Update app.py

app.py

@@ -62,47 +62,55 @@ class FastVideoGenerator:
     @staticmethod
     def clean_text(text):
         """Clean text to handle encoding issues"""
-        # Normalize unicode characters
-        text = unicodedata.normalize('NFKD', text)
-        # Replace special characters with standard ASCII
+        if not isinstance(text, str):
+            text = str(text)
+        
+        # Remove non-ASCII characters
         text = text.encode('ascii', 'ignore').decode('ascii')
-        # Replace common special characters
-        replacements = {
-            '–': '-',  # en dash
-            '—': '-',  # em dash
-            ''': "'",  # curly quote
-            ''': "'",  # curly quote
-            '"': '"',  # curly quote
-            '"': '"',  # curly quote
-            '…': '...' # ellipsis
-        }
-        for old, new in replacements.items():
-            text = text.replace(old, new)
-        return text
+        
+        # Replace common problematic characters
+        text = text.replace('–', '-')  # en dash
+        text = text.replace('—', '-')  # em dash
+        text = text.replace('"', '"')  # smart quotes
+        text = text.replace('"', '"')  # smart quotes
+        text = text.replace(''', "'")  # smart apostrophe
+        text = text.replace(''', "'")  # smart apostrophe
+        text = text.replace('…', '...') # ellipsis
+        
+        # Remove any remaining non-standard characters
+        text = re.sub(r'[^\x00-\x7F]+', '', text)
+        
+        return text.strip()
 
     @staticmethod
     @st.cache_data
     def generate_script_cached(prompt, style, length, temperature=0.7):
         """Cached script generation with proper text cleaning"""
-        style_prompts = {
-            'Professional': "Write a clear, professional video script about:",
-            'Creative': "Write an engaging, creative video script about:",
-            'Educational': "Write an informative educational video script about:"
-        }
-        
-        prompt = FastVideoGenerator.clean_text(prompt)
-        
-        with pipeline('text-generation', model='distilgpt2') as generator:
+        try:
+            style_prompts = {
+                'Professional': "Write a clear, professional video script about:",
+                'Creative': "Write an engaging, creative video script about:",
+                'Educational': "Write an informative educational video script about:"
+            }
+            
+            # Clean prompt before processing
+            prompt = FastVideoGenerator.clean_text(prompt)
+            
+            generator = pipeline('text-generation', model='distilgpt2')
             output = generator(
                 f"{style_prompts[style]} {prompt}. Make it {length} seconds long.",
                 max_length=min(length * 3, 1000),
                 num_return_sequences=1,
                 temperature=temperature
             )
-        
-        script = output[0]['generated_text']
-        script = script.replace(style_prompts[style], '').strip()
-        return FastVideoGenerator.clean_text(script)
+            
+            script = output[0]['generated_text']
+            script = script.replace(style_prompts[style], '').strip()
+            return FastVideoGenerator.clean_text(script)
+            
+        except Exception as e:
+            st.error(f"Script generation failed: {str(e)}")
+            raise
 
     def create_frame_fast(self, text, theme, frame_number, total_frames, size=(1280, 720)):
         """Create frame with cleaned text"""
@@ -145,16 +153,15 @@ class FastVideoGenerator:
         for i, chunk in enumerate(chunks):
             chunk_path = self.temp_dir / f"audio_chunk_{i}.mp3"
             try:
+                # Additional cleaning for gTTS
+                chunk = re.sub(r'[^\x00-\x7F]+', '', chunk)
                 tts = gTTS(text=chunk, lang='en', slow=False)
                 tts.save(str(chunk_path))
                 audio_paths.append(chunk_path)
             except Exception as e:
-                # If TTS fails, try with further cleaning
-                cleaned_chunk = re.sub(r'[^a-zA-Z0-9\s.,!?-]', '', chunk)
-                tts = gTTS(text=cleaned_chunk, lang='en', slow=False)
-                tts.save(str(chunk_path))
-                audio_paths.append(chunk_path)
-            
+                st.error(f"Audio generation failed for chunk {i}: {str(e)}")
+                raise
+        
         return audio_paths
 
     def create_optimized_video(self, script, theme, duration=30):