Update streamlit_app.py

streamlit_app.py

@@ -160,30 +160,61 @@ def _download_direct(url: str, dst: Path) -> Path:
 
 
 def _download_with_yt_dlp(url: str, dst: Path, password: str = "") -> Path:
-    """Fallback downloader using yt‑dlp."""
+    """Download via yt‑dlp with Streamlit progress and MP4‑first format."""
     tmpl = str(dst / "%(id)s.%(ext)s")
-    opts = {"outtmpl": tmpl, "format": "best"}
+    # Prefer MP4, fall back to best if not available
+    fmt = "bestvideo[ext=mp4]+bestaudio[ext=m4a]/bestvideo+bestaudio"
+    opts = {
+        "outtmpl": tmpl,
+        "format": fmt,
+        "quiet": True,               # we handle progress ourselves
+        "noprogress": True,
+        "nocheckcertificate": True,
+    }
     if password:
         opts["videopassword"] = password
 
+    progress_bar = st.empty()
+    status_text = st.empty()
+
+    def _progress_hook(d):
+        if d["status"] == "downloading":
+            total = d.get("total_bytes") or d.get("total_bytes_estimate")
+            downloaded = d.get("downloaded_bytes", 0)
+            if total:
+                pct = downloaded / total
+                progress_bar.progress(pct)
+                status_text.caption(f"Downloading… {pct:.0%}")
+        elif d["status"] == "finished":
+            progress_bar.progress(1.0)
+            status_text.caption("Download complete, processing…")
+
+    opts["progress_hooks"] = [_progress_hook]
+
     try:
         with yt_dlp.YoutubeDL(opts) as ydl:
             info = ydl.extract_info(url, download=True)
     except Exception as e:
         raise RuntimeError(f"yt‑dlp could not download the URL: {e}") from e
-
-    # yt‑dlp may return a dict with the final filename
-    if isinstance(info, dict) and "id" in info:
-        candidate = dst / f"{info['id']}.{info.get('ext', 'mp4')}"
-        if candidate.exists():
-            return _convert_to_mp4(candidate)
-
-    # Fallback: pick the newest file in the output folder
-    files = list(dst.iterdir())
-    if not files:
-        raise RuntimeError("yt‑dlp did not produce any files.")
-    newest = max(files, key=lambda p: p.stat().st_mtime)
-    return _convert_to_mp4(newest)
+    finally:
+        progress_bar.empty()
+        status_text.empty()
+
+    # yt‑dlp may have produced several files; pick the newest MP4
+    mp4_files = list(dst.glob("*.mp4"))
+    if not mp4_files:
+        raise RuntimeError("No MP4 file was created.")
+    newest = max(mp4_files, key=lambda p: p.stat().st_mtime)
+
+    # Optional cache: if a file with the same SHA‑256 already exists, reuse it
+    sha = _file_sha256(newest)
+    if sha:
+        for existing in dst.iterdir():
+            if existing != newest and _file_sha256(existing) == sha:
+                newest.unlink()          # remove duplicate
+                return existing
+
+    return newest
 
 
 def download_video(url: str, dst: Path, password: str = "") -> Path: