@node-llm/core 0.2.2 → 0.3.0

package/dist/providers/openai/Transcription.js

@@ -0,0 +1,161 @@
+import { handleOpenAIError } from "./Errors.js";
+import { AudioUtils } from "../../utils/audio.js";
+export class OpenAITranscription {
+    baseUrl;
+    apiKey;
+    constructor(baseUrl, apiKey) {
+        this.baseUrl = baseUrl;
+        this.apiKey = apiKey;
+    }
+    async execute(request) {
+        const model = request.model || "whisper-1";
+        if (model.startsWith("gpt-4o")) {
+            return this.transcribeViaChat(request);
+        }
+        return this.transcribeViaWhisper(request);
+    }
+    async transcribeViaWhisper(request) {
+        const formData = new FormData();
+        const { data, fileName, duration: estimatedDuration } = await AudioUtils.load(request.file);
+        const mimeType = fileName.endsWith(".wav") ? "audio/wav" : "audio/mpeg";
+        const file = new File([data], fileName, { type: mimeType });
+        formData.append("file", file);
+        formData.append("model", request.model || "whisper-1");
+        formData.append("response_format", "verbose_json");
+        if (request.prompt) {
+            formData.append("prompt", request.prompt);
+        }
+        if (request.language) {
+            formData.append("language", request.language);
+        }
+        const response = await fetch(`${this.baseUrl}/audio/transcriptions`, {
+            method: "POST",
+            headers: {
+                "Authorization": `Bearer ${this.apiKey}`,
+            },
+            body: formData,
+        });
+        if (!response.ok) {
+            await handleOpenAIError(response, request.model || "whisper-1");
+        }
+        const json = (await response.json());
+        return {
+            text: json.text,
+            model: json.model || request.model || "whisper-1",
+            duration: json.duration || estimatedDuration,
+            segments: json.segments?.map(s => ({
+                id: s.id,
+                start: s.start,
+                end: s.end,
+                text: s.text
+            }))
+        };
+    }
+    async transcribeViaChat(request) {
+        const { data, fileName, duration: estimatedDuration } = await AudioUtils.load(request.file);
+        const base64Audio = Buffer.from(data).toString("base64");
+        const model = request.model || "gpt-4o";
+        let actualModel = "gpt-4o-audio-preview";
+        let defaultPrompt = "Transcribe the audio exactly. Return only the transcription text.";
+        let isDiarization = false;
+        if (model.includes("diarize")) {
+            isDiarization = true;
+            const names = request.speakerNames?.join(", ") || "Speaker A, Speaker B";
+            defaultPrompt = `Transcribe the audio and identify different speakers. 
+      I will provide reference clips for specific voices. Please map the voices in the main audio to these names: ${names}.
+      Return the output as a JSON array of objects, each with 'speaker', 'text', 'start', and 'end' (in seconds).
+      Example: [{"speaker": "Alice", "text": "Hello", "start": 0.5, "end": 1.2}]`;
+        }
+        if (request.language) {
+            defaultPrompt += ` The audio is in ${request.language}.`;
+        }
+        const messagesContent = [
+            {
+                type: "text",
+                text: request.prompt
+                    ? `${defaultPrompt}\n\nContext for transcription: ${request.prompt}`
+                    : defaultPrompt
+            }
+        ];
+        if (request.speakerReferences && request.speakerNames) {
+            for (let i = 0; i < request.speakerReferences.length; i++) {
+                const refFile = request.speakerReferences[i];
+                if (!refFile)
+                    continue;
+                const name = request.speakerNames[i] || `Speaker ${i + 1}`;
+                const { data: refData } = await AudioUtils.load(refFile);
+                const refBase64 = Buffer.from(refData).toString("base64");
+                messagesContent.push({
+                    type: "text",
+                    text: `The following audio clip is the voice of ${name}:`
+                });
+                messagesContent.push({
+                    type: "input_audio",
+                    input_audio: {
+                        data: refBase64,
+                        format: refFile.endsWith(".wav") ? "wav" : "mp3"
+                    }
+                });
+            }
+            messagesContent.push({
+                type: "text",
+                text: "Now, transcribe this main audio file using identified names:"
+            });
+        }
+        messagesContent.push({
+            type: "input_audio",
+            input_audio: {
+                data: base64Audio,
+                format: fileName.endsWith(".wav") ? "wav" : "mp3"
+            }
+        });
+        const body = {
+            model: actualModel,
+            messages: [
+                {
+                    role: "user",
+                    content: messagesContent
+                }
+            ]
+        };
+        const response = await fetch(`${this.baseUrl}/chat/completions`, {
+            method: "POST",
+            headers: {
+                "Authorization": `Bearer ${this.apiKey}`,
+                "Content-Type": "application/json",
+            },
+            body: JSON.stringify(body),
+        });
+        if (!response.ok) {
+            await handleOpenAIError(response, actualModel);
+        }
+        const json = await response.json();
+        const content = json.choices[0]?.message?.content || "";
+        let text = content;
+        let segments = [];
+        if (isDiarization) {
+            try {
+                const parsed = JSON.parse(content);
+                if (Array.isArray(parsed)) {
+                    segments = parsed.map((s, i) => ({
+                        id: i,
+                        start: s.start || 0,
+                        end: s.end || 0,
+                        text: s.text || "",
+                        speaker: s.speaker
+                    }));
+                    text = segments.map(s => `${s.speaker}: ${s.text}`).join("\n");
+                }
+            }
+            catch (e) {
+                text = content;
+            }
+        }
+        return {
+            text,
+            model,
+            duration: estimatedDuration || 0,
+            segments
+        };
+    }
+}

package/dist/providers/openai/index.d.ts

@@ -1,2 +1,10 @@
+/**
+ * Idempotent registration of the OpenAI provider.
+ * Automatically called by LLM.configure({ provider: 'openai' })
+ */
 export declare function registerOpenAIProvider(): void;
+/**
+ * Alias for registerOpenAIProvider for internal use.
+ */
+export declare const ensureOpenAIRegistered: typeof registerOpenAIProvider;
 //# sourceMappingURL=index.d.ts.map

package/dist/providers/openai/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/providers/openai/index.ts"],"names":[],"mappings":"AAGA,wBAAgB,sBAAsB,SAUrC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/providers/openai/index.ts"],"names":[],"mappings":"AAKA;;;GAGG;AACH,wBAAgB,sBAAsB,SAcrC;AAED;;GAEG;AACH,eAAO,MAAM,sBAAsB,+BAAyB,CAAC"}

package/dist/providers/openai/index.js

@@ -1,6 +1,13 @@
 import { providerRegistry } from "../registry.js";
 import { OpenAIProvider } from "./OpenAIProvider.js";
+let registered = false;
+/**
+ * Idempotent registration of the OpenAI provider.
+ * Automatically called by LLM.configure({ provider: 'openai' })
+ */
 export function registerOpenAIProvider() {
+    if (registered)
+        return;
     providerRegistry.register("openai", () => {
         const apiKey = process.env.OPENAI_API_KEY;
         if (!apiKey) {
@@ -8,4 +15,9 @@ export function registerOpenAIProvider() {
         }
         return new OpenAIProvider({ apiKey });
     });
+    registered = true;
 }
+/**
+ * Alias for registerOpenAIProvider for internal use.
+ */
+export const ensureOpenAIRegistered = registerOpenAIProvider;

package/dist/transcription/Transcription.d.ts

@@ -0,0 +1,11 @@
+import { TranscriptionResponse, TranscriptionSegment } from "../providers/Provider.js";
+export declare class Transcription {
+    private readonly response;
+    constructor(response: TranscriptionResponse);
+    get text(): string;
+    get model(): string;
+    get segments(): TranscriptionSegment[];
+    get duration(): number | undefined;
+    toString(): string;
+}
+//# sourceMappingURL=Transcription.d.ts.map

package/dist/transcription/Transcription.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Transcription.d.ts","sourceRoot":"","sources":["../../src/transcription/Transcription.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,qBAAqB,EAAE,oBAAoB,EAAE,MAAM,0BAA0B,CAAC;AAEvF,qBAAa,aAAa;IACZ,OAAO,CAAC,QAAQ,CAAC,QAAQ;gBAAR,QAAQ,EAAE,qBAAqB;IAE5D,IAAI,IAAI,IAAI,MAAM,CAEjB;IAED,IAAI,KAAK,IAAI,MAAM,CAElB;IAED,IAAI,QAAQ,IAAI,oBAAoB,EAAE,CAErC;IAED,IAAI,QAAQ,IAAI,MAAM,GAAG,SAAS,CAEjC;IAED,QAAQ,IAAI,MAAM;CAGnB"}

package/dist/transcription/Transcription.js

@@ -0,0 +1,21 @@
+export class Transcription {
+    response;
+    constructor(response) {
+        this.response = response;
+    }
+    get text() {
+        return this.response.text;
+    }
+    get model() {
+        return this.response.model;
+    }
+    get segments() {
+        return this.response.segments || [];
+    }
+    get duration() {
+        return this.response.duration;
+    }
+    toString() {
+        return this.text;
+    }
+}

package/dist/utils/FileLoader.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"FileLoader.d.ts","sourceRoot":"","sources":["../../src/utils/FileLoader.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AA+BjD,qBAAa,UAAU;WACR,IAAI,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,WAAW,CAAC;CA+C1D"}
+{"version":3,"file":"FileLoader.d.ts","sourceRoot":"","sources":["../../src/utils/FileLoader.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AA0CjD,qBAAa,UAAU;WACR,IAAI,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,WAAW,CAAC;CA+C1D"}

package/dist/utils/FileLoader.js

@@ -13,6 +13,8 @@ const MIME_TYPES = {
     ".mp3": "audio/mpeg",
     ".json": "application/json",
     ".js": "text/javascript",
+    ".mjs": "text/javascript",
+    ".cjs": "text/javascript",
     ".ts": "text/typescript",
     ".rb": "text/x-ruby",
     ".py": "text/x-python",
@@ -23,9 +25,18 @@ const MIME_TYPES = {
     ".xml": "text/xml",
     ".yml": "text/yaml",
     ".yaml": "text/yaml",
+    ".csv": "text/csv",
+    ".go": "text/x-go",
+    ".java": "text/x-java",
+    ".c": "text/x-c",
+    ".cpp": "text/x-c++",
+    ".rs": "text/x-rust",
+    ".swift": "text/x-swift",
+    ".kt": "text/x-kotlin",
 };
 const TEXT_EXTENSIONS = new Set([
-    ".json", ".js", ".ts", ".rb", ".py", ".txt", ".md", ".html", ".css", ".xml", ".yml", ".yaml", ".env"
+    ".json", ".js", ".mjs", ".cjs", ".ts", ".rb", ".py", ".txt", ".md", ".html", ".css", ".xml", ".yml", ".yaml", ".env",
+    ".csv", ".go", ".java", ".c", ".cpp", ".rs", ".swift", ".kt"
 ]);
 export class FileLoader {
     static async load(filePath) {

package/dist/utils/audio.d.ts

@@ -0,0 +1,10 @@
+export interface AudioFileData {
+    data: Uint8Array;
+    fileName: string;
+    duration?: number;
+}
+export declare class AudioUtils {
+    static load(filePath: string): Promise<AudioFileData>;
+    static estimateDuration(data: Uint8Array, fileName: string): number | undefined;
+}
+//# sourceMappingURL=audio.d.ts.map

package/dist/utils/audio.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"audio.d.ts","sourceRoot":"","sources":["../../src/utils/audio.ts"],"names":[],"mappings":"AAGA,MAAM,WAAW,aAAa;IAC5B,IAAI,EAAE,UAAU,CAAC;IACjB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,qBAAa,UAAU;WACR,IAAI,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC;IAuB3D,MAAM,CAAC,gBAAgB,CAAC,IAAI,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS;CAkBhF"}

package/dist/utils/audio.js

@@ -0,0 +1,46 @@
+import fs from "fs";
+import path from "path";
+export class AudioUtils {
+    static async load(filePath) {
+        let data;
+        let fileName;
+        if (filePath.startsWith("http")) {
+            const response = await fetch(filePath);
+            if (!response.ok) {
+                throw new Error(`Failed to fetch remote audio file: ${response.statusText}`);
+            }
+            const arrayBuffer = await response.arrayBuffer();
+            data = new Uint8Array(arrayBuffer);
+            const urlPath = new URL(filePath).pathname;
+            fileName = path.basename(urlPath) || "audio.mp3";
+        }
+        else {
+            const buffer = fs.readFileSync(filePath);
+            data = new Uint8Array(buffer);
+            fileName = path.basename(filePath);
+        }
+        const duration = this.estimateDuration(data, fileName);
+        return { data, fileName, duration };
+    }
+    static estimateDuration(data, fileName) {
+        try {
+            const ext = path.extname(fileName).toLowerCase();
+            if (ext === ".wav") {
+                // Simple WAV duration: (DataSize / ByteRate)
+                const view = new DataView(data.buffer, data.byteOffset, data.length);
+                const byteRate = view.getUint32(28, true);
+                const dataSize = view.getUint32(40, true);
+                if (byteRate > 0)
+                    return dataSize / byteRate;
+            }
+            else if (ext === ".mp3") {
+                // Very rough MP3 estimate (assume 128kbps)
+                return data.length / (128000 / 8);
+            }
+        }
+        catch (e) {
+            // Ignore estimation errors
+        }
+        return undefined;
+    }
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@node-llm/core",
-  "version": "0.2.2",
+  "version": "0.3.0",
   "type": "module",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
@@ -13,17 +13,27 @@
   "description": "A provider-agnostic LLM core for Node.js, inspired by ruby-llm.",
   "author": "node-llm contributors",
   "license": "MIT",
+  "engines": {
+    "node": ">=20.0.0"
+  },
   "files": [
     "dist",
     "README.md"
   ],
+  "dependencies": {
+    "zod": "^3.23.8"
+  },
+  "devDependencies": {
+    "@pollyjs/adapter-fetch": "^6.0.7",
+    "@pollyjs/adapter-node-http": "^6.0.6",
+    "@pollyjs/core": "^6.0.6",
+    "@pollyjs/persister-fs": "^6.0.6"
+  },
   "scripts": {
     "build": "tsc -p tsconfig.json",
     "dev": "tsc -w",
-    "test": "vitest",
-    "prepublishOnly": "npm run build"
-  },
-  "dependencies": {
-    "zod": "^3.23.8"
+    "lint": "tsc --noEmit",
+    "test": "vitest run",
+    "test:watch": "vitest"
   }
-}
+}

package/dist/providers/openai/register.d.ts

@@ -1,2 +0,0 @@
-export declare function ensureOpenAIRegistered(): void;
-//# sourceMappingURL=register.d.ts.map

package/dist/providers/openai/register.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"register.d.ts","sourceRoot":"","sources":["../../../src/providers/openai/register.ts"],"names":[],"mappings":"AAKA,wBAAgB,sBAAsB,SAcrC"}

package/dist/providers/openai/register.js

@@ -1,15 +0,0 @@
-import { providerRegistry } from "../registry.js";
-import { OpenAIProvider } from "./OpenAIProvider.js";
-let registered = false;
-export function ensureOpenAIRegistered() {
-    if (registered)
-        return;
-    providerRegistry.register("openai", () => {
-        const apiKey = process.env.OPENAI_API_KEY;
-        if (!apiKey) {
-            throw new Error("OPENAI_API_KEY is not set");
-        }
-        return new OpenAIProvider({ apiKey });
-    });
-    registered = true;
-}

package/dist/tools/Tool.d.ts

@@ -1,8 +0,0 @@
-export type ToolHandler = (args: any) => any | Promise<any>;
-export interface Tool {
-    name: string;
-    description?: string;
-    parameters?: Record<string, any>;
-    handler: ToolHandler;
-}
-//# sourceMappingURL=Tool.d.ts.map

package/dist/tools/Tool.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"Tool.d.ts","sourceRoot":"","sources":["../../src/tools/Tool.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,WAAW,GAAG,CAAC,IAAI,EAAE,GAAG,KAAK,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC;AAE5D,MAAM,WAAW,IAAI;IACnB,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;IACjC,OAAO,EAAE,WAAW,CAAC;CACtB"}

package/dist/tools/Tool.js

@@ -1 +0,0 @@
-export {};

package/dist/tools/ToolSet.d.ts

@@ -1,15 +0,0 @@
-import { Tool, ToolHandler } from "./Tool.js";
-type ToolDefinition = ToolHandler | {
-    description?: string;
-    parameters?: Record<string, any>;
-    handler: ToolHandler;
-};
-export declare class ToolSet {
-    private tools;
-    constructor(defs?: Record<string, ToolDefinition>);
-    register(tool: Tool): void;
-    get(name: string): Tool | undefined;
-    list(): Tool[];
-}
-export {};
-//# sourceMappingURL=ToolSet.d.ts.map

package/dist/tools/ToolSet.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"ToolSet.d.ts","sourceRoot":"","sources":["../../src/tools/ToolSet.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,MAAM,WAAW,CAAC;AAE9C,KAAK,cAAc,GACf,WAAW,GACX;IACE,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;IACjC,OAAO,EAAE,WAAW,CAAC;CACtB,CAAC;AAEN,qBAAa,OAAO;IAClB,OAAO,CAAC,KAAK,CAA2B;gBAE5B,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,cAAc,CAAC;IAiBjD,QAAQ,CAAC,IAAI,EAAE,IAAI;IAInB,GAAG,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS;IAInC,IAAI,IAAI,IAAI,EAAE;CAGf"}

package/dist/tools/ToolSet.js

@@ -1,29 +0,0 @@
-export class ToolSet {
-    tools = new Map();
-    constructor(defs) {
-        if (defs) {
-            for (const [name, def] of Object.entries(defs)) {
-                if (typeof def === "function") {
-                    this.register({ name, handler: def });
-                }
-                else {
-                    this.register({
-                        name,
-                        handler: def.handler,
-                        description: def.description,
-                        parameters: def.parameters,
-                    });
-                }
-            }
-        }
-    }
-    register(tool) {
-        this.tools.set(tool.name, tool);
-    }
-    get(name) {
-        return this.tools.get(name);
-    }
-    list() {
-        return [...this.tools.values()];
-    }
-}

package/dist/tools/index.d.ts

@@ -1,2 +0,0 @@
-export { runCommand } from './runCommandTool.js';
-//# sourceMappingURL=index.d.ts.map

package/dist/tools/index.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/tools/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC"}

package/dist/tools/index.js

@@ -1 +0,0 @@
-export { runCommand } from './runCommandTool.js';

package/dist/tools/runCommandTool.d.ts

@@ -1,8 +0,0 @@
-/**
- * Executes a shell command and returns its stdout as a string.
- *
- * @param command - The command to execute.
- * @returns A promise that resolves with the command's standard output.
- */
-export declare function runCommand(command: string): Promise<string>;
-//# sourceMappingURL=runCommandTool.d.ts.map

package/dist/tools/runCommandTool.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"runCommandTool.d.ts","sourceRoot":"","sources":["../../src/tools/runCommandTool.ts"],"names":[],"mappings":"AAEA;;;;;GAKG;AACH,wBAAgB,UAAU,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAW3D"}

package/dist/tools/runCommandTool.js

@@ -1,19 +0,0 @@
-import { exec } from 'child_process';
-/**
- * Executes a shell command and returns its stdout as a string.
- *
- * @param command - The command to execute.
- * @returns A promise that resolves with the command's standard output.
- */
-export function runCommand(command) {
-    return new Promise((resolve, reject) => {
-        exec(command, { maxBuffer: 1024 * 1024 }, (error, stdout, stderr) => {
-            if (error) {
-                reject(error);
-                return;
-            }
-            // Trim trailing newlines for cleaner output
-            resolve(stdout.trim());
-        });
-    });
-}