Update app.py

app.py

@@ -2,16 +2,15 @@
 AI Graduate Job-Matching Platform
 Complete Hugging Face Spaces Deployment Version
 Developer: Najaf Ali Sharqi
-MoviePy-free (uses OpenCV + imageio)
+MoviePy-free (uses ffmpeg + tempfile)
 """
 
 import os
 import gradio as gr
 import tempfile
-import traceback
-from groq import Groq
-import imageio
+import subprocess
 import math
+from groq import Groq
 
 # ============================================================================
 # CONFIGURATION
@@ -20,23 +19,35 @@ import math
 client = Groq(api_key=os.getenv("GROQ_API_KEY"))
 
 TRANSCRIPTION_MODEL = "whisper-large-v3-turbo"
-CHAT_MODEL = "llama-3.3-70b-versatile"
 CHUNK_SIZE = 600  # 10 min in seconds
 
 # ============================================================================
-# AUDIO EXTRACTION AND CHUNKING
+# UTILITY FUNCTIONS
 # ============================================================================
 
+def get_video_duration(video_path):
+    """Get video duration using ffprobe (more reliable than imageio)"""
+    try:
+        result = subprocess.run(
+            ["ffprobe", "-v", "error", "-show_entries",
+             "format=duration", "-of",
+             "default=noprint_wrappers=1:nokey=1", video_path],
+            stdout=subprocess.PIPE,
+            stderr=subprocess.STDOUT,
+            text=True
+        )
+        duration = float(result.stdout.strip())
+        return duration
+    except Exception as e:
+        raise RuntimeError(f"Failed to get video duration: {str(e)}")
+
 def extract_audio_chunks(video_path, chunk_size=CHUNK_SIZE):
     """
     Extract audio from video and split into chunks
     Returns list of audio file paths
     """
     try:
-        reader = imageio.get_reader(video_path, 'ffmpeg')
-        fps = reader.get_meta_data()['fps']
-        nframes = reader.get_meta_data()['nframes']
-        duration = nframes / fps
+        duration = get_video_duration(video_path)
         num_chunks = math.ceil(duration / chunk_size)
         audio_chunks = []
 
@@ -44,7 +55,8 @@ def extract_audio_chunks(video_path, chunk_size=CHUNK_SIZE):
         temp_audio_full = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False)
         audio_full_path = temp_audio_full.name
         temp_audio_full.close()
-        os.system(f"ffmpeg -y -i '{video_path}' -q:a 0 -map a '{audio_full_path}'")
+
+        os.system(f"ffmpeg -y -i \"{video_path}\" -q:a 0 -map a \"{audio_full_path}\"")
 
         # split audio into chunks
         for i in range(num_chunks):
@@ -53,14 +65,14 @@ def extract_audio_chunks(video_path, chunk_size=CHUNK_SIZE):
             temp_chunk = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False)
             chunk_path = temp_chunk.name
             temp_chunk.close()
-            os.system(f"ffmpeg -y -i '{audio_full_path}' -ss {start} -to {end} -c copy '{chunk_path}'")
+            os.system(f"ffmpeg -y -i \"{audio_full_path}\" -ss {start} -to {end} -c copy \"{chunk_path}\"")
             audio_chunks.append(chunk_path)
 
         # remove full audio file
         if os.path.exists(audio_full_path):
             os.unlink(audio_full_path)
 
-        return audio_chunks, duration
+        return audio_chunks
     except Exception as e:
         raise RuntimeError(f"Audio extraction failed: {str(e)}")
 
@@ -69,9 +81,6 @@ def extract_audio_chunks(video_path, chunk_size=CHUNK_SIZE):
 # ============================================================================
 
 def transcribe_audio_chunks(audio_chunks, language="English"):
-    """
-    Transcribe a list of audio chunks
-    """
     transcript_text = ""
     for chunk in audio_chunks:
         try:
@@ -100,7 +109,7 @@ def process_video_ui(video_file, language):
     if not video_file:
         return "", "Please upload a video file." if lang=="English" else "براہ کرم ایک ویڈیو فائل اپ لوڈ کریں۔"
     try:
-        audio_chunks, duration = extract_audio_chunks(video_file.name)
+        audio_chunks = extract_audio_chunks(video_file.name)
         transcript = transcribe_audio_chunks(audio_chunks, language=lang)
         return transcript, "✅ Success / کامیاب"
     except Exception as e: