pushing version 2

helper.py

@@ -3,6 +3,8 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 from pypdf import PdfReader
 import requests
 import json
+import os
+import time
 
 
 def extract_text_from_pdf(pdf_path):
@@ -30,32 +32,81 @@ def embedding_function(texts):
 def generate_hypothetical_answer(query):
     import requests
     import json
+    import os
+    import time
     
-    # Ollama API endpoint (default is localhost:11434)
-    ollama_url = "http://localhost:11434/api/generate"
+    # Hugging Face API endpoint
+    api_url = "https://api-inference.huggingface.co/models/meta-llama/Llama-2-7b-chat-hf"
     
-    # Prepare the prompt
-    prompt = f"Generate a plausible answer to the question:\n\n{query}\n\nAnswer:"
+    # Get API token from environment variable
+    api_token = os.getenv("HUGGINGFACE_API_TOKEN")
+    if not api_token:
+        return "Error: HUGGINGFACE_API_TOKEN environment variable not set"
+    
+    # Headers for the API request
+    headers = {
+        "Authorization": f"Bearer {api_token}",
+        "Content-Type": "application/json"
+    }
+    
+    # Create a prompt for generating a hypothetical answer
+    prompt = f"""
+    Given the following query, generate a hypothetical answer that might be found in a document:
+    Query: {query}
+    
+    Hypothetical answer:
+    """
     
     # Prepare the request payload
     payload = {
-        "model": "llama2",  # or any other model you have pulled in Ollama
-        "prompt": prompt,
-        "stream": False
+        "inputs": prompt,
+        "parameters": {
+            "max_new_tokens": 256,
+            "temperature": 0.7,
+            "top_p": 0.95,
+            "do_sample": True
+        }
     }
     
     try:
-        # Make the API request to Ollama
-        response = requests.post(ollama_url, json=payload)
+        # Make the API request to Hugging Face
+        print("Sending request to Hugging Face API for hypothetical answer...")
+        start_time = time.time()
+        
+        # Set a longer timeout (5 minutes)
+        response = requests.post(api_url, headers=headers, json=payload, timeout=300)
         response.raise_for_status()  # Raise an exception for HTTP errors
         
+        end_time = time.time()
+        print(f"Received hypothetical answer from Hugging Face API in {end_time - start_time:.2f} seconds")
+        
         # Parse the response
         result = response.json()
         
         # Extract the generated text
-        generated_text = result.get("response", "")
+        if isinstance(result, list) and len(result) > 0:
+            generated_text = result[0].get("generated_text", "")
+        else:
+            generated_text = result.get("generated_text", "")
+            
         return generated_text.strip()
     
+    except requests.exceptions.Timeout:
+        print("Request to Hugging Face API timed out after 5 minutes")
+        return "The request timed out. The model is taking too long to respond. Please try again with a simpler query."
+    
+    except requests.exceptions.ConnectionError:
+        print("Could not connect to Hugging Face API")
+        return "Could not connect to the Hugging Face API. Please check your internet connection."
+    
+    except requests.exceptions.HTTPError as e:
+        print(f"HTTP error occurred: {e}")
+        if e.response.status_code == 401:
+            return "Authentication error. Please check your Hugging Face API token."
+        elif e.response.status_code == 429:
+            return "Rate limit exceeded. Please try again later."
+        return f"HTTP error occurred: {e}"
+    
     except Exception as e:
         print(f"Error generating hypothetical answer: {e}")
         return "Failed to generate a hypothetical answer."
@@ -63,49 +114,96 @@ def generate_hypothetical_answer(query):
 
 
 
-def query_llm_with_context(query,context,top_n=3):
-    # Get documents sorted by similarity
-    sorted_docs, sorted_scores = context
+def query_llm_with_context(query, context, top_n=3):
+    import requests
+    import json
+    import os
+    import time
+    
+    # Unpack the context tuple
+    documents, similarity_scores = context
     
     # Use only the top N documents
-    top_docs = sorted_docs[:top_n]
+    top_docs = documents[:top_n]
     
     # Create a context string by joining the top documents
-    context = "\n\n===Document Boundary===\n\n".join(top_docs)
+    context_text = "\n\n===Document Boundary===\n\n".join(top_docs)
     
     # Create a prompt with the context and query
     prompt = f"""
     Context information is below.
     ---------------------
-    {context}
+    {context_text}
     ---------------------
     
     Given the context information and not prior knowledge, answer the following query:
     Query: {query}
     """
     
-    # Call Ollama API instead of OpenAI
-    ollama_url = "http://localhost:11434/api/generate"
+    # Hugging Face API endpoint
+    api_url = "https://api-inference.huggingface.co/models/meta-llama/Llama-2-7b-chat-hf"
+    
+    # Get API token from environment variable
+    api_token = os.getenv("HUGGINGFACE_API_TOKEN")
+    if not api_token:
+        return "Error: HUGGINGFACE_API_TOKEN environment variable not set"
+    
+    # Headers for the API request
+    headers = {
+        "Authorization": f"Bearer {api_token}",
+        "Content-Type": "application/json"
+    }
     
     # Prepare the request payload
     payload = {
-        "model": "llama2",  # or any other model you have pulled in Ollama
-        "prompt": prompt,
-        "stream": False
+        "inputs": prompt,
+        "parameters": {
+            "max_new_tokens": 512,
+            "temperature": 0.7,
+            "top_p": 0.95,
+            "do_sample": True
+        }
     }
     
     try:
-        # Make the API request to Ollama
-        response = requests.post(ollama_url, json=payload)
+        # Make the API request to Hugging Face
+        print("Sending request to Hugging Face API...")
+        start_time = time.time()
+        
+        # Set a longer timeout (5 minutes)
+        response = requests.post(api_url, headers=headers, json=payload, timeout=300)
         response.raise_for_status()  # Raise an exception for HTTP errors
         
+        end_time = time.time()
+        print(f"Received response from Hugging Face API in {end_time - start_time:.2f} seconds")
+        
         # Parse the response
         result = response.json()
         
         # Extract the generated text
-        generated_text = result.get("response", "")
+        if isinstance(result, list) and len(result) > 0:
+            generated_text = result[0].get("generated_text", "")
+        else:
+            generated_text = result.get("generated_text", "")
+            
         return generated_text.strip()
     
+    except requests.exceptions.Timeout:
+        print("Request to Hugging Face API timed out after 5 minutes")
+        return "The request timed out. The model is taking too long to respond. Please try again with a simpler query or fewer context documents."
+    
+    except requests.exceptions.ConnectionError:
+        print("Could not connect to Hugging Face API")
+        return "Could not connect to the Hugging Face API. Please check your internet connection."
+    
+    except requests.exceptions.HTTPError as e:
+        print(f"HTTP error occurred: {e}")
+        if e.response.status_code == 401:
+            return "Authentication error. Please check your Hugging Face API token."
+        elif e.response.status_code == 429:
+            return "Rate limit exceeded. Please try again later."
+        return f"HTTP error occurred: {e}"
+    
     except Exception as e:
         print(f"Error querying LLM with context: {e}")
         return "Failed to generate an answer with the provided context."