Update app.py

app.py

@@ -10,8 +10,7 @@ import pandas as pd
 from langchain_google_genai import ChatGoogleGenerativeAI, GoogleGenerativeAIEmbeddings
 import json
 from datetime import datetime
-
-load_dotenv()
+from huggingface_hub import HfApi, Repository
 
 # Constants
 PAGE_TITLE = "PTE Assistant - Multi-Model Comparison"
@@ -23,6 +22,61 @@ MODELS = {
 }
 TRAINING_DATA_FILE = "training_data.json"
 
+class HuggingFaceStorage:
+    def __init__(self, repo_id, token_path=".env"):
+       
+        # Read token from .env file
+        load_dotenv(token_path)
+        self.token = os.getenv("HF_TOKEN")
+        if not self.token:
+            raise ValueError("HF_TOKEN not found in .env file")
+        
+        self.repo_id = repo_id
+        self.api = HfApi()
+        
+        # Initialize/clone the repository
+        self.repo = Repository(
+            local_dir="repo_cache",
+            clone_from=repo_id,
+            token=self.token,
+            git_user="JohnsonMLEngineer",
+            git_email="[email protected]"
+        )
+        
+    def load_json_file(self, filename="training_data.json"):
+        """Load JSON file from repository"""
+        json_path = f"repo_cache/{filename}"
+        try:
+            if os.path.exists(json_path) and os.path.getsize(json_path) > 0:
+                with open(json_path, 'r') as f:
+                    return json.load(f)
+            return []
+        except Exception as e:
+            st.error(f"Error loading data from HuggingFace: {str(e)}")
+            return []
+            
+    def save_json_file(self, data, filename="training_data.json"):
+        """Save JSON file to repository"""
+        try:
+            json_path = f"repo_cache/{filename}"
+            
+            # Create directory if it doesn't exist
+            os.makedirs(os.path.dirname(json_path), exist_ok=True)
+            
+            # Save file
+            with open(json_path, 'w') as f:
+                json.dump(data, f, indent=2)
+            
+            # Commit and push changes
+            self.repo.git_add(filename)
+            self.repo.git_commit(f"Update {filename}")
+            self.repo.git_push()
+            
+            return True
+        except Exception as e:
+            st.error(f"Error saving data to HuggingFace: {str(e)}")
+            return False
+
 # Initialize session state
 if "chat_history" not in st.session_state:
     st.session_state.chat_history = []
@@ -31,34 +85,21 @@ if "selected_responses" not in st.session_state:
 if "feedback_messages" not in st.session_state:
     st.session_state.feedback_messages = {}
 
-def load_training_data():
-    """Load existing training data or return empty list if file doesn't exist"""
-    try:
-        if os.path.exists(TRAINING_DATA_FILE) and os.path.getsize(TRAINING_DATA_FILE) > 0:
-            with open(TRAINING_DATA_FILE, 'r') as f:
-                return json.load(f)
-        return []
-    except json.JSONDecodeError:
-        st.warning("Found invalid training data file. Creating new one.")
-        return []
-    except Exception as e:
-        st.error(f"Error loading training data: {str(e)}")
-        return []
+def load_training_data(hf_storage):
+    """Load existing training data using HuggingFace storage"""
+    return hf_storage.load_json_file(TRAINING_DATA_FILE)
 
-def save_training_data(data):
-    """Save data for future training"""
+def save_training_data(data, hf_storage):
+    """Save training data using HuggingFace storage"""
     try:
         # Load existing data
-        existing_data = load_training_data()
+        existing_data = load_training_data(hf_storage)
         
         # Append new data
         existing_data.append(data)
         
         # Save updated data
-        with open(TRAINING_DATA_FILE, 'w') as f:
-            json.dump(existing_data, f, indent=2)
-            
-        return True
+        return hf_storage.save_json_file(existing_data, TRAINING_DATA_FILE)
     except Exception as e:
         st.error(f"Error saving training data: {str(e)}")
         return False
@@ -90,9 +131,9 @@ def initialize_rag_components():
     
     # Initialize vector store
     docsearch = PineconeVectorStore.from_existing_index(
-    index_name = "dataset2",
-    embedding = GoogleGenerativeAIEmbeddings(model="models/text-embedding-004")
-)
+        index_name="dataset2",
+        embedding=GoogleGenerativeAIEmbeddings(model="models/text-embedding-004")
+    )
     
     # Create retriever
     retriever = docsearch.as_retriever(search_type="similarity", search_kwargs={"k": 10})
@@ -106,31 +147,30 @@ def initialize_rag_components():
         ) for model_name, model_id in MODELS.items()
     }
     
-    # System prompt remains the same...
     system_prompt = """
     You are an advanced AI assistant specialized in PTE (Pearson Test of English) exam preparation. Your role is to provide expert guidance, explanations, and strategies to help students excel in all aspects of the PTE exam.
-Core Responsibilities:
-
-Provide accurate, detailed information about PTE exam structure, scoring, and recent updates.
-Offer tailored advice and strategies for each PTE section: Speaking, Writing, Reading, and Listening.
-Suggest effective study plans and time management techniques.
-Provide constructive feedback on practice responses (when given).
-
-Guidelines for Responses:
-
-Use the following retrieved context to inform your answers: {context}
-If the context doesn't provide sufficient information or
-If you don't know the answer or are unsure, clearly state this and suggest reliable resources for further information.
-Tailor your language complexity to the user's apparent level of understanding.
-Be concise yet thorough. Aim for clear, actionable advice.
-Use bullet points or numbered lists for step-by-step instructions or multiple tips.
-
-Ethical Considerations:
-Topic Limitation: If a question is outside the scope of the PTE exam, kindly inform the user that you are only equipped to address PTE-related topics.
-Never provide or encourage cheating methods.
-Emphasize the importance of genuine language skill development over exam tricks.
-Respect copyright; produce exact questions from official PTE materials.
-"""
+    Core Responsibilities:
+    
+    Provide accurate, detailed information about PTE exam structure, scoring, and recent updates.
+    Offer tailored advice and strategies for each PTE section: Speaking, Writing, Reading, and Listening.
+    Suggest effective study plans and time management techniques.
+    Provide constructive feedback on practice responses (when given).
+    
+    Guidelines for Responses:
+    
+    Use the following retrieved context to inform your answers: {context}
+    If the context doesn't provide sufficient information or
+    If you don't know the answer or are unsure, clearly state this and suggest reliable resources for further information.
+    Tailor your language complexity to the user's apparent level of understanding.
+    Be concise yet thorough. Aim for clear, actionable advice.
+    Use bullet points or numbered lists for step-by-step instructions or multiple tips.
+    
+    Ethical Considerations:
+    Topic Limitation: If a question is outside the scope of the PTE exam, kindly inform the user that you are only equipped to address PTE-related topics.
+    Never provide or encourage cheating methods.
+    Emphasize the importance of genuine language skill development over exam tricks.
+    Respect copyright; produce exact questions from official PTE materials.
+    """
 
     prompt = ChatPromptTemplate.from_messages([
         ("system", system_prompt),
@@ -171,7 +211,7 @@ def display_chat_history():
                             "selected_model": model_name,
                             "selected_response": response,
                             "custom_response": None
-                        }):
+                        }, st.session_state.hf_storage):
                             show_feedback_message(
                                 idx,
                                 f"✅ Response from {model_name} has been selected and saved successfully! This response will be used to improve future answers."
@@ -206,7 +246,7 @@ def display_chat_history():
                         "selected_model": "custom",
                         "selected_response": custom_response,
                         "custom_response": custom_response
-                    }):
+                    }, st.session_state.hf_storage):
                         show_feedback_message(
                             idx,
                             "✅ Your custom response has been submitted and saved successfully! This will help improve future responses."
@@ -217,10 +257,6 @@ def display_chat_history():
                             "❌ Failed to save your custom response. Please try again.",
                             is_error=True
                         )
-        #else:
-            # Display the selected response
-            #selected = st.session_state.selected_responses[idx]
-           # st.success(f"✅ Selected response from: {selected['selected_model']}")
         
         # Display any feedback messages for this interaction
         display_feedback(idx)
@@ -228,11 +264,18 @@ def display_chat_history():
         st.divider()
 
 def main():
+    load_dotenv()
     set_page_config()
 
     st.header("PTE Assistant - Multi-Model Comparison 🎓")
     st.subheader("Compare responses and select the best answer")
 
+    # Initialize HuggingFace storage
+    if 'hf_storage' not in st.session_state:
+        st.session_state.hf_storage = HuggingFaceStorage(
+            repo_id="JohnsonMLEngineer/Multi-Model_Comparison"  #  your actual repo ID
+        )
+
     # Initialize RAG chains for all models
     rag_chains = initialize_rag_components()
 
@@ -268,8 +311,5 @@ def main():
     # Display chat history with response selection
     display_chat_history()
 
-    
-     
-
 if __name__ == "__main__":
     main()