Update app.py

app.py

@@ -2,99 +2,217 @@ import gradio as gr
 import requests
 from urllib.parse import parse_qs, urlparse
 from transformers import pipeline
+import os
+import time
 
 # Backend API endpoints
-BACKEND_BASE_URL = "http://127.0.0.1:5000"  # Replace with your Flask backend URL
+BACKEND_BASE_URL = "http://127.0.0.1:5000"
 COMPLETE_APPOINTMENT_ENDPOINT = f"{BACKEND_BASE_URL}/complete_appointment"
+VALIDATE_APPOINTMENT_ENDPOINT = f"{BACKEND_BASE_URL}/validate_appointment"
+
+# Path to your pre-generated avatar video
+AVATAR_VIDEO_PATH = "http://127.0.0.1:5000/serve_video/ai_doctor_avatar.mp4"  # Replace with your actual video URL
+ # <source src="http://127.0.0.1:5000/serve_video/ai_doctor_avatar.mp4" type="video/mp4">
+
+# Load lightweight models for faster performance
+conversation_pipeline = pipeline(
+    "text-generation",
+    model="microsoft/DialoGPT-medium",
+    device="cpu"  # Change to "cuda" if GPU available
+)
+
+medical_qa_pipeline = pipeline(
+    "question-answering",
+    model="medicalai/ClinicalBERT_QA",
+    device="cpu"
+)
 
-# Function to extract `appointment_id` from query parameters
 def get_appointment_id():
     query_params = parse_qs(urlparse(gr.Request().url).query)
     return query_params.get("appointment_id", ["Unknown"])[0]
 
-# Load Hugging Face pipelines
-ner_pipeline = pipeline("ner", model="dslim/bert-base-NER")  # Named Entity Recognition
-sentiment_pipeline = pipeline("sentiment-analysis")  # Sentiment Analysis
-summarization_pipeline = pipeline("summarization", model="facebook/bart-large-cnn")  # Summarization
-conversation_pipeline = pipeline("text-generation", model="microsoft/DialoGPT-medium")  # Conversational AI
+def validate_appointment(appointment_id):
+    try:
+        response = requests.post(
+            VALIDATE_APPOINTMENT_ENDPOINT,
+            json={"appointment_id": appointment_id},
+            timeout=3
+        )
+        return response.json().get("status") == "success"
+    except Exception:
+        return False
 
-# Function to process user input
-def ai_doctor_interaction(user_input, chat_history=[]):
+def generate_response(user_input):
+    """Lightning-fast response generation"""
     try:
-        # Perform NER
-        ner_results = ner_pipeline(user_input)
-        entities = [f"{entity['entity_group']}: {entity['word']}" for entity in ner_results]
-
-        # Perform Sentiment Analysis
-        sentiment_results = sentiment_pipeline(user_input)
-        sentiment = f"{sentiment_results[0]['label']} (Confidence: {sentiment_results[0]['score']:.2f})"
-
-        # Perform Summarization
-        summary = summarization_pipeline(user_input, max_length=50, min_length=10, do_sample=False)[0]['summary_text']
-
-        # Generate AI Doctor response
-        ai_response = conversation_pipeline(user_input, max_length=100, num_return_sequences=1)[0]["generated_text"]
-
-        # Combine results
-        response = f"""
-        **Entities Identified**: {', '.join(entities)}  
-        **Sentiment**: {sentiment}  
-        **Summary**: {summary}  
-        **AI Doctor Response**: {ai_response}
-        """
-        chat_history.append(("You", user_input))
-        chat_history.append(("AI Doctor", response))
-        return chat_history, chat_history
-    except Exception as e:
-        return chat_history + [("System", f"Error: {str(e)}")], chat_history
-
-# Function to handle "End Call" button
-def end_call():
-    appointment_id = get_appointment_id()
-    if appointment_id == "Unknown":
-        return "Error: Appointment ID not found. Please try again."
-
-    # Send a request to the backend to mark the appointment as completed
+        # Try medical QA first
+        qa_result = medical_qa_pipeline({
+            "question": user_input,
+            "context": "Medical consultation between doctor and patient"
+        })
+        if qa_result['score'] > 0.25:
+            return qa_result['answer']
+        
+        # Fallback to conversational AI
+        return conversation_pipeline(
+            f"Patient: {user_input}\nDoctor:",
+            max_length=150,
+            num_return_sequences=1,
+            do_sample=True,
+            top_p=0.9,
+            temperature=0.7
+        )[0]["generated_text"].split("Doctor:")[-1].strip()
+    
+    except Exception:
+        return "I'm experiencing high demand. Could you please repeat your question?"
+
+def end_call(appointment_id):
     try:
-        response = requests.post(COMPLETE_APPOINTMENT_ENDPOINT, json={"appointment_id": appointment_id})
+        response = requests.post(
+            COMPLETE_APPOINTMENT_ENDPOINT,
+            json={"appointment_id": appointment_id},
+            timeout=3
+        )
         if response.status_code == 200:
-            return "Appointment marked as completed. Redirecting to the doctors' page..."
-        else:
-            return f"Error: {response.json().get('message', 'Failed to complete the appointment')}"
-    except Exception as e:
-        return f"Error: Unable to connect to the backend. Details: {str(e)}"
-
-# Gradio interface
-with gr.Blocks() as demo:
-    gr.Markdown("<h1 style='text-align: center; color: #4CAF50;'>AI Doctor Avatar</h1>")
-    gr.Markdown("<p style='text-align: center;'>Interact with the AI Doctor to resolve your health-related queries.</p>")
+            return "Consultation completed successfully. Thank you!"
+        return "Couldn't complete appointment. Please contact support."
+    except Exception:
+        return "Network error. Please check your connection."
+
+# Custom CSS for blazing fast UI
+custom_css = """
+:root {
+    --primary: #2d8cf0;
+    --secondary: #f8f9fa;
+    --accent: #ff6b6b;
+}
+
+.gradio-container {
+    font-family: 'Inter', sans-serif;
+    max-width: 1200px !important;
+}
+
+.avatar-container {
+    aspect-ratio: 9/16;
+    background: black;
+    border-radius: 12px;
+    overflow: hidden;
+}
+
+.video-container video {
+    object-fit: cover;
+    width: 100%;
+    height: 100%;
+}
+
+.chat-container {
+    height: 100%;
+    display: flex;
+    flex-direction: column;
+}
+
+.chatbot {
+    min-height: 500px;
+    flex-grow: 1;
+    border-radius: 12px;
+    background: var(--secondary);
+}
+
+.input-row {
+    margin-top: 0.5rem !important;
+}
+
+.primary-btn {
+    background: var(--primary) !important;
+}
+
+.end-btn {
+    background: var(--accent) !important;
+}
+
+/* Animation for smooth loading */
+@keyframes fadeIn {
+    from { opacity: 0; }
+    to { opacity: 1; }
+}
+
+.gradio-app {
+    animation: fadeIn 0.3s ease-in;
+}
+"""
+
+with gr.Blocks(css=custom_css, title="AI Doctor Consultation") as demo:
+    appointment_id = gr.State(value=get_appointment_id())
+    
+    # Header
+    gr.Markdown("""
+    <div style="text-align: center; margin-bottom: 1rem;">
+        <h1 style="margin: 0; color: #2d8cf0;">AI Doctor Consultation</h1>
+        <p style="margin: 0; color: #666;">Your health matters to us</p>
+    </div>
+    """)
     
-    with gr.Row():
-        with gr.Column(scale=1):
-            gr.Markdown("<h3 style='text-align: center;'>AI Doctor Avatar</h3>")
-            gr.HTML("""
-                <div style="position: relative; overflow: hidden; aspect-ratio: 16/9;">
-                    <video autoplay muted loop preload="none" style="width: 100%; height: auto; border-radius: 10px;">
-                        <source src="http://127.0.0.1:5000/serve_video/ai_doctor_avatar.mp4" type="video/mp4">
-                        Your browser does not support the video tag.
-                    </video>
-                </div>
-            """)
-        with gr.Column(scale=2):
-            chatbot = gr.Chatbot(label="AI Doctor Chat Interface", height=400)
-            user_input = gr.Textbox(label="Your Message", placeholder="Type your question here...", lines=2)
+    with gr.Row(equal_height=True):
+        # Left column - Avatar video
+        with gr.Column(scale=1, elem_classes=["avatar-container"]):
+            gr.Markdown("### Dr. AI Avatar")
+            video = gr.Video(
+                value=AVATAR_VIDEO_PATH,
+                autoplay=True,
+                interactive=False,
+                elem_classes=["video-container"]
+            )
+            
+        # Right column - Chat interface
+        with gr.Column(scale=2, elem_classes=["chat-container"]):
+            chatbot = gr.Chatbot(
+                label="Consultation Chat",
+                bubble_full_width=False,
+                show_copy_button=True,
+                avatar_images=(
+                    "https://i.imgur.com/8Km9tLL.png",  # User
+                    "https://i.imgur.com/3Q3ZQ2u.png"   # Doctor
+                )
+            )
+            
+            with gr.Row(elem_classes=["input-row"]):
+                user_input = gr.Textbox(
+                    placeholder="Describe your symptoms...",
+                    label="",
+                    container=False,
+                    autofocus=True,
+                    max_lines=3
+                )
+                submit_btn = gr.Button("Send", variant="primary", elem_classes=["primary-btn"])
+            
             with gr.Row():
-                submit_button = gr.Button("Send", variant="primary")
-                clear_button = gr.Button("Clear Chat", variant="secondary")
+                clear_btn = gr.Button("Clear Chat", variant="secondary")
+                end_btn = gr.Button("End Consultation", variant="stop", elem_classes=["end-btn"])
+                status = gr.Textbox(visible=False)
     
-    # Define interactions
-    submit_button.click(ai_doctor_interaction, inputs=[user_input, chatbot], outputs=[chatbot, chatbot])
-    clear_button.click(lambda: [], inputs=[], outputs=[chatbot])
-
-    # Add "End Call" button
-    end_call_button = gr.Button("End Call", variant="stop")
-    end_call_message = gr.Textbox(label="End Call Status", interactive=False)
-    end_call_button.click(end_call, outputs=[end_call_message])
+    # Event handlers
+    submit_btn.click(
+        fn=lambda msg, hist: (msg, hist + [(msg, generate_response(msg))]),
+        inputs=[user_input, chatbot],
+        outputs=[user_input, chatbot],
+        queue=True
+    ).then(
+        lambda: gr.update(autoplay=True),  # Ensure video keeps playing
+        outputs=video
+    )
+    
+    clear_btn.click(lambda: [], None, chatbot)
+    end_btn.click(
+        fn=end_call,
+        inputs=appointment_id,
+        outputs=status
+    )
 
-# Launch the Gradio app
-demo.launch()
+# Optimized launch settings
+demo.launch(
+    server_name="0.0.0.0",
+    server_port=7860,
+    share=False,
+    favicon_path="https://i.imgur.com/3Q3ZQ2u.png",
+    prevent_thread_lock=True
+)