Files
new-api/relay/helper/price.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

337 lines
12 KiB
Go

package helper
import (
"fmt"
"strings"
"github.com/QuantumNous/new-api/common"
"github.com/QuantumNous/new-api/logger"
"github.com/QuantumNous/new-api/model"
"github.com/QuantumNous/new-api/pkg/billingexpr"
relaycommon "github.com/QuantumNous/new-api/relay/common"
"github.com/QuantumNous/new-api/relaykit/types"
"github.com/QuantumNous/new-api/setting/billing_setting"
"github.com/QuantumNous/new-api/setting/operation_setting"
"github.com/QuantumNous/new-api/setting/ratio_setting"
"github.com/gin-gonic/gin"
)
func modelPriceNotConfiguredError(modelName string, userId int) error {
if model.IsAdmin(userId) {
return fmt.Errorf(
"模型 %s 的价格未配置。请前往「系统设置 → 运营设置」开启自用模式,或在「系统设置 → 分组与模型定价设置」中为该模型配置价格;"+
"Model %s price not configured. Go to System Settings → Operation Settings to enable self-use mode, or configure the model price in System Settings → Group & Model Pricing.",
modelName, modelName,
)
}
return fmt.Errorf(
"模型 %s 的价格尚未由管理员配置,暂时无法使用,请联系站点管理员开启该模型;"+
"Model %s has not been priced by the administrator yet. Please contact the site administrator to enable this model.",
modelName, modelName,
)
}
// https://docs.claude.com/en/docs/build-with-claude/prompt-caching#1-hour-cache-duration
const claudeCacheCreation1hMultiplier = 6 / 3.75
// defaultTieredPreConsumeMaxTokens is the fallback completion-token estimate
// used for tiered expression pre-consume when the client omits max_tokens, so
// the pre-consumed quota still reflects a plausible output cost in paid groups.
const defaultTieredPreConsumeMaxTokens = 8192
// HandleGroupRatio checks for "auto_group" in the context and updates the group ratio and relayInfo.UsingGroup if present
func HandleGroupRatio(ctx *gin.Context, relayInfo *relaycommon.RelayInfo) types.GroupRatioInfo {
groupRatioInfo := types.GroupRatioInfo{
GroupRatio: 1.0, // default ratio
GroupSpecialRatio: -1,
}
// check auto group
autoGroup, exists := ctx.Get("auto_group")
if exists {
logger.LogDebug(ctx, "final group: %s", autoGroup)
relayInfo.UsingGroup = autoGroup.(string)
}
// check user group special ratio
userGroupRatio, ok := ratio_setting.GetGroupGroupRatio(relayInfo.UserGroup, relayInfo.UsingGroup)
if ok {
// user group special ratio
groupRatioInfo.GroupSpecialRatio = userGroupRatio
groupRatioInfo.GroupRatio = userGroupRatio
groupRatioInfo.HasSpecialRatio = true
} else {
// normal group ratio
groupRatioInfo.GroupRatio = ratio_setting.GetGroupRatio(relayInfo.UsingGroup)
}
return groupRatioInfo
}
func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens int, meta *types.TokenCountMeta) (types.PriceData, error) {
modelPrice, usePrice := ratio_setting.GetModelPrice(info.OriginModelName, false)
groupRatioInfo := HandleGroupRatio(c, info)
// Check if this model uses tiered_expr billing
if billing_setting.GetBillingMode(info.OriginModelName) == billing_setting.BillingModeTieredExpr {
return modelPriceHelperTiered(c, info, promptTokens, meta, groupRatioInfo)
}
var preConsumedQuota int
var modelRatio float64
var completionRatio float64
var cacheRatio float64
var imageRatio float64
var cacheCreationRatio float64
var cacheCreationRatio5m float64
var cacheCreationRatio1h float64
var audioRatio float64
var audioCompletionRatio float64
var freeModel bool
if !usePrice {
preConsumedTokens := common.Max(promptTokens, common.PreConsumedQuota)
if meta.MaxTokens != 0 {
preConsumedTokens += meta.MaxTokens
}
var success bool
var matchName string
modelRatio, success, matchName = ratio_setting.GetModelRatio(info.OriginModelName)
if !success {
acceptUnsetRatio := false
if info.UserSetting.AcceptUnsetRatioModel {
acceptUnsetRatio = true
}
if !acceptUnsetRatio {
return types.PriceData{}, modelPriceNotConfiguredError(matchName, info.UserId)
}
}
completionRatio = ratio_setting.GetCompletionRatio(info.OriginModelName)
cacheRatio, _ = ratio_setting.GetCacheRatio(info.OriginModelName)
cacheCreationRatio, _ = ratio_setting.GetCreateCacheRatio(info.OriginModelName)
cacheCreationRatio5m = cacheCreationRatio
// 固定1h和5min缓存写入价格的比例
cacheCreationRatio1h = cacheCreationRatio * claudeCacheCreation1hMultiplier
imageRatio, _ = ratio_setting.GetImageRatio(info.OriginModelName)
audioRatio = ratio_setting.GetAudioRatio(info.OriginModelName)
audioCompletionRatio = ratio_setting.GetAudioCompletionRatio(info.OriginModelName)
ratio := modelRatio * groupRatioInfo.GroupRatio
quota, err := common.QuotaFromFloatStrict(float64(preConsumedTokens) * ratio)
if err != nil {
return types.PriceData{}, err
}
preConsumedQuota = quota
} else {
if meta.ImagePriceRatio != 0 {
modelPrice = modelPrice * meta.ImagePriceRatio
}
}
// check if free model pre-consume is disabled
if !operation_setting.GetQuotaSetting().EnableFreeModelPreConsume {
// if model price or ratio is 0, do not pre-consume quota
if groupRatioInfo.GroupRatio == 0 {
preConsumedQuota = 0
freeModel = true
} else if usePrice {
if modelPrice == 0 {
preConsumedQuota = 0
freeModel = true
}
} else {
if modelRatio == 0 {
preConsumedQuota = 0
freeModel = true
}
}
}
priceData := types.PriceData{
FreeModel: freeModel,
ModelPrice: modelPrice,
ModelRatio: modelRatio,
CompletionRatio: completionRatio,
GroupRatioInfo: groupRatioInfo,
UsePrice: usePrice,
CacheRatio: cacheRatio,
ImageRatio: imageRatio,
AudioRatio: audioRatio,
AudioCompletionRatio: audioCompletionRatio,
CacheCreationRatio: cacheCreationRatio,
CacheCreation5mRatio: cacheCreationRatio5m,
CacheCreation1hRatio: cacheCreationRatio1h,
QuotaToPreConsume: preConsumedQuota,
}
if usePrice {
for name, ratio := range meta.BillingRatios {
priceData.AddOtherRatio(name, ratio)
}
quotaToPreConsume := priceData.ApplyOtherRatiosToFloat(modelPrice * common.QuotaPerUnit * groupRatioInfo.GroupRatio)
quota, err := common.QuotaFromFloatStrict(quotaToPreConsume)
if err != nil {
return types.PriceData{}, err
}
priceData.QuotaToPreConsume = quota
}
if common.DebugEnabled {
logger.LogDebug(c, "model_price_helper result: %s", priceData.ToSetting())
}
info.PriceData = priceData
return priceData, nil
}
// ModelPriceHelperPerCall 按次/按量计费的 PriceHelper (MJ、Task)
func ModelPriceHelperPerCall(c *gin.Context, info *relaycommon.RelayInfo) (types.PriceData, error) {
groupRatioInfo := HandleGroupRatio(c, info)
modelPrice, success := ratio_setting.GetModelPrice(info.OriginModelName, true)
usePrice := success
var modelRatio float64
if !success {
defaultPrice, ok := ratio_setting.GetDefaultModelPriceMap()[info.OriginModelName]
if ok {
modelPrice = defaultPrice
usePrice = true
} else {
var ratioSuccess bool
var matchName string
modelRatio, ratioSuccess, matchName = ratio_setting.GetModelRatio(info.OriginModelName)
acceptUnsetRatio := false
if info.UserSetting.AcceptUnsetRatioModel {
acceptUnsetRatio = true
}
if !ratioSuccess && !acceptUnsetRatio {
return types.PriceData{}, modelPriceNotConfiguredError(matchName, info.UserId)
}
}
}
var quota int
freeModel := false
if usePrice {
var err error
quota, err = common.QuotaFromFloatStrict(modelPrice * common.QuotaPerUnit * groupRatioInfo.GroupRatio)
if err != nil {
return types.PriceData{}, err
}
if !operation_setting.GetQuotaSetting().EnableFreeModelPreConsume {
if groupRatioInfo.GroupRatio == 0 || modelPrice == 0 {
quota = 0
freeModel = true
}
}
} else {
// 按量计费:以模型倍率的一半作为预扣额度
var err error
quota, err = common.QuotaFromFloatStrict(modelRatio / 2 * common.QuotaPerUnit * groupRatioInfo.GroupRatio)
if err != nil {
return types.PriceData{}, err
}
modelPrice = -1
if !operation_setting.GetQuotaSetting().EnableFreeModelPreConsume {
if groupRatioInfo.GroupRatio == 0 || modelRatio == 0 {
quota = 0
freeModel = true
}
}
}
priceData := types.PriceData{
FreeModel: freeModel,
ModelPrice: modelPrice,
ModelRatio: modelRatio,
UsePrice: usePrice,
Quota: quota,
GroupRatioInfo: groupRatioInfo,
}
return priceData, nil
}
func HasModelBillingConfig(modelName string) bool {
if _, ok := ratio_setting.GetModelPrice(modelName, false); ok {
return true
}
if _, ok, _ := ratio_setting.GetModelRatio(modelName); ok {
return true
}
if billing_setting.GetBillingMode(modelName) != billing_setting.BillingModeTieredExpr {
return false
}
expr, ok := billing_setting.GetBillingExpr(modelName)
return ok && strings.TrimSpace(expr) != ""
}
func modelPriceHelperTiered(c *gin.Context, info *relaycommon.RelayInfo, promptTokens int, meta *types.TokenCountMeta, groupRatioInfo types.GroupRatioInfo) (types.PriceData, error) {
exprStr, ok := billing_setting.GetBillingExpr(info.OriginModelName)
if !ok {
return types.PriceData{}, fmt.Errorf("model %s is configured as tiered_expr but has no billing expression", info.OriginModelName)
}
estimatedCompletionTokens := meta.MaxTokens
if estimatedCompletionTokens == 0 && groupRatioInfo.GroupRatio != 0 {
estimatedCompletionTokens = defaultTieredPreConsumeMaxTokens
}
requestInput, err := ResolveIncomingBillingExprRequestInput(c, info)
if err != nil {
return types.PriceData{}, err
}
rawCost, trace, err := billingexpr.RunExprWithRequest(exprStr, billingexpr.TokenParams{
P: float64(promptTokens),
C: float64(estimatedCompletionTokens),
Len: float64(promptTokens),
}, requestInput)
if err != nil {
return types.PriceData{}, fmt.Errorf("model %s tiered expr run failed: %w", info.OriginModelName, err)
}
// Expression coefficients are $/1M tokens prices; convert to quota the same way per-call billing does.
quotaBeforeGroup := rawCost / 1_000_000 * common.QuotaPerUnit
preConsumedQuota, err := billingexpr.QuotaRoundStrict(quotaBeforeGroup * groupRatioInfo.GroupRatio)
if err != nil {
return types.PriceData{}, err
}
freeModel := false
if !operation_setting.GetQuotaSetting().EnableFreeModelPreConsume {
if groupRatioInfo.GroupRatio == 0 {
preConsumedQuota = 0
freeModel = true
}
}
exprHash := billingexpr.ExprHashString(exprStr)
snapshot := &billingexpr.BillingSnapshot{
BillingMode: billing_setting.BillingModeTieredExpr,
ModelName: info.OriginModelName,
ExprString: exprStr,
ExprHash: exprHash,
GroupRatio: groupRatioInfo.GroupRatio,
EstimatedPromptTokens: promptTokens,
EstimatedCompletionTokens: estimatedCompletionTokens,
EstimatedQuotaBeforeGroup: quotaBeforeGroup,
EstimatedQuotaAfterGroup: preConsumedQuota,
EstimatedTier: trace.MatchedTier,
QuotaPerUnit: common.QuotaPerUnit,
ExprVersion: billingexpr.ExprVersion(exprStr),
}
info.TieredBillingSnapshot = snapshot
info.BillingRequestInput = &requestInput
priceData := types.PriceData{
FreeModel: freeModel,
GroupRatioInfo: groupRatioInfo,
QuotaToPreConsume: preConsumedQuota,
}
logger.LogDebug(c, "model_price_helper_tiered result: model=%s preConsume=%d quotaBeforeGroup=%.2f groupRatio=%.2f tier=%s", info.OriginModelName, preConsumedQuota, quotaBeforeGroup, groupRatioInfo.GroupRatio, trace.MatchedTier)
info.PriceData = priceData
return priceData, nil
}