boos/BoosAPI
2
0
Fork 2
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: implement proxy billing details handling and resolveProxyLogBilling function

Browse Source
This commit is contained in:
cita
2026-03-06 18:28:45 +08:00
parent 3748e6dcb6
commit e88fc0a997
11 changed files with 380 additions and 217 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/server/routes/proxy/chat.ts
+10 -70
View File
@@ -5,7 +5,6 @@ import { fetch } from 'undici';
import { refreshModelsAndRebuildRoutes } from '../../services/modelService.js';
import { reportProxyAllFailed, reportTokenExpired } from '../../services/alertService.js';
import { isTokenExpiredError } from '../../services/alertRules.js';
import { buildProxyBillingDetails, estimateProxyCost } from '../../services/modelPricingService.js';
import { shouldRetryProxyRequest } from '../../services/proxyRetryPolicy.js';
import { resolveProxyUsageWithSelfLogFallback } from '../../services/proxyUsageFallbackService.js';
import { mergeProxyUsage, parseProxyUsage } from '../../services/proxyUsageParser.js';
@@ -38,6 +37,7 @@ import {
import { composeProxyLogMessage } from './logPathMeta.js';
import { executeEndpointFlow, withUpstreamPath } from './endpointFlow.js';
import { formatUtcSqlDateTime } from '../../services/localTimeService.js';
import { resolveProxyLogBilling } from './proxyBilling.js';
const MAX_RETRIES = 2;
const CLAUDE_SSE_EVENT_NAMES = new Set([
@@ -429,33 +429,13 @@ async function handleChatProxyRequest(
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
@@ -609,33 +589,13 @@ async function handleChatProxyRequest(
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
@@ -687,33 +647,13 @@ async function handleChatProxyRequest(
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
src/server/routes/proxy/completions.ts
+8 -46
View File
@@ -1,11 +1,10 @@
import { FastifyInstance, FastifyReply, FastifyRequest } from 'fastify';
import { fetch } from 'undici';
import { db, schema } from '../../db/index.js';
import { tokenRouter } from '../../services/tokenRouter.js';
import { tokenRouter } from '../../services/tokenRouter.js';
import { refreshModelsAndRebuildRoutes } from '../../services/modelService.js';
import { reportProxyAllFailed, reportTokenExpired } from '../../services/alertService.js';
import { isTokenExpiredError } from '../../services/alertRules.js';
import { buildProxyBillingDetails, estimateProxyCost } from '../../services/modelPricingService.js';
import { shouldRetryProxyRequest } from '../../services/proxyRetryPolicy.js';
import { resolveProxyUsageWithSelfLogFallback } from '../../services/proxyUsageFallbackService.js';
import { mergeProxyUsage, parseProxyUsage, pullSseDataEvents } from '../../services/proxyUsageParser.js';
@@ -13,6 +12,7 @@ import { ensureModelAllowedForDownstreamKey, getDownstreamRoutingPolicy, recordD
import { withExplicitProxyRequestInit } from '../../services/siteProxy.js';
import { composeProxyLogMessage } from './logPathMeta.js';
import { formatUtcSqlDateTime } from '../../services/localTimeService.js';
import { resolveProxyLogBilling } from './proxyBilling.js';
const MAX_RETRIES = 2;
@@ -160,32 +160,13 @@ export async function completionsProxyRoute(app: FastifyInstance) {
totalTokens: parsedUsage.totalTokens,
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
logProxy(
@@ -213,32 +194,13 @@ export async function completionsProxyRoute(app: FastifyInstance) {
totalTokens: parsedUsage.totalTokens,
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
src/server/routes/proxy/embeddings.ts
+5 -24
View File
@@ -1,11 +1,10 @@
import { FastifyInstance, FastifyReply, FastifyRequest } from 'fastify';
import { fetch } from 'undici';
import { db, schema } from '../../db/index.js';
import { tokenRouter } from '../../services/tokenRouter.js';
import { tokenRouter } from '../../services/tokenRouter.js';
import { refreshModelsAndRebuildRoutes } from '../../services/modelService.js';
import { reportProxyAllFailed, reportTokenExpired } from '../../services/alertService.js';
import { isTokenExpiredError } from '../../services/alertRules.js';
import { buildProxyBillingDetails, estimateProxyCost } from '../../services/modelPricingService.js';
import { shouldRetryProxyRequest } from '../../services/proxyRetryPolicy.js';
import { resolveProxyUsageWithSelfLogFallback } from '../../services/proxyUsageFallbackService.js';
import { parseProxyUsage } from '../../services/proxyUsageParser.js';
@@ -13,6 +12,7 @@ import { ensureModelAllowedForDownstreamKey, getDownstreamRoutingPolicy, recordD
import { withExplicitProxyRequestInit } from '../../services/siteProxy.js';
import { composeProxyLogMessage } from './logPathMeta.js';
import { formatUtcSqlDateTime } from '../../services/localTimeService.js';
import { resolveProxyLogBilling } from './proxyBilling.js';
const MAX_RETRIES = 2;
@@ -108,32 +108,13 @@ export async function embeddingsProxyRoute(app: FastifyInstance) {
totalTokens: parsedUsage.totalTokens,
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
src/server/routes/proxy/proxyBilling.test.ts
+66
View File
@@ -0,0 +1,66 @@
import { describe, expect, it } from 'vitest';
import { resolveProxyLogBilling } from './proxyBilling.js';
describe('resolveProxyLogBilling', () => {
it('uses self-log billing metadata for detail breakdown while preserving quota-derived total cost', async () => {
const result = await resolveProxyLogBilling({
site: {
id: 1,
url: 'https://anyrouter.top',
platform: 'anyrouter',
},
account: {
id: 2,
},
modelName: 'claude-haiku-4-5-20251001',
parsedUsage: {
promptTokens: 146638,
completionTokens: 172,
totalTokens: 146810,
cacheReadTokens: 0,
cacheCreationTokens: 0,
promptTokensIncludeCache: null,
},
resolvedUsage: {
promptTokens: 146638,
completionTokens: 172,
totalTokens: 146810,
recoveredFromSelfLog: true,
estimatedCostFromQuota: 0.083056,
selfLogBillingMeta: {
modelRatio: 2.5,
completionRatio: 5,
cacheRatio: 0.1,
cacheCreationRatio: 1.25,
groupRatio: 1,
cacheReadTokens: 145692,
cacheCreationTokens: 945,
promptTokensIncludeCache: true,
},
},
});
expect(result.estimatedCost).toBe(0.083056);
expect(result.billingDetails).toMatchObject({
usage: {
cacheReadTokens: 145692,
cacheCreationTokens: 945,
billablePromptTokens: 1,
},
pricing: {
modelRatio: 2.5,
completionRatio: 5,
cacheRatio: 0.1,
cacheCreationRatio: 1.25,
groupRatio: 1,
},
breakdown: {
inputPerMillion: 5,
outputPerMillion: 25,
cacheReadPerMillion: 0.5,
cacheCreationPerMillion: 6.25,
totalCost: 0.083057,
},
});
});
});
src/server/routes/proxy/proxyBilling.ts
+92
View File
@@ -0,0 +1,92 @@
import {
buildProxyBillingDetails,
estimateProxyCost,
type ProxyBillingDetails,
type ProxyBillingPricingOverride,
} from '../../services/modelPricingService.js';
import type { SelfLogBillingMeta } from '../../services/proxyUsageFallbackService.js';
interface ProxyBillingUsageSummary {
promptTokens: number;
completionTokens: number;
totalTokens: number;
cacheReadTokens: number;
cacheCreationTokens: number;
promptTokensIncludeCache: boolean | null;
}
interface ResolvedProxyUsageSummary {
promptTokens: number;
completionTokens: number;
totalTokens: number;
recoveredFromSelfLog: boolean;
estimatedCostFromQuota: number;
selfLogBillingMeta: SelfLogBillingMeta | null;
}
interface ResolveProxyLogBillingInput {
site: {
id: number;
url: string;
platform: string;
apiKey?: string | null;
};
account: {
id: number;
accessToken?: string | null;
apiToken?: string | null;
};
modelName: string;
parsedUsage: ProxyBillingUsageSummary;
resolvedUsage: ResolvedProxyUsageSummary;
}
function toPricingOverride(meta: SelfLogBillingMeta | null): ProxyBillingPricingOverride | null {
if (!meta) return null;
return {
modelRatio: meta.modelRatio,
completionRatio: meta.completionRatio,
cacheRatio: meta.cacheRatio,
cacheCreationRatio: meta.cacheCreationRatio,
groupRatio: meta.groupRatio,
};
}
export async function resolveProxyLogBilling(
input: ResolveProxyLogBillingInput,
): Promise<{ estimatedCost: number; billingDetails: ProxyBillingDetails | null }> {
const selfLogMeta = input.resolvedUsage.selfLogBillingMeta;
const billingPricingOverride = toPricingOverride(selfLogMeta);
const cacheReadTokens = selfLogMeta?.cacheReadTokens ?? input.parsedUsage.cacheReadTokens;
const cacheCreationTokens = selfLogMeta?.cacheCreationTokens ?? input.parsedUsage.cacheCreationTokens;
const promptTokensIncludeCache = selfLogMeta?.promptTokensIncludeCache
?? input.parsedUsage.promptTokensIncludeCache;
const billingInput = {
site: input.site,
account: input.account,
modelName: input.modelName,
promptTokens: input.resolvedUsage.promptTokens,
completionTokens: input.resolvedUsage.completionTokens,
totalTokens: input.resolvedUsage.totalTokens,
cacheReadTokens,
cacheCreationTokens,
promptTokensIncludeCache,
billingPricingOverride,
};
let estimatedCost = await estimateProxyCost(billingInput);
const billingDetails = await buildProxyBillingDetails(billingInput);
if (
input.resolvedUsage.estimatedCostFromQuota > 0
&& (input.resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)
) {
estimatedCost = input.resolvedUsage.estimatedCostFromQuota;
}
return {
estimatedCost,
billingDetails,
};
}
src/server/routes/proxy/responses.ts
+7 -45
View File
@@ -5,7 +5,6 @@ import { tokenRouter } from '../../services/tokenRouter.js';
import { refreshModelsAndRebuildRoutes } from '../../services/modelService.js';
import { reportProxyAllFailed, reportTokenExpired } from '../../services/alertService.js';
import { isTokenExpiredError } from '../../services/alertRules.js';
import { buildProxyBillingDetails, estimateProxyCost } from '../../services/modelPricingService.js';
import { shouldRetryProxyRequest } from '../../services/proxyRetryPolicy.js';
import { resolveProxyUsageWithSelfLogFallback } from '../../services/proxyUsageFallbackService.js';
import { mergeProxyUsage, parseProxyUsage } from '../../services/proxyUsageParser.js';
@@ -29,6 +28,7 @@ import { ensureModelAllowedForDownstreamKey, getDownstreamRoutingPolicy, recordD
import { composeProxyLogMessage } from './logPathMeta.js';
import { executeEndpointFlow, withUpstreamPath } from './endpointFlow.js';
import { formatUtcSqlDateTime } from '../../services/localTimeService.js';
import { resolveProxyLogBilling } from './proxyBilling.js';
const MAX_RETRIES = 2;
@@ -1475,32 +1475,13 @@ export async function responsesProxyRoute(app: FastifyInstance) {
totalTokens: parsedUsage.totalTokens,
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);
logProxy(
@@ -1541,32 +1522,13 @@ export async function responsesProxyRoute(app: FastifyInstance) {
totalTokens: parsedUsage.totalTokens,
},
});
let estimatedCost = await estimateProxyCost({
const { estimatedCost, billingDetails } = await resolveProxyLogBilling({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
parsedUsage,
resolvedUsage,
});
let billingDetails = await buildProxyBillingDetails({
site: selected.site,
account: selected.account,
modelName: selected.actualModel || requestedModel,
promptTokens: resolvedUsage.promptTokens,
completionTokens: resolvedUsage.completionTokens,
totalTokens: resolvedUsage.totalTokens,
cacheReadTokens: parsedUsage.cacheReadTokens,
cacheCreationTokens: parsedUsage.cacheCreationTokens,
promptTokensIncludeCache: parsedUsage.promptTokensIncludeCache,
});
if (resolvedUsage.estimatedCostFromQuota > 0 && (resolvedUsage.recoveredFromSelfLog || estimatedCost <= 0)) {
estimatedCost = resolvedUsage.estimatedCostFromQuota;
billingDetails = null;
}
tokenRouter.recordSuccess(selected.channel.id, latency, estimatedCost);
recordDownstreamCostUsage(request, estimatedCost);