app.py

import streamlit as st
from bs4 import BeautifulSoup
from langchain.embeddings import HuggingFaceEmbeddings
import pickle
import torch
import io

class CPU_Unpickler(pickle.Unpickler):
    def find_class(self, module, name):
        if module == 'torch.storage' and name == '_load_from_bytes':
            return lambda b: torch.load(io.BytesIO(b), map_location='cpu')
        else: return super().find_class(module, name)


@st.cache_resource
def get_hugging_face_model():
  model_name = "mchochlov/codebert-base-cd-ft"
  hf = HuggingFaceEmbeddings(model_name=model_name)
  return hf


@st.cache_resource
def get_db():
  with open("codesearchdb.pickle", "rb") as f:
    db = CPU_Unpickler(f).load()
  return db


def get_similar_links(query, db, embeddings):
  embedding_vector = embeddings.embed_query(query)
  docs_and_scores = db.similarity_search_by_vector(embedding_vector)
  hrefs = []
  for docs in docs_and_scores:
    html_doc = docs.page_content
    soup = BeautifulSoup(html_doc, 'html.parser')
    href = [a['href'] for a in soup.find_all('a', href=True)]
    hrefs.append(href)
  links = []
  for href_list in hrefs:
      for link in href_list:
        links.append(link)
  return links


embedding_vector = get_hugging_face_model()
db = get_db()
st.title("📒 DSASearch Engine 🤖 ")
text_input = st.text_input("Enter some text")
button = st.button("Find Similar Questions on Leetcode")
if text_input:
  query = text_input
  answer = get_similar_links(query, db, embedding_vector)
  for link in answer:
      st.write(link)
  st.balloons()

else:
  st.info("Please Input Valid Text")

st.markdown("""
        ### Created by Ashwin Rachha.
        Source Data : https://github.com/AshwinRachha/LeetCode-Solutions
        Medium Blog : https://medium.com/@ashwin_rachha/querying-a-code-database-to-find-similar-coding-problems-using-langchain-814730da6e6d
    """)