from huggingface_hub import InferenceClient
import os
import random
import json
from flask import Flask, jsonify
from flask import Flask, request, jsonify, redirect, url_for
from flask_cors import CORS
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
from sqlalchemy import create_engine, Column, Integer, String, Boolean, ForeignKey
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy import LargeBinary
from sqlalchemy.orm import sessionmaker, relationship
import json

HF_TOKEN = os.getenv("HF_TOKEN")

client = InferenceClient(model="mistralai/Mixtral-8x7B-Instruct-v0.1", token=HF_TOKEN)

connection_string = "postgresql://neondb_owner:Pl8cWUu0iLHn@ep-tiny-haze-a1w7wrrg.ap-southeast-1.aws.neon.tech/neondb?sslmode=require"

Base = declarative_base()

class Stream(Base):
    __tablename__ = 'streams'

    name = Column(String, primary_key=True, nullable=False)

class Mentor(Base):
    __tablename__ = 'mentors'

    id = Column(Integer, primary_key=True)
    mentor_name = Column(String)
    username = Column(String, unique=True)
    profile_photo = Column(LargeBinary)  
    description = Column(String)
    highest_degree = Column(String)
    expertise = Column(String)
    recent_project = Column(String)
    meeting_time = Column(String)
    fees = Column(String)
    stream_name = Column(String, ForeignKey('streams.name')) 
    country = Column(String)
    verified = Column(Boolean, default=False)
 
    stream = relationship("Stream", backref="mentors") 

 
    stream = relationship("Stream", backref="mentors") 

engine = create_engine(connection_string)
Session = sessionmaker(bind=engine)

app = Flask(__name__)
CORS(app)
    
@app.route('/')
def home():
    return jsonify({"message": "Welcome to the Recommendation API!"})
    

def format_prompt(message):
    # Generate a random user prompt and bot response pair
    user_prompt = "UserPrompt"
    bot_response = "BotResponse"

    return f"<s>[INST] {user_prompt} [/INST] {bot_response}</s> [INST] {message} [/INST]"


@app.route('/ai_mentor', methods=['POST'])
def ai_mentor():
    data = request.get_json()
    message = data.get('message')

    if not message:
        return jsonify({"message": "Missing message"}), 400

    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    # Define prompt for the conversation
    prompt = f""" prompt:
     You are an AI mentor providing concise and complete responses. Answer the user's question clearly and in a few sentences.
    User: {message}"""

    formatted_prompt = format_prompt(prompt)

    try:
        # Generate response from the Language Model
        response = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)

        return jsonify({"response": response}), 200
    except Exception as e:
        return jsonify({"message": f"Failed to process request: {str(e)}"}), 500


@app.route('/get_course', methods=['POST'])
def get_course():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    content = request.json
    # user_degree = content.get('degree')  # Uncomment this line
    user_stream = content.get('stream')

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )
    prompt = f""" prompt: 
    You need to act like as recommendation engine for degree recommendation for a student. Below are current details.
    Stream: {user_stream}
    Based on current details recommend the degree for higher education.
    Note: Output should be list in below format:
    [course1, course2, course3,...]
    Return only answer not prompt and unnecessary stuff, also dont add any special characters or punctuation marks
    """
    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": stream})


@app.route('/get_mentor', methods=['POST'])
def get_mentor():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    content = request.json
    user_stream = content.get('stream')

    session = Session()

    # Query verified mentors
    verified_mentors = session.query(Mentor).filter_by(verified=True).all()

    mentor_list = []
    for mentor in verified_mentors:
        mentor_info = {
            "id": mentor.id,
            "mentor_name": mentor.mentor_name,
            "profile_photo": mentor.profile_photo.decode('utf-8'),  # Decode binary photo to string
            "description": mentor.description,
            "highest_degree": mentor.highest_degree,
            "expertise": mentor.expertise,
            "recent_project": mentor.recent_project,
            "meeting_time": mentor.meeting_time,
            "fees": mentor.fees,
            "stream": mentor.stream,
            "country": mentor.country,
            "verified": mentor.verified
        }
        mentor_list.append(mentor_info)

    session.close()
    
    mentors_data= mentor_list

    temperature = float(temperature)
    if temperature < 1e-2:
        temperature = 1e-2
    top_p = float(top_p)

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    prompt = f""" prompt:
        You need to act as a recommendation engine for mentor recommendations based on the student's stream and a list of available mentors. 
        
        Stream: {user_stream}
        Mentor list: {mentors_data}
        
        Based on the provided details, recommend the mentors that relate to the student's stream. Dont choose mentor outside mentors list 
        
        Note: The output should be a valid list, containing only the mentor's name from attached mentor list. Dont give unnecessary explanations or additional details

    """
    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": stream})


@app.route('/get_streams', methods=['GET'])
def get_streams():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    prompt = """
    You are a recommendation engine. 
    List at least 40 branches of study (e.g., Computer Science, Chemical Engineering, Aerospace). 
    **Output should be a valid JSON array with double quotes, like this:**
    ["Computer Science", "Chemical Engineering", "Aerospace", ...]
    Do not add extra text, explanations, or newlines.
    """

    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)

    try:
        # Ensure the model's response is a valid JSON array
        cleaned_data = stream.strip()

        # Fix incomplete or malformed JSON
        if not cleaned_data.startswith("[") or not cleaned_data.endswith("]"):
            cleaned_data = cleaned_data.split("[", 1)[-1]  # Keep text after first [
            cleaned_data = "[" + cleaned_data  # Add missing opening bracket
            cleaned_data = cleaned_data.rsplit("]", 1)[0]  # Keep text before last ]
            cleaned_data = cleaned_data + "]"  # Add missing closing bracket

        # Parse JSON safely
        parsed_data = json.loads(cleaned_data)

        if not isinstance(parsed_data, list):  # Ensure it's a list
            raise ValueError("Response is not a valid list")

        return jsonify({"ans": parsed_data})  # Return clean JSON list
    except Exception as e:
        return jsonify({"error": "Invalid response format", "details": str(e)})



    

@app.route('/get_education_profiles', methods=['GET'])
def get_education_profiles():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )

    sectors = ["engineering", "medical", "arts", "commerce", "science", "management"]  # Example sectors
    prompt = f"""prompt:
    You need to act like a recommendation engine.
    List all education-related profiles in sectors like {', '.join(sectors)}.
    Note: Output should be a list in the below format:
    [profile1, profile2, profile3,...]
    Return only the answer, not the prompt or unnecessary stuff, and don't add any special characters or punctuation marks.
    """

    formatted_prompt = format_prompt(prompt)

    education_profiles = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": education_profiles})


@app.route('/get_certificate', methods=['POST'])
def get_certificate():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    content = request.json
    # user_degree = content.get('degree')  # Uncomment this line
    user_stream = content.get('stream')

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )
    prompt = f""" prompt: 
    You need to act like as recommendation engine for certification recommendation for a student. Below are current details.
    Stream: {user_stream}
    Based on current details recommend the certification 
    Note: Output should be list in below format:
    [course1, course2, course3,...]
    Return only answer not prompt and unnecessary stuff, also dont add any special characters or punctuation marks
    """
    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": stream})

@app.route('/get_three_streams', methods=['POST'])
def get_three_streams():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    content = request.json
    user_degree = content.get('degree')  # Uncomment this line
    

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )
    prompt = f""" prompt: 
    You need to act like as recommendation engine for stream recommendation for a student based on user degree. Below are details.
    Degree: {user_degree}
    Based on above degree details recommend only 3 the streams 
    Note: Output should be list in below format:
    [stream1, stream2, stream2]
    Return only answer not prompt and unnecessary stuff, also dont add any special characters or punctuation marks
    """
    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": stream})

@app.route('/get_competition', methods=['POST'])
def get_competition():
    temperature = 0.9
    max_new_tokens = 256
    top_p = 0.95
    repetition_penalty = 1.0

    content = request.json
    # user_degree = content.get('degree')  # Uncomment this line
    user_stream = content.get('stream')

    generate_kwargs = dict(
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=repetition_penalty,
        do_sample=True,
        seed=42,
    )
    prompt = f""" prompt: 
    You need to act like as recommendation engine for competition recommendation for a student. Below are current details.
    Stream: {user_stream}
    Based on current details recommend the competition 
    Note: Output should be list in below format:
    [course1, course2, course3,...]
    Return only answer not prompt and unnecessary stuff, also dont add any special characters or punctuation marks
    """
    formatted_prompt = format_prompt(prompt)

    stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=False, details=False, return_full_text=False)
    return jsonify({"ans": stream})


def validate_field(field_name, field_value):
    prompt = f"Check if the following {field_name} is valid: {field_value}. Return 'true' if valid, else 'false'."
    response = client.text_generation(prompt)
    return "true" in response.lower()

@app.route("/validate", methods=["POST"])
def validate():
    data = request.json
    if not data:
        return jsonify({"error": "No data provided"}), 400
    
    validation_results = {}
    for field, value in data.items():
        validation_results[field] = validate_field(field, value)
    
    return jsonify(validation_results)

    
if __name__ == '__main__':
    app.run(debug=True)