import gradio as gr
import pandas as pd
import faiss
from sentence_transformers import SentenceTransformer

#load the sentences data

path = "debate_hansard_sentences.txt"
file = open(path, "r")
sentences = file.readlines()

# remove duplicates
sentences = [word for word in list(set(sentences)) if type(word)]

sentences_dict = {idx:val for idx, val in enumerate(sentences)}
sentences_df = pd.DataFrame.from_dict(sentences_dict, orient='index', columns=['sentences'])

#load the index

index = faiss.read_index("semantic_search_vector.index") 

def question_answer(query):
    model = SentenceTransformer('bert-base-nli-mean-tokens')
    k = 4
    xq = model.encode([query])
    D, I = index.search(xq, k)  # search
    sentences_df["sentences"].iloc[I.tolist()[0]]
    a = list(sentences_df["sentences"].iloc[I.tolist()[0]])
    results = "\n".join(a)
    answer =  results
    return answer

iface = gr.Interface(
 fn=question_answer,
 inputs=gr.inputs.Textbox(lines=2, placeholder="Type your query here"), 
 outputs="text").launch()