@copilotkit/aimock 1.24.0 → 1.25.0

package/dist/server.js

@@ -1,4 +1,4 @@
-import { buildContentWithToolCallsChunks, buildContentWithToolCallsCompletion, buildTextChunks, buildTextCompletion, buildToolCallChunks, buildToolCallCompletion, extractOverrides, flattenHeaders, getTestId, isAudioResponse, isContentWithToolCallsResponse, isErrorResponse, isTextResponse, isToolCallResponse, readBody, resolveResponse, resolveStrictMode, serializeErrorResponse, strictOverrideField } from "./helpers.js";
+import { buildContentWithToolCallsChunks, buildContentWithToolCallsCompletion, buildTextChunks, buildTextCompletion, buildToolCallChunks, buildToolCallCompletion, buildUsageChunk, estimatePromptTokens, estimateTokens, extractOverrides, flattenHeaders, getTestId, isAudioResponse, isContentWithToolCallsResponse, isErrorResponse, isTextResponse, isToolCallResponse, readBody, resolveResponse, resolveStrictMode, serializeErrorResponse, strictOverrideField } from "./helpers.js";
 import { Logger } from "./logger.js";
 import { Journal } from "./journal.js";
 import { matchFixture } from "./router.js";
@@ -10,19 +10,20 @@ import { proxyAndRecord } from "./recorder.js";
 import { handleResponses } from "./responses.js";
 import { handleMessages } from "./messages.js";
 import { handleGemini } from "./gemini.js";
+import { handleGeminiEmbedContent } from "./gemini-embeddings.js";
 import { handleBedrock, handleBedrockStream } from "./bedrock.js";
 import { handleConverse, handleConverseStream } from "./bedrock-converse.js";
 import { handleGeminiInteractions, resetEventIdCounter, resetInteractionCounter } from "./gemini-interactions.js";
 import { handleEmbeddings } from "./embeddings.js";
-import { handleImages } from "./images.js";
-import { handleSpeech } from "./speech.js";
 import { handleTranscription } from "./transcription.js";
+import { handleImageEdit, handleImageVariations, handleImages } from "./images.js";
+import { handleSpeech } from "./speech.js";
 import { VideoStateMap, handleVideoCreate, handleVideoStatus } from "./video.js";
-import { handleElevenLabsAudio } from "./elevenlabs-audio.js";
+import { handleElevenLabsAudio, handleElevenLabsTTS } from "./elevenlabs-audio.js";
 import { falQueueStates, handleFal } from "./fal.js";
 import { falJobs, handleFalQueue } from "./fal-audio.js";
-import { handleOllama, handleOllamaGenerate } from "./ollama.js";
-import { handleCohere } from "./cohere.js";
+import { handleOllama, handleOllamaEmbeddings, handleOllamaGenerate } from "./ollama.js";
+import { handleCohere, handleCohereEmbed } from "./cohere.js";
 import { handleSearch } from "./search.js";
 import { handleRerank } from "./rerank.js";
 import { handleModeration } from "./moderation.js";
@@ -41,16 +42,21 @@ const GEMINI_LIVE_PATH = "/ws/google.ai.generativelanguage.v1beta.GenerativeServ
 const MESSAGES_PATH = "/v1/messages";
 const EMBEDDINGS_PATH = "/v1/embeddings";
 const COHERE_CHAT_PATH = "/v2/chat";
+const COHERE_EMBED_PATH = "/v2/embed";
 const SEARCH_PATH = "/search";
 const RERANK_PATH = "/v2/rerank";
 const MODERATIONS_PATH = "/v1/moderations";
 const IMAGES_PATH = "/v1/images/generations";
+const IMAGES_EDIT_PATH = "/v1/images/edit";
+const IMAGES_VARIATIONS_PATH = "/v1/images/variations";
 const SPEECH_PATH = "/v1/audio/speech";
 const TRANSCRIPTIONS_PATH = "/v1/audio/transcriptions";
+const TRANSLATIONS_PATH = "/v1/audio/translations";
 const VIDEOS_PATH = "/v1/videos";
 const VIDEOS_STATUS_RE = /^\/v1\/videos\/([^/]+)$/;
 const GEMINI_PREDICT_RE = /^\/v1beta\/models\/([^:]+):predict$/;
 const ELEVENLABS_SOUND_GENERATION_PATH = "/v1/sound-generation";
+const ELEVENLABS_TTS_RE = /^\/v1\/text-to-speech\/([^/]+)$/;
 const ELEVENLABS_MUSIC_RE = /^\/v1\/music(?:\/(.+))?$/;
 const FAL_QUEUE_SUBMIT_RE = /^\/fal\/queue\/submit\/(.+)$/;
 const FAL_QUEUE_REQUESTS_RE = /^\/fal\/queue\/requests\/(.+)$/;
@@ -63,7 +69,10 @@ const COMPAT_SUFFIXES = [
 	"/responses",
 	"/audio/speech",
 	"/audio/transcriptions",
-	"/images/generations"
+	"/audio/translations",
+	"/images/generations",
+	"/images/edit",
+	"/images/variations"
 ];
 /**
 * Normalize OpenAI-compatible paths with arbitrary prefixes.
@@ -83,6 +92,7 @@ function normalizeCompatPath(pathname, logger) {
 }
 const GEMINI_INTERACTIONS_PATH = "/v1beta/interactions";
 const GEMINI_PATH_RE = /^\/v1beta\/models\/([^:]+):(generateContent|streamGenerateContent)$/;
+const GEMINI_EMBED_RE = /^\/v1beta\/models\/([^:]+):embedContent$/;
 const AZURE_DEPLOYMENT_RE = /^\/openai\/deployments\/([^/]+)\/(chat\/completions|embeddings)$/;
 const BEDROCK_INVOKE_RE = /^\/model\/([^/]+)\/invoke$/;
 const BEDROCK_STREAM_RE = /^\/model\/([^/]+)\/invoke-with-response-stream$/;
@@ -91,6 +101,7 @@ const BEDROCK_CONVERSE_STREAM_RE = /^\/model\/([^/]+)\/converse-stream$/;
 const VERTEX_AI_RE = /^\/v1\/projects\/[^/]+\/locations\/[^/]+\/publishers\/google\/models\/([^/:]+):(generateContent|streamGenerateContent)$/;
 const OLLAMA_CHAT_PATH = "/api/chat";
 const OLLAMA_GENERATE_PATH = "/api/generate";
+const OLLAMA_EMBEDDINGS_PATH = "/api/embeddings";
 const OLLAMA_TAGS_PATH = "/api/tags";
 const HEALTH_PATH = "/health";
 const READY_PATH = "/ready";
@@ -433,6 +444,7 @@ async function handleCompletions(req, res, fixtures, journal, defaults, modelFal
 	const response = await resolveResponse(fixture, body);
 	const latency = fixture.latency ?? defaults.latency;
 	const chunkSize = Math.max(1, fixture.chunkSize ?? defaults.chunkSize);
+	const includeUsage = body.stream === true && body.stream_options?.include_usage === true;
 	if (isErrorResponse(response)) {
 		const status = response.status ?? 500;
 		journal.add({
@@ -445,7 +457,7 @@ async function handleCompletions(req, res, fixtures, journal, defaults, modelFal
 				fixture
 			}
 		});
-		writeErrorResponse(res, status, serializeErrorResponse(response));
+		writeErrorResponse(res, status, serializeErrorResponse(response), { retryAfter: response.retryAfter });
 		return;
 	}
 	if (isAudioResponse(response)) {
@@ -479,17 +491,28 @@ async function handleCompletions(req, res, fixtures, journal, defaults, modelFal
 			}
 		});
 		if (body.stream !== true) {
-			const completion = buildContentWithToolCallsCompletion(response.content, response.toolCalls, body.model, response.reasoning, overrides);
+			const completion = buildContentWithToolCallsCompletion(response.content, response.toolCalls, body.model, response.reasoning, overrides, body.messages);
 			res.writeHead(200, { "Content-Type": "application/json" });
 			res.end(JSON.stringify(completion));
 		} else {
 			const chunks = buildContentWithToolCallsChunks(response.content, response.toolCalls, body.model, chunkSize, response.reasoning, overrides);
+			const completionText = response.content + response.toolCalls.map((tc) => tc.name + tc.arguments).join("");
+			const usageChunk = includeUsage ? buildUsageChunk(chunks[0]?.id ?? "chatcmpl-unknown", overrides?.model ?? body.model, chunks[0]?.created ?? Math.floor(Date.now() / 1e3), overrides?.usage ? {
+				prompt_tokens: overrides.usage.prompt_tokens ?? 0,
+				completion_tokens: overrides.usage.completion_tokens ?? 0,
+				total_tokens: overrides.usage.total_tokens ?? (overrides.usage.prompt_tokens ?? 0) + (overrides.usage.completion_tokens ?? 0)
+			} : {
+				prompt_tokens: estimatePromptTokens(body.messages),
+				completion_tokens: estimateTokens(completionText),
+				total_tokens: estimatePromptTokens(body.messages) + estimateTokens(completionText)
+			}, overrides?.systemFingerprint) : void 0;
 			const interruption = createInterruptionSignal(fixture);
 			if (!await writeSSEStream(res, chunks, {
 				latency,
 				streamingProfile: fixture.streamingProfile,
 				signal: interruption?.signal,
-				onChunkSent: interruption?.tick
+				onChunkSent: interruption?.tick,
+				usageChunk
 			})) {
 				if (!res.writableEnded) res.destroy();
 				journalEntry.response.interrupted = true;
@@ -513,17 +536,27 @@ async function handleCompletions(req, res, fixtures, journal, defaults, modelFal
 			}
 		});
 		if (body.stream !== true) {
-			const completion = buildTextCompletion(response.content, body.model, response.reasoning, overrides);
+			const completion = buildTextCompletion(response.content, body.model, response.reasoning, overrides, body.messages);
 			res.writeHead(200, { "Content-Type": "application/json" });
 			res.end(JSON.stringify(completion));
 		} else {
 			const chunks = buildTextChunks(response.content, body.model, chunkSize, response.reasoning, overrides);
+			const usageChunk = includeUsage ? buildUsageChunk(chunks[0]?.id ?? "chatcmpl-unknown", overrides?.model ?? body.model, chunks[0]?.created ?? Math.floor(Date.now() / 1e3), overrides?.usage ? {
+				prompt_tokens: overrides.usage.prompt_tokens ?? 0,
+				completion_tokens: overrides.usage.completion_tokens ?? 0,
+				total_tokens: overrides.usage.total_tokens ?? (overrides.usage.prompt_tokens ?? 0) + (overrides.usage.completion_tokens ?? 0)
+			} : {
+				prompt_tokens: estimatePromptTokens(body.messages),
+				completion_tokens: estimateTokens(response.content),
+				total_tokens: estimatePromptTokens(body.messages) + estimateTokens(response.content)
+			}, overrides?.systemFingerprint) : void 0;
 			const interruption = createInterruptionSignal(fixture);
 			if (!await writeSSEStream(res, chunks, {
 				latency,
 				streamingProfile: fixture.streamingProfile,
 				signal: interruption?.signal,
-				onChunkSent: interruption?.tick
+				onChunkSent: interruption?.tick,
+				usageChunk
 			})) {
 				if (!res.writableEnded) res.destroy();
 				journalEntry.response.interrupted = true;
@@ -547,17 +580,28 @@ async function handleCompletions(req, res, fixtures, journal, defaults, modelFal
 			}
 		});
 		if (body.stream !== true) {
-			const completion = buildToolCallCompletion(response.toolCalls, body.model, overrides);
+			const completion = buildToolCallCompletion(response.toolCalls, body.model, overrides, body.messages);
 			res.writeHead(200, { "Content-Type": "application/json" });
 			res.end(JSON.stringify(completion));
 		} else {
 			const chunks = buildToolCallChunks(response.toolCalls, body.model, chunkSize, overrides);
+			const completionText = response.toolCalls.map((tc) => tc.name + tc.arguments).join("");
+			const usageChunk = includeUsage ? buildUsageChunk(chunks[0]?.id ?? "chatcmpl-unknown", overrides?.model ?? body.model, chunks[0]?.created ?? Math.floor(Date.now() / 1e3), overrides?.usage ? {
+				prompt_tokens: overrides.usage.prompt_tokens ?? 0,
+				completion_tokens: overrides.usage.completion_tokens ?? 0,
+				total_tokens: overrides.usage.total_tokens ?? (overrides.usage.prompt_tokens ?? 0) + (overrides.usage.completion_tokens ?? 0)
+			} : {
+				prompt_tokens: estimatePromptTokens(body.messages),
+				completion_tokens: estimateTokens(completionText),
+				total_tokens: estimatePromptTokens(body.messages) + estimateTokens(completionText)
+			}, overrides?.systemFingerprint) : void 0;
 			const interruption = createInterruptionSignal(fixture);
 			if (!await writeSSEStream(res, chunks, {
 				latency,
 				streamingProfile: fixture.streamingProfile,
 				signal: interruption?.signal,
-				onChunkSent: interruption?.tick
+				onChunkSent: interruption?.tick,
+				usageChunk
 			})) {
 				if (!res.writableEnded) res.destroy();
 				journalEntry.response.interrupted = true;
@@ -686,6 +730,61 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 				if (await handler.handleRequest(req, res, subPath)) return;
 			}
 		}
+		if (pathname === OLLAMA_CHAT_PATH && req.method === "POST") {
+			try {
+				await handleOllama(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
+		if (pathname === OLLAMA_GENERATE_PATH && req.method === "POST") {
+			try {
+				await handleOllamaGenerate(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
+		if (pathname === OLLAMA_EMBEDDINGS_PATH && req.method === "POST") {
+			try {
+				await handleOllamaEmbeddings(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
+		if (pathname === OLLAMA_TAGS_PATH && req.method === "GET") {
+			setCorsHeaders(res);
+			const modelIds = /* @__PURE__ */ new Set();
+			for (const f of fixtures) if (f.match.model && typeof f.match.model === "string") modelIds.add(f.match.model);
+			const models = (modelIds.size > 0 ? [...modelIds] : DEFAULT_MODELS).map((name) => ({
+				name,
+				model: name,
+				modified_at: (/* @__PURE__ */ new Date()).toISOString(),
+				size: 0,
+				digest: "",
+				details: {}
+			}));
+			res.writeHead(200, { "Content-Type": "application/json" });
+			res.end(JSON.stringify({ models }));
+			return;
+		}
 		let azureDeploymentId;
 		const azureMatch = pathname.match(AZURE_DEPLOYMENT_RE);
 		if (azureMatch && req.method === "POST") {
@@ -832,6 +931,19 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
+		if (pathname === COHERE_EMBED_PATH && req.method === "POST") {
+			try {
+				await handleCohereEmbed(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
 		if (pathname === EMBEDDINGS_PATH && req.method === "POST") {
 			try {
 				const deploymentId = azureDeploymentId;
@@ -870,6 +982,32 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
+		if (pathname === IMAGES_EDIT_PATH && req.method === "POST") {
+			try {
+				await handleImageEdit(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
+		if (pathname === IMAGES_VARIATIONS_PATH && req.method === "POST") {
+			try {
+				await handleImageVariations(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
 		if (pathname === SPEECH_PATH && req.method === "POST") {
 			try {
 				await handleSpeech(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
@@ -896,6 +1034,19 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
+		if (pathname === TRANSLATIONS_PATH && req.method === "POST") {
+			try {
+				await handleTranscription(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders, "translation");
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
 		if (pathname === VIDEOS_PATH && req.method === "POST") {
 			try {
 				await handleVideoCreate(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders, videoStates);
@@ -948,6 +1099,21 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
+		const geminiEmbedMatch = pathname.match(GEMINI_EMBED_RE);
+		if (geminiEmbedMatch && req.method === "POST") {
+			const embedModel = geminiEmbedMatch[1];
+			try {
+				await handleGeminiEmbedContent(req, res, await readBody(req), embedModel, fixtures, journal, defaults, setCorsHeaders);
+			} catch (err) {
+				const msg = err instanceof Error ? err.message : "Internal error";
+				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
+					message: msg,
+					type: "server_error"
+				} }));
+				else if (!res.writableEnded) res.destroy();
+			}
+			return;
+		}
 		const geminiMatch = pathname.match(GEMINI_PATH_RE);
 		if (geminiMatch && req.method === "POST") {
 			const geminiModel = geminiMatch[1];
@@ -1050,22 +1216,9 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
-		if (pathname === OLLAMA_CHAT_PATH && req.method === "POST") {
-			try {
-				await handleOllama(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
-			} catch (err) {
-				const msg = err instanceof Error ? err.message : "Internal error";
-				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
-					message: msg,
-					type: "server_error"
-				} }));
-				else if (!res.writableEnded) res.destroy();
-			}
-			return;
-		}
-		if (pathname === OLLAMA_GENERATE_PATH && req.method === "POST") {
+		if (pathname === SEARCH_PATH && req.method === "POST") {
 			try {
-				await handleOllamaGenerate(req, res, await readBody(req), fixtures, journal, defaults, setCorsHeaders);
+				await handleSearch(req, res, await readBody(req), serviceFixtures?.search ?? [], journal, defaults, setCorsHeaders);
 			} catch (err) {
 				const msg = err instanceof Error ? err.message : "Internal error";
 				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
@@ -1076,25 +1229,9 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
-		if (pathname === OLLAMA_TAGS_PATH && req.method === "GET") {
-			setCorsHeaders(res);
-			const modelIds = /* @__PURE__ */ new Set();
-			for (const f of fixtures) if (f.match.model && typeof f.match.model === "string") modelIds.add(f.match.model);
-			const models = (modelIds.size > 0 ? [...modelIds] : DEFAULT_MODELS).map((name) => ({
-				name,
-				model: name,
-				modified_at: (/* @__PURE__ */ new Date()).toISOString(),
-				size: 0,
-				digest: "",
-				details: {}
-			}));
-			res.writeHead(200, { "Content-Type": "application/json" });
-			res.end(JSON.stringify({ models }));
-			return;
-		}
-		if (pathname === SEARCH_PATH && req.method === "POST") {
+		if (pathname === RERANK_PATH && req.method === "POST") {
 			try {
-				await handleSearch(req, res, await readBody(req), serviceFixtures?.search ?? [], journal, defaults, setCorsHeaders);
+				await handleRerank(req, res, await readBody(req), serviceFixtures?.rerank ?? [], journal, defaults, setCorsHeaders);
 			} catch (err) {
 				const msg = err instanceof Error ? err.message : "Internal error";
 				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
@@ -1105,9 +1242,9 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
-		if (pathname === RERANK_PATH && req.method === "POST") {
+		if (pathname === MODERATIONS_PATH && req.method === "POST") {
 			try {
-				await handleRerank(req, res, await readBody(req), serviceFixtures?.rerank ?? [], journal, defaults, setCorsHeaders);
+				await handleModeration(req, res, await readBody(req), serviceFixtures?.moderation ?? [], journal, defaults, setCorsHeaders);
 			} catch (err) {
 				const msg = err instanceof Error ? err.message : "Internal error";
 				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
@@ -1118,9 +1255,10 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
-		if (pathname === MODERATIONS_PATH && req.method === "POST") {
+		if (pathname === ELEVENLABS_SOUND_GENERATION_PATH && req.method === "POST") {
+			setCorsHeaders(res);
 			try {
-				await handleModeration(req, res, await readBody(req), serviceFixtures?.moderation ?? [], journal, defaults, setCorsHeaders);
+				await handleElevenLabsAudio(req, res, await readBody(req), fixtures, defaults, journal, "sound-generation");
 			} catch (err) {
 				const msg = err instanceof Error ? err.message : "Internal error";
 				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {
@@ -1131,10 +1269,12 @@ async function createServer(fixtures, options, mounts, serviceFixtures) {
 			}
 			return;
 		}
-		if (pathname === ELEVENLABS_SOUND_GENERATION_PATH && req.method === "POST") {
+		const elevenLabsTTSMatch = pathname.match(ELEVENLABS_TTS_RE);
+		if (elevenLabsTTSMatch && req.method === "POST") {
 			setCorsHeaders(res);
+			const voiceId = elevenLabsTTSMatch[1];
 			try {
-				await handleElevenLabsAudio(req, res, await readBody(req), fixtures, defaults, journal, "sound-generation");
+				await handleElevenLabsTTS(req, res, await readBody(req), fixtures, defaults, journal, voiceId);
 			} catch (err) {
 				const msg = err instanceof Error ? err.message : "Internal error";
 				if (!res.headersSent) writeErrorResponse(res, 500, JSON.stringify({ error: {