Update config to load .env, add HF token support, and clean up settings

app/config.py

@@ -1,38 +1,38 @@
-import os
-from dotenv import load_dotenv
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
-# Load environment variables from .env
-load_dotenv()
-
 class Settings(BaseSettings):
     """Application settings loaded from environment variables."""
 
     # ───────────────────────────────────────────────────────────────────────────
     # Google API Keys
     # ───────────────────────────────────────────────────────────────────────────
-    pagespeed_api_key: str = os.getenv("PAGESPEED_API_KEY", "")
-    gemini_api_key: str = os.getenv("GEMINI_API_KEY", "")
-    
+    pagespeed_api_key: str
+    gemini_api_key: str
+
     # ───────────────────────────────────────────────────────────────────────────
     # Chat & RAG Configuration
     # ───────────────────────────────────────────────────────────────────────────
-    groq_api_key: str = os.getenv("GROQ_API_KEY", "")
-    vectorstore_base_path: str = os.getenv("VECTORSTORE_BASE_PATH", "./vectorstores")
+    groq_api_key: str
+    vectorstore_base_path: str = "./vectorstores"
+
+    # ───────────────────────────────────────────────────────────────────────────
+    # Hugging Face Hub
+    # ───────────────────────────────────────────────────────────────────────────
+    huggingfacehub_api_token: str
 
     # ───────────────────────────────────────────────────────────────────────────
     # MongoDB Configuration (Local)
     # ───────────────────────────────────────────────────────────────────────────
-    mongo_uri: str = os.getenv("MONGO_URI", "mongodb://localhost:27017")
-    mongo_chat_db: str = os.getenv("MONGO_CHAT_DB", "Education_chatbot")
-    mongo_chat_collection: str = os.getenv("MONGO_CHAT_COLLECTION", "chat_histories")
+    mongo_uri: str = "mongodb://localhost:27017"
+    mongo_chat_db: str = "Education_chatbot"
+    mongo_chat_collection: str = "chat_histories"
 
     # ───────────────────────────────────────────────────────────────────────────
     # FastAPI Server Configuration
     # ───────────────────────────────────────────────────────────────────────────
-    host: str = os.getenv("HOST", "0.0.0.0")
-    port: int = int(os.getenv("PORT", "8000"))
-    debug: bool = os.getenv("DEBUG", "False").lower() == "true"
+    host: str = "0.0.0.0"
+    port: int = 8000
+    debug: bool = False
 
     # ───────────────────────────────────────────────────────────────────────────
     # App Metadata (unchanged)
@@ -44,10 +44,14 @@ class Settings(BaseSettings):
         "using Google's APIs and Gemini AI"
     )
 
+    # ───────────────────────────────────────────────────────────────────────────
+    # Tell Pydantic to load from .env and ignore extras
+    # ───────────────────────────────────────────────────────────────────────────
     model_config = SettingsConfigDict(
         env_file=".env",
-        env_file_encoding="utf-8"
+        env_file_encoding="utf-8",
+        extra="ignore",
     )
 
-# Instantiate settings
+# Single shared Settings instance
 settings = Settings()

app/rag/chat_history.py

@@ -0,0 +1,80 @@
+import time
+from typing import List, Dict, Any
+from pymongo import ReturnDocument
+
+from app.config import settings
+from .db import mongo_client, chat_collection_name
+from .embeddings import get_llm
+from langchain.prompts import ChatPromptTemplate
+from .logging_config import logger
+
+# Get the actual collection object
+db = mongo_client[settings.mongo_chat_db]
+coll = db[chat_collection_name]
+
+# LLM & summarization prompt
+llm = get_llm()
+summarization_prompt = ChatPromptTemplate.from_messages([
+    ("system", "Summarize the following conversation into a concise summary:"),
+    ("human", "{chat_history}")
+])
+
+class ChatHistoryManager:
+    @staticmethod
+    def create_session(chat_id: str) -> None:
+        """Ensure a document exists for this chat_id with empty messages."""
+        coll.update_one(
+            {"session_id": chat_id},
+            {"$setOnInsert": {"session_id": chat_id, "messages": []}},
+            upsert=True
+        )
+        logger.info("Initialized chat session %s", chat_id)
+
+    @staticmethod
+    def get_messages(chat_id: str) -> List[Dict[str, Any]]:
+        """Return the messages array for this session (or empty if none)."""
+        doc = coll.find_one({"session_id": chat_id}, {"_id": 0, "messages": 1})
+        return doc.get("messages", []) if doc else []
+
+    @staticmethod
+    def add_message(chat_id: str, role: str, content: str) -> None:
+        """Append a new {role,content,timestamp} entry to the messages array."""
+        entry = {
+            "type": role,
+            "content": content,
+            "timestamp": time.time()
+        }
+        coll.update_one(
+            {"session_id": chat_id},
+            {"$push": {"messages": entry}}
+        )
+        logger.debug("Appended %s message to %s", role, chat_id)
+
+    @staticmethod
+    def summarize_if_needed(chat_id: str, threshold: int = 10) -> bool:
+        """
+        If message count > threshold, summarize and replace all messages
+        with a single "ai" summary entry.
+        """
+        messages = ChatHistoryManager.get_messages(chat_id)
+        if len(messages) <= threshold:
+            return False
+
+        # Flatten for summarization
+        chat_text = "\n".join(f"{m['type'].upper()}: {m['content']}" for m in messages)
+
+        # Run summarization
+        summary_chain = summarization_prompt | llm
+        result = summary_chain.invoke({"chat_history": chat_text})
+        summary = getattr(result, "content", result)
+
+        # Replace entire messages array with the summary
+        coll.find_one_and_update(
+            {"session_id": chat_id},
+            {"$set": {"messages": [
+                {"type": "ai", "content": summary, "timestamp": time.time()}
+            ]}},
+            return_document=ReturnDocument.AFTER
+        )
+        logger.info("Summarized chat %s down to one message", chat_id)
+        return True

app/rag/embeddings.py

@@ -2,6 +2,10 @@ import os
 from langchain_community.embeddings import HuggingFaceBgeEmbeddings
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain.prompts import ChatPromptTemplate
+from dotenv import load_dotenv
+
+load_dotenv()  # now os.getenv(...) will pick up values from your .env file
+
 
 def get_llm():
     """

app/rag/routes.py

@@ -22,6 +22,9 @@ from .utils import (
 )
 from .logging_config import logger
 
+from .chat_history import ChatHistoryManager
+from .logging_config import logger
+
 router = APIRouter(prefix="/rag", tags=["rag"])
 
 @router.post("/ingest/{user_id}", response_model=IngestResponse)
@@ -92,32 +95,32 @@ async def create_chat_session(user_id: str):
         logger.error("Error creating chat for user_id=%s: %s", user_id, e, exc_info=True)
         raise HTTPException(status_code=500, detail=f"Failed to create chat session: {e}")
 
+
 @router.post("/chat/{user_id}/{chat_id}", response_model=ChatResponse)
 async def chat_with_user(user_id: str, chat_id: str, body: ChatRequest):
-    """
-    Send a user question to the RAG chain and return the LLM answer.
-    - Loads the FAISS index for user_id (404 if not found).
-    - Retrieves (or initializes) the MongoDBChatMessageHistory for chat_id.
-    - Runs the ConversationalRetrievalChain to get an answer.
-    - Returns the answer, plus re‐stores chat history in Mongo automatically.
-    """
-    question = body.question
-    logger.info("Received chat request: user_id=%s, chat_id=%s, question='%s'", user_id, chat_id, question)
+    question = body.question.strip()
+    logger.info("Chat request user=%s chat=%s question=%s", user_id, chat_id, question)
 
     try:
-        # 1. Build the RAG chain (or 404 if no vectorstore)
-        chain = build_rag_chain(user_id, chat_id)
+        # 1) Ensure session exists
+        ChatHistoryManager.create_session(chat_id)
+
+        # 2) Summarize long histories
+        ChatHistoryManager.summarize_if_needed(chat_id, threshold=10)
 
-        # 2. Call the chain
-        result = chain.invoke({"question": question})
-        # Some chains use "answer", some use "output_text"
-        answer = result.get("answer") or result.get("output_text") or None
+        # 3) Record the user message
+        ChatHistoryManager.add_message(chat_id, role="human", content=question)
 
-        if answer is None:
-            logger.error("Chain returned no 'answer' or 'output_text': %s", result)
-            raise Exception("Failed to retrieve answer from chain.")
+        # 4) Build and invoke the RAG chain
+        chain = build_rag_chain(user_id, chat_id)
+        history = ChatHistoryManager.get_messages(chat_id)
+        result = chain.invoke({"question": question, "chat_history": history})
+        answer = result.get("answer") or result.get("output_text")
+        if not answer:
+            raise Exception("No answer returned from chain")
 
-        logger.info("Chain answered for chat_id=%s: %s", chat_id, answer)
+        # 5) Record the AI response
+        ChatHistoryManager.add_message(chat_id, role="ai", content=answer)
 
         return ChatResponse(
             success=True,
@@ -126,11 +129,11 @@ async def chat_with_user(user_id: str, chat_id: str, body: ChatRequest):
             chat_id=chat_id,
             user_id=user_id
         )
+
     except HTTPException:
-        # Re‐raise known HTTPExceptions (e.g. 404 from build_rag_chain)
         raise
     except Exception as e:
-        logger.error("Error in chat endpoint for user_id=%s, chat_id=%s: %s", user_id, chat_id, e, exc_info=True)
+        logger.error("Error chatting user=%s chat=%s: %s", user_id, chat_id, e, exc_info=True)
         return ChatResponse(
             success=False,
             answer=None,

requirements.txt

@@ -10,3 +10,6 @@ langchain_community
 faiss-cpu 
 pymongo 
 langchain-mongodb
+huggingface_hub
+python_dotenv
+sentence_transformers