if lastMessage.Role == message.Role && lastMessage.Role != "tool" {
if lastMessage.IsStringContent() && message.IsStringContent() {
fmtMessage.SetStringContent(strings.Trim(fmt.Sprintf("%s %s", lastMessage.StringContent(), message.StringContent()), "\""))
formatMessages = formatMessages[:len(formatMessages)-1]
}
}
if fmtMessage.Content == nil || (fmtMessage.IsStringContent() && fmtMessage.StringContent() == "") {
fmtMessage.SetStringContent("...")
}
formatMessages = append(formatMessages, fmtMessage)
lastMessage = fmtMessage
}
claudeMessages := make([]dto.ClaudeMessage, 0)
isFirstMessage := true
var systemMessages []dto.ClaudeMediaMessage
for _, message := range formatMessages {
if message.Role == "system" {
if message.IsStringContent() {
if text := message.StringContent(); text != "" {
systemMessages = append(systemMessages, dto.ClaudeMediaMessage{
Type: "text",
Text: kitutil.GetPointer[string](text),
})
}
} else {
for _, ctx := range message.ParseContent() {
if ctx.Type == "text" && ctx.Text != "" {
systemMessages = append(systemMessages, dto.ClaudeMediaMessage{
Type: "text",
Text: kitutil.GetPointer[string](ctx.Text),
})
}
}
}
continue
}
if isFirstMessage {
isFirstMessage = false
if message.Role != "user" {
claudeMessage := dto.ClaudeMessage{
Role: "user",
Content: []dto.ClaudeMediaMessage{
{
Type: "text",
Text: kitutil.GetPointer[string]("..."),
},
},
}
claudeMessages = append(claudeMessages, claudeMessage)
}
}
claudeMessage := dto.ClaudeMessage{
Role: message.Role,
}
if message.Role == "tool" {
if len(claudeMessages) > 0 && claudeMessages[len(claudeMessages)-1].Role == "user" {
lastClaudeMessage := claudeMessages[len(claudeMessages)-1]
if content, ok := lastClaudeMessage.Content.(string); ok {
lastClaudeMessage.Content = []dto.ClaudeMediaMessage{
{
Type: "text",
Text: kitutil.GetPointer[string](content),
},
}
}
lastClaudeMessage.Content = append(lastClaudeMessage.Content.([]dto.ClaudeMediaMessage), dto.ClaudeMediaMessage{
Type: "tool_result",
ToolUseId: message.ToolCallId,
Content: message.Content,
})
claudeMessages[len(claudeMessages)-1] = lastClaudeMessage
continue
}
claudeMessage.Role = "user"
claudeMessage.Content = []dto.ClaudeMediaMessage{
{
Type: "tool_result",
ToolUseId: message.ToolCallId,
Content: message.Content,
},
}
} else if message.IsStringContent() && message.ToolCalls == nil {
text := message.StringContent()
if text == "" {
text = "..."
}
claudeMessage.Content = text
} else {
claudeMediaMessages := make([]dto.ClaudeMediaMessage, 0)
for _, mediaMessage := range message.ParseContent() {
switch mediaMessage.Type {
case "text":
if mediaMessage.Text != "" {
claudeMediaMessages = append(claudeMediaMessages, dto.ClaudeMediaMessage{
Type: "text",
Text: kitutil.GetPointer[string](mediaMessage.Text),
})
}
default:
source := mediaMessage.ToFileSource()
if source == nil {
continue
}
base64Data, mimeType, err := relaymedia.ResolveBase64Data(c, source, "formatting image for Claude")
if err != nil {
return nil, fmt.Errorf("get file data failed: %s", err.Error())
}
claudeMediaMessage := dto.ClaudeMediaMessage{
Source: &dto.ClaudeMessageSource{
Type: "base64",
},
}
if strings.HasPrefix(mimeType, "application/pdf") {
claudeMediaMessage.Type = "document"
} else {
claudeMediaMessage.Type = "image"
}
claudeMediaMessage.Source.MediaType = mimeType
claudeMediaMessage.Source.Data = base64Data
claudeMediaMessages = append(claudeMediaMessages, claudeMediaMessage)
continue
}
}
if message.ToolCalls != nil {
for _, toolCall := range message.ParseToolCalls() {
inputObj := make(map[string]any)
if args := toolCall.Function.Arguments; args != "" {
if err := kitutil.Unmarshal([]byte(args), &inputObj); err != nil {
kitutil.LogInfo("tool call function arguments is not a map[string]any: " + fmt.Sprintf("%v", toolCall.Function.Arguments))
}
}
claudeMediaMessages = append(claudeMediaMessages, dto.ClaudeMediaMessage{
Type: "tool_use",
Id: toolCall.ID,
Name: toolCall.Function.Name,
Input: inputObj,
})
}
}
claudeMessage.Content = claudeMediaMessages
}
claudeMessages = append(claudeMessages, claudeMessage)
}
if len(systemMessages) > 0 {
claudeRequest.System = systemMessages
}
claudeRequest.Prompt = ""
claudeRequest.Messages = claudeMessages
// Checked last so every injection path (default hook, thinking adapter
// floor) has had its chance to satisfy the required field.
if claudeRequest.MaxTokens == nil {
return nil, sharedclaude.ErrMissingMaxTokens
}
return &claudeRequest, nil
}