feat: find segment with highest integral jump in heatmap

This commit is contained in:
Kilo Code Cloud
2026-01-14 20:01:48 +00:00
parent 44415040d5
commit 3c263a78c6
3 changed files with 24 additions and 45 deletions

View File

@@ -2,14 +2,12 @@ export interface CliArgs {
url?: string; url?: string;
output: string; output: string;
format: string; format: string;
skipStartSeconds: number;
} }
export function parseArgs(): CliArgs { export function parseArgs(): CliArgs {
const args: CliArgs = { const args: CliArgs = {
output: "./downloads", output: "./downloads",
format: "best", format: "best",
skipStartSeconds: 30,
}; };
const rawArgs = Bun.argv; const rawArgs = Bun.argv;
@@ -24,9 +22,6 @@ export function parseArgs(): CliArgs {
} else if (arg === "-f" || arg === "--format") { } else if (arg === "-f" || arg === "--format") {
args.format = nextArg || "best"; args.format = nextArg || "best";
i++; i++;
} else if (arg === "-s" || arg === "--skip-start") {
args.skipStartSeconds = parseFloat(nextArg || "30");
i++;
} else if (arg === "-h" || arg === "--help") { } else if (arg === "-h" || arg === "--help") {
console.log(`YouTube Most Watched Segment Downloader console.log(`YouTube Most Watched Segment Downloader
@@ -36,16 +31,13 @@ Arguments:
<url> YouTube video URL (required) <url> YouTube video URL (required)
Options: Options:
-o, --output <dir> Output directory (default: ./downloads) -o, --output <dir> Output directory (default: ./downloads)
-f, --format <fmt> Video format (default: best) -f, --format <fmt> Video format (default: best)
-s, --skip-start <sec> Skip first N seconds (default: 30) -h, --help Show this help message
Higher values skip more of the intro
-h, --help Show this help message
Examples: Examples:
yt-segments "https://www.youtube.com/watch?v=abc123" yt-segments "https://www.youtube.com/watch?v=abc123"
yt-segments "https://youtu.be/abc123" -o ./videos -f mp4 yt-segments "https://youtu.be/abc123" -o ./videos -f mp4
yt-segments "https://www.youtube.com/watch?v=abc123" -s 60
`); `);
process.exit(0); process.exit(0);
} else if (!arg.startsWith("-") && !arg.includes("bun")) { } else if (!arg.startsWith("-") && !arg.includes("bun")) {

View File

@@ -6,7 +6,6 @@ export interface DownloadOptions {
url: string; url: string;
outputDir: string; outputDir: string;
format: string; format: string;
skipStartSeconds: number;
} }
interface RawHeatmapSegment { interface RawHeatmapSegment {
@@ -23,7 +22,7 @@ interface ProcessedSegment {
start: number; start: number;
end: number; end: number;
intensity: number; intensity: number;
integral: number; integralJump: number;
} }
interface VideoInfo { interface VideoInfo {
@@ -143,9 +142,8 @@ function getIntensity(segment: RawHeatmapSegment): number {
return segment.intensity ?? segment.heat ?? segment.value ?? 0; return segment.intensity ?? segment.heat ?? segment.value ?? 0;
} }
function findSegmentByIntegral( function findHighestIntegralJump(
segments: RawHeatmapSegment[], segments: RawHeatmapSegment[]
skipStartSeconds: number
): ProcessedSegment | null { ): ProcessedSegment | null {
// Convert to processed format and filter valid segments // Convert to processed format and filter valid segments
const validSegments = segments const validSegments = segments
@@ -166,32 +164,25 @@ function findSegmentByIntegral(
return null; return null;
} }
// Calculate the primitive (integral) for each segment // Calculate integral jump for each segment (intensity × duration)
// The integral represents cumulative watch time contribution
let cumulativeIntegral = 0;
const withIntegral = validSegments.map(seg => { const withIntegral = validSegments.map(seg => {
const segmentDuration = seg.end - seg.start; const segmentDuration = seg.end - seg.start;
const segmentIntegral = seg.intensity * segmentDuration; const integralJump = seg.intensity * segmentDuration;
cumulativeIntegral += segmentIntegral;
return { return {
...seg, ...seg,
integral: cumulativeIntegral, integralJump,
}; };
}); });
// Sort by integral value (highest contribution first) // Find the segment with the highest integral jump (biggest bump in the integral)
// The integral tells us which segment contributed most to total watch time // This is the segment that contributed most to the total integral
withIntegral.sort((a, b) => b.integral - a.integral); return withIntegral.reduce((max, current) => {
return current.integralJump > max.integralJump ? current : max;
// Find the highest integral segment that starts after skipStartSeconds });
const candidate = withIntegral.find(seg => seg.start >= skipStartSeconds);
// If all segments are in the skipped region, return the highest one anyway
return candidate || withIntegral[0];
} }
export async function downloadMostWatchedSegment(options: DownloadOptions): Promise<void> { export async function downloadMostWatchedSegment(options: DownloadOptions): Promise<void> {
const { url, outputDir, format, skipStartSeconds } = options; const { url, outputDir, format } = options;
// Create output directory if it doesn't exist // Create output directory if it doesn't exist
if (!existsSync(outputDir)) { if (!existsSync(outputDir)) {
@@ -218,8 +209,8 @@ export async function downloadMostWatchedSegment(options: DownloadOptions): Prom
console.log(`\nHeatmap data found: ${info.heatmap.length} segments`); console.log(`\nHeatmap data found: ${info.heatmap.length} segments`);
// Find the segment with highest integral (using primitive function) // Find segment with highest integral jump (biggest bump)
const topSegment = findSegmentByIntegral(info.heatmap, skipStartSeconds); const topSegment = findHighestIntegralJump(info.heatmap);
if (!topSegment) { if (!topSegment) {
console.log("No valid segments found. Downloading full video..."); console.log("No valid segments found. Downloading full video...");
@@ -228,11 +219,11 @@ export async function downloadMostWatchedSegment(options: DownloadOptions): Prom
return; return;
} }
console.log(`\nSegment with highest integral (primitive):`); console.log(`\nSegment with highest integral jump:`);
console.log(` Time: ${formatTime(topSegment.start)} - ${formatTime(topSegment.end)}`); console.log(` Time: ${formatTime(topSegment.start)} - ${formatTime(topSegment.end)}`);
console.log(` Duration: ${formatTime(topSegment.end - topSegment.start)}`); console.log(` Duration: ${formatTime(topSegment.end - topSegment.start)}`);
console.log(` Intensity: ${(topSegment.intensity * 100).toFixed(1)}%`); console.log(` Intensity: ${(topSegment.intensity * 100).toFixed(1)}%`);
console.log(` Integral: ${topSegment.integral.toFixed(4)}`); console.log(` Integral Jump: ${topSegment.integralJump.toFixed(4)}`);
// Download the segment // Download the segment
const outputPath = join(outputDir, `${safeTitle}_most_watched.%(ext)s`); const outputPath = join(outputDir, `${safeTitle}_most_watched.%(ext)s`);
@@ -243,13 +234,12 @@ export async function downloadMostWatchedSegment(options: DownloadOptions): Prom
// Save segment info // Save segment info
const segmentInfoPath = join(outputDir, `${safeTitle}_segment_info.txt`); const segmentInfoPath = join(outputDir, `${safeTitle}_segment_info.txt`);
const segmentInfo = `# ${info.title}\n\n` + const segmentInfo = `# ${info.title}\n\n` +
`Most watched segment (highest integral from YouTube heatmap):\n` + `Segment with highest integral jump (biggest bump in heatmap):\n` +
` Start: ${formatTime(topSegment.start)} (${topSegment.start.toFixed(1)}s)\n` + ` Start: ${formatTime(topSegment.start)} (${topSegment.start.toFixed(1)}s)\n` +
` End: ${formatTime(topSegment.end)} (${topSegment.end.toFixed(1)}s)\n` + ` End: ${formatTime(topSegment.end)} (${topSegment.end.toFixed(1)}s)\n` +
` Duration: ${formatTime(topSegment.end - topSegment.start)}\n` + ` Duration: ${formatTime(topSegment.end - topSegment.start)}\n` +
` Intensity: ${(topSegment.intensity * 100).toFixed(1)}%\n` + ` Intensity: ${(topSegment.intensity * 100).toFixed(1)}%\n` +
` Integral: ${topSegment.integral.toFixed(4)}\n\n` + ` Integral Jump: ${topSegment.integralJump.toFixed(4)}\n`;
`Note: This segment had the highest integral value (cumulative watch contribution).\n`;
writeFileSync(segmentInfoPath, segmentInfo); writeFileSync(segmentInfoPath, segmentInfo);
console.log(`\nSegment info saved to: ${segmentInfoPath}`); console.log(`\nSegment info saved to: ${segmentInfoPath}`);

View File

@@ -10,17 +10,15 @@ async function main() {
console.error("Error: YouTube URL is required"); console.error("Error: YouTube URL is required");
console.log("Usage: yt-segments <url> [options]"); console.log("Usage: yt-segments <url> [options]");
console.log("Options:"); console.log("Options:");
console.log(" -o, --output <dir> Output directory (default: ./downloads)"); console.log(" -o, --output <dir> Output directory (default: ./downloads)");
console.log(" -f, --format <fmt> Video format (default: best)"); console.log(" -f, --format <fmt> Video format (default: best)");
console.log(" -s, --skip-start <sec> Skip first N seconds (default: 30)"); console.log(" -h, --help Show help");
console.log(" -h, --help Show help");
process.exit(1); process.exit(1);
} }
console.log(`Downloading most watched segment from: ${args.url}`); console.log(`Downloading most watched segment from: ${args.url}`);
console.log(`Output directory: ${args.output}`); console.log(`Output directory: ${args.output}`);
console.log(`Format: ${args.format}`); console.log(`Format: ${args.format}`);
console.log(`Skip first ${args.skipStartSeconds} seconds`);
console.log(""); console.log("");
try { try {
@@ -28,7 +26,6 @@ async function main() {
url: args.url, url: args.url,
outputDir: args.output, outputDir: args.output,
format: args.format, format: args.format,
skipStartSeconds: args.skipStartSeconds,
}); });
} catch (error) { } catch (error) {
console.error("Error:", error instanceof Error ? error.message : error); console.error("Error:", error instanceof Error ? error.message : error);