@markitdownjs/audio 0.1.0

package/dist/__tests__/audio-converter.test.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=audio-converter.test.d.ts.map

package/dist/__tests__/audio-converter.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"audio-converter.test.d.ts","sourceRoot":"","sources":["../../src/__tests__/audio-converter.test.ts"],"names":[],"mappings":""}

package/dist/__tests__/audio-converter.test.js

@@ -0,0 +1,59 @@
+import { describe, it, expect } from 'vitest';
+import { AudioConverter } from '../audio-converter.js';
+describe('AudioConverter', () => {
+    const converter = new AudioConverter();
+    it('should detect audio by MIME type', async () => {
+        expect(await converter.canConvert({ data: '', mimeType: 'audio/mpeg' })).toBe(true);
+        expect(await converter.canConvert({ data: '', mimeType: 'audio/wav' })).toBe(true);
+        expect(await converter.canConvert({ data: '', mimeType: 'audio/mp4' })).toBe(true);
+        expect(await converter.canConvert({ data: '', mimeType: 'audio/ogg' })).toBe(true);
+        expect(await converter.canConvert({ data: '', mimeType: 'audio/flac' })).toBe(true);
+    });
+    it('should detect audio by extension', async () => {
+        expect(await converter.canConvert({ data: '', fileName: 'test.mp3' })).toBe(true);
+        expect(await converter.canConvert({ data: '', fileName: 'test.wav' })).toBe(true);
+        expect(await converter.canConvert({ data: '', fileName: 'test.m4a' })).toBe(true);
+        expect(await converter.canConvert({ data: '', fileName: 'test.ogg' })).toBe(true);
+        expect(await converter.canConvert({ data: '', fileName: 'test.flac' })).toBe(true);
+    });
+    it('should reject non-audio files', async () => {
+        expect(await converter.canConvert({ data: '', fileName: 'test.txt' })).toBe(false);
+        expect(await converter.canConvert({ data: '', mimeType: 'text/plain' })).toBe(false);
+    });
+    it('should return placeholder when no provider is set', async () => {
+        const result = await converter.convert({
+            data: new Uint8Array([0, 1, 2, 3]),
+            mimeType: 'audio/mpeg',
+        });
+        expect(result.markdown).toContain('Transcription unavailable');
+        expect(result.ast).toBeDefined();
+        expect(result.ast.type).toBe('document');
+    });
+    it('should use provider when set', async () => {
+        converter.setProvider({
+            async transcribe() {
+                return 'Hello from transcription';
+            },
+        });
+        const result = await converter.convert({
+            data: new Uint8Array([0, 1, 2, 3]),
+            mimeType: 'audio/mpeg',
+        });
+        expect(result.markdown).toBe('Hello from transcription');
+    });
+    it('should pass mimeType to provider', async () => {
+        let receivedMimeType = '';
+        converter.setProvider({
+            async transcribe(_data, mimeType) {
+                receivedMimeType = mimeType;
+                return 'done';
+            },
+        });
+        await converter.convert({
+            data: 'audio data',
+            mimeType: 'audio/wav',
+        });
+        expect(receivedMimeType).toBe('audio/wav');
+    });
+});
+//# sourceMappingURL=audio-converter.test.js.map

package/dist/__tests__/audio-converter.test.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"audio-converter.test.js","sourceRoot":"","sources":["../../src/__tests__/audio-converter.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,EAAE,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAC9C,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAEvD,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;IAC9B,MAAM,SAAS,GAAG,IAAI,cAAc,EAAE,CAAC;IAEvC,EAAE,CAAC,kCAAkC,EAAE,KAAK,IAAI,EAAE;QAChD,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACpF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACnF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACnF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACnF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACtF,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,kCAAkC,EAAE,KAAK,IAAI,EAAE;QAChD,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACrF,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,+BAA+B,EAAE,KAAK,IAAI,EAAE;QAC7C,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACnF,MAAM,CAAC,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACvF,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mDAAmD,EAAE,KAAK,IAAI,EAAE;QACjE,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,OAAO,CAAC;YACrC,IAAI,EAAE,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAClC,QAAQ,EAAE,YAAY;SACvB,CAAC,CAAC;QACH,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,SAAS,CAAC,2BAA2B,CAAC,CAAC;QAC/D,MAAM,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,WAAW,EAAE,CAAC;QACjC,MAAM,CAAC,MAAM,CAAC,GAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC5C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,8BAA8B,EAAE,KAAK,IAAI,EAAE;QAC5C,SAAS,CAAC,WAAW,CAAC;YACpB,KAAK,CAAC,UAAU;gBACd,OAAO,0BAA0B,CAAC;YACpC,CAAC;SACF,CAAC,CAAC;QACH,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,OAAO,CAAC;YACrC,IAAI,EAAE,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YAClC,QAAQ,EAAE,YAAY;SACvB,CAAC,CAAC;QACH,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,0BAA0B,CAAC,CAAC;IAC3D,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,kCAAkC,EAAE,KAAK,IAAI,EAAE;QAChD,IAAI,gBAAgB,GAAG,EAAE,CAAC;QAC1B,SAAS,CAAC,WAAW,CAAC;YACpB,KAAK,CAAC,UAAU,CAAC,KAAK,EAAE,QAAQ;gBAC9B,gBAAgB,GAAG,QAAQ,CAAC;gBAC5B,OAAO,MAAM,CAAC;YAChB,CAAC;SACF,CAAC,CAAC;QACH,MAAM,SAAS,CAAC,OAAO,CAAC;YACtB,IAAI,EAAE,YAAY;YAClB,QAAQ,EAAE,WAAW;SACtB,CAAC,CAAC;QACH,MAAM,CAAC,gBAAgB,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}

package/dist/audio-converter.d.ts

@@ -0,0 +1,14 @@
+import type { Converter, ConversionInput, ConversionResult } from "@markitdownjs/shared";
+export interface AudioTranscriptionProvider {
+    transcribe(audioData: Uint8Array, mimeType: string): Promise<string>;
+}
+export declare class AudioConverter implements Converter {
+    readonly id = "audio";
+    readonly supportedMimeTypes: string[];
+    readonly supportedExtensions: string[];
+    private provider;
+    setProvider(provider: AudioTranscriptionProvider): void;
+    canConvert(input: ConversionInput): Promise<boolean>;
+    convert(input: ConversionInput): Promise<ConversionResult>;
+}
+//# sourceMappingURL=audio-converter.d.ts.map

package/dist/audio-converter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"audio-converter.d.ts","sourceRoot":"","sources":["../src/audio-converter.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,SAAS,EACT,eAAe,EACf,gBAAgB,EAEjB,MAAM,sBAAsB,CAAC;AAG9B,MAAM,WAAW,0BAA0B;IACzC,UAAU,CAAC,SAAS,EAAE,UAAU,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;CACtE;AAED,qBAAa,cAAe,YAAW,SAAS;IAC9C,QAAQ,CAAC,EAAE,WAAW;IACtB,QAAQ,CAAC,kBAAkB,WAMzB;IACF,QAAQ,CAAC,mBAAmB,WAA6C;IAEzE,OAAO,CAAC,QAAQ,CAA2C;IAE3D,WAAW,CAAC,QAAQ,EAAE,0BAA0B,GAAG,IAAI;IAIjD,UAAU,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,OAAO,CAAC;IAepD,OAAO,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,gBAAgB,CAAC;CAoDjE"}

package/dist/audio-converter.js

@@ -0,0 +1,76 @@
+import { createNode, readInputData } from "@markitdownjs/shared";
+export class AudioConverter {
+    id = "audio";
+    supportedMimeTypes = [
+        "audio/mpeg",
+        "audio/wav",
+        "audio/mp4",
+        "audio/ogg",
+        "audio/flac",
+    ];
+    supportedExtensions = [".mp3", ".wav", ".m4a", ".ogg", ".flac"];
+    provider = null;
+    setProvider(provider) {
+        this.provider = provider;
+    }
+    async canConvert(input) {
+        if (input.mimeType && this.supportedMimeTypes.includes(input.mimeType)) {
+            return true;
+        }
+        if (input.fileName) {
+            const ext = input.fileName.slice(input.fileName.lastIndexOf(".")).toLowerCase();
+            if (this.supportedExtensions.includes(ext)) {
+                return true;
+            }
+        }
+        return false;
+    }
+    async convert(input) {
+        const startTime = performance.now();
+        const data = await readInputData(input.data);
+        const mimeType = input.mimeType ?? "audio/mpeg";
+        let transcription;
+        if (this.provider) {
+            transcription = await this.provider.transcribe(data, mimeType);
+        }
+        else {
+            transcription =
+                "[Transcription unavailable: no AudioTranscriptionProvider configured. Call setProvider() to inject a transcription service.]";
+        }
+        const textNode = createNode({
+            type: "text",
+            value: transcription,
+        });
+        const paragraphNode = createNode({
+            type: "paragraph",
+            children: [textNode],
+        });
+        const documentNode = createNode({
+            type: "document",
+            children: [paragraphNode],
+        });
+        const endTime = performance.now();
+        const stats = {
+            startTime,
+            endTime,
+            duration: endTime - startTime,
+            inputSize: data.length,
+            outputSize: transcription.length,
+        };
+        return {
+            markdown: transcription,
+            metadata: {
+                wordCount: transcription.split(/\s+/).filter(Boolean).length,
+            },
+            assets: [],
+            tables: [],
+            images: [],
+            headings: [],
+            ast: documentNode,
+            format: "markdown",
+            converterId: this.id,
+            stats,
+        };
+    }
+}
+//# sourceMappingURL=audio-converter.js.map

package/dist/audio-converter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"audio-converter.js","sourceRoot":"","sources":["../src/audio-converter.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAC;AAMjE,MAAM,OAAO,cAAc;IAChB,EAAE,GAAG,OAAO,CAAC;IACb,kBAAkB,GAAG;QAC5B,YAAY;QACZ,WAAW;QACX,WAAW;QACX,WAAW;QACX,YAAY;KACb,CAAC;IACO,mBAAmB,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC;IAEjE,QAAQ,GAAsC,IAAI,CAAC;IAE3D,WAAW,CAAC,QAAoC;QAC9C,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC;IAC3B,CAAC;IAED,KAAK,CAAC,UAAU,CAAC,KAAsB;QACrC,IAAI,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC,kBAAkB,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvE,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;YACnB,MAAM,GAAG,GAAG,KAAK,CAAC,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;YAChF,IAAI,IAAI,CAAC,mBAAmB,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;gBAC3C,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QAED,OAAO,KAAK,CAAC;IACf,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,KAAsB;QAClC,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,EAAE,CAAC;QACpC,MAAM,IAAI,GAAG,MAAM,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC7C,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,IAAI,YAAY,CAAC;QAEhD,IAAI,aAAqB,CAAC;QAC1B,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAClB,aAAa,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;QACjE,CAAC;aAAM,CAAC;YACN,aAAa;gBACX,8HAA8H,CAAC;QACnI,CAAC;QAED,MAAM,QAAQ,GAAG,UAAU,CAAC;YAC1B,IAAI,EAAE,MAAe;YACrB,KAAK,EAAE,aAAa;SACrB,CAAC,CAAC;QAEH,MAAM,aAAa,GAAG,UAAU,CAAC;YAC/B,IAAI,EAAE,WAAoB;YAC1B,QAAQ,EAAE,CAAC,QAAQ,CAAC;SACrB,CAAC,CAAC;QAEH,MAAM,YAAY,GAAG,UAAU,CAAC;YAC9B,IAAI,EAAE,UAAmB;YACzB,QAAQ,EAAE,CAAC,aAAa,CAAC;SAC1B,CAAC,CAAC;QAEH,MAAM,OAAO,GAAG,WAAW,CAAC,GAAG,EAAE,CAAC;QAClC,MAAM,KAAK,GAAoB;YAC7B,SAAS;YACT,OAAO;YACP,QAAQ,EAAE,OAAO,GAAG,SAAS;YAC7B,SAAS,EAAE,IAAI,CAAC,MAAM;YACtB,UAAU,EAAE,aAAa,CAAC,MAAM;SACjC,CAAC;QAEF,OAAO;YACL,QAAQ,EAAE,aAAa;YACvB,QAAQ,EAAE;gBACR,SAAS,EAAE,aAAa,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,MAAM;aAC7D;YACD,MAAM,EAAE,EAAE;YACV,MAAM,EAAE,EAAE;YACV,MAAM,EAAE,EAAE;YACV,QAAQ,EAAE,EAAE;YACZ,GAAG,EAAE,YAAY;YACjB,MAAM,EAAE,UAAU;YAClB,WAAW,EAAE,IAAI,CAAC,EAAE;YACpB,KAAK;SACN,CAAC;IACJ,CAAC;CACF"}

package/dist/index.d.ts

@@ -0,0 +1,3 @@
+export { AudioConverter } from "./audio-converter.js";
+export type { AudioTranscriptionProvider } from "./audio-converter.js";
+//# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,YAAY,EAAE,0BAA0B,EAAE,MAAM,sBAAsB,CAAC"}

package/dist/index.js

@@ -0,0 +1,2 @@
+export { AudioConverter } from "./audio-converter.js";
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

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

package/package.json

@@ -0,0 +1,31 @@
+{
+  "name": "@markitdownjs/audio",
+  "version": "0.1.0",
+  "description": "Audio file converter for MarkItDownJS",
+  "license": "MIT",
+  "type": "module",
+  "main": "./dist/index.js",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "import": "./dist/index.js",
+      "types": "./dist/index.d.ts"
+    }
+  },
+  "files": [
+    "dist"
+  ],
+  "dependencies": {
+    "@markitdownjs/shared": "0.1.0"
+  },
+  "devDependencies": {
+    "typescript": "^5.5.0"
+  },
+  "scripts": {
+    "build": "tsc --project tsconfig.json",
+    "dev": "tsc --watch --project tsconfig.json",
+    "typecheck": "tsc --noEmit",
+    "clean": "rm -rf dist tsconfig.tsbuildinfo"
+  }
+}