fix: properly extract YouTube heatmap data for most watched segments
This commit is contained in:
Kilo Code Cloud
@@ -6,20 +6,18 @@ export interface DownloadOptions {
|
||||
url: string;
|
||||
outputDir: string;
|
||||
format: string;
|
||||
extractChapters: boolean;
|
||||
}
|
||||
|
||||
interface MostWatchedSegment {
|
||||
start: number;
|
||||
end: number;
|
||||
intensity?: number;
|
||||
interface HeatmapSegment {
|
||||
start_seconds: number;
|
||||
end_seconds: number;
|
||||
intensity: number;
|
||||
}
|
||||
|
||||
interface VideoInfo {
|
||||
title: string;
|
||||
chapters: Array<{ title: string; start_time: number; end_time: number }>;
|
||||
heatmap?: Array<{ start_seconds: number; end_seconds: number; intensity: number }>;
|
||||
duration: number;
|
||||
heatmap?: HeatmapSegment[];
|
||||
}
|
||||
|
||||
async function getVideoInfo(url: string): Promise<VideoInfo> {
|
||||
@@ -27,6 +25,8 @@ async function getVideoInfo(url: string): Promise<VideoInfo> {
|
||||
const ytDlp = spawn("yt-dlp", [
|
||||
"--dump-json",
|
||||
"--no-download",
|
||||
"--compat-option",
|
||||
"no-youtube-channel-redirect",
|
||||
url,
|
||||
]);
|
||||
|
||||
@@ -50,11 +50,14 @@ async function getVideoInfo(url: string): Promise<VideoInfo> {
|
||||
try {
|
||||
const info = JSON.parse(stdout);
|
||||
|
||||
// Extract heatmap data from YouTube's internal API
|
||||
// The heatmap shows what segments were re-watched the most
|
||||
const heatmapData = info.heatmap;
|
||||
|
||||
resolve({
|
||||
title: info.title || "video",
|
||||
chapters: info.chapters || [],
|
||||
heatmap: info.heatmap || [],
|
||||
duration: info.duration || 0,
|
||||
heatmap: heatmapData,
|
||||
});
|
||||
} catch (parseError) {
|
||||
reject(new Error(`Failed to parse video info: ${parseError}`));
|
||||
@@ -67,12 +70,15 @@ async function getVideoInfo(url: string): Promise<VideoInfo> {
|
||||
});
|
||||
}
|
||||
|
||||
async function downloadSection(
|
||||
async function downloadSegment(
|
||||
url: string,
|
||||
outputPath: string,
|
||||
section: string,
|
||||
startTime: number,
|
||||
endTime: number,
|
||||
format: string
|
||||
): Promise<void> {
|
||||
const section = `*${startTime.toFixed(3)}-${endTime.toFixed(3)}`;
|
||||
|
||||
return new Promise((resolve, reject) => {
|
||||
const ytDlp = spawn("yt-dlp", [
|
||||
"-f", format,
|
||||
@@ -114,29 +120,7 @@ function formatTime(seconds: number): string {
|
||||
return `${mins}:${secs.toString().padStart(2, "0")}`;
|
||||
}
|
||||
|
||||
function getMostWatchedSegments(
|
||||
heatmap: Array<{ start_seconds: number; end_seconds: number; intensity: number }>,
|
||||
duration: number,
|
||||
topN: number = 1
|
||||
): MostWatchedSegment[] {
|
||||
if (!heatmap || heatmap.length === 0) {
|
||||
return [];
|
||||
}
|
||||
|
||||
// Sort by intensity (most watched first)
|
||||
const sorted = [...heatmap].sort((a, b) => b.intensity - a.intensity);
|
||||
|
||||
// Get top N segments
|
||||
const topSegments = sorted.slice(0, topN);
|
||||
|
||||
return topSegments.map((segment) => ({
|
||||
start: segment.start_seconds,
|
||||
end: segment.end_seconds,
|
||||
intensity: segment.intensity,
|
||||
}));
|
||||
}
|
||||
|
||||
export async function downloadVideoSegments(options: DownloadOptions): Promise<void> {
|
||||
export async function downloadMostWatchedSegment(options: DownloadOptions): Promise<void> {
|
||||
const { url, outputDir, format } = options;
|
||||
|
||||
// Create output directory if it doesn't exist
|
||||
@@ -144,65 +128,57 @@ export async function downloadVideoSegments(options: DownloadOptions): Promise<v
|
||||
mkdirSync(outputDir, { recursive: true });
|
||||
}
|
||||
|
||||
// Get video info including most watched segments from YouTube API
|
||||
console.log("Fetching video information from YouTube API...");
|
||||
// Get video info with heatmap data from YouTube
|
||||
console.log("Fetching video information from YouTube...");
|
||||
const info = await getVideoInfo(url);
|
||||
const safeTitle = sanitizeFilename(info.title);
|
||||
|
||||
console.log(`Video: ${info.title}`);
|
||||
console.log(`Duration: ${formatTime(info.duration)}`);
|
||||
|
||||
// Try to get most watched segments from heatmap data
|
||||
const mostWatchedSegments = getMostWatchedSegments(info.heatmap || [], info.duration);
|
||||
|
||||
// If no heatmap data, fall back to chapters
|
||||
if (mostWatchedSegments.length === 0 && info.chapters.length > 0) {
|
||||
console.log("\nNo most watched segments found. Falling back to chapters...");
|
||||
const chapter = info.chapters[0]; // Download first chapter as most relevant
|
||||
const outputPath = join(outputDir, `${safeTitle}_most_watched.%(ext)s`);
|
||||
const section = `*${formatTime(chapter.start_time || 0)}-${formatTime(chapter.end_time || 60)}`;
|
||||
|
||||
console.log(`Downloading chapter: ${chapter.title || "First Chapter"}`);
|
||||
await downloadSection(url, outputPath, section, format);
|
||||
return;
|
||||
}
|
||||
|
||||
if (mostWatchedSegments.length === 0) {
|
||||
console.log("No segments found. Downloading full video...");
|
||||
// Check for heatmap data - this shows what was re-watched the most
|
||||
if (!info.heatmap || info.heatmap.length === 0) {
|
||||
console.log("\nNo heatmap data available for this video.");
|
||||
console.log("The video may not have enough view data to determine most watched segments.");
|
||||
console.log("Downloading full video instead...");
|
||||
|
||||
const outputPath = join(outputDir, `${safeTitle}.%(ext)s`);
|
||||
await downloadSection(url, outputPath, "*", format);
|
||||
await downloadSegment(url, outputPath, 0, info.duration, format);
|
||||
return;
|
||||
}
|
||||
|
||||
// Find the most watched segment (highest intensity)
|
||||
const mostWatched = info.heatmap.reduce((max, current) => {
|
||||
return current.intensity > max.intensity ? current : max;
|
||||
});
|
||||
|
||||
console.log(`\nHeatmap data found: ${info.heatmap.length} segments`);
|
||||
console.log(`Most watched segment intensity: ${(mostWatched.intensity * 100).toFixed(1)}%`);
|
||||
console.log(`Segment: ${formatTime(mostWatched.start_seconds)} - ${formatTime(mostWatched.end_seconds)}`);
|
||||
|
||||
// Download the most watched segment
|
||||
const topSegment = mostWatchedSegments[0];
|
||||
const outputPath = join(outputDir, `${safeTitle}_most_watched.%(ext)s`);
|
||||
const section = `*${formatTime(topSegment.start)}-${formatTime(topSegment.end)}`;
|
||||
|
||||
console.log(`\nMost watched segment: ${formatTime(topSegment.start)} - ${formatTime(topSegment.end)}`);
|
||||
console.log(`Duration: ${formatTime(topSegment.end - topSegment.start)}`);
|
||||
console.log(`Intensity: ${((topSegment.intensity || 0) * 100).toFixed(1)}%`);
|
||||
console.log(`\nDownloading most watched segment...`);
|
||||
|
||||
await downloadSection(url, outputPath, section, format);
|
||||
await downloadSegment(
|
||||
url,
|
||||
outputPath,
|
||||
mostWatched.start_seconds,
|
||||
mostWatched.end_seconds,
|
||||
format
|
||||
);
|
||||
|
||||
// Save segment info
|
||||
const segmentInfoPath = join(outputDir, `${safeTitle}_most_watched.txt`);
|
||||
let segmentInfo = `# ${info.title}\n\n`;
|
||||
segmentInfo += `Most watched segment:\n`;
|
||||
segmentInfo += ` Start: ${formatTime(topSegment.start)}\n`;
|
||||
segmentInfo += ` End: ${formatTime(topSegment.end)}\n`;
|
||||
segmentInfo += ` Duration: ${formatTime(topSegment.end - topSegment.start)}\n`;
|
||||
segmentInfo += ` Intensity: ${((topSegment.intensity || 0) * 100).toFixed(1)}%\n`;
|
||||
|
||||
if (mostWatchedSegments.length > 1) {
|
||||
segmentInfo += `\nOther top segments:\n`;
|
||||
for (let i = 1; i < mostWatchedSegments.length; i++) {
|
||||
const seg = mostWatchedSegments[i];
|
||||
segmentInfo += ` ${formatTime(seg.start)} - ${formatTime(seg.end)} (${((seg.intensity || 0) * 100).toFixed(1)}%)\n`;
|
||||
}
|
||||
}
|
||||
const segmentInfoPath = join(outputDir, `${safeTitle}_segment_info.txt`);
|
||||
const segmentInfo = `# ${info.title}\n\n` +
|
||||
`Most watched segment (from YouTube heatmap):\n` +
|
||||
` Start: ${formatTime(mostWatched.start_seconds)} (${mostWatched.start_seconds}s)\n` +
|
||||
` End: ${formatTime(mostWatched.end_seconds)} (${mostWatched.end_seconds}s)\n` +
|
||||
` Duration: ${formatTime(mostWatched.end_seconds - mostWatched.start_seconds)}\n` +
|
||||
` Intensity: ${(mostWatched.intensity * 100).toFixed(1)}%\n\n` +
|
||||
`Note: This segment had the highest re-watch rate according to YouTube's analytics.\n`;
|
||||
|
||||
writeFileSync(segmentInfoPath, segmentInfo);
|
||||
console.log(`Segment info saved to: ${segmentInfoPath}`);
|
||||
console.log(`\nSegment info saved to: ${segmentInfoPath}`);
|
||||
console.log("Download complete!");
|
||||
}
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
#!/usr/bin/env bun
|
||||
|
||||
import { downloadVideoSegments } from "./downloader.js";
|
||||
import { downloadMostWatchedSegment } from "./downloader.js";
|
||||
import { parseArgs } from "./args.js";
|
||||
|
||||
async function main() {
|
||||
@@ -17,20 +17,17 @@ async function main() {
|
||||
process.exit(1);
|
||||
}
|
||||
|
||||
console.log(`Downloading most watched segment(s) from: ${args.url}`);
|
||||
console.log(`Downloading most watched segment from: ${args.url}`);
|
||||
console.log(`Output directory: ${args.output}`);
|
||||
console.log(`Format: ${args.format}`);
|
||||
console.log(`Segments to download: ${args.segments}`);
|
||||
console.log("");
|
||||
|
||||
try {
|
||||
await downloadVideoSegments({
|
||||
await downloadMostWatchedSegment({
|
||||
url: args.url,
|
||||
outputDir: args.output,
|
||||
format: args.format,
|
||||
extractChapters: true,
|
||||
});
|
||||
console.log("\nDownload complete!");
|
||||
} catch (error) {
|
||||
console.error("Error:", error instanceof Error ? error.message : error);
|
||||
process.exit(1);
|
||||
|
||||
Reference in New Issue
Block a user