Update app.py

app.py

@@ -1,10 +1,14 @@
 import gradio as gr
 import PyPDF2
+import fitz  # PyMuPDF
 import io
 import os
 from groq import Groq
 import tempfile
 import traceback
+import base64
+from io import BytesIO
+from PIL import Image
 
 def extract_text_from_pdf(pdf_file):
     """Extract text from uploaded PDF file"""
@@ -30,6 +34,74 @@ def extract_text_from_pdf(pdf_file):
     except Exception as e:
         return None, f"Error reading PDF: {str(e)}"
 
+def extract_images_from_pdf(pdf_file_path):
+    """Extract all images from PDF and return them as PIL Images with page info"""
+    images = []
+    
+    # Open the PDF
+    pdf_document = fitz.open(pdf_file_path)
+    
+    for page_num in range(len(pdf_document)):
+        page = pdf_document.load_page(page_num)
+        
+        # Get image list from the page
+        image_list = page.get_images(full=True)
+        
+        for img_index, img in enumerate(image_list):
+            # Get the XREF of the image
+            xref = img[0]
+            
+            # Extract the image bytes
+            base_image = pdf_document.extract_image(xref)
+            image_bytes = base_image["image"]
+            
+            # Convert to PIL Image
+            pil_image = Image.open(BytesIO(image_bytes))
+            
+            images.append({
+                'image': pil_image,
+                'page': page_num + 1,
+                'index': img_index + 1,
+                'format': base_image["ext"]
+            })
+    
+    pdf_document.close()
+    return images
+
+def convert_pdf_pages_to_images(pdf_file_path, dpi=150):
+    """Convert each PDF page to an image for comprehensive analysis"""
+    images = []
+    
+    # Open the PDF
+    pdf_document = fitz.open(pdf_file_path)
+    
+    for page_num in range(len(pdf_document)):
+        page = pdf_document.load_page(page_num)
+        
+        # Convert page to image
+        mat = fitz.Matrix(dpi/72, dpi/72)  # zoom factor
+        pix = page.get_pixmap(matrix=mat)
+        
+        # Convert to PIL Image
+        img_data = pix.tobytes("png")
+        pil_image = Image.open(BytesIO(img_data))
+        
+        images.append({
+            'image': pil_image,
+            'page': page_num + 1,
+            'type': 'full_page'
+        })
+    
+    pdf_document.close()
+    return images
+
+def encode_image_to_base64(pil_image):
+    """Convert PIL Image to base64 string"""
+    buffered = BytesIO()
+    pil_image.save(buffered, format="PNG")
+    img_str = base64.b64encode(buffered.getvalue()).decode()
+    return f"data:image/png;base64,{img_str}"
+
 def summarize_with_groq(api_key, text):
     """Generate summary using Groq API"""
     try:
@@ -44,11 +116,8 @@ def summarize_with_groq(api_key, text):
         
         # System prompt for summarization
         system_prompt = """You are a highly capable language model specialized in document summarization. Your task is to read and understand the full content of a multi-page PDF document and generate a clear, accurate, and detailed summary of the entire document.
-
 Focus on capturing all main ideas, key points, arguments, findings, and conclusions presented throughout the document. If the document is technical, legal, academic, or contains structured sections (e.g., introduction, methods, results, discussion), maintain the logical flow and structure while expressing the content in a comprehensive and accessible manner.
-
 Avoid unnecessary simplification. Include important details, supporting evidence, and nuanced insights that reflect the depth of the original material. Do not copy the text verbatim.
-
 Output only the summary. Do not explain your process. Use a neutral, professional, and informative tone. The summary should provide a full understanding of the document to someone who has not read it."""
 
         # Create completion with optimal hyperparameters
@@ -80,8 +149,83 @@ Output only the summary. Do not explain your process. Use a neutral, professiona
             error_msg += "\n\nPlease check your Groq API key and ensure it's valid."
         return error_msg
 
-def process_pdf_and_summarize(api_key, pdf_file, progress=gr.Progress()):
-    """Main function to process PDF and generate summary"""
+def analyze_images_with_groq(api_key, images):
+    """Analyze images using Groq API"""
+    if not api_key:
+        return "❌ Please enter your Groq API key."
+    
+    try:
+        client = Groq(api_key=api_key)
+        
+        results = []
+        
+        for idx, img_data in enumerate(images):
+            # Encode image to base64
+            base64_image = encode_image_to_base64(img_data['image'])
+            
+            # Prepare messages for the API call
+            messages = [
+                {
+                    "role": "system",
+                    "content": """You are an advanced language model with strong capabilities in visual and textual understanding. Your task is to analyze all images, diagrams, and flowcharts within a PDF document. This includes:
+
+1. Extracting and interpreting text from images and flowcharts.
+2. Understanding the visual structure, logic, and relationships depicted in diagrams.
+3. Summarizing the content and purpose of each visual element in a clear and informative manner.
+
+After processing, be ready to answer user questions about any of the images or flowcharts, including their meaning, structure, data, process flows, or relationships shown.
+
+Be accurate, concise, and visually aware. Clearly explain visual content in text form. Do not guess if visual data is unclear or ambiguous — instead, state what is observable.
+
+Use a neutral, helpful tone. Do not include irrelevant information or commentary unrelated to the visual content. When summarizing or answering questions, assume the user may not have access to the original image or diagram."""
+                },
+                {
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": "text",
+                            "text": f"Please analyze this image from page {img_data.get('page', 'unknown')} of the PDF document. Provide a detailed analysis of all visual elements, text, diagrams, flowcharts, and their relationships."
+                        },
+                        {
+                            "type": "image_url",
+                            "image_url": {
+                                "url": base64_image
+                            }
+                        }
+                    ]
+                }
+            ]
+            
+            # Make API call with optimal parameters
+            completion = client.chat.completions.create(
+                model="llama-3.3-70b-versatile",
+                messages=messages,
+                temperature=0.2,
+                max_completion_tokens=2048,
+                top_p=0.85,
+                stream=False
+            )
+            
+            analysis = completion.choices[0].message.content
+            
+            page_info = f"Page {img_data.get('page', 'N/A')}"
+            if 'index' in img_data:
+                page_info += f", Image {img_data['index']}"
+            elif 'type' in img_data and img_data['type'] == 'full_page':
+                page_info += " (Full Page)"
+            
+            results.append(f"## 📄 {page_info}\n\n{analysis}\n\n---\n")
+        
+        if not results:
+            return "⚠️ No images found in the PDF document."
+        
+        return "\n".join(results)
+        
+    except Exception as e:
+        return f"❌ Error analyzing images: {str(e)}"
+
+def process_pdf_text_summary(api_key, pdf_file, progress=gr.Progress()):
+    """Process PDF and generate text summary"""
     try:
         if not api_key or not api_key.strip():
             return "❌ Please enter your Groq API key", "", ""
@@ -114,14 +258,78 @@ def process_pdf_and_summarize(api_key, pdf_file, progress=gr.Progress()):
         error_traceback = traceback.format_exc()
         return f"❌ Unexpected error: {str(e)}\n\nTraceback:\n{error_traceback}", "", ""
 
-def clear_all():
-    """Clear all fields"""
+def process_pdf_image_analysis(api_key, pdf_file, analysis_method, progress=gr.Progress()):
+    """Process PDF and analyze images"""
+    if pdf_file is None:
+        return "⚠️ Please upload a PDF file."
+    
+    if not api_key or api_key.strip() == "":
+        return "⚠️ Please enter your Groq API key."
+    
+    try:
+        progress(0.1, desc="Processing PDF file...")
+        
+        # Create temporary file for PDF processing
+        with tempfile.NamedTemporaryFile(delete=False, suffix='.pdf') as tmp_file:
+            # Write uploaded file content to temporary file
+            if hasattr(pdf_file, 'read'):
+                tmp_file.write(pdf_file.read())
+            else:
+                with open(pdf_file.name, 'rb') as f:
+                    tmp_file.write(f.read())
+            tmp_file_path = tmp_file.name
+        
+        progress(0.3, desc="Extracting images...")
+        
+        images_to_analyze = []
+        
+        if analysis_method == "Extract embedded images only":
+            # Extract only embedded images
+            images_to_analyze = extract_images_from_pdf(tmp_file_path)
+            if not images_to_analyze:
+                return "⚠️ No embedded images found in the PDF. Try 'Full page analysis' to analyze the entire content."
+        
+        elif analysis_method == "Full page analysis":
+            # Convert each page to image for comprehensive analysis
+            images_to_analyze = convert_pdf_pages_to_images(tmp_file_path)
+        
+        else:  # Both methods
+            # First try embedded images
+            embedded_images = extract_images_from_pdf(tmp_file_path)
+            # Then add full page analysis
+            page_images = convert_pdf_pages_to_images(tmp_file_path)
+            images_to_analyze = embedded_images + page_images
+        
+        # Clean up temporary file
+        os.unlink(tmp_file_path)
+        
+        if not images_to_analyze:
+            return "⚠️ No visual content found in the PDF document."
+        
+        progress(0.7, desc="Analyzing images with AI...")
+        
+        # Analyze images with Groq
+        analysis_result = analyze_images_with_groq(api_key, images_to_analyze)
+        
+        progress(1.0, desc="Analysis complete!")
+        
+        return analysis_result
+        
+    except Exception as e:
+        return f"❌ Error processing PDF: {str(e)}"
+
+def clear_all_text():
+    """Clear all text analysis fields"""
     return "", None, "", "", ""
 
+def clear_all_image():
+    """Clear all image analysis fields"""
+    return "", None, "Full page analysis", ""
+
 # Custom CSS for better styling
 css = """
 .gradio-container {
-    max-width: 1200px !important;
+    max-width: 1400px !important;
     margin: auto !important;
 }
 .main-header {
@@ -141,19 +349,51 @@ css = """
     border-left: 4px solid #007bff;
     margin: 1rem 0;
 }
+.feature-box {
+    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    padding: 1.5rem;
+    border-radius: 1rem;
+    color: white;
+    margin: 1rem 0;
+}
 """
 
 # Create Gradio interface
-with gr.Blocks(css=css, title="PDF Summarizer with Groq AI") as demo:
+with gr.Blocks(css=css, title="Advanced PDF Analyzer with Groq AI") as demo:
     # Header
     gr.HTML("""
     <div class="main-header">
-        <h1>📄 PDF Summarizer with Groq AI</h1>
-        <p>Upload any PDF document and get an AI-powered summary using Groq's Llama model</p>
+        <h1>🚀 Advanced PDF Analyzer with Groq AI</h1>
+        <p>Comprehensive PDF analysis tool - Extract text summaries and analyze images/diagrams using state-of-the-art AI models</p>
+    </div>
+    """)
+    
+    # Feature overview
+    gr.HTML("""
+    <div class="feature-box">
+        <h3>✨ What this tool can do:</h3>
+        <div style="display: grid; grid-template-columns: 1fr 1fr; gap: 1rem; margin-top: 1rem;">
+            <div>
+                <h4>📝 Text Analysis:</h4>
+                <ul>
+                    <li>Extract and summarize text content</li>
+                    <li>Generate comprehensive document summaries</li>
+                    <li>Maintain logical structure and key insights</li>
+                </ul>
+            </div>
+            <div>
+                <h4>🖼️ Visual Analysis:</h4>
+                <ul>
+                    <li>Analyze embedded images and diagrams</li>
+                    <li>Process flowcharts and technical drawings</li>
+                    <li>Extract text from images (OCR)</li>
+                </ul>
+            </div>
+        </div>
     </div>
     """)
     
-    # Info box
+    # API Key section
     gr.HTML("""
     <div class="info-box">
         <strong>🔑 How to get your Groq API Key:</strong><br>
@@ -165,62 +405,115 @@ with gr.Blocks(css=css, title="PDF Summarizer with Groq AI") as demo:
     </div>
     """)
     
-    with gr.Row():
-        with gr.Column(scale=1):
-            # Input section
-            gr.Markdown("## 🔧 Configuration")
-            api_key_input = gr.Textbox(
-                label="Groq API Key",
-                placeholder="Enter your Groq API key here...",
-                type="password"
-            )
-            gr.Markdown("*Your API key is not stored and only used for this session*")
-            
-            pdf_file_input = gr.File(
-                label="Upload PDF Document",
-                file_types=[".pdf"]
-            )
-            gr.Markdown("*Upload any PDF file to summarize*")
-            
+    # Global API key input
+    api_key_input = gr.Textbox(
+        label="🔑 Groq API Key",
+        placeholder="Enter your Groq API key here (used for both text and image analysis)...",
+        type="password",
+        info="Your API key is not stored and only used for this session"
+    )
+    
+    # Tabs for different functionalities
+    with gr.Tabs():
+        # Text Summary Tab
+        with gr.TabItem("📝 Text Summary", elem_id="text-tab"):
             with gr.Row():
-                summarize_btn = gr.Button("📋 Generate Summary", variant="primary", size="lg")
-                clear_btn = gr.Button("🗑️ Clear All", variant="secondary")
-        
-        with gr.Column(scale=2):
-            # Output section
-            gr.Markdown("## 📊 Results")
-            status_output = gr.Textbox(
-                label="Status",
-                interactive=False,
-                show_label=True
-            )
-            
-            with gr.Tabs():
-                with gr.TabItem("📝 Summary"):
-                    summary_output = gr.Textbox(
-                        label="AI Generated Summary",
-                        lines=15,
-                        interactive=False,
-                        placeholder="Your PDF summary will appear here..."
+                with gr.Column(scale=1):
+                    gr.Markdown("## 📄 Text Analysis")
+                    
+                    pdf_file_text = gr.File(
+                        label="Upload PDF Document",
+                        file_types=[".pdf"]
                     )
+                    
+                    with gr.Row():
+                        summarize_btn = gr.Button("📋 Generate Text Summary", variant="primary", size="lg")
+                        clear_text_btn = gr.Button("🗑️ Clear All", variant="secondary")
                 
-                with gr.TabItem("📄 Extracted Text Preview"):
-                    text_preview_output = gr.Textbox(
-                        label="Extracted Text (First 500 characters)",
-                        lines=10,
+                with gr.Column(scale=2):
+                    gr.Markdown("## 📊 Text Analysis Results")
+                    status_text_output = gr.Textbox(
+                        label="Status",
                         interactive=False,
-                        placeholder="Preview of extracted text will appear here..."
+                        show_label=True
+                    )
+                    
+                    with gr.Tabs():
+                        with gr.TabItem("📝 Summary"):
+                            summary_output = gr.Textbox(
+                                label="AI Generated Summary",
+                                lines=15,
+                                interactive=False,
+                                placeholder="Your PDF summary will appear here...",
+                                show_copy_button=True
+                            )
+                        
+                        with gr.TabItem("📄 Extracted Text Preview"):
+                            text_preview_output = gr.Textbox(
+                                label="Extracted Text (First 500 characters)",
+                                lines=10,
+                                interactive=False,
+                                placeholder="Preview of extracted text will appear here...",
+                                show_copy_button=True
+                            )
+        
+        # Image Analysis Tab
+        with gr.TabItem("🖼️ Image Analysis", elem_id="image-tab"):
+            with gr.Row():
+                with gr.Column(scale=1):
+                    gr.Markdown("## 🔍 Visual Analysis")
+                    
+                    pdf_file_image = gr.File(
+                        label="Upload PDF Document",
+                        file_types=[".pdf"]
+                    )
+                    
+                    analysis_method = gr.Radio(
+                        choices=[
+                            "Extract embedded images only",
+                            "Full page analysis", 
+                            "Both (embedded + full pages)"
+                        ],
+                        value="Full page analysis",
+                        label="Analysis Method",
+                        info="Choose how to analyze the PDF content"
+                    )
+                    
+                    with gr.Row():
+                        analyze_images_btn = gr.Button("🔍 Analyze Images", variant="primary", size="lg")
+                        clear_image_btn = gr.Button("🗑️ Clear All", variant="secondary")
+                
+                with gr.Column(scale=2):
+                    gr.Markdown("## 📊 Image Analysis Results")
+                    image_analysis_output = gr.Textbox(
+                        label="Visual Analysis Results",
+                        lines=20,
+                        max_lines=50,
+                        show_copy_button=True,
+                        placeholder="Image analysis results will appear here..."
                     )
     
     # Usage instructions
     gr.HTML("""
     <div class="info-box">
         <strong>📋 Usage Instructions:</strong><br>
-        1. Enter your Groq API key in the field above<br>
-        2. Upload a PDF document (any size, any content)<br>
-        3. Click "Generate Summary" to process your document<br>
-        4. View the AI-generated summary and extracted text preview<br>
-        5. Use "Clear All" to reset all fields
+        <h4>For Text Summary:</h4>
+        1. Enter your Groq API key above<br>
+        2. Go to "Text Summary" tab<br>
+        3. Upload a PDF document<br>
+        4. Click "Generate Text Summary"<br>
+        
+        <h4>For Image Analysis:</h4>
+        1. Enter your Groq API key above<br>
+        2. Go to "Image Analysis" tab<br>
+        3. Upload a PDF document<br>
+        4. Choose analysis method<br>
+        5. Click "Analyze Images"<br>
+        
+        <h4>Analysis Methods:</h4>
+        • <strong>Extract embedded images:</strong> Analyzes only images embedded within the PDF<br>
+        • <strong>Full page analysis:</strong> Converts each page to image for comprehensive analysis (recommended)<br>
+        • <strong>Both:</strong> Combines both methods for maximum coverage
     </div>
     """)
     
@@ -228,31 +521,44 @@ with gr.Blocks(css=css, title="PDF Summarizer with Groq AI") as demo:
     gr.HTML("""
     <div style="margin-top: 2rem; padding: 1rem; background-color: #e9ecef; border-radius: 0.5rem;">
         <strong>🤖 Model Information:</strong><br>
-        • Model: Llama-3.3-70B-Versatile (via Groq)<br>
-        • Temperature: 0.3 (focused, factual summaries)<br>
-        • Max Tokens: 2048 (comprehensive summaries)<br>
-        • Top-p: 0.9 (balanced creativity and accuracy)
+        • <strong>Text Analysis:</strong> Llama-3.3-70B-Versatile (optimized for summarization)<br>
+        • <strong>Image Analysis:</strong> Llama-3.3-70B-Versatile (with vision capabilities)<br>
+        • <strong>Temperature:</strong> 0.2-0.3 (focused, factual analysis)<br>
+        • <strong>Max Tokens:</strong> 2048 (comprehensive outputs)
     </div>
     """)
     
-    # Event handlers
+    # Event handlers for text summary
     summarize_btn.click(
-        fn=process_pdf_and_summarize,
-        inputs=[api_key_input, pdf_file_input],
-        outputs=[status_output, text_preview_output, summary_output],
+        fn=process_pdf_text_summary,
+        inputs=[api_key_input, pdf_file_text],
+        outputs=[status_text_output, text_preview_output, summary_output],
+        show_progress=True
+    )
+    
+    clear_text_btn.click(
+        fn=clear_all_text,
+        outputs=[api_key_input, pdf_file_text, status_text_output, text_preview_output, summary_output]
+    )
+    
+    # Event handlers for image analysis
+    analyze_images_btn.click(
+        fn=process_pdf_image_analysis,
+        inputs=[api_key_input, pdf_file_image, analysis_method],
+        outputs=[image_analysis_output],
         show_progress=True
     )
     
-    clear_btn.click(
-        fn=clear_all,
-        outputs=[api_key_input, pdf_file_input, status_output, text_preview_output, summary_output]
+    clear_image_btn.click(
+        fn=clear_all_image,
+        outputs=[api_key_input, pdf_file_image, analysis_method, image_analysis_output]
     )
 
 # Launch the app
 if __name__ == "__main__":
-    print("🚀 Starting PDF Summarizer App...")
+    print("🚀 Starting Advanced PDF Analyzer App...")
     print("📋 Make sure you have the required packages installed:")
-    print("   pip install gradio groq PyPDF2")
+    print("   pip install -r requirements.txt")
     print("\n🔑 Don't forget to get your Groq API key from: https://console.groq.com/")
     
     demo.launch(