import tensorflow as tf
from tensorflow import keras
import pandas as pd
import numpy as np
print("keep in mind: the ai will periodicaly relearn from your chat and its own data, please dont close the ai's console window while it relearns, as the file could get corrupted, and uterly unusable, forcing you to retrain it from scratch, sincirely: the data-finder team(cross)")
data_location = "data/"

# Define the structure of the neural network
def build_model(input_shape):
    model = keras.Sequential([
        keras.layers.Dense(64, activation='relu', input_shape=input_shape),
        keras.layers.Dense(64, activation='relu'),
        keras.layers.Dense(1, activation='sigmoid')
    ])
    model.compile(optimizer='adam',
                  loss='binary_crossentropy',
                  metrics=['accuracy'])
    return model

# Load the training data
def load_data(train_file):
    df = pd.read_csv(train_file)
    print(df)
    X = df['content'].values
    y = df['expected_output'].values
    return X, y

# Convert text data to vectors using an embedding layer
def preprocess_data(X_train, y_train, X_test, y_test, max_features=20000, maxlen=100):
    tokenizer = keras.preprocessing.text.Tokenizer(num_words=max_features)
    tokenizer.fit_on_texts(list(X_train))
    X_train = tokenizer.texts_to_sequences(X_train)
    X_test = tokenizer.texts_to_sequences(X_test)
    X_train = keras.preprocessing.sequence.pad_sequences(X_train, maxlen=maxlen)
    X_test = keras.preprocessing.sequence.pad_sequences(X_test, maxlen=maxlen)
    y_train = np.array(y_train)
    y_test = np.array(y_test)
    return X_train, y_train, X_test, y_test

# Train the neural network on the training data
def train(model, X_train, y_train, X_test, y_test, epochs=10, batch_size=32):
    model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=epochs, batch_size=batch_size)

# Generate a response to user input using the trained model
def generate_response(model, tokenizer, input_text):
    input_vector = tokenizer.texts_to_sequences([input_text])
    input_vector = keras.preprocessing.sequence.pad_sequences(input_vector, maxlen=100)
    output_vector = model.predict(input_vector)
    output_text = "I'm not sure what you're asking. Can you please clarify?"
    if output_vector[0][0] >= 0.5:
        output_text = "Yes, I am Chihiro Fujisaki."
    return output_text

# Main function to run the AI
def run():
    train_file = f"{data_location}initial_data.csv"
    X, y = load_data(train_file)
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
    X_train, y_train, X_test, y_test = preprocess_data(X_train, y_train, X_test, y_test)
    model = build_model(input_shape=(X_train.shape[1],))
    train(model, X_train, y_train, X_test, y_test)
    tokenizer = keras.preprocessing.text.Tokenizer(num_words=20000)
    tokenizer.fit_on_texts(X)
    while True:
        input_text = input("You: ")
        output_text = generate_response(model, tokenizer, input_text)
        print(f"AI: {output_text}")

if __name__ == '__main__':
    run()