src/query.py

from langchain_community.vectorstores import FAISS
from langchain_huggingface import HuggingFaceEmbeddings

# Load embeddings and vector DB
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
db = FAISS.load_local("vectorStore", embeddings, allow_dangerous_deserialization=True)

# Ask a query
query = "how many cyclone struck odisha in 2015"
results = db.similarity_search_with_score(query, k=5)

# Combine results into one paragraph
# answer = " ".join([doc.page_content for doc in results])


# Apply similarity threshold
THRESHOLD = 0.75  

filtered = []
for doc, score in results:
    print(f"🔎 Retrieved (distance={score:.4f}): {doc.metadata}")  # debug
    if score <= THRESHOLD:   # <-- check for "closer than threshold"
        filtered.append(doc)

if not filtered:
    answer = "I don't know. This information is not available in my knowledge base."
else:
    answer = "\n\n".join([doc.page_content for doc in filtered])


print(f"\n🔍 Query: {query}")
print(f"\n📝 Answer:\n{answer}")

# for i, doc in enumerate(results, 1):
#     print(f"\nResult {i}:")
#     print(f"Source: {doc.metadata}")
#     print(f"Content: {doc.page_content[:300]}...")




# it gives query aswell as metadata  also
# print(f"\n🔍 Query: {query}")
# for i, doc in enumerate(results, 2):
#    print(f"\n--- Result {i} ---")
#    print(doc.page_content[:700])   # show first 500 characters
#    print("Source:", doc.metadata)