Update app.py

app.py

@@ -72,11 +72,11 @@ def search_web(query: str, serpapi_key: str) -> str:
         
         if "error" in results:
             logger.error("SerpApi error: %s", results["error"])
-            return f"Error during search: {results['error']}"
+            return None
         
         if "organic_results" not in results or not results["organic_results"]:
             logger.info("No search results found for query: %s", query)
-            return f"No results found for query: {query}"
+            return None
         
         formatted_results = []
         for item in results["organic_results"][:5]:
@@ -87,11 +87,11 @@ def search_web(query: str, serpapi_key: str) -> str:
         
         formatted_output = "\n".join(formatted_results)
         logger.info("Successfully retrieved search results for query: %s", query)
-        return f"Search results for {query}:\n{formatted_output}"
+        return formatted_output
     
     except Exception as e:
         logger.error("Unexpected error during search: %s", str(e))
-        return f"Unexpected error during search: {str(e)}"
+        return None
 
 # Define helper function for progress HTML
 def html_with_progress(label, progress):
@@ -338,32 +338,7 @@ async def update_audio_preview(audio_file):
     return None
 
 # Async function to generate lecture materials and audio
-async def on_generate(api_service, api_key, serpapi_key, title, lecture_content_description, instructions, lecture_type, speaker_audio, num_slides):
-    if not serpapi_key:
-        yield (
-            f"""
-            <div style="display: flex; flex-direction: column; justify-content: center; align-items: center; height: 100%; min-height: 700px; padding: 20px; text-align: center; border: 1px solid #ddd; border-radius: 8px;">
-                <h2 style="color: #d9534f;">SerpApi key required</h2>
-                <p style="margin-top: 20px;">Please provide a valid SerpApi key and try again.</p>
-            </div>
-            """,
-            []
-        )
-        return
-
-    if tts is None:
-        yield (
-            f"""
-            <div style="display: flex; flex-direction: column; justify-content: center; align-items: center; height: 100%; min-height: 700px; padding: 20px; text-align: center; border: 1px solid #ddd; border-radius: 8px;">
-                <h2 style="color: #d9534f;">TTS model initialization failed</h2>
-                <p style="margin-top: 20px;">The TTS model failed to initialize at startup.</p>
-                <p>Please ensure the Coqui TTS model is properly installed and try restarting the application.</p>
-            </div>
-            """,
-            []
-        )
-        return
-
+async def on_generate(api_service, api_key, serpapi_key, title, lecture_content_description, lecture_type, speaker_audio, num_slides):
     model_client = get_model_client(api_service, api_key)
     
     total_slides = num_slides  # Use exactly the number of slides from input
@@ -433,10 +408,9 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
     initial_message = f"""
     Lecture Title: {title}
     Lecture Content Description: {lecture_content_description}
-    Additional Instructions: {instructions}
     Audience: {lecture_type}
     Number of Slides: {total_slides}
-    Please start by researching the topic.
+    Please start by researching the topic, or proceed without research if search is unavailable.
     """
     logger.info("Starting lecture generation for title: %s with %d slides", title, total_slides)
     
@@ -451,7 +425,11 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
     
     try:
         logger.info("Research Agent starting...")
-        task_result = await Console(swarm.run_stream(task=initial_message))
+        if serpapi_key:
+            task_result = await Console(swarm.run_stream(task=initial_message))
+        else:
+            logger.warning("No SerpApi key provided, bypassing research phase")
+            task_result = await Console(swarm.run_stream(task=f"{initial_message}\nNo search available, proceed with slide generation."))
         logger.info("Swarm execution completed")
         
         slide_retry_count = 0
@@ -574,13 +552,13 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                     scripts = extracted_json
                     logger.info("Script Agent generated scripts for %d slides", len(scripts))
                     for i, script in enumerate(scripts):
-                        script_file = os.path.join(OUTPUT_DIR, f"slide_{i+1}_raw_script.txt")
+                        script_file = os.path.join(OUTPUT_DIR, f"slide_{i+1}_script.txt")
                         try:
                             with open(script_file, "w", encoding="utf-8") as f:
                                 f.write(script)
-                            logger.info("Saved raw script to %s", script_file)
+                            logger.info("Saved script to %s", script_file)
                         except Exception as e:
-                            logger.error("Error saving raw script to %s: %s", script_file, str(e))
+                            logger.error("Error saving script to %s: %s", script_file, str(e))
                     progress = 75
                     label = "Scripts generated and saved. Reviewing..."
                     yield (
@@ -625,7 +603,7 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                 f"""
                 <div style="display: flex; flex-direction: column; justify-content: center; align-items: center; height: 100%; min-height: 700px; padding: 20px; text-align: center; border: 1px solid #ddd; border-radius: 8px;">
                     <h2 style="color: #d9534f;">{error_message}</h2>
-                    <p style="margin-top: 20px;">Please try again with a different model (e.g., Anthropic-claude-3-sonnet-20240229) or simplify the topic/instructions.</p>
+                    <p style="margin-top: 20px;">Please try again with a different model or adjust your inputs.</p>
                 </div>
                 """,
                 []
@@ -708,15 +686,15 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             try:
                 with open(script_file, "w", encoding="utf-8") as f:
                     f.write(cleaned_script or "")
-                logger.info("Saved cleaned script to %s: %s", script_file, cleaned_script)
+                logger.info("Saved script to %s: %s", script_file, cleaned_script)
             except Exception as e:
-                logger.error("Error saving cleaned script to %s: %s", script_file, str(e))
+                logger.error("Error saving script to %s: %s", script_file, str(e))
             
             if not cleaned_script:
                 logger.error("Skipping audio for slide %d due to empty or invalid script", i + 1)
                 audio_files.append(None)
                 progress = 90 + ((i + 1) / len(scripts)) * 10
-                label = f"Generating speech for slide {i + 1}/{len(scripts)}..."
+                label = f"Generated speech for slide {i + 1}/{len(scripts)}..."
                 yield (
                     html_with_progress(label, progress),
                     []
@@ -770,10 +748,10 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
         # Generate audio timeline with playable audio elements
         audio_timeline = ""
         for i, audio_file in enumerate(audio_files):
-            if audio_file:
-                audio_timeline += f'<audio id="audio-{i+1}" controls src="{audio_file}" style="margin: 0 5px;"></audio>'
+            if audio_file and os.path.exists(audio_file):
+                audio_timeline += f'<audio id="audio-{i+1}" controls src="file://{audio_file}" style="display: inline-block; margin: 0 10px; width: 200px;"></audio>'
             else:
-                audio_timeline += f'<span id="audio-{i+1}" style="margin: 0 5px;">slide_{i+1}.mp3 (not generated)</span>'
+                audio_timeline += f'<span id="audio-{i+1}" style="display: inline-block; margin: 0 10px;">slide_{i+1}.mp3 (not generated)</span>'
         
         slides_info = json.dumps({"slides": markdown_slides, "audioFiles": audio_files})
         
@@ -782,8 +760,8 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             <div id="slide-content" style="flex: 1; overflow: auto; padding: 20px; text-align: center;">
                 <!-- Slides will be rendered here -->
             </div>
-            <div style="padding: 20px;">
-                <div style="text-align: center; margin-bottom: 10px;">
+            <div style="padding: 20px; text-align: center;">
+                <div style="display: flex; justify-content: center; margin-bottom: 10px;">
                     {audio_timeline}
                 </div>
                 <div style="display: flex; justify-content: center; margin-bottom: 10px;">
@@ -799,16 +777,19 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             let currentSlide = 0;
             const totalSlides = lectureData.slides.length;
             let audioElements = [];
-            let isPlayingAll = false;
 
             for (let i = 0; i < totalSlides; i++) {{
-                const audio = document.getElementById(`audio-${{i+1}}`);
+                const audio = document.getElementById(`audio-${i+1}`);
                 audioElements.push(audio);
             }}
 
             function renderSlide() {{
                 const slideContent = document.getElementById('slide-content');
-                slideContent.innerHTML = lectureData.slides[currentSlide];
+                if (lectureData.slides[currentSlide]) {{
+                    slideContent.innerHTML = lectureData.slides[currentSlide];
+                }} else {{
+                    slideContent.innerHTML = '<h2>No slide content available</h2>';
+                }}
             }}
 
             function updateSlide() {{
@@ -836,51 +817,27 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             }}
 
             function playAll() {{
-                if (isPlayingAll) {{
-                    audioElements.forEach(audio => {{
-                        if (audio && audio.pause) audio.pause();
-                    }});
-                    isPlayingAll = false;
-                    document.getElementById('play-btn').innerText = '⏯';
-                    return;
-                }}
-
-                isPlayingAll = true;
-                document.getElementById('play-btn').innerText = '⏸';
-                currentSlide = 0;
-                updateSlide();
-                playCurrentSlide();
-            }}
-
-            function playCurrentSlide() {{
-                if (!isPlayingAll || currentSlide >= totalSlides) {{
-                    isPlayingAll = false;
-                    document.getElementById('play-btn').innerText = '⏯';
-                    return;
-                }}
-
-                const audio = audioElements[currentSlide];
-                if (audio && audio.play) {{
-                    audio.play().then(() => {{
-                        audio.addEventListener('ended', () => {{
-                            currentSlide++;
-                            if (currentSlide < totalSlides) {{
-                                updateSlide();
-                                playCurrentSlide();
-                            }} else {{
-                                isPlayingAll = false;
-                                document.getElementById('play-btn').innerText = '⏯';
-                            }}
-                        }}, {{ once: true }});
-                    }}).catch(e => {{
-                        console.error('Audio play failed:', e);
-                        currentSlide++;
-                        playCurrentSlide();
-                    }});
-                }} else {{
-                    currentSlide++;
-                    playCurrentSlide();
+                let index = 0;
+                function playNext() {{
+                    if (index >= totalSlides) return;
+                    const audio = audioElements[index];
+                    if (audio && audio.play) {{
+                        audio.play().then(() => {{
+                            audio.addEventListener('ended', () => {{
+                                index++;
+                                playNext();
+                            }}, {{ once: true }});
+                        }}).catch(e => {{
+                            console.error('Audio play failed:', e);
+                            index++;
+                            playNext();
+                        }});
+                    }} else {{
+                        index++;
+                        playNext();
+                    }}
                 }}
+                playNext();
             }}
 
             // Attach event listeners
@@ -920,7 +877,6 @@ with gr.Blocks(title="Agent Feynman") as demo:
             with gr.Group():
                 title = gr.Textbox(label="Lecture Title", placeholder="e.g. Introduction to AI")
                 lecture_content_description = gr.Textbox(label="Lecture Content Description", placeholder="e.g. Focus on recent advancements")
-                instructions = gr.Textbox(label="Additional Instructions", placeholder="e.g. Include examples")
                 lecture_type = gr.Dropdown(["Conference", "University", "High school"], label="Audience", value="University")
                 api_service = gr.Dropdown(
                     choices=[
@@ -933,7 +889,7 @@ with gr.Blocks(title="Agent Feynman") as demo:
                     value="Google-gemini-1.5-flash"
                 )
                 api_key = gr.Textbox(label="Model Provider API Key", type="password", placeholder="Not required for Ollama")
-                serpapi_key = gr.Textbox(label="SerpApi Key", type="password", placeholder="Enter your SerpApi key")
+                serpapi_key = gr.Textbox(label="SerpApi Key", type="password", placeholder="Enter your SerpApi key (optional)")
                 num_slides = gr.Slider(1, 20, step=1, label="Number of Slides", value=3)
                 speaker_audio = gr.Audio(label="Speaker sample audio (MP3 or WAV)", type="filepath", elem_id="speaker-audio")
                 generate_btn = gr.Button("Generate Lecture")
@@ -955,7 +911,7 @@ with gr.Blocks(title="Agent Feynman") as demo:
     
     generate_btn.click(
         fn=on_generate,
-        inputs=[api_service, api_key, serpapi_key, title, lecture_content_description, instructions, lecture_type, speaker_audio, num_slides],
+        inputs=[api_service, api_key, serpapi_key, title, lecture_content_description, lecture_type, speaker_audio, num_slides],
         outputs=[slide_display, file_output]
     )