debug.py

import os
import pickle
import time
from flask import Flask, render_template, request, jsonify, Response,session, send_file
from googletrans import Translator
from location import LocationExtractor
from date_extractor import extract_date
from map_visualizer import MapVisualizer
import pandas as pd
import ee
import speech_recognition as sr
from gtts import gTTS
from flask_cors import CORS
from google.transliteration import transliterate_word
import json
from gtts import gTTS
from geopy.distance import geodesic
from shapely.geometry import shape, Point
from flask import Flask, send_from_directory
from dotenv import load_dotenv
from groq import Groq
from reservoir import reservoirinfo
import threading
from gesture_detection import GestureControl
import cv2
import keyboard
from sign_language import SignLanguageRecognition
import numpy as np
import face_recognition
from PIL import Image
from keras.models import load_model
import google.generativeai as genai
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
import subprocess
from dotenv import load_dotenv
from groq import Groq
from io import BytesIO
from gtts import gTTS
from langchain.chains import ConversationChain, LLMChain
from langchain_core.prompts import (
    ChatPromptTemplate,
    HumanMessagePromptTemplate,
    MessagesPlaceholder,
)
from langchain_core.messages import SystemMessage
from langchain.chains.conversation.memory import ConversationBufferWindowMemory
from langchain_groq import ChatGroq
import logging

service_account = 'isronrsc@isro-407105.iam.gserviceaccount.com'
credentials = ee.ServiceAccountCredentials(service_account, 'isro-407105-31fe627b6f09.json')
ee.Initialize(credentials)

app = Flask(__name__, static_folder='static')
CORS(app) 

csv_file_path = 'ISROP.csv'
df = pd.read_csv(csv_file_path)
asset_ids = df['ROI_path'].tolist()

map_visualizer = MapVisualizer()
translator = Translator()
audio_text = ""
error = ""
browser_opened = False

gesture_control = GestureControl()
sign_language_recognition = SignLanguageRecognition()
cam = None
webcam_thread = None

with open('static/data/displaytext.json', 'r', encoding='utf-8') as file:
    displayText = json.load(file)


dataset_path = "static/data/dataset"

face_detector = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_frontalface_default.xml')
model1 = load_model('mobile_net_v2_firstmodel.h5')
max_emotion = None
max_count = 0

load_dotenv()

# Load the trained face recognition model
with open('static/data/dataset/trained_faces.pkl', 'rb') as f:
    known_face_encodings, known_face_names = pickle.load(f)

def text_to_speech(text, lang, dest):
    speech = gTTS(text=text, lang=lang)
    speech.save(dest)

# Function to recognize faces
def recognize_face_encoding(unknown_encoding, known_face_encodings, tolerance=0.12):
    name = ""
    min_distance = float('inf')
    
    for idx, known_encoding in enumerate(known_face_encodings):
        distance = face_recognition.face_distance([known_encoding], unknown_encoding)
        flat_distance = distance.flatten()
        print(np.max(flat_distance), known_face_names[idx], min_distance)
        
        if np.max(flat_distance) < min_distance:
            min_distance = np.max(flat_distance)
            name = known_face_names[idx]
    
    if min_distance <= tolerance:
        return name
    

@app.route('/qr', methods=['GET', 'POST'])
def qrscan():
    if request.method == 'POST':
        data = request.get_json()['data']

        timestamp = str(int(time.time()))
        qroutput_file_path = f"static/audio/qroutput/qroutput_{timestamp}.mp3"
        text_to_speech(data, "en", qroutput_file_path)

        return jsonify({'qroutput_file_path': qroutput_file_path})

    return render_template('qr.html')

def predict_emotion(face_image):
    face_image = cv2.imdecode(np.frombuffer(face_image, np.uint8), cv2.IMREAD_COLOR)
    final_image = cv2.resize(face_image, (224, 224))
    final_image = np.expand_dims(final_image, axis=0)
    final_image = final_image / 255.0

    predictions = model1.predict(final_image)

    emotion_labels = ["Angry", "Disgust", "Fear", "Happy", "Surprise", "Sad", "Neutral"]
    predicted_emotion = emotion_labels[np.argmax(predictions)]

    return predicted_emotion

def initialize_bot(current_mood):
    api_key = os.getenv('GENAI_API_KEY')
    genai.configure(api_key = api_key)
    
    title_template = PromptTemplate(input_variables = ['topic'], template = '')
    if current_mood == 'Happy' or current_mood=='Surprise':    
        title_template = PromptTemplate(
        input_variables = ['topic',],
        template = 'You are Kriti, an intelligent and helpful AI chatbot created by NRSC ISRO, designed by dedicated student groups. You are tasked to answer any and all user queries. Greet the user and introduce yourself. The user looks happy from their expressions, set your tone accordingly and compliment them. This is their query{topic}'
        )   
    elif current_mood == 'Sad' or current_mood == 'Fear':
        title_template = PromptTemplate(
        input_variables = ['topic'],
        template = 'You are Kriti, an intelligent and helpful AI chatbot created by NRSC ISRO, designed by dedicated student groups. You are tasked to answer any and all user queries. Greet the user and introduce yourself. The user looks sad from their expressions, comfort them and ask them a bit about it and set your tone accordingly. This is their query {topic}'
        )  
    elif current_mood == 'Angry' or current_mood == 'Disgust':
        title_template = PromptTemplate(
        input_variables = ['topic'],
        template = 'You are Kriti, an intelligent and helpful AI chatbot created by NRSC ISRO, designed by dedicated student groups. You are tasked to answer any and all user queries. Greet the user and introduce yourself. The user looks irritated and unhappy from their expressions, Set your tone accordingly. This is their query {topic}'
        ) 
    elif current_mood == 'Neutral':
        title_template = PromptTemplate(
        input_variables = ['topic'],
        template = 'You are Kriti, an intelligent and helpful AI chatbot created by NRSC ISRO, designed by dedicated student groups. You are tasked to answer any and all user queries. Greet the user in a human-ly manner. This is their query {topic}'
        ) 

    llm = ChatGoogleGenerativeAI(model = 'gemini-pro',google_api_key = api_key, temperature=0.5)
    answer_chain = LLMChain(llm = llm, prompt = title_template, verbose = False)
    
    return answer_chain

def bot_answer(question,current_mood):
    answer_chain = initialize_bot(current_mood)
    bot_response = answer_chain.run(topic = question, current_mood = current_mood)
    return bot_response

def generate_video():
    global webcam_displaying
    video_capture = cv2.VideoCapture("http://127.0.0.1:8000/")

    while webcam_displaying:
        ret, frame = video_capture.read()
        if not ret:
            break

        _, buffer = cv2.imencode('.jpg', frame)
        frame = buffer.tobytes()

        yield (b'--frame\r\n'
               b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')

    video_capture.release()

@app.route('/video_feed')
def video_feed():
    global webcam_displaying
    webcam_displaying = True

    return Response(generate_video(), mimetype='multipart/x-mixed-replace; boundary=frame')

@app.route('/recognize', methods=['GET'])
def recognize_face():
    global detect_face, webcam_displaying, known_face_encodings, known_face_names, max_count, max_emotion

    face_images = []
    capture_interval = 1
    start_time = time.time()

    detect_face = "Welcome! "
    
    # Capture video from webcam
    video_capture = cv2.VideoCapture("http://127.0.0.1:8000/", cv2.CAP_DSHOW)

    while webcam_displaying:
        ret, frame = video_capture.read()
        if ret:
            gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
            faces = face_detector.detectMultiScale(gray, 1.1, 4)

            if len(faces) > 0:
                largest_face = max(faces, key=lambda rect: rect[2] * rect[3])
                x, y, w, h = largest_face

                def expand_roi(x, y, w, h, scale_w, scale_h, img_shape):
                    new_x = max(int(x - w * (scale_w - 1) / 2), 0)
                    new_y = max(int(y - h * (scale_h - 1) / 2), 0)
                    new_w = min(int(w * scale_w), img_shape[1] - new_x)
                    new_h = min(int(h * scale_h), img_shape[0] - new_y)
                    return new_x, new_y, new_w, new_h

                scale_w = 1.3
                scale_h = 1.5

                new_x, new_y, new_w, new_h = expand_roi(x, y, w, h, scale_w, scale_h, frame.shape)
                roi_color = frame[new_y:new_y+new_h, new_x:new_x+new_w]

                if time.time() - start_time >= capture_interval:
                    face_images.append(cv2.imencode('.png', roi_color)[1].tobytes())
                    if len(face_images) > 5:
                        face_images.pop(0)
                    start_time = time.time()
                    
                emotion_counts = {"Angry": 0, "Disgust": 0, "Fear": 0, "Happy": 0, "Surprise": 0, "Sad": 0, "Neutral": 0}
                if len(face_images) >= 4:
                        for face_image in face_images:
                            predicted_emotion = predict_emotion(face_image)
                            emotion_counts[predicted_emotion] += 1

                        max_emotion = max(emotion_counts, key=emotion_counts.get)
                        max_count = emotion_counts[max_emotion]

            if detect_face == "Welcome! ":
                rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

                # Detect faces in the frame
                face_locations = face_recognition.face_locations(rgb_frame)
                face_encodings = face_recognition.face_encodings(rgb_frame, face_locations)
                copy_frame = frame.copy()
                
                for face_encoding, face_location in zip(face_encodings, face_locations):
                    print(face_location)
                    name = recognize_face_encoding(face_encoding, known_face_encodings)

                    if name:
                        if detect_face == "Welcome! ":
                            detect_face += name
                        else:
                            detect_face += ", " + name

                    top, right, bottom, left = face_location

                    cv2.rectangle(copy_frame, (left, top), (right, bottom), (0, 255, 0), 2)
                    cv2.putText(copy_frame, name, (left + 6, bottom - 6), cv2.FONT_HERSHEY_SIMPLEX, 0.75, (0, 255, 0), 2)
                
                cv2.imwrite("static/assets/display.png", copy_frame)
            if len(face_images) >= 4:
                webcam_displaying = False

    video_capture.release()

    if len(detect_face.split(',')) > 1:
        detect_face = detect_face[0:detect_face.rfind(',')] + " and " + detect_face[detect_face.rfind(',')+1:]

    print("Max emotion: ", max_emotion)
    query = 'Who are you?'
    response = bot_answer(query,max_emotion)
    if max_emotion not in ["Happy","Sad","Neutral","Angry",""]:
        max_emotion = "Neutral"

    timestamp = str(int(time.time()))
    reconize_file_path = f"static/audio/recognize/recognize_{timestamp}.mp3"
    text_to_speech(detect_face + response, "en", reconize_file_path)

    return jsonify({'names': detect_face, 'image': "static/assets/display.png", 'reconize_file_path': reconize_file_path, 'emotion': max_emotion, 'response': response})

@app.route('/command', methods=['GET', 'POST'])
def handle_command():
    global detect_face, max_count, max_emotion

    detect_face = ""
    max_count = 0
    max_emotion = None

    return render_template('command.html')

# API endpoint URL
# url = 'http://mt-ulca-api.rb-aai.in/getbulksyncULCA'

# # Request payload
# payload = {
#     "input": [
#         {
#             "source": text
#         }
#     ],
#     "config": {
#         "language": {
#             "sourceLanguage": "en",
#             "targetLanguage": "hi"
#         }
#     }
# }

# response = requests.post(url, json=payload)

# # Check the response
# if response.status_code == 200:
#     return response.json()['output'][0]['target']
# else:
#     print(f'Error: {response.status_code}, {response.text}')

def convert_text_to_text(text, s_lang, d_lang):
    numeral_mappings = {
        # Telugu numerals
        '౧': '1', '౨': '2', '౩': '3', '౪': '4', '౫': '5',
        '౬': '6', '౭': '7', '౮': '8', '౯': '9', '౦': '0',
        # Tamil numerals
        '௧': '1', '௨': '2', '௩': '3', '௪': '4', '௫': '5',
        '௬': '6', '௭': '7', '௮': '8', '௯': '9', '௦': '0',
        # Malayalam numerals
        '൧': '1', '൨': '2', '൩': '3', '൪': '4', '൫': '5',
        '൬': '6', '൭': '7', '൮': '8', '൯': '9', '൦': '0',
        # Gujarati numerals
        '૧': '1', '૨': '2', '૩': '3', '૪': '4', '૫': '5',
        '૬': '6', '૭': '7', '૮': '8', '૯': '9', '૦': '0',
        # Bengali numerals
        '১': '1', '২': '2', '৩': '3', '৪': '4', '৫': '5',
        '৬': '6', '৭': '7', '৮': '8', '৯': '9', '০': '0',
        # Kannada numerals
        '೧': '1', '೨': '2', '೩': '3', '೪': '4', '೫': '5',
        '೬': '6', '೭': '7', '೮': '8', '೯': '9', '೦': '0',
    }

    for numeral, replacement in numeral_mappings.items():
        text = text.replace(numeral, replacement)

    return translator.translate(text=text, src=s_lang, dest=d_lang).text

def text_to_speech(text, lang, dest):
    speech = gTTS(text=text, lang=lang)
    speech.save(dest)

@app.route('/transliterate', methods=['POST'])
def transliterate():
    text_to_transliterate = request.get_json()['text']
    lang_code = displayText[request.get_json()['lang']]['code']

    if text_to_transliterate:
        suggestions = transliterate_word(text_to_transliterate, lang_code=lang_code)        
        return jsonify({'translation': suggestions})
    
    return jsonify({'translation': []})
    
@app.route('/', methods=['GET', 'POST'])
def index():
    global audio_text

    directories = ['static/assets/plot', 'static/audio/command', 'static/audio/qroutput', 'static/audio/recognize', 'static/audio/status', 'static/audio/conclusion']

    for directory in directories:
        for filename in os.listdir(directory):
            if (filename.startswith('doyoumean') or filename.startswith('conclusion') or filename.startswith('status') or filename.startswith('qroutput') or filename.startswith('recognize') or filename.startswith('chart')):
                file_path = os.path.join(directory, filename)
                os.remove(file_path)

    if request.method == 'POST':
        try:
            lang = request.form['lang']
            if lang == "english":
                if audio_text == "":
                    user_text = request.form['user_text']
                else:
                    user_text = audio_text

                if user_text != "":
                    user_text = request.form['user_text']
                    print(user_text)
                    location_extractor = LocationExtractor()
                    selected_roi_name, exact = location_extractor.extract_entities(input_text=user_text)

                    if exact != "exact":
                        timestamp = str(int(time.time()))
                        audio_file_path = f"static/audio/doyoumean/doyoumean_{timestamp}.mp3"

                        text_to_speech(displayText[lang]['dym'] + selected_roi_name, "en", audio_file_path)

                        while not os.path.exists(audio_file_path):
                            time.sleep(0.1)

                        return jsonify({'selection': displayText[lang]['dym'] + selected_roi_name, 
                                        'selected_roi_name': selected_roi_name,
                                        'user_text': user_text,
                                        'audio_file_path': audio_file_path
                                        });
                    
                    return jsonify({
                        'selected_roi_name': selected_roi_name,
                        'user_text': user_text,
                    });
                    
                if audio_text == "" and user_text == "":
                       return jsonify({'status': displayText[lang]['inputIssue']})
                
                return jsonify({'status': displayText[lang]['error']})
                    
            else:
                if audio_text == "":
                    user_text = convert_text_to_text(request.form['user_text'], displayText[lang]['code'], "en")
                else:
                    user_text = convert_text_to_text(audio_text, displayText[lang]['code'], "en")

                if user_text != "":
                    user_text = convert_text_to_text(request.form['user_text'], displayText[lang]['code'], "en")
                    print(user_text)
                    location_extractor = LocationExtractor()
                    selected_roi_name, exact = location_extractor.extract_entities(input_text=user_text)

                    if exact != "exact":
                        timestamp = str(int(time.time()))
                        audio_file_path = f"static/audio/doyoumean/doyoumean_{timestamp}.mp3"

                        converted_roi_name = convert_text_to_text(selected_roi_name, 'en', displayText[lang]['code'])
                        text_to_speech(displayText[lang]['dym'] + converted_roi_name, displayText[lang]['code'], audio_file_path)

                        while not os.path.exists(audio_file_path):
                            time.sleep(0.1)

                        return jsonify({'selection': displayText[lang]['dym'] + converted_roi_name, 
                                        'selected_roi_name': selected_roi_name,
                                        'user_text': user_text,
                                        'audio_file_path': audio_file_path
                                        });
                    
                    return jsonify({
                        'selected_roi_name': selected_roi_name,
                        'user_text': user_text,
                    });
            
                if audio_text == "" and user_text == "":
                       return jsonify({'status': displayText[lang]['inputIssue']})
                
                return jsonify({'status': displayText[lang]['error']})
                   
        except Exception as e:
            return jsonify({'status': 'Exception found', 'message': str(e)})

    # For GET requests, render the template without map_html and conclusion
    return render_template('index.html')

@app.route('/date', methods=['POST'])
def date_gatherer():
    global audio_text
    
    lang = request.get_json()['lang']
    user_text = request.get_json()['user_text']
    start_date, end_date = extract_date(convert_text_to_text(user_text, displayText[lang]['code'], "en"))
    audio_text = ""

    text_to_speech(displayText[lang]['query'] + user_text + ".", displayText[lang]['code'], "static/audio/languages/" + lang + "/step2.mp3")

    return jsonify({
        'start_date': start_date,
        'end_date': end_date
    })

@app.route('/map', methods=['POST'])
def map__reservior():
    global audio_text
    
    lang = request.get_json()['lang']
    start_date, end_date, selected_roi_name = request.get_json()['start_date'], request.get_json()['end_date'], request.get_json()['selected_roi_name']

    # Run the analysis in the background
    static_map, conclusion, chart_file_path, status, maperror = map_visualizer.run_analysis(
        asset_ids, selected_roi_name, start_date, end_date, csv_file_path, lang
    )
    audio_text = ""

    # file_name = 'static_map.html'

    # # Write the HTML content to the file
    # with open(file_name, 'w') as file:
    #     file.write(static_map.to_html())

    # print(f"HTML file '{file_name}' has been created.")

    if status == "" and maperror == "":
        return jsonify({
            'map_html': static_map,
            'conclusion': conclusion,
            'conclusion_file_path': "",
            'chart_file_path': chart_file_path,
            'status': status,
            'error': displayText[lang]['reservoirIssue']
        })

    if static_map != None:
        # Convert Earth Engine map to HTML code
        map_html = static_map.to_html()
        converted_conclusion = ""

        timestamp = str(int(time.time()))
        conclusion_file_path = f"static/audio/conclusion/conclusion_{timestamp}.mp3"

        if lang == "english":
            converted_conclusion = conclusion
            text_to_speech(conclusion, 'en', conclusion_file_path)
        else:
            for con in conclusion.split(". "):
                converted_conclusion += convert_text_to_text(con, "en", displayText[lang]['code']) + ". "
                
            text_to_speech(converted_conclusion, displayText[lang]['code'], conclusion_file_path)

        return jsonify({
            'map_html': map_html,
            'conclusion': converted_conclusion,
            'conclusion_file_path': conclusion_file_path,
            'chart_file_path': chart_file_path,
            'status': status,
            'error': maperror
        })
    
    else:
        converted_status = ""

        timestamp = str(int(time.time()))
        status_file_path = f"static/audio/status/status_{timestamp}.mp3"

        if lang == "english":
            converted_status = status
            text_to_speech(status, 'en', status_file_path)
        else:
            for con in status.split(". "):
                converted_status += convert_text_to_text(con, "en", displayText[lang]['code']) + ". "
                
            text_to_speech(converted_status, displayText[lang]['code'], status_file_path)

        return jsonify({
            'map_html': static_map,
            'conclusion': conclusion,
            'conclusion_file_path': status_file_path,
            'chart_file_path': chart_file_path,
            'status': converted_status,
            'error': maperror
        })

@app.route('/timelapse', methods=['POST'])
def timelapse():
    selected_roi_name = request.get_json()['selected_roi_name']

    total_chunks, video_thumb_url, num_chunk = map_visualizer.timelapse(
        asset_ids, selected_roi_name
    )

    return jsonify({
        "total_chunks": total_chunks,
        "video_thumb_url": video_thumb_url,
        "num_chunk": num_chunk
    })

@app.route('/process', methods=['POST'])
def processchunk():
    total_chunks, num_chunk = request.get_json()['total_chunks'], request.get_json()['num_chunk']

    total_chunks, video_thumb_url, num_chunk = map_visualizer.process_each_chunk(
        total_chunks, num_chunk
    )

    return jsonify({
        "video_thumb_url": video_thumb_url,
        "num_chunk": num_chunk
    })

@app.route('/about', methods=['GET'])
def about():
    return render_template('about.html')

@app.route('/info', methods=['GET'])
def info():
    return render_template('info.html')

@app.route('/reservoirs', methods=['POST'])
def reservoirs():
    return reservoirinfo(request.get_json()['type'])

@app.route('/startstop', methods=['POST'])
def startstop():
    lang = request.get_json()['lang']
    type_audio = request.get_json()['type_audio']

    global audio_text
    global error
    
    recognizer = sr.Recognizer()
    audio_text = ""
    error = ""
    user_ended = True

    if type_audio == "start":
        user_ended = False
    else:
        user_ended = True

    stop = False

    while not stop and not user_ended:
        with sr.Microphone() as source:
            recognizer.adjust_for_ambient_noise(source)  # Adjust for ambient noise
            try:
                audio = recognizer.listen(source)
                print("Recognizing...")
                audio_text = recognizer.recognize_google(audio, language=displayText[lang]['audioCode'])
                print("You said:", audio_text)

                if audio_text != "" or user_ended:
                    stop = True
            except sr.UnknownValueError:
                print("Could not understand audio")
                error = displayText[lang]['audioIssue']
                stop = True
            except sr.RequestError as e:
                print(f"Could not request results from API {e}")
                error = displayText[lang]['apiIssue']
                stop = True
            except sr.WaitTimeoutError:
                print("Microphone timeout. Stopping...")

    return jsonify({'message': audio_text, "error": error})


@app.route('/start_webcam', methods=['GET'])
def api_start_webcam():
    global cam, webcam_thread

    # Start webcam
    cam = cv2.VideoCapture(0)
    gesture_control.set_webcam(cam)

    # Start the gesture control in a separate thread
    webcam_thread = threading.Thread(target=gesture_control.start)
    webcam_thread.start()

    print("Webcam started")
    return jsonify("Webcam started successfully")

@app.route('/stop_webcam', methods=['GET'])
def api_stop_webcam():
    global cam, webcam_thread

    # Stop both recognition processes
    gesture_control.stop()
    sign_language_recognition.stop()

    # Join the webcam thread if it is alive
    if webcam_thread is not None and webcam_thread.is_alive():
        webcam_thread.join()

    # Release the webcam
    if cam is not None:
        cam.release()
        cam = None

    print("Webcam stopped")
    return jsonify("Webcam stopped successfully")

@app.route('/start_sign', methods=['GET'])
def api_start_sign():
    global cam, webcam_thread

    # Stop gesture control if it is running
    gesture_control.stop()
    
    cam = cv2.VideoCapture(0)
    sign_language_recognition.set_webcam(cam)

    # Start sign language recognition in a separate thread
    webcam_thread = threading.Thread(target=sign_language_recognition.start)
    webcam_thread.start()

    print("Sign language recognition started")
    return jsonify("Sign started successfully")

@app.route('/stop_sign', methods=['GET'])
def api_stop_sign():
    global cam, webcam_thread

    # Stop the sign language recognition process
    sign_language_recognition.stop()

    # Restart gesture control
    cam = cv2.VideoCapture(0)
    gesture_control.set_webcam(cam)
    webcam_thread = threading.Thread(target=gesture_control.start)
    webcam_thread.start()

    print("Sign language recognition stopped")
    return jsonify("Sign stopped successfully")

@app.route('/typed_text', methods=['GET'])
def get_typed_text():
    global sign_language_recognition
    return jsonify({"typed_text": sign_language_recognition.typed_text})

# Load reservoir data
reservoirs = pd.read_csv('ISROP.csv')  # Ensure this CSV file is correctly formatted


# Function to find the nearest reservoir within 10 km
def find_nearest_reservoir(lat, lon):
    min_distance = float('inf')
    nearest_reservoir = None
    for index, row in reservoirs.iterrows():
        reservoir_location = (row['Latitude'], row['Longitude'])  # Ensure these column names match your CSV
        distance = geodesic((lat, lon), reservoir_location).km
        if distance < 10 and distance < min_distance:  # Limit search within 10 km radius
            min_distance = distance
            nearest_reservoir = row
    return nearest_reservoir

# Function to calculate a buffer around a reservoir boundary
def calculate_buffer(roi_name):
    # Get the Earth Engine asset path for the reservoir based on roi_name
    asset_id = f'projects/isro-407105/assets/{roi_name}'
    try:
        ee_shapefile = ee.FeatureCollection(asset_id)
        # Buffer distance in meters (10 km = 10000 meters)
        buffered_shapefile = ee_shapefile.geometry().buffer(10000)
        return buffered_shapefile
    except Exception as e:
        print(f"Error calculating buffer for {roi_name}: {e}")
        return None


# Function to get the GeoJSON of the reservoir shapefile from Google Earth Engine
def get_reservoir_geojson(roi_name):
    # Get the Earth Engine asset path for the reservoir based on roi_name
    asset_id = f'projects/isro-407105/assets/{roi_name}'
    try:
        ee_shapefile = ee.FeatureCollection(asset_id)
        geojson = ee_shapefile.getInfo()
        return geojson
    except Exception as e:
        print(f"Error fetching GeoJSON for {roi_name}: {e}")
        return None
    
# # Route to serve reservoir.html
@app.route('/reservoir')
def reservoir():
    return render_template('reservoir.html')
    # return send_from_directory('templates', 'reservoir.html')

# Route to find nearest reservoir
@app.route('/find_nearest_reservoir', methods=['POST'])
def find_nearest_reservoir_endpoint():
    request_data = request.get_json()
    lat = float(request_data['lat'])
    lon = float(request_data['lon'])
    
    # Calculate buffers and find the nearest reservoir
    nearest_reservoir = None
    min_distance = float('inf')
    for index, row in reservoirs.iterrows():
        reservoir_location = (row['Latitude'], row['Longitude'])  # Ensure these column names match your CSV
        distance = geodesic((lat, lon), reservoir_location).km
        
        # Calculate buffer for the reservoir
        buffered_shapefile = calculate_buffer(row['ROI_Name'])
        
        if buffered_shapefile is not None:
            # Check if the click point is within the buffered area
            if buffered_shapefile.contains(ee.Geometry.Point(lon, lat)):
                # Update nearest reservoir if closer
                if distance < min_distance:
                    min_distance = distance
                    nearest_reservoir = row
    
    if nearest_reservoir is not None:
        geojson = get_reservoir_geojson(nearest_reservoir['ROI_Name'])
        if geojson:
            return jsonify({
                'name': nearest_reservoir['ROI_Name'],
                'state': nearest_reservoir['State'],
                'latitude': nearest_reservoir['Latitude'],
                'longitude': nearest_reservoir['Longitude'],
                'color': nearest_reservoir['Color'],
                'geojson': geojson,  # Return the GeoJSON data
                'distance': min_distance
            })
        else:
            return jsonify({
                'name': nearest_reservoir['ROI_Name'],
                'state': nearest_reservoir['State'],
                'latitude': nearest_reservoir['Latitude'],
                'longitude': nearest_reservoir['Longitude'],
                'color': nearest_reservoir['Color'],
                'error': f"Shapefile for {nearest_reservoir['ROI_Name']} not found in Earth Engine."
            })
    else:
        return jsonify({'error': 'No reservoir found within 10 km.'})
    
# Route to handle chat query
@app.route('/search')
def search():
    query = request.args.get('query', '')
    # Render the chat page or handle the query as needed
    return render_template('index.html', query=query)


client = Groq()

data = {}


#@app.route('/voice_assistant')
#def voice_assistant():
#   return render_template('voice_assistant.html')

@app.route('/chat', methods=['POST'])
def chat():
    user_input = request.json.get('message')
    target_language = request.json.get('language')

    try:
    
        response_text = process_query(user_input)
        
        if not response_text:
            chat_completion = client.chat.completions.create(
                messages=[
                    {
                        "role": "user",
                        "content": user_input,
                    }
                ],
                model="gemma2-9b-it",
            )
            response_text = chat_completion.choices[0].message.content
                        
            if not response_text.strip():
                response_text = "Sorry, I don't have a response for that."
        
        
        if target_language != 'en':
            translation = translate_text(response_text, target_language)
            response_text = translation
        
        return jsonify({'response': response_text})
    
    except Exception as e:
        return jsonify({'error': str(e)}), 500

def process_query(input_text):
    max_query_length = 500  
    responses = []

    while input_text:
        query_chunk = input_text[:max_query_length]
        input_text = input_text[max_query_length:]
        
        chat_completion = client.chat.completions.create(
            messages=[{"role": "user", "content": query_chunk}],
            model="gemma2-9b-it"
        )
        response_text = chat_completion.choices[0].message.content
        responses.append(response_text)

    return ' '.join(responses)

def translate_text(text, target_language):
    
    return text

app.secret_key = 'supersecretkey'  # Needed for session management

# Define the Groq API key
groq_api_key = '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'  # Replace with your actual API key

# Initialize Groq Langchain chat object and conversation memory
model = 'llama-3.3-70b-versatile'
conversational_memory_length = 5
memory = ConversationBufferWindowMemory(k=conversational_memory_length, memory_key="chat_history", return_messages=True)

system_prompt = """
You are Kriti, an intelligent and helpful AI chatbot created by ISRO, designed by dedicated student groups. Provide accurate, concise, and helpful information in brief responses. Only refer to your identity when explicitly asked.
"""

@app.route("/voiceassistant", methods=["GET", "POST"])
def va():
    if request.method == "POST":
        user_input = request.get_json().get("question")
        if user_input:
            if 'chat_history' not in session:
                session['chat_history'] = []

            for message in session.get("chat_history", []):
                memory.save_context(
                    {"input": message["human"]},
                    {"output": message["AI"]}
                )
            
            groq_chat = ChatGroq(groq_api_key=groq_api_key, model_name=model)
            prompt = ChatPromptTemplate.from_messages(
                [
                    SystemMessage(content=system_prompt),
                    MessagesPlaceholder(variable_name="chat_history"),
                    HumanMessagePromptTemplate.from_template("{human_input}"),
                ]
            )

            conversation = LLMChain(
                llm=groq_chat,
                prompt=prompt,
                verbose=True,
                memory=memory,
            )

            response = conversation.predict(human_input=user_input)
            message = {"human": user_input, "AI": response}
            session["chat_history"].append(message)

            return jsonify({"response": response})
    
    # For GET requests, render the HTML template
    return render_template("voiceassistant.html", chat_history=session.get("chat_history", []))

@app.route("/voiceassistant/audio")
def audio():
    response = request.args.get("response", "")
    tts = gTTS(text=response, lang='en')
    audio = BytesIO()
    tts.write_to_fp(audio)
    audio.seek(0)
    return send_file(audio, mimetype="audio/mp3")


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