open-agents-ai 0.163.0 → 0.165.0

package/dist/index.js

@@ -11674,6 +11674,25 @@ async function loadTranscribeCli() {
   }
   return null;
 }
+function isYouTubeUrl(url) {
+  return /(?:youtube\.com\/(?:watch|shorts|live|embed|v\/)|youtu\.be\/)/i.test(url);
+}
+function ensureYtDlp() {
+  try {
+    execSync10("yt-dlp --version", { timeout: 5e3, stdio: "pipe" });
+    return true;
+  } catch {
+    try {
+      execSync10("pip3 install --user yt-dlp 2>/dev/null || pip install --user yt-dlp 2>/dev/null", {
+        timeout: 6e4,
+        stdio: "pipe"
+      });
+      return true;
+    } catch {
+      return false;
+    }
+  }
+}
 function formatTime(seconds) {
   const m = Math.floor(seconds / 60);
   const s = Math.floor(seconds % 60);
@@ -11762,23 +11781,69 @@ var init_transcribe_tool = __esm({
             model,
             format: "json",
             diarize,
-            wordTimestamps: false
+            wordTimestamps: true
+            // Always get timestamps for structured output
           });
           const transcriptDir = join19(this.workingDir, ".oa", "transcripts");
           mkdirSync7(transcriptDir, { recursive: true });
-          const outFile = join19(transcriptDir, `${basename4(filePath)}.txt`);
-          writeFileSync7(outFile, result.text, "utf-8");
+          const fileBase = basename4(filePath).replace(/\.[^.]+$/, "");
+          const timestamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-").slice(0, 19);
+          const txtFile = join19(transcriptDir, `${fileBase}-${timestamp}.txt`);
+          writeFileSync7(txtFile, result.text, "utf-8");
+          const jsonFile = join19(transcriptDir, `${fileBase}-${timestamp}.json`);
+          const structured = {
+            source: filePath,
+            model,
+            language: result.language,
+            duration: result.duration,
+            wordCount: result.wordCount,
+            speakers: result.speakers,
+            transcribedAt: (/* @__PURE__ */ new Date()).toISOString(),
+            segments: result.segments.map((seg) => ({
+              start: seg.start,
+              end: seg.end,
+              text: seg.text,
+              speaker: seg.speaker || void 0
+            })),
+            fullText: result.text
+          };
+          writeFileSync7(jsonFile, JSON.stringify(structured, null, 2), "utf-8");
+          try {
+            const memDir = join19(this.workingDir, ".oa", "memory");
+            mkdirSync7(memDir, { recursive: true });
+            const memFile = join19(memDir, "transcripts.json");
+            let memEntries = [];
+            try {
+              memEntries = JSON.parse(readFileSync13(memFile, "utf-8"));
+            } catch {
+            }
+            memEntries.push({
+              source: basename4(filePath),
+              timestamp: (/* @__PURE__ */ new Date()).toISOString(),
+              duration: result.duration,
+              wordCount: result.wordCount,
+              language: result.language,
+              summary: result.text.slice(0, 200),
+              jsonPath: jsonFile,
+              txtPath: txtFile
+            });
+            if (memEntries.length > 50)
+              memEntries = memEntries.slice(-50);
+            writeFileSync7(memFile, JSON.stringify(memEntries, null, 2), "utf-8");
+          } catch {
+          }
           const lines = [
             `Transcription of: ${basename4(filePath)}`,
             `Model: ${model} | Language: ${result.language} | Duration: ${result.duration ? `${result.duration.toFixed(1)}s` : "unknown"}`,
-            `Words: ${result.wordCount} | Saved to: ${outFile}`,
+            `Words: ${result.wordCount} | Segments: ${result.segments.length}`,
+            `Saved: ${txtFile} (text) + ${jsonFile} (structured with timestamps)`,
             ""
           ];
           if (result.speakers.length > 1) {
             lines.push(`Speakers: ${result.speakers.join(", ")}`);
             lines.push("");
           }
-          if (diarize && result.segments.length > 0) {
+          if (result.segments.length > 0) {
             for (const seg of result.segments) {
               const ts = `[${formatTime(seg.start)} \u2192 ${formatTime(seg.end)}]`;
               const speaker = seg.speaker ? `${seg.speaker}: ` : "";
@@ -11832,13 +11897,13 @@ var init_transcribe_tool = __esm({
     };
     TranscribeUrlTool = class {
       name = "transcribe_url";
-      description = "Download an audio or video file from a URL and transcribe it to text. Supports direct links to MP3, WAV, MP4, and other media files. The file is downloaded to a temp location, transcribed locally, then cleaned up.";
+      description = "Download and transcribe audio/video from a URL. Supports YouTube links (youtube.com/watch?v=..., youtu.be/...) and direct media URLs (MP3, WAV, MP4, etc.). YouTube audio is extracted via yt-dlp (auto-installed). Transcription is local via faster-whisper (no cloud API).";
       parameters = {
         type: "object",
         properties: {
           url: {
             type: "string",
-            description: "URL of the audio or video file to download and transcribe"
+            description: "URL to transcribe \u2014 YouTube links (youtube.com/watch?v=..., youtu.be/...) or direct media files (MP3, WAV, MP4)"
           },
           model: {
             type: "string",
@@ -11866,25 +11931,57 @@ var init_transcribe_tool = __esm({
         }
         const tmpDir = join19(this.workingDir, ".oa", "tmp");
         mkdirSync7(tmpDir, { recursive: true });
-        const urlPath = new URL(url).pathname;
-        let ext = extname3(urlPath).toLowerCase();
-        if (!ext || !AUDIO_EXTS.has(ext) && !VIDEO_EXTS.has(ext)) {
-          ext = ".mp3";
-        }
-        const tmpFile = join19(tmpDir, `download-${Date.now()}${ext}`);
+        const tmpBase = join19(tmpDir, `download-${Date.now()}`);
+        let tmpFile = "";
         try {
-          try {
-            execSync10(`curl -sL -o "${tmpFile}" "${url}"`, {
-              timeout: 12e4,
-              stdio: ["pipe", "pipe", "pipe"]
-            });
-          } catch {
-            execSync10(`wget -q -O "${tmpFile}" "${url}"`, {
-              timeout: 12e4,
-              stdio: ["pipe", "pipe", "pipe"]
-            });
+          if (isYouTubeUrl(url)) {
+            if (!ensureYtDlp()) {
+              return {
+                success: false,
+                output: "",
+                error: "yt-dlp not found and auto-install failed. Install manually: pip3 install yt-dlp",
+                durationMs: performance.now() - start
+              };
+            }
+            tmpFile = `${tmpBase}.mp3`;
+            try {
+              execSync10(`yt-dlp -x --audio-format mp3 --audio-quality 5 -o "${tmpBase}.%(ext)s" "${url}" 2>&1`, { timeout: 3e5, stdio: ["pipe", "pipe", "pipe"] });
+              if (!existsSync16(tmpFile)) {
+                const { readdirSync: rd } = __require("node:fs");
+                const files = rd(tmpDir).filter((f) => f.startsWith(`download-`) && f !== ".gitkeep");
+                const match = files.find((f) => f.includes(basename4(tmpBase)));
+                if (match)
+                  tmpFile = join19(tmpDir, match);
+              }
+            } catch (dlErr) {
+              const errMsg = dlErr instanceof Error ? dlErr.message : String(dlErr);
+              return {
+                success: false,
+                output: "",
+                error: `yt-dlp failed: ${errMsg.slice(0, 200)}. Is the video available and not age-restricted?`,
+                durationMs: performance.now() - start
+              };
+            }
+          } else {
+            const urlPath = new URL(url).pathname;
+            let ext = extname3(urlPath).toLowerCase();
+            if (!ext || !AUDIO_EXTS.has(ext) && !VIDEO_EXTS.has(ext)) {
+              ext = ".mp3";
+            }
+            tmpFile = `${tmpBase}${ext}`;
+            try {
+              execSync10(`curl -sL -o "${tmpFile}" "${url}"`, {
+                timeout: 12e4,
+                stdio: ["pipe", "pipe", "pipe"]
+              });
+            } catch {
+              execSync10(`wget -q -O "${tmpFile}" "${url}"`, {
+                timeout: 12e4,
+                stdio: ["pipe", "pipe", "pipe"]
+              });
+            }
           }
-          if (!existsSync16(tmpFile)) {
+          if (!tmpFile || !existsSync16(tmpFile)) {
             return {
               success: false,
               output: "",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.163.0",
+  "version": "0.165.0",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",