glin-profanity 3.1.5 → 3.2.2

package/dist/multimodal/audio.d.cts

@@ -0,0 +1,255 @@
+import { F as Filter } from '../Filter-D34Wsmrj.cjs';
+import { L as Language, F as FilterConfig, C as CheckProfanityResult } from '../types-B9c_ik4k.cjs';
+
+/**
+ * Audio Pipeline Utilities for glin-profanity
+ *
+ * Provides utilities for checking profanity in audio content.
+ * This module does NOT include speech-to-text - users bring their own
+ * transcription from Whisper, Google STT, Azure Speech, etc.
+ *
+ * @example
+ * ```typescript
+ * import { createAudioPipeline } from 'glin-profanity/audio';
+ * import OpenAI from 'openai';
+ *
+ * const openai = new OpenAI();
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audioBuffer) => {
+ *     const response = await openai.audio.transcriptions.create({
+ *       file: audioBuffer,
+ *       model: 'whisper-1',
+ *     });
+ *     return response.text;
+ *   },
+ * });
+ *
+ * const result = await pipeline.checkAudio(audioFile);
+ * console.log(result.containsProfanity);
+ * ```
+ *
+ * @packageDocumentation
+ * @module glin-profanity/audio
+ */
+
+/**
+ * Transcription function type
+ * Users provide their own transcription implementation
+ */
+type TranscriberFunction = (audio: AudioInput) => Promise<string>;
+/**
+ * Audio input types
+ */
+type AudioInput = Buffer | Uint8Array | Blob | File | string;
+/**
+ * Audio pipeline configuration
+ */
+interface AudioPipelineConfig {
+    /** Custom transcription function (REQUIRED) */
+    transcriber: TranscriberFunction;
+    /** Languages for profanity detection */
+    languages?: Language[];
+    /** Enable leetspeak detection */
+    detectLeetspeak?: boolean;
+    /** Enable Unicode normalization */
+    normalizeUnicode?: boolean;
+    /** Custom filter configuration */
+    filterConfig?: Partial<FilterConfig>;
+}
+/**
+ * Audio check result
+ */
+interface AudioCheckResult {
+    /** Whether profanity was found */
+    containsProfanity: boolean;
+    /** Transcribed text from audio */
+    transcribedText: string;
+    /** Profane words found */
+    profaneWords: string[];
+    /** Full profanity check result */
+    profanityResult: CheckProfanityResult;
+    /** Processing time in milliseconds */
+    processingTimeMs: number;
+    /** Transcription time in milliseconds */
+    transcriptionTimeMs: number;
+    /** Profanity check time in milliseconds */
+    checkTimeMs: number;
+}
+/**
+ * Segment result for timestamped audio
+ */
+interface AudioSegmentResult {
+    /** Segment index */
+    index: number;
+    /** Start time in seconds */
+    startTime: number;
+    /** End time in seconds */
+    endTime: number;
+    /** Transcribed text for this segment */
+    text: string;
+    /** Whether this segment contains profanity */
+    containsProfanity: boolean;
+    /** Profane words in this segment */
+    profaneWords: string[];
+}
+/**
+ * Creates an audio profanity checking pipeline
+ *
+ * @example
+ * ```typescript
+ * // With OpenAI Whisper
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audio) => {
+ *     const formData = new FormData();
+ *     formData.append('file', audio);
+ *     formData.append('model', 'whisper-1');
+ *
+ *     const response = await fetch('https://api.openai.com/v1/audio/transcriptions', {
+ *       method: 'POST',
+ *       headers: { 'Authorization': `Bearer ${process.env.OPENAI_API_KEY}` },
+ *       body: formData,
+ *     });
+ *     const data = await response.json();
+ *     return data.text;
+ *   },
+ * });
+ *
+ * // With Google Cloud Speech-to-Text
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audio) => {
+ *     // Your Google STT implementation
+ *     return transcribedText;
+ *   },
+ * });
+ * ```
+ */
+declare function createAudioPipeline(config: AudioPipelineConfig): {
+    /**
+     * Check audio for profanity
+     */
+    checkAudio(audio: AudioInput): Promise<AudioCheckResult>;
+    /**
+     * Check multiple audio files
+     */
+    checkMultiple(audios: AudioInput[]): Promise<AudioCheckResult[]>;
+    /**
+     * Check pre-transcribed text (if you already have transcription)
+     */
+    checkTranscript(text: string): CheckProfanityResult;
+    /**
+     * Check timestamped segments (for Whisper with timestamps)
+     *
+     * @example
+     * ```typescript
+     * const segments = [
+     *   { startTime: 0, endTime: 5, text: 'Hello everyone' },
+     *   { startTime: 5, endTime: 10, text: 'This is a test' },
+     * ];
+     * const results = pipeline.checkSegments(segments);
+     * const flaggedSegments = results.filter(s => s.containsProfanity);
+     * ```
+     */
+    checkSegments(segments: Array<{
+        startTime: number;
+        endTime: number;
+        text: string;
+    }>): AudioSegmentResult[];
+    /**
+     * Censor transcribed text
+     */
+    censorTranscript(text: string, replacement?: string): string;
+    /**
+     * Get the underlying filter instance
+     */
+    getFilter(): Filter;
+};
+/**
+ * Creates a transcriber function for OpenAI Whisper API
+ *
+ * @example
+ * ```typescript
+ * const transcriber = createWhisperTranscriber({
+ *   apiKey: process.env.OPENAI_API_KEY,
+ *   model: 'whisper-1',
+ * });
+ *
+ * const pipeline = createAudioPipeline({ transcriber });
+ * ```
+ */
+declare function createWhisperTranscriber(config: {
+    apiKey: string;
+    model?: string;
+    baseUrl?: string;
+    language?: string;
+}): TranscriberFunction;
+/**
+ * Creates a transcriber function for Google Cloud Speech-to-Text
+ *
+ * @example
+ * ```typescript
+ * const transcriber = createGoogleSTTTranscriber({
+ *   apiKey: process.env.GOOGLE_API_KEY,
+ *   languageCode: 'en-US',
+ * });
+ *
+ * const pipeline = createAudioPipeline({ transcriber });
+ * ```
+ */
+declare function createGoogleSTTTranscriber(config: {
+    apiKey: string;
+    languageCode?: string;
+    enableAutomaticPunctuation?: boolean;
+    profanityFilter?: boolean;
+}): TranscriberFunction;
+/**
+ * Real-time audio stream checker
+ * For live audio moderation (e.g., voice chat, podcasts)
+ *
+ * @example
+ * ```typescript
+ * const streamChecker = createRealtimeChecker({
+ *   transcriber: myTranscriber,
+ *   onProfanityDetected: (result) => {
+ *     console.log('Profanity detected:', result.profaneWords);
+ *     // Trigger beep, mute, or warning
+ *   },
+ * });
+ *
+ * // Feed audio chunks as they arrive
+ * audioStream.on('data', (chunk) => {
+ *   streamChecker.processChunk(chunk);
+ * });
+ *
+ * // Get summary when done
+ * const summary = streamChecker.getSummary();
+ * ```
+ */
+declare function createRealtimeChecker(config: {
+    transcriber: TranscriberFunction;
+    onProfanityDetected?: (result: AudioCheckResult) => void;
+    bufferDurationMs?: number;
+    languages?: Language[];
+    detectLeetspeak?: boolean;
+}): {
+    /**
+     * Process an audio chunk
+     */
+    processChunk(chunk: AudioInput): Promise<AudioCheckResult>;
+    /**
+     * Get summary of all processed chunks
+     */
+    getSummary(): {
+        totalChunks: number;
+        flaggedChunks: number;
+        cleanChunks: number;
+        flagRate: number;
+        allProfaneWords: string[];
+        fullTranscript: string;
+    };
+    /**
+     * Reset the checker state
+     */
+    reset(): void;
+};
+
+export { type AudioCheckResult, type AudioInput, type AudioPipelineConfig, type AudioSegmentResult, CheckProfanityResult, FilterConfig, Language, type TranscriberFunction, createAudioPipeline, createGoogleSTTTranscriber, createRealtimeChecker, createWhisperTranscriber };

package/dist/multimodal/audio.d.ts

@@ -0,0 +1,255 @@
+import { F as Filter } from '../Filter-BGcyIAvO.js';
+import { L as Language, F as FilterConfig, C as CheckProfanityResult } from '../types-B9c_ik4k.js';
+
+/**
+ * Audio Pipeline Utilities for glin-profanity
+ *
+ * Provides utilities for checking profanity in audio content.
+ * This module does NOT include speech-to-text - users bring their own
+ * transcription from Whisper, Google STT, Azure Speech, etc.
+ *
+ * @example
+ * ```typescript
+ * import { createAudioPipeline } from 'glin-profanity/audio';
+ * import OpenAI from 'openai';
+ *
+ * const openai = new OpenAI();
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audioBuffer) => {
+ *     const response = await openai.audio.transcriptions.create({
+ *       file: audioBuffer,
+ *       model: 'whisper-1',
+ *     });
+ *     return response.text;
+ *   },
+ * });
+ *
+ * const result = await pipeline.checkAudio(audioFile);
+ * console.log(result.containsProfanity);
+ * ```
+ *
+ * @packageDocumentation
+ * @module glin-profanity/audio
+ */
+
+/**
+ * Transcription function type
+ * Users provide their own transcription implementation
+ */
+type TranscriberFunction = (audio: AudioInput) => Promise<string>;
+/**
+ * Audio input types
+ */
+type AudioInput = Buffer | Uint8Array | Blob | File | string;
+/**
+ * Audio pipeline configuration
+ */
+interface AudioPipelineConfig {
+    /** Custom transcription function (REQUIRED) */
+    transcriber: TranscriberFunction;
+    /** Languages for profanity detection */
+    languages?: Language[];
+    /** Enable leetspeak detection */
+    detectLeetspeak?: boolean;
+    /** Enable Unicode normalization */
+    normalizeUnicode?: boolean;
+    /** Custom filter configuration */
+    filterConfig?: Partial<FilterConfig>;
+}
+/**
+ * Audio check result
+ */
+interface AudioCheckResult {
+    /** Whether profanity was found */
+    containsProfanity: boolean;
+    /** Transcribed text from audio */
+    transcribedText: string;
+    /** Profane words found */
+    profaneWords: string[];
+    /** Full profanity check result */
+    profanityResult: CheckProfanityResult;
+    /** Processing time in milliseconds */
+    processingTimeMs: number;
+    /** Transcription time in milliseconds */
+    transcriptionTimeMs: number;
+    /** Profanity check time in milliseconds */
+    checkTimeMs: number;
+}
+/**
+ * Segment result for timestamped audio
+ */
+interface AudioSegmentResult {
+    /** Segment index */
+    index: number;
+    /** Start time in seconds */
+    startTime: number;
+    /** End time in seconds */
+    endTime: number;
+    /** Transcribed text for this segment */
+    text: string;
+    /** Whether this segment contains profanity */
+    containsProfanity: boolean;
+    /** Profane words in this segment */
+    profaneWords: string[];
+}
+/**
+ * Creates an audio profanity checking pipeline
+ *
+ * @example
+ * ```typescript
+ * // With OpenAI Whisper
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audio) => {
+ *     const formData = new FormData();
+ *     formData.append('file', audio);
+ *     formData.append('model', 'whisper-1');
+ *
+ *     const response = await fetch('https://api.openai.com/v1/audio/transcriptions', {
+ *       method: 'POST',
+ *       headers: { 'Authorization': `Bearer ${process.env.OPENAI_API_KEY}` },
+ *       body: formData,
+ *     });
+ *     const data = await response.json();
+ *     return data.text;
+ *   },
+ * });
+ *
+ * // With Google Cloud Speech-to-Text
+ * const pipeline = createAudioPipeline({
+ *   transcriber: async (audio) => {
+ *     // Your Google STT implementation
+ *     return transcribedText;
+ *   },
+ * });
+ * ```
+ */
+declare function createAudioPipeline(config: AudioPipelineConfig): {
+    /**
+     * Check audio for profanity
+     */
+    checkAudio(audio: AudioInput): Promise<AudioCheckResult>;
+    /**
+     * Check multiple audio files
+     */
+    checkMultiple(audios: AudioInput[]): Promise<AudioCheckResult[]>;
+    /**
+     * Check pre-transcribed text (if you already have transcription)
+     */
+    checkTranscript(text: string): CheckProfanityResult;
+    /**
+     * Check timestamped segments (for Whisper with timestamps)
+     *
+     * @example
+     * ```typescript
+     * const segments = [
+     *   { startTime: 0, endTime: 5, text: 'Hello everyone' },
+     *   { startTime: 5, endTime: 10, text: 'This is a test' },
+     * ];
+     * const results = pipeline.checkSegments(segments);
+     * const flaggedSegments = results.filter(s => s.containsProfanity);
+     * ```
+     */
+    checkSegments(segments: Array<{
+        startTime: number;
+        endTime: number;
+        text: string;
+    }>): AudioSegmentResult[];
+    /**
+     * Censor transcribed text
+     */
+    censorTranscript(text: string, replacement?: string): string;
+    /**
+     * Get the underlying filter instance
+     */
+    getFilter(): Filter;
+};
+/**
+ * Creates a transcriber function for OpenAI Whisper API
+ *
+ * @example
+ * ```typescript
+ * const transcriber = createWhisperTranscriber({
+ *   apiKey: process.env.OPENAI_API_KEY,
+ *   model: 'whisper-1',
+ * });
+ *
+ * const pipeline = createAudioPipeline({ transcriber });
+ * ```
+ */
+declare function createWhisperTranscriber(config: {
+    apiKey: string;
+    model?: string;
+    baseUrl?: string;
+    language?: string;
+}): TranscriberFunction;
+/**
+ * Creates a transcriber function for Google Cloud Speech-to-Text
+ *
+ * @example
+ * ```typescript
+ * const transcriber = createGoogleSTTTranscriber({
+ *   apiKey: process.env.GOOGLE_API_KEY,
+ *   languageCode: 'en-US',
+ * });
+ *
+ * const pipeline = createAudioPipeline({ transcriber });
+ * ```
+ */
+declare function createGoogleSTTTranscriber(config: {
+    apiKey: string;
+    languageCode?: string;
+    enableAutomaticPunctuation?: boolean;
+    profanityFilter?: boolean;
+}): TranscriberFunction;
+/**
+ * Real-time audio stream checker
+ * For live audio moderation (e.g., voice chat, podcasts)
+ *
+ * @example
+ * ```typescript
+ * const streamChecker = createRealtimeChecker({
+ *   transcriber: myTranscriber,
+ *   onProfanityDetected: (result) => {
+ *     console.log('Profanity detected:', result.profaneWords);
+ *     // Trigger beep, mute, or warning
+ *   },
+ * });
+ *
+ * // Feed audio chunks as they arrive
+ * audioStream.on('data', (chunk) => {
+ *   streamChecker.processChunk(chunk);
+ * });
+ *
+ * // Get summary when done
+ * const summary = streamChecker.getSummary();
+ * ```
+ */
+declare function createRealtimeChecker(config: {
+    transcriber: TranscriberFunction;
+    onProfanityDetected?: (result: AudioCheckResult) => void;
+    bufferDurationMs?: number;
+    languages?: Language[];
+    detectLeetspeak?: boolean;
+}): {
+    /**
+     * Process an audio chunk
+     */
+    processChunk(chunk: AudioInput): Promise<AudioCheckResult>;
+    /**
+     * Get summary of all processed chunks
+     */
+    getSummary(): {
+        totalChunks: number;
+        flaggedChunks: number;
+        cleanChunks: number;
+        flagRate: number;
+        allProfaneWords: string[];
+        fullTranscript: string;
+    };
+    /**
+     * Reset the checker state
+     */
+    reset(): void;
+};
+
+export { type AudioCheckResult, type AudioInput, type AudioPipelineConfig, type AudioSegmentResult, CheckProfanityResult, FilterConfig, Language, type TranscriberFunction, createAudioPipeline, createGoogleSTTTranscriber, createRealtimeChecker, createWhisperTranscriber };