import pickle
import warnings
import numpy as np
import pandas as pd
from models import scrapper
warnings.filterwarnings('ignore')
from nltk import PorterStemmer
from nltk import TweetTokenizer
from deep_translator import GoogleTranslator

pd.options.display.max_colwidth = 1000

def isfloat(num):
    try:
        float(num)
        return True
    except ValueError:
        return False

def preprocessing(text):
    # Translating the text into english
    if not isfloat(text) and not text.isdigit():
        text = GoogleTranslator(source='auto', target='en').translate(text)
    
    #removing punctuations
    preprocessed_text = text.replace("[^a-zA-Z#\s]", "")

    #Removing short words
    preprocessed_text = ' '.join([w for w in preprocessed_text.split() if len(w) > 3])

    #Tokenization
    Tokenizer = TweetTokenizer()
    preprocessed_text = Tokenizer.tokenize(str(preprocessed_text))

    #Stemming
    ps = PorterStemmer()
    preprocessed_text = [ps.stem(letter) for letter in preprocessed_text]

    #Stiching the tokens back
    processed_text = ' '.join(preprocessed_text)
    
    return processed_text

def lr_IR():
    preprocessed_data = []
    original = []
    for i in scrapper.scrapper_func():
        preprocessed_data.append(preprocessing(i))
        original.append(i)

    post_vectorizer = pickle.load(open('models/facebook_whatsapp_posts_vectorizer.pkl', 'rb'))

    # Transforming user input into numerical representation
    post_vectors = post_vectorizer.transform(preprocessed_data)

    post_vectors.todense()
    model = pickle.load(open('models/fb_wa_lr_trained_model.pkl', 'rb'))
    prediction = model.predict(post_vectors)

    combined_list = list(zip(original, preprocessed_data, prediction))

    return combined_list

def lr_BS():
    combined_list = lr_IR()
    original = []
    texts = []
    labels = []

    for i, j, z in combined_list:
        if z == 1:
            original.append(i)
            texts.append(j)
            labels.append(z)
        else:
            continue

    post_vectorizer = pickle.load(open('models/facebook_whatsapp_bs_vectorizer.pkl', 'rb'))

    # Transforming user input into numerical representation
    post_vectors = post_vectorizer.transform(texts)

    post_vectors.todense()
    model = pickle.load(open('models/fb_wa_lr_trained_bs_model.pkl', 'rb'))
    prediction = model.predict(post_vectors)

    updated_prediction = []
    for value in prediction:
        if value == 0:
            updated_prediction.append('Buyer')
        else:
            updated_prediction.append('Seller')


    original_list = []
    for i in original:
        if not isfloat(i) and not i.isdigit():
            original = GoogleTranslator(source='auto', target='en').translate(i)
        original_list.append(original)


    combined_list = list(zip(original_list, updated_prediction))

    return combined_list