fix: extract video file and output as mkv

src/cli/downloader.ts

@@ -77,11 +77,14 @@ async function getVideoInfo(url: string): Promise<VideoInfo> {
   });
 }
 
-async function downloadFullVideo(url: string, outputPath: string): Promise<string> {
+async function downloadFullVideo(url: string, outputDir: string, safeTitle: string): Promise<string> {
+  // Download to a temp pattern, yt-dlp will fill in the filename
+  const tempPattern = join(outputDir, `${safeTitle}_temp_%(id)s.%(ext)s`);
+
   return new Promise((resolve, reject) => {
     // Use minimal options - let yt-dlp use its config file
     const ytDlp = spawn("yt-dlp", [
-      "-o", outputPath,
+      "-o", tempPattern,
       url,
     ]);
 
@@ -96,7 +99,7 @@ async function downloadFullVideo(url: string, outputPath: string): Promise<strin
         reject(new Error(`yt-dlp failed: ${stderr}`));
         return;
       }
-      resolve(outputPath);
+      resolve(tempPattern);
     });
 
     ytDlp.on("error", (err) => {
@@ -105,6 +108,33 @@ async function downloadFullVideo(url: string, outputPath: string): Promise<strin
   });
 }
 
+async function findVideoFile(outputDir: string, safeTitle: string): Promise<string | null> {
+  // Video extensions to look for (not subtitles)
+  const videoExtensions = ["mp4", "mkv", "webm", "mov", "avi", "m4v"];
+  const subtitleExtensions = ["vtt", "srt", "ass", "lrc"];
+
+  return new Promise((resolve) => {
+    const glob = spawn("find", [outputDir, "-name", `${safeTitle}_temp_*`, "-type", "f"]);
+    let output = "";
+    glob.stdout.on("data", (data) => { output += data.toString(); });
+    glob.on("close", () => {
+      const files = output.split("\n").filter(f => f.length > 0);
+      
+      // Find video file (not subtitle)
+      for (const file of files) {
+        const ext = file.split(".").pop()?.toLowerCase();
+        if (ext && videoExtensions.includes(ext)) {
+          resolve(file);
+          return;
+        }
+      }
+
+      // If no video file, return null
+      resolve(null);
+    });
+  });
+}
+
 async function extractSegment(
   inputPath: string,
   outputPath: string,
@@ -180,17 +210,14 @@ function clusterHighIntensitySegments(
   segments: RawSegment[],
   threshold: number
 ): ProcessedSegment[] {
-  // Filter to only high-intensity segments
   const highIntensity = segments.filter(seg => seg.intensity >= threshold);
 
   if (highIntensity.length === 0) {
     return [];
   }
 
-  // Sort by start time
   highIntensity.sort((a, b) => a.start - b.start);
 
-  // Cluster adjacent/overlapping segments
   const clusters: ProcessedSegment[] = [];
   let currentCluster: RawSegment | null = null;
 
@@ -200,15 +227,12 @@ function clusterHighIntensitySegments(
       continue;
     }
 
-    // Check if this segment is adjacent or overlapping with current cluster
     const gap = seg.start - currentCluster.end;
-    const maxGap = 10; // Allow up to 10 second gap
+    const maxGap = 10;
 
-    if (gap <= maxGap && gap >= -1) { // -1 allows for small overlaps
+    if (gap <= maxGap && gap >= -1) {
-      // Merge into current cluster
       currentCluster.end = Math.max(currentCluster.end, seg.end);
     } else {
-      // Finalize current cluster and start new one
       const totalIntensity = highIntensity
         .filter(s => s.start >= currentCluster!.start && s.end <= currentCluster!.end)
         .reduce((sum, s) => sum + s.intensity, 0);
@@ -227,7 +251,6 @@ function clusterHighIntensitySegments(
     }
   }
 
-  // Don't forget the last cluster
   if (currentCluster) {
     const totalIntensity = highIntensity
       .filter(s => s.start >= currentCluster.start && s.end <= currentCluster.end)
@@ -244,9 +267,7 @@ function clusterHighIntensitySegments(
     });
   }
 
-  // Sort clusters by total intensity (highest first)
   clusters.sort((a, b) => b.totalIntensity - a.totalIntensity);
-
   return clusters;
 }
 
@@ -315,42 +336,31 @@ export async function downloadMostWatchedSegment(options: DownloadOptions): Prom
 
   console.log("");
 
-  // Download the top segment
   const topSegment = topSegments[0];
 
   // Step 1: Download full video (uses your yt-dlp config)
-  const fullVideoPath = join(outputDir, `${safeTitle}_full_temp.%(ext)s`);
   console.log(`Downloading full video (using your yt-dlp config)...`);
   
-  await downloadFullVideo(url, fullVideoPath);
+  await downloadFullVideo(url, outputDir, safeTitle);
 
-  // Find the actual file (yt-dlp may have changed extension)
+  // Find video file (not subtitle)
-  // The downloaded path should already be correct, but let's handle the pattern
+  const videoPath = await findVideoFile(outputDir, safeTitle);
-  const tempFiles = await new Promise<string[]>((resolve) => {
-    const glob = spawn("find", [outputDir, "-name", `${safeTitle}_full_temp.*`, "-type", "f"]);
-    let output = "";
-    glob.stdout.on("data", (data) => { output += data.toString(); });
-    glob.on("close", () => {
-      resolve(output.split("\n").filter(f => f.length > 0));
-    });
-  });
 
-  if (tempFiles.length === 0) {
+  if (!videoPath) {
-    throw new Error("Could not find downloaded video file");
+    throw new Error("No video file found. Your yt-dlp config may be downloading subtitles instead.");
   }
 
-  const actualFullPath = tempFiles[0];
+  console.log(`Downloaded video: ${videoPath}`);
-  console.log(`Downloaded to: ${actualFullPath}`);
 
-  // Step 2: Extract segment with ffmpeg
+  // Step 2: Extract segment with ffmpeg to mkv
-  const outputPath = join(outputDir, `${safeTitle}_segment.${actualFullPath.split(".").pop()}`);
+  const outputPath = join(outputDir, `${safeTitle}_segment.mkv`);
   console.log(`Extracting segment: ${formatTime(topSegment.start)} - ${formatTime(topSegment.end)}`);
 
-  await extractSegment(actualFullPath, outputPath, topSegment.start, topSegment.end);
+  await extractSegment(videoPath, outputPath, topSegment.start, topSegment.end);
 
-  // Clean up temp file
+  // Clean up temp files
   try {
-    unlinkSync(actualFullPath);
+    unlinkSync(videoPath);
   } catch {
     // Ignore cleanup errors
   }