Update src/gradio_app.py

src/gradio_app.py

@@ -2,23 +2,20 @@ import gradio as gr
 import os
 import requests
 import soundfile as sf
+import tempfile
+import re as regex
+import subprocess
 
 # --- Configuration ---
 MAX_SIZE_MB = "50"
 MAX_SECONDS = 60
-# Replace with your actual key if your instance requires it
 LIBRE_API_KEY = "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
 TRANSLATE_URL = "https://imsidag-community-libretranslate-kabyle.hf.space/translate"
 
 # --- Translation Logic ---
 def translate_to_english(text):
-    """
-    Sends the Kabyle transcript to a LibreTranslate instance.
-    Expects application/x-www-form-urlencoded data.
-    """
     if not text or any(symbol in text for symbol in ["⚠️", "❌"]):
         return ""
-    
     payload = {
         'q': text,
         'source': 'kab',
@@ -26,9 +23,7 @@ def translate_to_english(text):
         'format': 'text',
         'api_key': LIBRE_API_KEY
     }
-    
     try:
-        # requests.post with 'data=' automatically uses x-www-form-urlencoded
         response = requests.post(TRANSLATE_URL, data=payload, timeout=15)
         response.raise_for_status()
         data = response.json()
@@ -36,12 +31,69 @@ def translate_to_english(text):
     except Exception as e:
         return f"❌ Translation Error: {str(e)}"
 
-# --- Transcription & Processing Logic ---
-def process_audio(audio_file):
+# --- YouTube Download Logic ---
+def download_youtube_audio(url: str) -> str:
     """
-    Handles the full pipeline: Validation -> Transcription -> Translation.
+    Downloads audio from a YouTube URL using yt-dlp.
+    Returns the path to the downloaded audio file.
     """
-    # 1. Validation
+    if not url or not url.strip():
+        raise ValueError("Please provide a YouTube URL.")
+
+    # Basic URL validation
+    youtube_pattern = regex.compile(
+        r'(https?://)?(www\.)?(youtube|youtu|youtube-nocookie)\.(com|be)/'
+        r'(watch\?v=|embed/|v/|.+\?v=)?([^&=%\?]{11})'
+    )
+    if not youtube_pattern.match(url.strip()):
+        raise ValueError("Invalid YouTube URL.")
+
+    tmp_dir = tempfile.gettempdir()
+    output_template = os.path.join(tmp_dir, "youtube_%(id)s.%(ext)s")
+
+    # yt-dlp command: download best audio, convert to wav with ffmpeg
+    cmd = [
+        "yt-dlp",
+        "-f", "bestaudio/best",
+        "--extract-audio",
+        "--audio-format", "wav",
+        "--audio-quality", "0",
+        "--no-playlist",
+        "--max-filesize", f"{MAX_SIZE_MB}M",
+        "-o", output_template,
+        url.strip()
+    ]
+
+    result = subprocess.run(cmd, capture_output=True, text=True)
+    if result.returncode != 0:
+        raise RuntimeError(f"yt-dlp failed:\n{result.stderr}")
+
+    # Find the downloaded file
+    video_id = None
+    for line in result.stderr.splitlines() + result.stdout.splitlines():
+        m = regex.search(r'\[download\] Destination: .+youtube_([a-zA-Z0-9_-]+)\.wav', line)
+        if m:
+            video_id = m.group(1)
+            break
+
+    if not video_id:
+        # Fallback: glob the temp dir
+        import glob
+        files = glob.glob(os.path.join(tmp_dir, "youtube_*.wav"))
+        if not files:
+            raise FileNotFoundError("Download completed but audio file not found.")
+        audio_path = max(files, key=os.path.getctime)
+    else:
+        audio_path = os.path.join(tmp_dir, f"youtube_{video_id}.wav")
+
+    if not os.path.exists(audio_path):
+        raise FileNotFoundError("Downloaded audio file not found.")
+
+    return audio_path
+
+# --- Unified Processing Logic ---
+def process_audio(audio_file):
+    """Handles validation -> Transcription -> Translation."""
     if audio_file is None or (isinstance(audio_file, str) and audio_file.strip() == ""):
         return "⚠️ Please upload an audio file first.", ""
 
@@ -53,19 +105,35 @@ def process_audio(audio_file):
         except Exception as e:
             return f"❌ Error reading audio info: {str(e)}", ""
 
-    # 2. Transcription
     try:
         from inference_file import inference
         transcript = inference(audio_file)
-        
-        # 3. Translation
         translation = translate_to_english(transcript)
-        
         return transcript, translation
-
     except Exception as e:
         return f"❌ Error during processing: {str(e)}", ""
 
+def process_youtube(url):
+    """Downloads YouTube audio then runs the ASR pipeline."""
+    if not url or not url.strip():
+        return "⚠️ Please enter a YouTube URL.", ""
+
+    try:
+        audio_path = download_youtube_audio(url)
+    except Exception as e:
+        return f"❌ Download Error: {str(e)}", ""
+
+    # Run the same pipeline
+    transcript, translation = process_audio(audio_path)
+
+    # Optional cleanup to save disk space
+    try:
+        if os.path.exists(audio_path):
+            os.remove(audio_path)
+    except Exception:
+        pass
+
+    return transcript, translation
 
 # --- Build Gradio UI ---
 with gr.Blocks(title="🎙️ Mmeslay") as demo:
@@ -74,51 +142,72 @@ with gr.Blocks(title="🎙️ Mmeslay") as demo:
         # 🎙️ Mmeslay by [G1ya777](https://github.com/G1ya777/Mmeslay)
         ### Kabyle ASR & Translation
         *Powered by Squeezeformer (ASR) and LibreTranslate (NMT)*
-        
-        Upload a Kabyle audio file or record directly to get a transcript and English translation.
+
+        Upload a Kabyle audio file, record directly, **or paste a YouTube link** to get a transcript and English translation.
         """
     )
 
-    with gr.Row():
-        with gr.Column(scale=1):
-            audio_input = gr.Audio(
-                label="Input Audio",
-                type="filepath",
-                sources=["upload", "microphone"],
-                format="mp3",
-            )
-            transcribe_btn = gr.Button("🚀 Transcribe & Translate", variant="primary", size="lg")
-
-        with gr.Column(scale=2):
-            text_output = gr.Textbox(
-                label="Transcription (Kabyle)",
-                lines=5
-            )
-            translation_output = gr.Textbox(
-                label="LibreTranslate (English)",
-                lines=5,
-                placeholder="English LibreTranslate translation will appear here..."
-            )
-
-    # Map the button click to the dual-output function
-    transcribe_btn.click(
-        fn=process_audio,
-        inputs=audio_input,
-        outputs=[text_output, translation_output],
-    )
+    with gr.Tab("🎧 Audio Upload / Record"):
+        with gr.Row():
+            with gr.Column(scale=1):
+                audio_input = gr.Audio(
+                    label="Input Audio",
+                    type="filepath",
+                    sources=["upload", "microphone"],
+                    format="mp3",
+                )
+                transcribe_btn = gr.Button("🚀 Transcribe & Translate", variant="primary", size="lg")
 
-    gr.Examples(
-        examples=[
-            "ressources/examples/e1.mp3",
-            "ressources/examples/e2.mp3",
-        ],
-        inputs=audio_input,
-    )
+            with gr.Column(scale=2):
+                text_output_1 = gr.Textbox(label="Transcription (Kabyle)", lines=5)
+                translation_output_1 = gr.Textbox(
+                    label="LibreTranslate (English)",
+                    lines=5,
+                    placeholder="English LibreTranslate translation will appear here..."
+                )
+
+        transcribe_btn.click(
+            fn=process_audio,
+            inputs=audio_input,
+            outputs=[text_output_1, translation_output_1],
+        )
+
+        gr.Examples(
+            examples=[
+                "ressources/examples/e1.mp3",
+                "ressources/examples/e2.mp3",
+            ],
+            inputs=audio_input,
+        )
+
+    with gr.Tab("📺 YouTube Link"):
+        with gr.Row():
+            with gr.Column(scale=1):
+                youtube_url = gr.Textbox(
+                    label="YouTube URL",
+                    placeholder="https://www.youtube.com/watch?v=...",
+                    lines=1
+                )
+                youtube_btn = gr.Button("🚀 Download & Transcribe", variant="primary", size="lg")
+
+            with gr.Column(scale=2):
+                text_output_2 = gr.Textbox(label="Transcription (Kabyle)", lines=5)
+                translation_output_2 = gr.Textbox(
+                    label="LibreTranslate (English)",
+                    lines=5,
+                    placeholder="English LibreTranslate translation will appear here..."
+                )
+
+        youtube_btn.click(
+            fn=process_youtube,
+            inputs=youtube_url,
+            outputs=[text_output_2, translation_output_2],
+        )
 
     gr.Markdown(
         """
         ---
-        Developed by [G1ya777](https://github.com/G1ya777/Mmeslay). 
+        Developed by [G1ya777](https://github.com/G1ya777/Mmeslay).
         Examples from Tatoeba (CC BY licenses).
         """
     )
@@ -131,5 +220,5 @@ if __name__ == "__main__":
         server_name="0.0.0.0",
         server_port=port,
         max_file_size=f"{MAX_SIZE_MB}mb",
-        theme=gr.themes.Soft(),  
+        theme=gr.themes.Soft(),
     )