Files
new-api/service/log_info_generate.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

320 lines
12 KiB
Go

package service
import (
"encoding/base64"
"fmt"
"strings"
"github.com/QuantumNous/new-api/common"
"github.com/QuantumNous/new-api/constant"
"github.com/QuantumNous/new-api/logger"
"github.com/QuantumNous/new-api/pkg/billingexpr"
relaycommon "github.com/QuantumNous/new-api/relay/common"
"github.com/QuantumNous/new-api/relaykit/dto"
"github.com/QuantumNous/new-api/relaykit/types"
"github.com/gin-gonic/gin"
)
// attachQuotaSaturationToOther nests a quota saturation marker under
// other.admin_info.quota_saturation. Nesting under admin_info makes it
// admin-only for free, since model.formatUserLogs strips the whole admin_info
// object for non-admin viewers. Creates admin_info if absent. No-op when the
// clamp is nil (the common case: no saturation happened).
func attachQuotaSaturationToOther(other map[string]interface{}, clamp *common.QuotaClamp) {
if clamp == nil || other == nil {
return
}
adminInfo, ok := other["admin_info"].(map[string]interface{})
if !ok || adminInfo == nil {
adminInfo = map[string]interface{}{}
other["admin_info"] = adminInfo
}
adminInfo["quota_saturation"] = clamp.AuditMap()
}
// attachQuotaSaturation records the request's quota clamp (if any) onto the
// consume log's other.admin_info and emits a request-correlated backend audit
// line. Called right before RecordConsumeLog on the text/audio/wss paths.
func attachQuotaSaturation(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil {
return
}
clamp := relayInfo.QuotaClamp
if clamp == nil {
return
}
attachQuotaSaturationToOther(other, clamp)
logger.LogWarn(ctx, fmt.Sprintf("quota saturation on consume log: op=%s kind=%s original=%g clamped=%d user=%d model=%s",
clamp.Op, clamp.Kind, clamp.Original, clamp.Clamped, relayInfo.UserId, relayInfo.OriginModelName))
}
func appendRequestPath(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if other == nil {
return
}
if ctx != nil && ctx.Request != nil && ctx.Request.URL != nil {
if path := ctx.Request.URL.Path; path != "" {
other["request_path"] = path
return
}
}
if relayInfo != nil && relayInfo.RequestURLPath != "" {
path := relayInfo.RequestURLPath
if idx := strings.Index(path, "?"); idx != -1 {
path = path[:idx]
}
other["request_path"] = path
}
}
func GenerateTextOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelRatio, groupRatio, completionRatio float64,
cacheTokens int, cacheRatio float64, modelPrice float64, userGroupRatio float64) map[string]interface{} {
other := make(map[string]interface{})
other["model_ratio"] = modelRatio
other["group_ratio"] = groupRatio
other["completion_ratio"] = completionRatio
other["cache_tokens"] = cacheTokens
other["cache_ratio"] = cacheRatio
other["model_price"] = modelPrice
other["user_group_ratio"] = userGroupRatio
other["frt"] = float64(relayInfo.FirstResponseTime.UnixMilli() - relayInfo.StartTime.UnixMilli())
if relayInfo.ReasoningEffort != "" {
other["reasoning_effort"] = relayInfo.ReasoningEffort
}
if relayInfo.IsModelMapped {
other["is_model_mapped"] = true
other["upstream_model_name"] = relayInfo.UpstreamModelName
}
isSystemPromptOverwritten := common.GetContextKeyBool(ctx, constant.ContextKeySystemPromptOverride)
if isSystemPromptOverwritten {
other["is_system_prompt_overwritten"] = true
}
adminInfo := make(map[string]interface{})
adminInfo["use_channel"] = ctx.GetStringSlice("use_channel")
isMultiKey := common.GetContextKeyBool(ctx, constant.ContextKeyChannelIsMultiKey)
if isMultiKey {
adminInfo["is_multi_key"] = true
adminInfo["multi_key_index"] = common.GetContextKeyInt(ctx, constant.ContextKeyChannelMultiKeyIndex)
}
isLocalCountTokens := common.GetContextKeyBool(ctx, constant.ContextKeyLocalCountTokens)
if isLocalCountTokens {
adminInfo["local_count_tokens"] = isLocalCountTokens
}
AppendChannelAffinityAdminInfo(ctx, adminInfo)
other["admin_info"] = adminInfo
appendRequestPath(ctx, relayInfo, other)
appendRequestConversionChain(relayInfo, other)
appendFinalRequestFormat(relayInfo, other)
appendBillingInfo(relayInfo, other)
appendParamOverrideInfo(relayInfo, other)
appendStreamStatus(relayInfo, other)
return other
}
func appendParamOverrideInfo(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil || other == nil || len(relayInfo.ParamOverrideAudit) == 0 {
return
}
other["po"] = relayInfo.ParamOverrideAudit
}
func appendStreamStatus(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil || other == nil || !relayInfo.IsStream || relayInfo.StreamStatus == nil {
return
}
ss := relayInfo.StreamStatus
status := "ok"
if !ss.IsNormalEnd() || ss.HasErrors() {
status = "error"
}
streamInfo := map[string]interface{}{
"status": status,
"end_reason": string(ss.EndReason),
}
if ss.EndError != nil {
streamInfo["end_error"] = ss.EndError.Error()
}
if ss.ErrorCount > 0 {
streamInfo["error_count"] = ss.ErrorCount
messages := make([]string, 0, len(ss.Errors))
for _, e := range ss.Errors {
messages = append(messages, e.Message)
}
streamInfo["errors"] = messages
}
other["stream_status"] = streamInfo
}
func appendBillingInfo(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil || other == nil {
return
}
// billing_source: "wallet" or "subscription"
if relayInfo.BillingSource != "" {
other["billing_source"] = relayInfo.BillingSource
}
if relayInfo.UserSetting.BillingPreference != "" {
other["billing_preference"] = relayInfo.UserSetting.BillingPreference
}
if relayInfo.BillingSource == "subscription" {
if relayInfo.SubscriptionId != 0 {
other["subscription_id"] = relayInfo.SubscriptionId
}
if relayInfo.SubscriptionPreConsumed > 0 {
other["subscription_pre_consumed"] = relayInfo.SubscriptionPreConsumed
}
// post_delta: settlement delta applied after actual usage is known (can be negative for refund)
if relayInfo.SubscriptionPostDelta != 0 {
other["subscription_post_delta"] = relayInfo.SubscriptionPostDelta
}
if relayInfo.SubscriptionPlanId != 0 {
other["subscription_plan_id"] = relayInfo.SubscriptionPlanId
}
if relayInfo.SubscriptionPlanTitle != "" {
other["subscription_plan_title"] = relayInfo.SubscriptionPlanTitle
}
// Compute "this request" subscription consumed + remaining
consumed := relayInfo.SubscriptionPreConsumed + relayInfo.SubscriptionPostDelta
usedFinal := relayInfo.SubscriptionAmountUsedAfterPreConsume + relayInfo.SubscriptionPostDelta
if consumed < 0 {
consumed = 0
}
if usedFinal < 0 {
usedFinal = 0
}
if relayInfo.SubscriptionAmountTotal > 0 {
remain := relayInfo.SubscriptionAmountTotal - usedFinal
if remain < 0 {
remain = 0
}
other["subscription_total"] = relayInfo.SubscriptionAmountTotal
other["subscription_used"] = usedFinal
other["subscription_remain"] = remain
}
if consumed > 0 {
other["subscription_consumed"] = consumed
}
// Wallet quota is not deducted when billed from subscription.
other["wallet_quota_deducted"] = 0
}
}
func appendRequestConversionChain(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil || other == nil {
return
}
if len(relayInfo.RequestConversionChain) == 0 {
return
}
chain := make([]string, 0, len(relayInfo.RequestConversionChain))
for _, f := range relayInfo.RequestConversionChain {
switch f {
case types.RelayFormatOpenAI:
chain = append(chain, "OpenAI Compatible")
case types.RelayFormatClaude:
chain = append(chain, "Claude Messages")
case types.RelayFormatGemini:
chain = append(chain, "Google Gemini")
case types.RelayFormatOpenAIResponses:
chain = append(chain, "OpenAI Responses")
default:
chain = append(chain, string(f))
}
}
if len(chain) == 0 {
return
}
other["request_conversion"] = chain
}
func appendFinalRequestFormat(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
if relayInfo == nil || other == nil {
return
}
if relayInfo.GetFinalRequestRelayFormat() == types.RelayFormatClaude {
// claude indicates the final upstream request format is Claude Messages.
// Frontend log rendering uses this to keep the original Claude input display.
other["claude"] = true
}
}
func GenerateWssOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice, userGroupRatio float64) map[string]interface{} {
info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, 0, 0.0, modelPrice, userGroupRatio)
info["ws"] = true
info["audio_input"] = usage.InputTokenDetails.AudioTokens
info["audio_output"] = usage.OutputTokenDetails.AudioTokens
info["text_input"] = usage.InputTokenDetails.TextTokens
info["text_output"] = usage.OutputTokenDetails.TextTokens
info["audio_ratio"] = audioRatio
info["audio_completion_ratio"] = audioCompletionRatio
return info
}
func GenerateAudioOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice, userGroupRatio float64) map[string]interface{} {
info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, 0, 0.0, modelPrice, userGroupRatio)
info["audio"] = true
info["audio_input"] = usage.PromptTokensDetails.AudioTokens
info["audio_output"] = usage.CompletionTokenDetails.AudioTokens
info["text_input"] = usage.PromptTokensDetails.TextTokens
info["text_output"] = usage.CompletionTokenDetails.TextTokens
info["audio_ratio"] = audioRatio
info["audio_completion_ratio"] = audioCompletionRatio
return info
}
func GenerateClaudeOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, modelRatio, groupRatio, completionRatio float64,
cacheTokens int, cacheRatio float64,
cacheCreationTokens int, cacheCreationRatio float64,
cacheCreationTokens5m int, cacheCreationRatio5m float64,
cacheCreationTokens1h int, cacheCreationRatio1h float64,
modelPrice float64, userGroupRatio float64) map[string]interface{} {
info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, cacheTokens, cacheRatio, modelPrice, userGroupRatio)
info["claude"] = true
info["cache_creation_tokens"] = cacheCreationTokens
info["cache_creation_ratio"] = cacheCreationRatio
if cacheCreationTokens5m != 0 {
info["cache_creation_tokens_5m"] = cacheCreationTokens5m
info["cache_creation_ratio_5m"] = cacheCreationRatio5m
}
if cacheCreationTokens1h != 0 {
info["cache_creation_tokens_1h"] = cacheCreationTokens1h
info["cache_creation_ratio_1h"] = cacheCreationRatio1h
}
return info
}
func GenerateMjOtherInfo(relayInfo *relaycommon.RelayInfo, priceData types.PriceData) map[string]interface{} {
other := make(map[string]interface{})
other["model_price"] = priceData.ModelPrice
other["group_ratio"] = priceData.GroupRatioInfo.GroupRatio
if priceData.GroupRatioInfo.HasSpecialRatio {
other["user_group_ratio"] = priceData.GroupRatioInfo.GroupSpecialRatio
}
appendRequestPath(nil, relayInfo, other)
return other
}
// InjectTieredBillingInfo overlays tiered billing fields onto an existing
// module-specific other map. Call this after GenerateTextOtherInfo /
// GenerateClaudeOtherInfo / etc. when the request used tiered_expr billing.
func InjectTieredBillingInfo(other map[string]interface{}, relayInfo *relaycommon.RelayInfo, result *billingexpr.TieredResult) {
if relayInfo == nil || other == nil {
return
}
snap := relayInfo.TieredBillingSnapshot
if snap == nil {
return
}
other["billing_mode"] = "tiered_expr"
other["expr_b64"] = base64.StdEncoding.EncodeToString([]byte(snap.ExprString))
if result != nil {
other["matched_tier"] = result.MatchedTier
}
}