app.py

import json
import pickle
import streamlit as st
# from haystack.document_stores import FAISSDocumentStore
from haystack.document_stores import InMemoryDocumentStore
from transformers import DPRContextEncoder, DPRContextEncoderTokenizer
from haystack.nodes import DensePassageRetriever
from haystack.nodes import FARMReader
from haystack.pipelines import ExtractiveQAPipeline

st.title("DPR on Supreme Court Judgements (Capital Gain)")

# with open("responses.json", 'r') as f:
#   data = json.load(f)

# documents = [
#     {
#         "content": doc["text"],
#         "meta": {
#             "name": doc["title"],
#             "url": doc["url"]
#         }
#     } for doc in data
# ]

# document_store = FAISSDocumentStore(embedding_dim=768, faiss_index_factory_str="Flat", sql_url="sqlite:///faiss_document_store.d")
with open("inmemory_document_store.pkl", "rb") as f:
    document_store = pickle.load(f)
# document_store.write_documents(documents)

# document_store = FAISSDocumentStore.load(index_path="./faiss_index", config_path="./faiss_index.json")

retriever = DensePassageRetriever(
    document_store=document_store,
    query_embedding_model="facebook/dpr-question_encoder-single-nq-base",
    passage_embedding_model="facebook/dpr-ctx_encoder-single-nq-base",
)

# document_store.update_embeddings(retriever)
# document_store.save(index_path="./faiss_index", config_path="./faiss_index.json")
# with open("inmemory_document_store.pkl", "wb") as f:
#     pickle.dump(document_store, f)


reader = FARMReader(model_name_or_path="deepset/bert-base-cased-squad2")

pipeline = ExtractiveQAPipeline(reader=reader, retriever=retriever)

query = st.text_input("Enter your query:", "")

if query:
    with st.spinner("Searching..."):
        results = pipeline.run(query=query, params={"Retriever": {"top_k": 5}})
        for answer in results['answers']:
            st.markdown(f"=====================\nAnswer: {answer.answer}\nContext: {answer.context}\nScore: {answer.score}")

# query = st.text_input("Enter Question")
# query = "What is the subject matter of the petition in the Sadanand S. Varde case?"
# result = pipeline.run(query=query, params={"Retriever": {"top_k": 10}, "Reader": {"top_k": 5}})
# for answer in result['answers']:
#    print(f"=====================\nAnswer: {answer.answer}\nContext: {answer.context}\nScore: {answer.score}")