app.py

import streamlit as st
import pandas as pd
from appStore.prep_data import process_giz_worldwide
from appStore.prep_utils import create_documents, get_client
from appStore.embed import hybrid_embed_chunks
from appStore.search import hybrid_search
from torch import cuda
# get the device to be used eithe gpu or cpu
device = 'cuda' if cuda.is_available() else 'cpu'


st.set_page_config(page_title="SEARCH IATI",layout='wide')
st.title("SEARCH IATI Database")
var=st.text_input("enter keyword")

####################  Create the embeddings collection and save ######################
# the steps below need to be performed only once and then commented out any unnecssary compute over-run
##### First we process and create the chunks for relvant data source
#chunks = process_giz_worldwide()
##### Convert to langchain documents
#temp_doc = create_documents(chunks,'chunks')
##### Embed and store docs, check if collection exist then you need to update the collection
collection_name = "giz_worldwide"
#hybrid_embed_chunks(docs= temp_doc, collection_name = collection_name)

################### Hybrid Search ######################################################
client = get_client()
print(client.get_collections())





button=st.button("search")
#found_docs = vectorstore.similarity_search(var)
#print(found_docs)
# results= get_context(vectorstore, f"find the relvant paragraphs for: {var}")
if button:
    results = hybrid_search(client, var, collection_name)
    st.write(f"Showing Top 10 results for query:{var}")
    st.write(f"Semantic: {len(results[0])}")
    st.write(results[0])
    st.write(f"Semantic: {len(results[1])}")
    st.write(results[1])

    #  for i in results: 
    #      st.subheader(str(i.metadata['id'])+":"+str(i.metadata['title_main']))
    #      st.caption(f"Status:{str(i.metadata['status'])}, Country:{str(i.metadata['country_name'])}")
    #      st.write(i.page_content)
    #      st.divider()