interview_double.py

import streamlit as st
from streamlit_option_menu import option_menu
from app_utils import switch_page
from PIL import Image
from streamlit_lottie import st_lottie
from typing import Literal
from dataclasses import dataclass
import json
import base64
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationChain, RetrievalQA
from langchain.prompts.prompt import PromptTemplate
from langchain.text_splitter import NLTKTextSplitter
from langchain.vectorstores import FAISS
import nltk
from prompts.prompts import templates
from langchain_google_genai import ChatGoogleGenerativeAI
import getpass
import os
from langchain_google_genai import GoogleGenerativeAIEmbeddings



if "GOOGLE_API_KEY" not in os.environ:
    os.environ["GOOGLE_API_KEY"] = "AIzaSyCA4__JMC_ZIQ9xQegIj5LOMLhSSrn3pMw"

im = Image.open("icon.png")

def app():
    lan = st.selectbox("#### Language", ["English", "中文"])

    if lan == "English":
        home_title = "AI Interviewer"
        home_introduction = "Welcome to AI Interviewer, empowering your interview preparation with generative AI."

        st.markdown(
            "<style>#MainMenu{visibility:hidden;}</style>",
            unsafe_allow_html=True
        )
        st.image(im, width=100)
        st.markdown(f"""# {home_title}""", unsafe_allow_html=True)
        st.markdown("""\n""")
        # st.markdown("#### Greetings")
        st.markdown("Welcome to AI Interviewer! 👏 AI Interviewer is your personal interviewer powered by generative AI that conducts mock interviews."
                    "You can upload your resume and enter job descriptions, and AI Interviewer will ask you customized questions. Additionally, you can configure your own Interviewer!")
        st.markdown("""\n""")
        role = st.text_input("Enter your role")
        if role:
            st.markdown(f"Your role is {role}")

            llm = ChatGoogleGenerativeAI(
                model="gemini-pro")
            llm = ChatGoogleGenerativeAI(model="gemini-pro")
            prompt = f"Provide the tech stack and responsibilities for the top 3 job recommendations based on the role: {role}. " + """
            For each job recommendation, list the required tech stack and associated responsibilities without giving any title or role name. 
            Ensure the information is detailed and precise.

For above each job recommendation, list the required tech stack and associated responsibilities in the following format too:

[
    {
        "tech_stack": ["tech1", "tech2", ...],
        "responsibilities": ["resp1", "resp2", ...]
    },
    {
        "tech_stack": ["tech1", "tech2", ...],
        "responsibilities": ["resp1", "resp2", ...]
    },
    ...
]
"""
            try:
                analysis = llm.invoke(prompt)
                st.write(analysis.content)
                job_recommendations = json.loads(analysis.content)
            except json.JSONDecodeError:
                st.error("Failed to parse the LLM response. Please ensure the LLM is returning a structured JSON-like response.")
                return
            except Exception as e:
                st.error(f"An error occurred: {e}")
                return

            if job_recommendations:
                # Display Selector Boxes
                options = [f"Tech Stack: {rec['tech_stack']}, Responsibilities: {rec['responsibilities']}" for rec in job_recommendations]
                selected_option = st.selectbox("Select your preferred tech stack and responsibilities", options)

                # Form Submission
                submit_button = st.button(label='Submit')

                if submit_button:
                   selected_index = options.index(selected_option)
                   selected_rec = job_recommendations[selected_index]
                   tech_stack = ", ".join(selected_rec['tech_stack'])
                   responsibilities = ", ".join(selected_rec['responsibilities'])
                   llm2 = ChatGoogleGenerativeAI(model="gemini-pro")
                   prompt = f"""Tech stack: {tech_stack}\nResponsibilities: {responsibilities}
create a job description based on tech stack, responsibilities and give tech stack, responsibilities and qualifications for job description
example -
Tech stack: all technical stack here
Qualifications: all qualifications here
Responsibilities: all responsibilities here
"""
                   try:
                        response = llm2.invoke(prompt)
                        st.write(response.content)
                        jd = response.content
                   except Exception as e:
                        st.error(f"An error occurred: {e}")
                        return

                   if jd:
                        # Save the jd into a json file
                        with open("job_description.json", "w") as f:
                            json.dump(jd, f)
                            st.success("Job description saved successfully!")

if __name__ == "__main__":
     app()