import pandas as pd
import numpy as np
from spectro_gen import generate_aiff_spectrogram
from sklearn.model_selection import train_test_split
import tensorflow as tf

def sunflower_data():
    audio_samples = pd.read_csv('./outputs/audio.csv',names=['word','voice','rate','variant','file'])
    sunflowers = audio_samples.loc[audio_samples['word'] == 'sunflowers'].reset_index(drop=True)
    sunflowers.loc[:,'file'] = sunflowers.loc[:,'file'].apply(lambda x:'outputs/'+x).apply(generate_aiff_spectrogram)
    y_data = sunflowers['variant'].apply(lambda x:x=='normal').values
    max_samples = sunflowers['file'].apply(lambda x:x.shape[0]).max()
    sample_size = sunflowers['file'][0].shape[1]
    sample_count = sunflowers['file'].shape[0]
    sunflowers['file'][0].shape[0]
    def append_zeros(spgr):
        orig = spgr.shape[0]
        return np.lib.pad(spgr,[(0, max_samples-orig), (0,0)],'median')
    pad_sun = sunflowers['file'].apply(append_zeros).values
    x_data = np.vstack(pad_sun).reshape((sample_count,max_samples,sample_size,))
    # x_data.shape
    # y_data.shape
    # train_test_split(x_data,y_data,test_size=0.33)[].shape
    # len(train_test_split(x_data,y_data,test_size=0.33))
    # sunflowers.loc[:,'file'][0]
    # generate_aiff_spectrogram('outputs/sunflowers-Alex-150-normal-589.aiff')
    # sunflowers[sunflowers['variant'] == 'phoneme']
    # sunflowers[sunflowers['variant'] == 'normal']
    # for s in sunflowers.values:
    #     print(s)
    return train_test_split(x_data,y_data,test_size=0.33)


def sunflower_pairs_data():
    audio_samples = pd.read_csv('./outputs/audio.csv',names=['word','voice','rate','variant','file'])
    sunflowers = audio_samples.loc[audio_samples['word'] == 'sunflowers'].reset_index(drop=True)
    sunflowers.loc[:,'file'] = sunflowers.loc[:,'file'].apply(lambda x:'outputs/'+x).apply(generate_aiff_spectrogram)
    y_data = sunflowers['variant'].apply(lambda x:x=='normal').values
    max_samples = sunflowers['file'].apply(lambda x:x.shape[0]).max()
    sample_size = sunflowers['file'][0].shape[1]
    sunflowers_pos = sunflowers[sunflowers['variant'] == 'normal'].reset_index(drop=True)
    sunflowers_neg = sunflowers[sunflowers['variant'] == 'phoneme'].reset_index(drop=True)
    def append_zeros(spgr):
        return np.lib.pad(spgr,[(0, max_samples-spgr.shape[0]), (0,0)],'median')
    def create_data(sf):
        sample_count = sf['file'].shape[0]
        pad_sun = sf['file'].apply(append_zeros).values
        x_data = np.vstack(pad_sun).reshape((sample_count,max_samples,sample_size))
        return x_data
    x_data_pos = create_data(sunflowers_pos)
    x_data_neg = create_data(sunflowers_neg)
    x_pos_train, x_pos_test, x_neg_train, x_neg_test =train_test_split(x_data_pos,x_data_neg,test_size=0.33)
    tr_y = np.array(x_pos_train.shape[0]*[[1,0]])
    te_y = np.array(x_pos_test.shape[0]*[[1,0]])
    tr_pairs = np.array([x_pos_train,x_neg_train]).reshape(x_pos_train.shape[0],2,max_samples,sample_size)
    te_pairs = np.array([x_pos_test,x_neg_test]).reshape(x_pos_test.shape[0],2,max_samples,sample_size)
    # x_data.shape
    # y_data.shape
    # train_test_split(x_data,y_data,test_size=0.33)[].shape
    # len(train_test_split(x_data,y_data,test_size=0.33))
    # sunflowers.loc[:,'file'][0]
    # generate_aiff_spectrogram('outputs/sunflowers-Alex-150-normal-589.aiff')
    # sunflowers[sunflowers['variant'] == 'phoneme']
    # sunflowers[sunflowers['variant'] == 'normal']
    # for s in sunflowers.values:
    #     print(s)
    #return train_test_split(x_data,y_data,test_size=0.33)
    return tr_pairs,te_pairs,tr_y,te_y