KaneAI - World’s First E2E Software Testing Agent.Book a Demo Join TestMu Conference

How to use init_new method in autotest

Best Python code snippet using autotest_python

datapreparation.py

Source:datapreparation.py

1from __future__ import division2import numpy as np3import matplotlib.pyplot as plt4import pandas as pd5import pretty_midi6import os7import pypianoroll as pproll8import sys9import IPython10import fluidsynth11import torch12### need also to install fluidsynth to be able to synthesize midi file to audio (pip install fluidsynth)13def piano_roll_to_pretty_midi(piano_roll, fs=100, program=2):14    '''Convert a Piano Roll array into a PrettyMidi object15     with a single instrument.16    Parameters17    ----------18    piano_roll : np.ndarray, shape=(128,frames), dtype=int19        Piano roll of one instrument20    fs : int21        Sampling frequency of the columns, i.e. each column is spaced apart22        by ``1./fs`` seconds.23    program : int24        The program number of the instrument.25    Returns26    -------27    midi_object : pretty_midi.PrettyMIDI28        A pretty_midi.PrettyMIDI class instance describing29        the piano roll.30    '''31    notes, frames = piano_roll.shape32    pm = pretty_midi.PrettyMIDI()33    instrument = pretty_midi.Instrument(program=program)34    # pad 1 column of zeros so we can acknowledge inital and ending events35    piano_roll = np.pad(piano_roll, [(0, 0), (1, 1)], 'constant')36    # use changes in velocities to find note on / note off events37    velocity_changes = np.nonzero(np.diff(piano_roll).T)38    # keep track on velocities and note on times39    prev_velocities = np.zeros(notes, dtype=int)40    note_on_time = np.zeros(notes)41    for time, note in zip(*velocity_changes):42        # use time + 1 because of padding above43        velocity = piano_roll[note, time + 1]44        time = time / fs45        if velocity > 0:46            if prev_velocities[note] == 0:47                note_on_time[note] = time48                prev_velocities[note] = velocity49        else:50            pm_note = pretty_midi.Note(51                velocity=prev_velocities[note],52                pitch=note,53                start=note_on_time[note],54                end=time)55            instrument.notes.append(pm_note)56            prev_velocities[note] = 057    pm.instruments.append(instrument)58    return pm59def midfile_to_piano_roll(filepath,fs=5):60    """ convert a mid file to a piano roll matrix and saves it in a csv file61        input: path to mid file62        output: path to piano_roll csv file63    """64    pm = pretty_midi.PrettyMIDI(filepath)65    pr=pm.get_piano_roll(fs)66    df = pd.DataFrame(pr)67    df.to_csv(filepath[:-3]+"csv")68    return filepath[:-3]+"csv"69def piano_roll_to_mid_file(pianoroll_matrix,fname,fs=5):70    """ input: piano roll matrix with shape (number of notes, time steps)71        output: string with path to mid file72    """73    piano_roll_to_pretty_midi(pianoroll_matrix,fs).write(fname)74    return os.path.join(os.getcwd(),fname)75    76    77def midfile_to_piano_roll_ins(filepath,instrument_n=0,fs=5):78    """ convert mid file to piano_roll csv matrix, but selecting a specific instrument in the mid file79        input: path to mid file, intrument to select in midfile80        output: path to piano_roll csv file81    """82    pm = pretty_midi.PrettyMIDI(filepath)83    pr=pm.instruments[instrument_n].get_piano_roll(fs)84    df = pd.DataFrame(pr)85    df.to_csv(filepath[:-3]+str(instrument_n)+".csv")86    return filepath[:-3]+str(instrument_n)+".csv"87def load_all_dataset(dirpath,binarize=True):88    """ given a diretory finds all the csv in the diretory an load them as numpy arrays89        input: path a diretory90        output: list of numpy arrays91    """92    if(binarize):93        return [(pd.read_csv(os.path.join(dirpath, file)).values>0).astype(int) for file in sorted(os.listdir(dirpath)) if file.endswith(".csv")]94    else:95        return [pd.read_csv(os.path.join(dirpath, file)).values for file in sorted(os.listdir(dirpath)) if file.endswith(".csv")]96def load_all_dataset_names(dirpath):97    """ given a diretory finds all the csv in the d98iretory an split the first part99        of the name of the file to return as a tag for the associated numpy array100        input: path a diretory101        output: list of strings102    """103    return [file.split('_')[0] for file in sorted(os.listdir(dirpath)) if file.endswith(".csv")]104def get_max_length(dataset):105    """ find the maximum length of piano roll matrices in a list of matrices106        input: list of numpy 2d arrays107        output: maximun shape[1] of the list of arrays108        109    """110    return np.max([x.shape[1] for x in dataset])111def get_numkeys(dataset):112    """ return all the number of keys present in the piano roll matrices 113    (typically it should all have the same number of keys and it should be 128)114    input: list of numpy 2d arrays115    output: unique shape[0] of the list of arrays116        117    """118    return np.unique([x.shape[0] for x in dataset])119def visualize_piano_roll(pianoroll_matrix,fs=5):120    """ input: piano roll matrix with shape (number of notes, time steps)121        effect: generates a nice graph with the piano roll visualization122    """123    if(pianoroll_matrix.shape[0]==128):124        pianoroll_matrix=pianoroll_matrix.T.astype(float)125    track = pproll.Track(pianoroll=pianoroll_matrix, program=0, is_drum=False, name='piano roll')   126    # Plot the piano-roll127    fig, ax = track.plot(beat_resolution=fs)128    plt.show()129def embed_play_v1(piano_roll_matrix,fs=5):130    return IPython.display.Audio(data=piano_roll_to_pretty_midi(piano_roll_matrix,fs).synthesize(),rate=44100)131def generate_round(model,tag,n,k=1,init=None):132    if(init is None):133        init = torch.zeros(size=(k,1,model.input_size)).cuda()134    else:135        k = init.shape[0]136    res = init137    hidden = None138    for i in xrange(n//k):139        init,hidden = model.forward(init,tag,hidden)140        #init = torch.round(torch.exp(init))141        init = torch.round(init/torch.max(init))142        res = torch.cat ( ( res, init ) )143    return res144def generate_smooth(model,tag,n,init):145    res = init146    hidden = None147    for i in xrange(n):148        init_new,hidden = model.forward(init,tag,hidden)149        #init = torch.round(torch.exp(init))150        init_new = init_new[-1:]151        init_new = torch.round(init_new/torch.max(init_new))152        res = torch.cat ( ( res, init_new ) )153        init = torch.cat( (init[1:], init_new) )154    return res155def gen_music(model,length=1000,init=None,composer=0,fs=5):156    if(init is None):157        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),length,1)158    else:159        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),length,1,init)160    res = ( song.squeeze(1).detach().cpu().numpy()).astype(int).T161    visualize_piano_roll(res,fs)162    return embed_play_v1(res,fs)163def gen_music_initkeys(model,length=1000,initkeys=40,composer=0,fs=5):164    init = torch.zeros(size=(1,1,model.input_size)).cuda()165    init[0,0,initkeys]=1166    song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),length,1,init)167    res = ( song.squeeze(1).detach().cpu().numpy()).astype(int).T168    visualize_piano_roll(res,fs)169    return embed_play_v1(res,fs)170    #return song171    172def gen_music_pianoroll(model,length=1000,init=None,composer=0,fs=5):173    if(init is None):174        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),length,1)175    else:176        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),length,1,init)177    res = ( song.squeeze(1).detach().cpu().numpy()).astype(int).T178    return res179def gen_music_seconds(model,init,composer=0,fs=5,gen_seconds=10,init_seconds=5):180    if(init is None):181        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),gen_seconds*fs,1)182    else:183        init_index = int(init_seconds*fs) 184        song=generate_round(model, torch.LongTensor([composer]).unsqueeze(1).cuda(),(gen_seconds-init_seconds+1)*fs,1,init[1:(init_index+1)])185    res = ( song.squeeze(1).detach().cpu().numpy()).astype(float).T186    visualize_piano_roll(res,fs)187    return embed_play_v1(res,fs)188def gen_music_seconds_smooth(model,init,composer=0,fs=5,gen_seconds=10,init_seconds=5):189    init_index = int(init_seconds*fs) 190    tag = torch.LongTensor([composer]).unsqueeze(1).cuda()191    song=generate_smooth(model,tag,(gen_seconds-init_seconds+1)*fs,init[1:(init_index+1)])192    res = ( song.squeeze(1).detach().cpu().numpy()).astype(float).T193    visualize_piano_roll(res,fs)194    return embed_play_v1(res,fs)195    196    ...

gradient_descent.py

Source:gradient_descent.py

1#!/usr/bin/env python32# -*- coding: utf-8 -*-3"""4Created on Fri Nov  2 15:04:43 20185@author: vadim6"""7import random8def func(point, f):9    x, y = point10    if f == 1:11        return (1-x) ** 2 + 100 * (y - x ** 2) ** 2 # Rosenbrock12    elif f == 2:13        return (x ** 2 + y - 11) ** 2 + (x + y ** 2 - 7) ** 2 # Himmelblau14    elif f == 3:15        return (x - 4) ** 2 + y ** 216def part_x(point, f):17    x, y = point18    if f == 1:19        return -2 + 2 * x - 400 * x * y + 400 * x ** 320    elif f == 2:21        return 4 * x ** 3 + 2 * y ** 2 + 4 * x * y - 42 * x - 1422    elif f == 3:23        return 2 * x - 824    25def part_y(point, f):26    x, y = point27    if f == 1:28        return 200 * y - 200 * x ** 229    elif f == 2:30        return 4 * y ** 3 + 2 * x ** 2 + 4 * x * y - 26 * y - 2231    elif f == 3:32        return 2 * y33    34def output(point, f):35    x, y = point36    digits = 037    return 'Point: (' + str((round(x, digits), round(y, digits))) + ', F: ' + str(round(f, 0))38    39def GD(init, f, iters = 0):40    rate = 0.000001 # learning rate41    accuracy = 0.0000000000000142    x, y = init43    x_new = x - rate * part_x(init, f)44    y_new = y - rate * part_y(init, f)45    init_new = (x_new, y_new)46    if abs(func(init, f) - func(init_new, f)) < accuracy or iters > 2000:47        return init_new48    else:49        return GD(init_new, f, iters + 1)50#Rosenbrock function51init = (1.1, 1.1)52x, y = GD(init, 1)53f = func((x, y), 1)54print(output((x,y), f) + '\n')55#Himmelblau function56results = []57for _ in range(100):58    init = (random.randint(-5, 5), random.randint(-5, 5))59    x, y = GD(init, 2)60    x = round(x, 2)61    y = round(y, 2)62    results.append((x, y))63results = set(results)64for _, point in enumerate(results):65    f = func(point, 2)...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.