Update app.py

app.py

@@ -320,7 +320,7 @@ def generate_markdown_slides(slides, title, speaker="Prof. AI Feynman", date="Ap
 """
             markdown_slides.append(slide_md.strip())
         
-        logger.info(f"Generated Markdown slides for: {title}")
+        logger.info(f"Generated Markdown slides for: {title}: {markdown_slides}")
         return markdown_slides
     except Exception as e:
         logger.error(f"Failed to generate Markdown slides: {str(e)}")
@@ -368,7 +368,7 @@ Example output for 2 slides:
         model_client=model_client,
         handoffs=["feynman_agent"],
         system_message=f"""
-You are a Script Agent. Access the JSON array of {total_slides} slides from the conversation history. Generate a narration script (1-2 sentences) for each of the {total_slides} slides, summarizing its content in a clear, academically inclined tone as a professor would deliver it. Avoid using non-verbal fillers such as "um," "you know," or "like." Output ONLY a JSON array wrapped in ```json ... ``` with exactly {total_slides} strings, one script per slide, in the same order. Ensure the JSON is valid and complete. After outputting, use the handoff_to_script_agent tool. If scripts cannot be generated, retry once.
+You are a Script Agent. Access the JSON array of {total_slides} slides from the conversation history. Generate a narration script (1-2 sentences) for each of the {total_slides} slides, summarizing its content in a clear, academically inclined tone as a professor would deliver it. Avoid using non-verbal fillers such as "um," "you know," or "like." Output ONLY a JSON array wrapped in ```json ... ``` with exactly {total_slides} strings, one script per slide, in the same order. Ensure the JSON is valid and complete. After outputting, use the handoff_to_feynman_agent tool. If scripts cannot be generated, retry once.
 Example for 3 slides:
 ```json
 [
@@ -502,7 +502,7 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                 extracted_json = extract_json_from_message(message)
                 if extracted_json:
                     slides = extracted_json
-                    logger.info("Slide Agent generated %d slides", len(slides))
+                    logger.info("Slide Agent generated %d slides: %s", len(slides), slides)
                     if len(slides) != total_slides:
                         if slide_retry_count < max_retries:
                             slide_retry_count += 1
@@ -547,7 +547,7 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                 extracted_json = extract_json_from_message(message)
                 if extracted_json:
                     scripts = extracted_json
-                    logger.info("Script Agent generated scripts for %d slides", len(scripts))
+                    logger.info("Script Agent generated scripts for %d slides: %s", len(scripts), scripts)
                     for i, script in enumerate(scripts):
                         script_file = os.path.join(OUTPUT_DIR, f"slide_{i+1}_script.txt")
                         try:
@@ -661,7 +661,7 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             return
         
         audio_files = []
-        audio_file_paths = []
+        audio_urls = []
         validated_speaker_wav = await validate_and_convert_speaker_audio(speaker_audio)
         if not validated_speaker_wav:
             logger.error("Invalid speaker audio after conversion, skipping TTS")
@@ -691,9 +691,9 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             if not cleaned_script:
                 logger.error("Skipping audio for slide %d due to empty or invalid script", i + 1)
                 audio_files.append(None)
-                audio_file_paths.append(None)
+                audio_urls.append(None)
                 progress = 90 + ((i + 1) / len(scripts)) * 10
-                label = f"Generating speech for slide {i + 1}/{len(scripts)}..."
+                label = f"Generated audio for slide {i + 1}/{len(scripts)}..."
                 yield (
                     html_with_progress(label, progress),
                     []
@@ -715,11 +715,12 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                     if not success:
                         raise RuntimeError("TTS generation failed")
                     
-                    logger.info("Generating speech for slide %d: %s", i + 1, audio_file)
+                    logger.info("Generated audio for slide %d: %s", i + 1, audio_file)
                     audio_files.append(audio_file)
-                    audio_file_paths.append(audio_file)
+                    # Use Gradio's file serving URL
+                    audio_urls.append(f"/gradio_api/file={audio_file}")
                     progress = 90 + ((i + 1) / len(scripts)) * 10
-                    label = f"Generating speech for slide {i + 1}/{len(scripts)}..."
+                    label = f"Generated audio for slide {i + 1}/{len(scripts)}..."
                     yield (
                         html_with_progress(label, progress),
                         []
@@ -731,9 +732,9 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
                     if attempt == max_audio_retries:
                         logger.error("Max retries reached for slide %d, skipping", i + 1)
                         audio_files.append(None)
-                        audio_file_paths.append(None)
+                        audio_urls.append(None)
                         progress = 90 + ((i + 1) / len(scripts)) * 10
-                        label = f"Generating speech for slide {i + 1}/{len(scripts)}..."
+                        label = f"Generated audio for slide {i + 1}/{len(scripts)}..."
                         yield (
                             html_with_progress(label, progress),
                             []
@@ -748,18 +749,16 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
         
         # Generate audio timeline with playable audio elements
         audio_timeline = ""
-        audio_urls = []
-        for i, audio_file in enumerate(audio_file_paths):
-            if audio_file and os.path.exists(audio_file):
-                # Use the file path directly; Gradio will serve it via gr.File
-                audio_urls.append(audio_file)
-                audio_timeline += f'<audio id="audio-{i+1}" controls src="" style="display: inline-block; margin: 0 10px; width: 200px;"></audio>'
+        for i, audio_url in enumerate(audio_urls):
+            if audio_url:
+                audio_timeline += f'<audio id="audio-{i+1}" controls src="{audio_url}" style="display: inline-block; margin: 0 10px; width: 200px;"></audio>'
             else:
                 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_urls})
         
         html_output = f"""
+        <script src="https://cdn.jsdelivr.net/npm/[email protected]/marked.min.js"></script>
         <div id="lecture-container" style="height: 700px; border: 1px solid #ddd; border-radius: 8px; display: flex; flex-direction: column; justify-content: space-between;">
             <div id="slide-content" style="flex: 1; overflow: auto; padding: 20px; text-align: center; background-color: #fff; color: #333;">
                 <!-- Slides will be rendered here -->
@@ -782,20 +781,20 @@ Example: 'Received {total_slides} slides and {total_slides} scripts. Lecture is
             const totalSlides = lectureData.slides.length;
             let audioElements = [];
 
-            // Populate audio elements and set their sources
+            // Populate audio elements
             for (let i = 0; i < totalSlides; i++) {{
                 const audio = document.getElementById(`audio-${{i+1}}`);
-                if (audio && lectureData.audioFiles[i]) {{
-                    audio.src = lectureData.audioFiles[i];
-                }}
                 audioElements.push(audio);
             }}
 
             function renderSlide() {{
                 const slideContent = document.getElementById('slide-content');
                 if (lectureData.slides[currentSlide]) {{
-                    slideContent.innerHTML = lectureData.slides[currentSlide].replace(/\\n/g, '<br>');
-                    console.log("Rendering slide:", lectureData.slides[currentSlide]);
+                    const markdownText = lectureData.slides[currentSlide];
+                    const htmlContent = marked.parse(markdownText);
+                    slideContent.innerHTML = htmlContent;
+                    console.log("Rendering slide:", markdownText);
+                    console.log("Rendered HTML:", htmlContent);
                 }} else {{
                     slideContent.innerHTML = '<h2>No slide content available</h2>';
                     console.log("No slide content for index:", currentSlide);
@@ -926,4 +925,4 @@ with gr.Blocks(title="Agent Feynman") as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(allowed_paths=[OUTPUT_DIR], max_file_size="5mb")