Update main.py

main.py

@@ -1,5 +1,6 @@
 import os
 import zipfile
+import logging
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 
@@ -9,6 +10,13 @@ from langchain_groq import ChatGroq
 from langchain.chains import RetrievalQA
 from langchain.prompts import PromptTemplate
 
+# Configure logging
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
+)
+logger = logging.getLogger(__name__)
+
 app = FastAPI()
 
 # === Globals ===
@@ -25,60 +33,55 @@ class QueryRequest(BaseModel):
 def load_components():
     global llm, embeddings, vectorstore, retriever, chain
 
-    # 1) Init LLM & Embeddings
-    llm = ChatGroq(
-        model="meta-llama/llama-4-scout-17b-16e-instruct",
-        temperature=0,
-        max_tokens=1024,
-        api_key=os.getenv("API_KEY"),
-    )
-    embeddings = HuggingFaceEmbeddings(
-        model_name="intfloat/multilingual-e5-large",
-        model_kwargs={"device": "cpu"},
-        encode_kwargs={"normalize_embeddings": True},
-    )
-
-    # 2) Unzip & Load both FAISS vectorstores
-    # — First index
-    zip1 = "faiss_index.zip"
-    dir1 = "faiss_index"
-    if not os.path.exists(dir1):
-        with zipfile.ZipFile(zip1, 'r') as z:
-            z.extractall(dir1)
-        print("✅ Unzipped FAISS index 1.")
-    vs1 = FAISS.load_local(
-        dir1,
-        embeddings,
-        allow_dangerous_deserialization=True
-    )
-    print("✅ FAISS index 1 loaded.")
-
-    # — Second index
-    zip2 = "faiss_index(1).zip"
-    dir2 = "faiss_index_extra"
-    if not os.path.exists(dir2):
-        with zipfile.ZipFile(zip2, 'r') as z:
-            z.extractall(dir2)
-        print("✅ Unzipped FAISS index 2.")
-    vs2 = FAISS.load_local(
-        dir2,
-        embeddings,
-        allow_dangerous_deserialization=True
-    )
-    print("✅ FAISS index 2 loaded.")
-
-    # 3) Merge them
-    vs1.merge_from(vs2)
-    vectorstore = vs1
-    print("✅ Merged FAISS indexes into a single vectorstore.")
-
-    # 4) Create retriever & QA chain
-    retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
-    prompt = PromptTemplate(
-        template="""
+    try:
+        # 1) Init LLM & Embeddings
+        llm = ChatGroq(
+            model="meta-llama/llama-4-scout-17b-16e-instruct",
+            temperature=0,
+            max_tokens=1024,
+            api_key=os.getenv("API_KEY"),
+        )
+        embeddings = HuggingFaceEmbeddings(
+            model_name="intfloat/multilingual-e5-large",
+            model_kwargs={"device": "cpu"},
+            encode_kwargs={"normalize_embeddings": True},
+        )
+
+        # 2) Unzip & Load both FAISS vectorstores
+        for zip_name, dir_name in [("faiss_index.zip", "faiss_index"), ("faiss_index(1).zip", "faiss_index_extra")]:
+            if not os.path.exists(dir_name):
+                with zipfile.ZipFile(zip_name, 'r') as z:
+                    z.extractall(dir_name)
+                logger.info(f"Unzipped {zip_name} to {dir_name}.")
+            else:
+                logger.info(f"Directory {dir_name} already exists.")
+
+        vs1 = FAISS.load_local(
+            "faiss_index",
+            embeddings,
+            allow_dangerous_deserialization=True
+        )
+        logger.info("FAISS index 1 loaded.")
+
+        vs2 = FAISS.load_local(
+            "faiss_index_extra",
+            embeddings,
+            allow_dangerous_deserialization=True
+        )
+        logger.info("FAISS index 2 loaded.")
+
+        # 3) Merge them
+        vs1.merge_from(vs2)
+        vectorstore = vs1
+        logger.info("Merged FAISS indexes into a single vectorstore.")
+
+        # 4) Create retriever & QA chain
+        retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
+        prompt = PromptTemplate(
+            template="""
 You are an expert assistant on Islamic knowledge.
-Use **only** the information in the “Retrieved context” to answer the user’s question.
-Do **not** add any outside information, personal opinions, or conjecture—if the answer is not contained in the context, reply with “لا أعلم”.
+Use **only** the information in the “Retrieved context” to answer general questions related to Islam.
+Do **not** add any outside information, personal opinions, or conjecture—if the answer is not contained in the context, reply with "I don't know".
 Be concise, accurate, and directly address the user’s question.
 
 Retrieved context:
@@ -89,16 +92,20 @@ User’s question:
 
 Your response:
 """,
-        input_variables=["context", "question"],
-    )
-    chain = RetrievalQA.from_chain_type(
-        llm=llm,
-        chain_type="stuff",
-        retriever=retriever,
-        return_source_documents=False,
-        chain_type_kwargs={"prompt": prompt},
-    )
-    print("✅ QA chain ready.")
+            input_variables=["context", "question"],
+        )
+        chain = RetrievalQA.from_chain_type(
+            llm=llm,
+            chain_type="stuff",
+            retriever=retriever,
+            return_source_documents=False,
+            chain_type_kwargs={"prompt": prompt},
+        )
+        logger.info("QA chain ready.")
+
+    except Exception as e:
+        logger.error("Error loading components", exc_info=True)
+        raise
 
 @app.get("/")
 def root():
@@ -107,7 +114,10 @@ def root():
 @app.post("/query")
 def query(request: QueryRequest):
     try:
+        logger.info(f"Received query: %s", request.question)
         result = chain.invoke({"query": request.question})
-        return {"answer": result["result"]}
+        logger.info("Query processed successfully.")
+        return {"answer": result.get("result")}
     except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
+        logger.error("Error processing query", exc_info=True)
+        raise HTTPException(status_code=500, detail="Internal server error.")