Update app.py

app.py

@@ -9,8 +9,9 @@ import torch
 from sentence_transformers import CrossEncoder
 import gradio as gr
 from datetime import datetime 
-from huggingface_hub import hf_hub_download, HfApi
+from huggingface_hub import hf_hub_download, HfApi, CommitOperationAdd
 from pathlib import Path
+import tempfile
 
 # Load environment variables and initialize clients
 load_dotenv()
@@ -165,10 +166,7 @@ def get_context(message):
     return context
 
 def log_conversation(timestamp, user_message, assistant_response, model_name, context, error=None):
-    """Log conversation details to a JSON file"""
-    log_dir = Path("logs")
-    log_dir.mkdir(exist_ok=True)
-    
+    """Log conversation details to JSON file - local directory or HuggingFace Dataset repository"""
     # Create a log entry
     log_entry = {
         "timestamp": timestamp,
@@ -179,28 +177,82 @@ def log_conversation(timestamp, user_message, assistant_response, model_name, co
         "error": str(error) if error else None
     }
     
-    # Use the current date for the log file name
+    # Check if running on Hugging Face Spaces
+    is_hf_space = os.getenv('SPACE_ID') is not None
     current_date = datetime.now().strftime("%Y-%m-%d")
-    log_file = log_dir / f"conversation_log_{current_date}.json"
     
-    try:
-        # Load existing logs if file exists
-        if log_file.exists():
-            with open(log_file, 'r', encoding='utf-8') as f:
-                logs = json.load(f)
-        else:
-            logs = []
-        
-        # Append new log entry
-        logs.append(log_entry)
-        
-        # Write updated logs
-        with open(log_file, 'w', encoding='utf-8') as f:
-            json.dump(logs, f, ensure_ascii=False, indent=2)
+    if is_hf_space:
+        try:
+            # Initialize Hugging Face API
+            api = HfApi(token=hf_token)
+            filename = f"conversation_logs/daily_{current_date}.json"
             
-        print(f"\n✍️ Logged conversation to {log_file}")
-    except Exception as e:
-        print(f"\n⚠️ Error logging conversation: {str(e)}")
+            try:
+                # Try to download existing file
+                existing_file = api.hf_hub_download(
+                    repo_id="Mr-Geo/bas_chat_logs",
+                    filename=filename,
+                    repo_type="dataset",
+                    token=hf_token
+                )
+                # Load existing logs
+                with open(existing_file, 'r', encoding='utf-8') as f:
+                    logs = json.load(f)
+            except Exception:
+                # File doesn't exist yet, start with empty list
+                logs = []
+            
+            # Append new log entry
+            logs.append(log_entry)
+            
+            # Create temporary file with updated logs
+            with tempfile.NamedTemporaryFile(mode='w', encoding='utf-8', delete=False, suffix='.json') as temp_file:
+                json.dump(logs, temp_file, ensure_ascii=False, indent=2)
+                temp_file_path = temp_file.name
+            
+            # Push to the dataset repository
+            api.create_commit(
+                repo_id="Mr-Geo/bas_chat_logs",
+                repo_type="dataset",
+                operations=[
+                    CommitOperationAdd(
+                        path_in_repo=filename,
+                        path_or_fileobj=temp_file_path,
+                        commit_message=f"Update conversation logs for {current_date}"
+                    )
+                ],
+                commit_message=f"Update conversation logs for {current_date}"
+            )
+            
+            # Clean up temporary file
+            os.unlink(temp_file_path)
+            
+        except Exception as e:
+            print(f"\n⚠️ Error logging conversation to HuggingFace: {str(e)}")
+    else:
+        # Local environment - save to file
+        try:
+            log_dir = Path("logs")
+            log_dir.mkdir(exist_ok=True)
+            
+            log_file = log_dir / f"conversation_log_{current_date}.json"
+            
+            # Load existing logs if file exists
+            if log_file.exists():
+                with open(log_file, 'r', encoding='utf-8') as f:
+                    logs = json.load(f)
+            else:
+                logs = []
+            
+            # Append new log entry
+            logs.append(log_entry)
+            
+            # Write updated logs
+            with open(log_file, 'w', encoding='utf-8') as f:
+                json.dump(logs, f, ensure_ascii=False, indent=2)
+                
+        except Exception as e:
+            print(f"\n⚠️ Error logging conversation locally: {str(e)}")
 
 def chat_response(message, history, model_name):
     """Chat response function for Gradio interface"""
@@ -241,7 +293,7 @@ Context: {context}"""
         # Add history and current message
         if history:
             for h in history:
-                messages.append({"role": "user", "content": str(h[0])})
+                messages.append({"role": "user", "content": f"{str(h[0])} at BAS"})
                 if h[1]:  # If there's a response
                     messages.append({"role": "assistant", "content": str(h[1])})
         
@@ -279,7 +331,7 @@ Context: {context}"""
                         final_response = f"""<details>
 <summary>🤔 <u>Click to see 'thinking' process</u></summary>
 <div style="font-size: 0.9em;">
-<i>💭{thinking_process}💭</i>
+<i>💭{thinking_process}</i>
 </div>
 <hr style="margin: 0; height: 2px;">
 </details>
@@ -294,14 +346,12 @@ Context: {context}"""
                     final_response += content
                     yield final_response
         
-        
         log_conversation(timestamp, message, final_response, model_name, context)
         print("\n=== LLM Response End ===\n")
 
     except Exception as e:
         error_msg = f"An error occurred: {str(e)}"
         print(f"\nERROR: {error_msg}")
-        
         log_conversation(datetime.now().strftime("%Y-%m-%d %H:%M:%S"), 
                        message, error_msg, model_name, context, error=e)
         yield error_msg
@@ -325,7 +375,7 @@ if __name__ == "__main__":
         
         with demo:
             gr.Markdown("# 🌍❄️British Antarctic Survey Website Chat Assistant 🧊🤖")
-            gr.Markdown("Accesses text data from 11,982 unique BAS URLs (6GB [Vector Database](https://huggingface.co/datasets/Mr-Geo/chroma_db/tree/main/) 📚 extracted 02/02/2025) | Created with open source technologies: [Gradio](https://gradio.app) for the interface 🎨, [Groq](https://groq.com) for LLM processing ⚡, and [Chroma](https://www.trychroma.com/) as the vector database 💻🔍")
+            gr.Markdown("Accesses text data from 11,982 unique BAS URLs (6GB [Vector Database](https://huggingface.co/datasets/Mr-Geo/chroma_db/tree/main/) 📚 extracted 02/02/2025) Created with open source technologies: [Gradio](https://gradio.app) for the interface 🎨, [Groq](https://groq.com) for LLM processing ⚡, and [Chroma](https://www.trychroma.com/) as the vector database 💻")
             model_selector = gr.Dropdown(
                 choices=[
                     "llama-3.1-8b-instant",
@@ -336,7 +386,7 @@ if __name__ == "__main__":
                 ],
                 value="llama-3.1-8b-instant",
                 label="Select AI Large Language Model 🤖",
-                info="Choose which AI model to use for responses - Select 'deepseek' for the 'thinking' AI search 💭"
+                info="Choose which AI model to use for responses (all models running on [GroqCloud](https://groq.com/groqrack/)"
             )
             
             chatbot = gr.Chatbot(height=600)