api update

app.py

@@ -12,8 +12,9 @@ import time
 from enum import Enum
 import gradio as gr
 from huggingface_hub import InferenceClient
+from gradio_client import Client, handle_file  # <-- REQUIRED IMPORT
 
-# Environment variables
+# --- Environment variables ---
 HF_TOKEN = os.getenv("HF_TOKEN")  # Required for Llama model access
 MODEL_NAME = "meta-llama/Llama-3.1-8B-Instruct"
 MAX_PROMPT_LENGTH = int(os.getenv("MAX_PROMPT_LENGTH", "2000"))
@@ -21,7 +22,7 @@ RECOMMENDATION_TIMEOUT = int(os.getenv("RECOMMENDATION_TIMEOUT", "60"))
 MAX_RETRIES = int(os.getenv("MAX_RETRIES", "2"))
 ENABLE_CACHING = os.getenv("ENABLE_CACHING", "true").lower() == "true"
 
-# Logging setup
+# --- Logging setup ---
 logging.basicConfig(
     level=logging.INFO,
     format='%(asctime)s [%(levelname)s] %(name)s: %(message)s',
@@ -29,7 +30,7 @@ logging.basicConfig(
 )
 logger = logging.getLogger("hindi_emotion_recommendation")
 
-# Initialize HuggingFace Inference Client
+# --- Initialize HuggingFace Inference Client (for Llama 3.1) ---
 try:
     client = InferenceClient(token=HF_TOKEN)
     logger.info(f"✓ HuggingFace client initialized with model: {MODEL_NAME}")
@@ -37,11 +38,24 @@ except Exception as e:
     logger.error(f"Failed to initialize HuggingFace client: {str(e)}")
     client = None
 
-# Cache for recommendations
+# --- Initialize Audio Analysis Client (for your Emotion API) ---
+# This client points to your *other* Hugging Face Space
+try:
+    audio_analysis_client = Client("JustNikunj/Sentimental_Analysis")
+    logger.info("✓ Audio Analysis API client initialized for 'JustNikunj/Sentimental_Analysis'")
+except Exception as e:
+    logger.error(f"Failed to initialize Audio Analysis API client: {str(e)}")
+    audio_analysis_client = None
+
+# --- Cache for recommendations ---
 recommendation_cache = {}
 CACHE_TTL_SECONDS = 3600
 
-# Response validation enums
+# =========================================================================
+# === START: ORIGINAL HELPER FUNCTIONS (FROM YOUR MAIN CODE) ===
+# =========================================================================
+
+# --- Response validation enums ---
 class ValidationStatus(str, Enum):
     VALID = "valid"
     WARNING = "warning"
@@ -162,6 +176,7 @@ class ResponseValidator:
         
         return recommendation
 
+# --- Caching functions ---
 def get_cache_key(emotion_result: dict) -> str:
     """Generate cache key from emotion result"""
     cache_data = {
@@ -190,6 +205,7 @@ def save_to_cache(cache_key: str, data: Dict[str, Any]):
     if ENABLE_CACHING:
         recommendation_cache[cache_key] = (data, time.time())
 
+# --- LLM Prompting functions ---
 @lru_cache(maxsize=1)
 def load_few_shot_examples() -> str:
     """Load few-shot examples"""
@@ -345,6 +361,15 @@ def get_fallback_recommendation(emotion_result: dict) -> str:
     
     return "यदि आपको सहायता चाहिए तो किसी विश्वसनीय व्यक्ति से संपर्क करें। आपकी भावनाएं महत्वपूर्ण हैं।"
 
+# =========================================================================
+# === END: ORIGINAL HELPER FUNCTIONS ===
+# =========================================================================
+
+
+# --- CORE LOGIC: Function to process emotion data and get LLM response ---
+# This is your original 'process_emotion_analysis' function.
+# It is now called by the new 'get_recommendation_from_audio' function.
+
 def process_emotion_analysis(
     transcript: str,
     sentiment: str,
@@ -355,7 +380,7 @@ def process_emotion_analysis(
     is_mental_health: bool,
     is_grief_loss: bool,
     is_relationship: bool
-) -> Tuple[str, str, str, str, str]:
+) -> Tuple[str, str, str, str, str]: # Returns 5 values for the 5 Gradio outputs
     """Process emotion analysis and generate recommendation"""
     
     start_time = time.time()
@@ -377,7 +402,7 @@ def process_emotion_analysis(
                 'is_relationship_distress': is_relationship
             }
         },
-        'prosodic_features': {}
+        'prosodic_features': {} # Prosodic features are not used in the prompt
     }
     
     # Check cache
@@ -392,6 +417,7 @@ def process_emotion_analysis(
         cached = True
     else:
         # Generate new recommendation
+        logger.info("Generating new LLM recommendation...")
         action = get_llama_recommendation(emotion_result)
         
         # Validate the recommendation
@@ -407,6 +433,7 @@ def process_emotion_analysis(
             if action != original_action:
                 enhanced = True
                 logger.info("Recommendation auto-enhanced")
+                # Re-validate after enhancement
                 validation_result = ResponseValidator.validate_recommendation(action, emotion_result)
         
         # Cache the result
@@ -441,12 +468,95 @@ def process_emotion_analysis(
 **Processing Time:** {processing_time}ms
 **Cached:** {'Yes ♻️' if cached else 'No 🆕'}
 **Enhanced:** {'Yes 🔧' if enhanced else 'No'}
-**Confidence:** {confidence:.2%}
 """
     
-    return action, risk_level, validation_info, metadata
+    # This string will show the raw API data in the UI
+    api_debug_info = f"""
+**Transcript:** {transcript}
+**Emotion:** {primary_emotion} (Conf: {confidence:.1%})
+**Sentiment:** {sentiment}
+**Crisis:** {is_crisis}
+**Mental Health:** {is_mental_health}
+"""
+    
+    # Return 5 values to match the Gradio output components
+    return action, risk_level, validation_info, metadata, api_debug_info
+
+
+# --- NEW: MASTER FUNCTION TO CONNECT API AND LLM ---
+def get_recommendation_from_audio(audio_filepath: str) -> Tuple[str, str, str, str, str]:
+    """
+    This is the new master function for the Gradio interface.
+    1. Takes audio input from Gradio.
+    2. Calls the 'JustNikunj/Sentimental_Analysis' API.
+    3. Parses the result.
+    4. Calls 'process_emotion_analysis' to get the LLM recommendation.
+    """
+    if not audio_analysis_client:
+        error_msg = "Audio Analysis API client ('JustNikunj/Sentimental_Analysis') is not initialized."
+        logger.error(error_msg)
+        return "Error: " + error_msg, "🔴 ERROR", error_msg, "", ""
+        
+    if not audio_filepath:
+        logger.warning("No audio file provided.")
+        return "No audio provided. Please record or upload.", "⚪️ N/A", "No input", "", ""
+
+    logger.info(f"Processing audio file: {audio_filepath}")
+    
+    try:
+        # --- STEP 1: Call Audio Analysis API ---
+        logger.info("Calling audio analysis API...")
+        api_result = audio_analysis_client.predict(
+            audio_filepath=handle_file(audio_filepath),
+            api_name="/predict"
+        )
+        logger.info("✓ Audio analysis complete.")
+
+        # --- STEP 2: Check for API errors ---
+        if not isinstance(api_result, dict) or api_result.get('status') != 'success':
+            error_msg = api_result.get('message', 'Unknown API error')
+            logger.error(f"Audio API failed: {error_msg}")
+            # Show the raw error in the "API Debug" box
+            return f"Error from Audio API: {error_msg}", "🔴 ERROR", error_msg, "", json.dumps(api_result, indent=2, ensure_ascii=False)
+
+        # --- STEP 3: Parse the successful API result ---
+        transcript = api_result['transcription']
+        sentiment = api_result['sentiment']['dominant']
+        primary_emotion = api_result['emotion']['primary']
+        secondary_emotion = api_result['emotion'].get('secondary')
+        confidence = float(api_result['emotion']['confidence'])
+        
+        situations = api_result['analysis']['situations']
+        is_crisis = situations['is_crisis']
+        is_mental_health = situations['is_mental_health_distress']
+        is_grief_loss = situations['is_grief_loss']
+        is_relationship = situations['is_relationship_distress']
+        
+        logger.info(f"✓ API Success. Transcript: '{transcript}', Emotion: {primary_emotion}")
+
+        # --- STEP 4: Call the LLM Recommendation Logic ---
+        # This calls your original function with the data we just got
+        return process_emotion_analysis(
+            transcript=transcript,
+            sentiment=sentiment,
+            primary_emotion=primary_emotion,
+            secondary_emotion=secondary_emotion,
+            confidence=confidence,
+            is_crisis=is_crisis,
+            is_mental_health=is_mental_health,
+            is_grief_loss=is_grief_loss,
+            is_relationship=is_relationship
+        )
+
+    except Exception as e:
+        logger.error(f"Error in 'get_recommendation_from_audio': {str(e)}")
+        import traceback
+        tb_str = traceback.format_exc()
+        logger.error(tb_str)
+        return f"A system error occurred: {str(e)}", "🔴 ERROR", str(e), "", tb_str
+
 
-# Gradio Interface
+# --- MODIFIED: Gradio Interface ---
 def create_interface():
     """Create Gradio interface"""
     
@@ -461,58 +571,33 @@ def create_interface():
         AI-powered emotional analysis and action recommendations for Indian women's support.
         Powered by **Meta Llama 3.1 8B Instruct** with intelligent validation and enhancement.
         
-        ### Important Helplines:
-        - 🚨 **Emergency/Police:** 112
-        - 👩 **Women's Helpline:** 181, 1091
-        - 🧠 **Mental Health:** 9152987821 (Vandrevala), 08046110007 (NIMHANS)
-        - 💙 **Suicide Prevention:** 9820466726 (AASRA)
+        This app **calls the 'JustNikunj/Sentimental_Analysis' API** to get emotion data from your audio.
         """)
         
         with gr.Row():
             with gr.Column(scale=1):
-                gr.Markdown("### 📝 Input Emotion Analysis")
-                
-                transcript = gr.Textbox(
-                    label="Transcript (Hindi/English)",
-                    placeholder="मुझे बहुत डर लग रहा है...",
-                    lines=3
-                )
+                gr.Markdown("### 🎙️ Input Audio")
                 
-                sentiment = gr.Dropdown(
-                    label="Sentiment",
-                    choices=["Positive", "Negative", "Neutral"],
-                    value="Negative"
+                # --- THIS IS THE NEW INPUT ---
+                audio_input = gr.Audio(
+                    label="Record or Upload Audio",
+                    sources=["microphone", "upload"],
+                    type="filepath" # Send the file path to the function
                 )
                 
-                with gr.Row():
-                    primary_emotion = gr.Dropdown(
-                        label="Primary Emotion",
-                        choices=["fear", "sadness", "anger", "joy", "surprise", "disgust", "neutral", "despair", "anxiety"],
-                        value="sadness"
-                    )
-                    
-                    secondary_emotion = gr.Dropdown(
-                        label="Secondary Emotion",
-                        choices=["", "distress", "frustration", "hopelessness", "worry", "relief"],
-                        value=""
-                    )
-                
-                confidence = gr.Slider(
-                    label="Confidence Score",
-                    minimum=0.0,
-                    maximum=1.0,
-                    value=0.8,
-                    step=0.05
-                )
-                
-                gr.Markdown("### 🎯 Situation Flags")
+                submit_btn = gr.Button("Generate Recommendation 🚀", variant="primary", size="lg")
                 
-                is_crisis = gr.Checkbox(label="🚨 Crisis Situation", value=False)
-                is_mental_health = gr.Checkbox(label="🧠 Mental Health Distress", value=False)
-                is_grief_loss = gr.Checkbox(label="💔 Grief/Loss", value=False)
-                is_relationship = gr.Checkbox(label="👥 Relationship Distress", value=False)
+                gr.Markdown("### ℹ️ About This App")
+                gr.Markdown("""
+                This system uses:
+                - **'JustNikunj/Sentimental_Analysis'** for audio-to-emotion analysis.
+                - **Meta Llama 3.1 8B Instruct** for generating empathetic recommendations.
+                - **Multi-criteria validation** to ensure quality and safety.
+                - **Auto-enhancement** to add missing critical information.
+                - **Caching** for faster repeated queries.
                 
-                submit_btn = gr.Button("Generate Recommendation 🚀", variant="primary", size="lg")
+                **Note:** This is an AI assistant. In emergencies, always call emergency services immediately.
+                """)
             
             with gr.Column(scale=1):
                 gr.Markdown("### 💡 Recommendation Output")
@@ -535,54 +620,42 @@ def create_interface():
                 metadata = gr.Markdown(
                     label="Metadata"
                 )
+                
+                # Added an output to show the data from the audio API
+                api_debug = gr.Markdown(
+                    label="Raw Analysis from Audio API"
+                )
         
-        # Example inputs
-        gr.Markdown("### 📚 Example Inputs")
-        gr.Examples(
-            examples=[
-                ["मुझे बचाओ! कोई मुझे मार रहा है।", "Negative", "fear", "distress", 0.95, True, False, False, False],
-                ["मैं बहुत अकेला और उदास महसूस कर रहा हूँ।", "Negative", "sadness", "neutral", 0.78, False, True, False, False],
-                ["मेरी पत्नी ने मुझे छोड़ दिया है।", "Negative", "sadness", "distress", 0.82, False, False, False, True],
-                ["मैं अपने जीवन को खत्म करना चाहती हूं।", "Negative", "despair", "hopelessness", 0.92, True, True, False, False],
-                ["आज मौसम बहुत अच्छा है।", "Positive", "joy", "", 0.85, False, False, False, False],
-            ],
-            inputs=[transcript, sentiment, primary_emotion, secondary_emotion, confidence, is_crisis, is_mental_health, is_grief_loss, is_relationship],
-        )
-        
-        # Connect button
+        # Connect button to the new master function
         submit_btn.click(
-            fn=process_emotion_analysis,
-            inputs=[
-                transcript, sentiment, primary_emotion, secondary_emotion, 
-                confidence, is_crisis, is_mental_health, is_grief_loss, is_relationship
-            ],
-            outputs=[recommendation, risk_level, validation, metadata]
+            fn=get_recommendation_from_audio, # <--- NEW FUNCTION
+            inputs=[audio_input],              # <--- NEW INPUT
+            outputs=[recommendation, risk_level, validation, metadata, api_debug] # <--- 5 OUTPUTS
         )
         
+        gr.Markdown("### Important Helplines")
         gr.Markdown("""
-        ---
-        ### ℹ️ About
-        
-        This system uses:
-        - **Meta Llama 3.1 8B Instruct** for generating empathetic recommendations
-        - **Multi-criteria validation** to ensure quality and safety
-        - **Auto-enhancement** to add missing critical information
-        - **Caching** for faster repeated queries
-        
-        **Note:** This is an AI assistant. In emergencies, always call emergency services immediately.
+        - 🚨 **Emergency/Police:** 112
+        - 👩 **Women's Helpline:** 181, 1091
+        - 🧠 **Mental Health:** 9152987821 (Vandrevala), 08046110007 (NIMHANS)
+        - 💙 **Suicide Prevention:** 9820466726 (AASRA)
         """)
     
     return demo
 
-# Launch the app
+# --- Launch the app ---
 if __name__ == "__main__":
+    if not HF_TOKEN:
+        logger.warning("HF_TOKEN environment variable is not set. Llama 3.1 calls may fail.")
+    if not audio_analysis_client:
+        logger.warning("Could not connect to 'JustNikunj/Sentimental_Analysis'. The app will not work.")
+        
     logger.info("Starting Gradio interface...")
-    logger.info(f"Model: {MODEL_NAME}")
-    logger.info(f"HF Token Available: {'Yes' if HF_TOKEN else 'No'}")
+    logger.info(f"Recommendation Model: {MODEL_NAME}")
     
     demo = create_interface()
     demo.launch(
         server_name="0.0.0.0",
         server_port=7860,
-        share=False
+        share=False # Set to True if you want a public link from your local machine
     )