import openai

def generate_questions(role, num_questions=5):
    prompt = f"""Generate {num_questions} interview questions for a {role} position.
    Include a mix of behavioral, technical, and situational questions.

    Questions:
    1. """

    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[{"role": "user", "content": prompt}]
    )

    questions = response.choices[0].message.content.strip().split('\n')
    questions = [q.strip() for q in questions if q.strip()]

    return questions

# Example
questions = generate_questions("Data Analyst")
for i, q in enumerate(questions):
    print(f"{i+1}. {q}")

# Question templates for different roles
question_templates = {
    "Data Analyst": [
        "Tell me about a time you used data to solve a business problem.",
        "How do you approach cleaning messy data?",
        "Explain a dashboard you built and its impact.",
        "Describe your experience with SQL and Python.",
        "How do you communicate insights to non-technical stakeholders?"
    ],
    "Software Engineer": [
        "Explain a complex technical problem you solved.",
        "How do you approach debugging?",
        "Describe your experience with system design.",
        "How do you stay updated with new technologies?",
        "Tell me about a project you led."
    ]
}

def get_template_questions(role):
    return question_templates.get(role, question_templates["Data Analyst"])

import openai

def transcribe_audio(audio_file):
    with open(audio_file, "rb") as f:
        response = openai.Audio.transcribe(
            model="whisper-1",
            file=f
        )
    return response["text"]

# Alternative: Use local Whisper model
# import whisper
# model = whisper.load_model("base")
# result = model.transcribe("audio.wav")
# print(result["text"])

# Record audio from microphone
import sounddevice as sd
import soundfile as sf

def record_audio(duration=10, sample_rate=16000):
    print(f"Recording for {duration} seconds...")
    recording = sd.rec(int(duration * sample_rate), samplerate=sample_rate, channels=1)
    sd.wait()
    sf.write("recording.wav", recording, sample_rate)
    print("Recording saved.")
    return "recording.wav"

# Usage in app
def get_voice_answer():
    audio_file = record_audio(duration=30)
    text = transcribe_audio(audio_file)
    return text

def evaluate_answer(question, answer):
    prompt = f"""Evaluate this interview answer.

    Question: {question}

    Answer: {answer}

    Provide:
    1. A score from 1-10
    2. Strengths of the answer
    3. Areas for improvement
    4. A sample improved answer

    Format your response as a structured evaluation."""

    response = openai.ChatCompletion.create(
        model="gpt-4",
        messages=[{"role": "user", "content": prompt}]
    )

    return response.choices[0].message.content

# Example
question = "Tell me about a time you used data to solve a business problem."
answer = "I used SQL to analyze customer data and found a trend..."
feedback = evaluate_answer(question, answer)
print(feedback)

# Structured feedback template
def get_feedback(question, answer):
    evaluation = evaluate_answer(question, answer)

    # Parse evaluation into structured format
    feedback = {
        "score": extract_score(evaluation),
        "strengths": extract_strengths(evaluation),
        "improvements": extract_improvements(evaluation),
        "sample_answer": extract_sample(evaluation)
    }
    return feedback

def extract_score(evaluation):
    # Extract score from evaluation text
    import re
    match = re.search(r'Score:?\s*(\d+)/10', evaluation)
    if match:
        return int(match.group(1))
    return 7  # default

# Helper functions for parsing
def extract_strengths(evaluation):
    # Extract strengths section
    pass

def extract_improvements(evaluation):
    # Extract areas for improvement
    pass

import streamlit as st

st.title("AI Interview Preparation Agent")

# Sidebar for settings
st.sidebar.title("Settings")
role = st.sidebar.selectbox("Select Role", ["Data Analyst", "Software Engineer", "Data Scientist"])
question_type = st.sidebar.radio("Question Type", ["Technical", "Behavioral", "Mixed"])

# Initialize session state
if 'questions' not in st.session_state:
    st.session_state.questions = []
if 'current_q' not in st.session_state:
    st.session_state.current_q = 0

# Generate questions
if st.button("Start Interview"):
    st.session_state.questions = generate_questions(role)
    st.session_state.current_q = 0
    st.session_state.answers = []

# Interview loop
if st.session_state.questions:
    q_index = st.session_state.current_q
    if q_index < len(st.session_state.questions):
        st.subheader(f"Question {q_index + 1}")
        st.write(st.session_state.questions[q_index])

        # Input methods
        input_method = st.radio("How would you like to answer?", ["Text", "Voice"])

        if input_method == "Voice":
            if st.button("Record Answer"):
                with st.spinner("Recording..."):
                    answer = get_voice_answer()
                    st.write(f"Your answer: {answer}")
                    st.session_state.answer = answer
        else:
            answer = st.text_area("Your answer:")

        # Submit and evaluate
        if st.button("Submit Answer"):
            if 'answer' in locals():
                feedback = evaluate_answer(st.session_state.questions[q_index], answer)
                st.session_state.feedback = feedback
                st.session_state.answers.append(answer)
                st.session_state.current_q += 1
            elif st.session_state.get('answer'):
                feedback = evaluate_answer(st.session_state.questions[q_index], st.session_state.answer)
                st.session_state.feedback = feedback
                st.session_state.answers.append(st.session_state.answer)
                st.session_state.current_q += 1

    else:
        # End of interview
        st.success("🎉 Interview Complete!")
        st.subheader("Your Answers")
        for i, ans in enumerate(st.session_state.answers):
            with st.expander(f"Question {i+1}"):
                st.write(st.session_state.questions[i])
                st.write("Your answer:", ans)

        if st.button("Get Overall Feedback"):
            overall = evaluate_overall(st.session_state.questions, st.session_state.answers)
            st.write(overall)

# Deploy on Streamlit Cloud
# Requirements:
# streamlit, openai, sounddevice, soundfile, whisper

# Environment variables:
# OPENAI_API_KEY

# Run:
# streamlit run app.py