@openhoo/hoopilot 1.2.0 → 2.0.0

package/dist/codexx.js

@@ -3,7 +3,7 @@ import {
   buildCodexxInvocation,
   main,
   verifyCodexxModel
-} from "./chunk-JU6F5L34.js";
+} from "./chunk-6ALEIJJM.js";
 export {
   buildCodexxInvocation,
   main,

package/dist/index.d.ts

@@ -51,6 +51,8 @@ declare class MetricsRegistry {
  * the shared upstream connection instead of draining it in the background.
  */
 declare function observeResponseUsage(response: Response, fallbackModel: string, onUsage: (model: string, usage: TokenUsage) => void, signal?: AbortSignal, onOutcome?: (extracted: boolean) => void): Response;
+/** Extract and record token usage from an already-buffered response body. */
+declare function recordResponseTextUsage(text: string, isSse: boolean, fallbackModel: string, onUsage: (model: string, usage: TokenUsage) => void, onOutcome?: (extracted: boolean) => void): void;
 
 type FetchLike = (input: string | URL | Request, init?: RequestInit) => Promise<Response>;
 interface Logger {
@@ -79,8 +81,8 @@ interface HoopilotLoggerOptions {
     base?: LogFields;
     colorize?: boolean;
     env?: NodeJS.ProcessEnv;
-    format?: LogFormat | string;
-    level?: LogLevel | string;
+    format?: LogFormat | (string & {});
+    level?: LogLevel | (string & {});
     stream?: {
         write(message: string): unknown;
     };
@@ -103,11 +105,11 @@ interface HoopilotServerOptions extends CopilotAuthOptions {
     apiKey?: string;
     host?: string;
     logger?: HoopilotLogger;
-    logFormat?: LogFormat | string;
-    logLevel?: LogLevel | string;
+    logFormat?: LogFormat | (string & {});
+    logLevel?: LogLevel | (string & {});
     metrics?: MetricsRegistry;
     port?: number;
-    streamingProxyMode?: StreamingProxyMode | string;
+    streamingProxyMode?: StreamingProxyMode | (string & {});
 }
 interface StartedHoopilotServer {
     server: Bun.Server<undefined>;
@@ -193,10 +195,28 @@ interface GithubRateLimitSnapshot {
     retryAfterSeconds?: number;
     used?: number;
 }
+/** Request-latency summary for one route, in milliseconds. */
+interface RouteLatency {
+    avgMs: number;
+    count: number;
+}
+/**
+ * Aggregate request-latency summary derived from the duration histogram. `avgMs`
+ * is exact; the percentiles are estimated from the histogram buckets (Prometheus-
+ * style linear interpolation), so they are approximate.
+ */
+interface LatencySnapshot {
+    avgMs: number;
+    byRoute: Record<string, RouteLatency>;
+    count: number;
+    p50Ms: number;
+    p95Ms: number;
+}
 /** A point-in-time JSON view of the in-process metrics. */
 interface MetricsSnapshot {
     githubRateLimit: Record<string, GithubRateLimitSnapshot>;
     inFlight: number;
+    latency: LatencySnapshot;
     requests: {
         byRoute: Record<string, number>;
         byStatus: Record<string, number>;
@@ -221,6 +241,14 @@ interface MetricsSnapshot {
     };
     uptimeSeconds: number;
 }
+/** JSON body returned by the proxy's `/v1/usage` route. */
+interface UsageResponseBody {
+    copilot: CopilotUsage | null;
+    copilot_error?: string;
+    object: "usage";
+    proxy: MetricsSnapshot;
+    version: string;
+}
 
 interface AnthropicStreamOptions {
     model: string;
@@ -234,6 +262,7 @@ declare function responsesResponseToAnthropicMessage(response: JsonObject, fallb
 declare function responsesStreamToAnthropicStream(stream: ReadableStream<Uint8Array>, options: AnthropicStreamOptions): ReadableStream<Uint8Array>;
 declare function estimateAnthropicMessageTokens(request: JsonObject): JsonObject;
 
+declare const DEFAULT_COPILOT_API_BASE_URL = "https://api.githubcopilot.com";
 declare class CopilotAuthError extends Error {
     constructor(message: string);
 }
@@ -335,6 +364,9 @@ interface ResponseStreamOptions {
     model: string;
     responseId?: string;
 }
+declare class OpenAICompatibilityError extends Error {
+    constructor(message: string);
+}
 declare function responsesRequestToChatCompletion(request: JsonObject): JsonObject;
 declare function normalizeChatCompletionRequest(request: JsonObject): JsonObject;
 declare function completionsRequestToChatCompletion(request: JsonObject): JsonObject;
@@ -367,4 +399,4 @@ declare function extractTokenUsage(usage: unknown): TokenUsage | undefined;
 declare function createHoopilotHandler(options?: HoopilotServerOptions): (request: Request) => Promise<Response>;
 declare function startHoopilotServer(options?: HoopilotServerOptions): StartedHoopilotServer;
 
-export { AnthropicCompatibilityError, COPILOT_USAGE_API_VERSION, type CopilotAccess, CopilotAuth, CopilotAuthError, type CopilotAuthOptions, CopilotClient, type CopilotQuota, type CopilotUsage, DEFAULT_GITHUB_API_BASE_URL, DEFAULT_LOG_FORMAT, DEFAULT_LOG_LEVEL, DEFAULT_MODEL, type FetchLike, type GithubRateLimit, type GithubRateLimitSnapshot, type HoopilotLogger, type HoopilotLoggerOptions, type HoopilotServerOptions, type JsonObject, type LogFields, type LogFormat, type LogLevel, type LogMethod, type Logger, MetricsRegistry, type MetricsSnapshot, type ModelTokenTotals, PROMETHEUS_CONTENT_TYPE, type RequestObservation, type StartedHoopilotServer, type TokenUsage, anthropicMessagesToResponsesRequest, applyCopilotHeaders, applyGithubApiHeaders, authStorePath, chatCompletionToCompletion, chatCompletionToResponse, completionStreamFromChatStream, completionsRequestToChatCompletion, createHoopilotHandler, createHoopilotLogger, estimateAnthropicMessageTokens, extractTokenUsage, fallbackModels, githubCopilotDeviceLogin, noopLogger, normalizeChatCompletionRequest, normalizeCopilotUsage, normalizeModelsResponse, normalizeRequestedModel, observeResponseUsage, parseLogFormat, parseLogLevel, parseRateLimitHeaders, readStoredCopilotAuth, responsesCompactionResult, responsesRequestToChatCompletion, responsesResponseToAnthropicMessage, responsesStreamFromChatStream, responsesStreamToAnthropicStream, startHoopilotServer, writeStoredCopilotAuth };
+export { AnthropicCompatibilityError, COPILOT_USAGE_API_VERSION, type CopilotAccess, CopilotAuth, CopilotAuthError, type CopilotAuthOptions, CopilotClient, type CopilotQuota, type CopilotUsage, DEFAULT_COPILOT_API_BASE_URL, DEFAULT_GITHUB_API_BASE_URL, DEFAULT_LOG_FORMAT, DEFAULT_LOG_LEVEL, DEFAULT_MODEL, type FetchLike, type GithubRateLimit, type GithubRateLimitSnapshot, type HoopilotLogger, type HoopilotLoggerOptions, type HoopilotServerOptions, type JsonObject, type LatencySnapshot, type LogFields, type LogFormat, type LogLevel, type LogMethod, type Logger, MetricsRegistry, type MetricsSnapshot, type ModelTokenTotals, OpenAICompatibilityError, PROMETHEUS_CONTENT_TYPE, type RequestObservation, type RouteLatency, type StartedHoopilotServer, type StreamingProxyMode, type TokenUsage, type UsageResponseBody, anthropicMessagesToResponsesRequest, applyCopilotHeaders, applyGithubApiHeaders, authStorePath, chatCompletionToCompletion, chatCompletionToResponse, completionStreamFromChatStream, completionsRequestToChatCompletion, createHoopilotHandler, createHoopilotLogger, estimateAnthropicMessageTokens, extractTokenUsage, fallbackModels, githubCopilotDeviceLogin, noopLogger, normalizeChatCompletionRequest, normalizeCopilotUsage, normalizeModelsResponse, normalizeRequestedModel, observeResponseUsage, parseLogFormat, parseLogLevel, parseRateLimitHeaders, readStoredCopilotAuth, recordResponseTextUsage, responsesCompactionResult, responsesRequestToChatCompletion, responsesResponseToAnthropicMessage, responsesStreamFromChatStream, responsesStreamToAnthropicStream, startHoopilotServer, writeStoredCopilotAuth };