import { downloadAudioFromUrl } from "./sources/rss.js";
import { searchYouTube, downloadAudio as downloadYouTubeAudio } from "./sources/youtube.js";
import { transcribeAudio } from "./transcription.js";
import { saveTranscript } from "./file-manager.js";
export interface EpisodeInfo {
podcastName: string;
title: string;
pubDate: string;
audioUrl?: string;
}
export interface ScrapeResult {
transcriptPath: string;
transcriptPreview: string;
source: 'rss' | 'youtube';
}
/**
* Custom error types for better error handling
*/
export class AudioDownloadError extends Error {
constructor(message: string, public originalError?: Error) {
super(message);
this.name = 'AudioDownloadError';
}
}
export class AudioNotFoundError extends Error {
constructor(message: string) {
super(message);
this.name = 'AudioNotFoundError';
}
}
/**
* Search YouTube for a podcast episode
*/
async function searchYouTubeForEpisode(
podcastName: string,
episodeTitle: string
): Promise<string | null> {
console.log(`\nš Searching YouTube for: "${podcastName} ${episodeTitle}"`);
const query = `${podcastName} ${episodeTitle}`;
const results = await searchYouTube(query, 5);
if (results.length === 0) {
return null;
}
// Simple heuristic: Pick the first result that:
// 1. Contains the podcast name in the title or uploader
// 2. Is longer than 5 minutes (likely full episode, not clip)
const podcastNameLower = podcastName.toLowerCase();
for (const result of results) {
const titleLower = result.title.toLowerCase();
const uploaderLower = result.uploader.toLowerCase();
const isLikelyMatch = titleLower.includes(podcastNameLower) || uploaderLower.includes(podcastNameLower);
const isFullEpisode = result.duration > 300; // 5 minutes
if (isLikelyMatch && isFullEpisode) {
console.log(`ā Found match: "${result.title}" by ${result.uploader} (${Math.floor(result.duration / 60)} min)`);
return result.url;
}
}
// If no perfect match, return the first long video
const firstLongVideo = results.find(r => r.duration > 300);
if (firstLongVideo) {
console.log(`ā ļø No perfect match, using: "${firstLongVideo.title}" (${Math.floor(firstLongVideo.duration / 60)} min)`);
return firstLongVideo.url;
}
return null;
}
/**
* Download audio from RSS with automatic YouTube fallback
*/
async function downloadAudioWithFallback(
episodeInfo: EpisodeInfo
): Promise<{ audioPath: string; source: 'rss' | 'youtube' }> {
// Try RSS first if we have an audio URL
if (episodeInfo.audioUrl) {
try {
console.log("š„ Attempting download from RSS feed...");
const audioPath = await downloadAudioFromUrl(
episodeInfo.audioUrl,
`${episodeInfo.podcastName.replace(/[^a-z0-9]/gi, '-')}-${Date.now()}.mp3`
);
console.log(`ā Successfully downloaded from RSS`);
return { audioPath, source: 'rss' };
} catch (error) {
console.log(`ā RSS download failed: ${error instanceof Error ? error.message : 'Unknown error'}`);
console.log(`š Attempting YouTube fallback...`);
}
} else {
console.log(`ā ļø No audio URL provided, trying YouTube...`);
}
// Fallback to YouTube
const youtubeUrl = await searchYouTubeForEpisode(
episodeInfo.podcastName,
episodeInfo.title
);
if (!youtubeUrl) {
throw new AudioNotFoundError(
`Could not find episode on YouTube: "${episodeInfo.title}"`
);
}
console.log(`š„ Downloading from YouTube...`);
const result = await downloadYouTubeAudio(youtubeUrl);
console.log(`ā Successfully downloaded from YouTube`);
return { audioPath: result.audioPath, source: 'youtube' };
}
/**
* Scrape and transcribe a podcast episode with automatic fallback
*/
export async function scrapeEpisode(episodeInfo: EpisodeInfo): Promise<ScrapeResult> {
console.log(`\n${'='.repeat(60)}`);
console.log(`š» Scraping: ${episodeInfo.title}`);
console.log(`${'='.repeat(60)}\n`);
// Download audio (with fallback)
const { audioPath, source } = await downloadAudioWithFallback(episodeInfo);
console.log(`Audio file: ${audioPath}`);
// Transcribe
console.log(`\nšļø Transcribing audio (this may take a few minutes)...`);
const transcriptResult = await transcribeAudio(audioPath);
const transcript = transcriptResult.text;
console.log(`ā Transcription complete: ${transcript.length.toLocaleString()} characters`);
// Save transcript
console.log(`\nš¾ Saving transcript...`);
const transcriptPath = saveTranscript(
episodeInfo.podcastName,
episodeInfo.title,
episodeInfo.pubDate,
transcript
);
console.log(`ā Saved to: ${transcriptPath}`);
// Generate preview
const transcriptPreview = transcript.substring(0, 500);
console.log(`\n${'='.repeat(60)}`);
console.log(`ā
Successfully scraped episode (source: ${source.toUpperCase()})`);
console.log(`${'='.repeat(60)}\n`);
return {
transcriptPath,
transcriptPreview,
source,
};
}
