amine@192: import os
amine@221: import math
amine@88: from random import random
amine@192: from tempfile import TemporaryDirectory
amine@252: from array import array as array_
amine@221: from unittest import TestCase
amine@264: from mock import patch
amine@86: from genty import genty, genty_dataset
amine@207: from auditok import split, AudioRegion, AudioParameterError
amine@215: from auditok.core import _duration_to_nb_windows
amine@212: from auditok.util import AudioDataSource
amine@210: from auditok.io import (
amine@210:     _normalize_use_channel,
amine@210:     _extract_selected_channel,
amine@210:     get_audio_source,
amine@210: )
amine@86: 
amine@86: 
amine@264: def _make_random_length_regions(byte_seq, sampling_rate, sample_width, channels):
amine@88:     regions = []
amine@88:     for b in byte_seq:
amine@88:         duration = round(random() * 10, 6)
amine@95:         data = b * int(duration * sampling_rate) * sample_width * channels
amine@244:         region = AudioRegion(data, sampling_rate, sample_width, channels)
amine@88:         regions.append(region)
amine@88:     return regions
amine@88: 
amine@88: 
amine@86: @genty
amine@215: class TestFunctions(TestCase):
amine@215:     @genty_dataset(
amine@221:         zero_duration=(0, 1, None, 0),
amine@221:         multiple=(0.3, 0.1, round, 3),
amine@221:         not_multiple_ceil=(0.35, 0.1, math.ceil, 4),
amine@221:         not_multiple_floor=(0.35, 0.1, math.floor, 3),
amine@221:         small_duration=(0.05, 0.1, round, 0),
amine@221:         small_duration_ceil=(0.05, 0.1, math.ceil, 1),
amine@233:         with_round_error=(0.3, 0.1, math.floor, 3, {"epsilon": 1e-6}),
amine@221:         negative_duration=(-0.5, 0.1, math.ceil, ValueError),
amine@221:         negative_analysis_window=(0.5, -0.1, math.ceil, ValueError),
amine@215:     )
amine@221:     def test_duration_to_nb_windows(
amine@232:         self, duration, analysis_window, round_fn, expected, kwargs=None
amine@221:     ):
amine@221:         if expected == ValueError:
amine@215:             with self.assertRaises(expected):
amine@221:                 _duration_to_nb_windows(duration, analysis_window, round_fn)
amine@215:         else:
amine@232:             if kwargs is None:
amine@232:                 kwargs = {}
amine@221:             result = _duration_to_nb_windows(
amine@232:                 duration, analysis_window, round_fn, **kwargs
amine@221:             )
amine@215:             self.assertEqual(result, expected)
amine@215: 
amine@215: 
amine@215: @genty
amine@207: class TestSplit(TestCase):
amine@207:     @genty_dataset(
amine@264:         simple=(0.2, 5, 0.2, False, False, {"eth": 50}, [(2, 16), (17, 31), (34, 76)]),
amine@214:         short_max_dur=(
amine@214:             0.3,
amine@214:             2,
amine@214:             0.2,
amine@214:             False,
amine@214:             False,
amine@214:             {"eth": 50},
amine@214:             [(2, 16), (17, 31), (34, 54), (54, 74), (74, 76)],
amine@214:         ),
amine@214:         long_min_dur=(3, 5, 0.2, False, False, {"eth": 50}, [(34, 76)]),
amine@214:         long_max_silence=(0.2, 80, 10, False, False, {"eth": 50}, [(2, 76)]),
amine@214:         zero_max_silence=(
amine@214:             0.2,
amine@214:             5,
amine@214:             0.0,
amine@214:             False,
amine@214:             False,
amine@214:             {"eth": 50},
amine@214:             [(2, 14), (17, 24), (26, 29), (34, 76)],
amine@214:         ),
amine@207:         low_energy_threshold=(
amine@207:             0.2,
amine@207:             5,
amine@207:             0.2,
amine@207:             False,
amine@207:             False,
amine@207:             {"energy_threshold": 40},
amine@207:             [(0, 50), (50, 76)],
amine@207:         ),
amine@264:         high_energy_threshold=(0.2, 5, 0.2, False, False, {"energy_threshold": 60}, []),
amine@207:         trim_leading_and_trailing_silence=(
amine@207:             0.2,
amine@207:             10,  # use long max_dur
amine@207:             0.5,  # and a max_silence longer than any inter-region silence
amine@207:             True,
amine@207:             False,
amine@207:             {"eth": 50},
amine@207:             [(2, 76)],
amine@207:         ),
amine@207:         drop_trailing_silence=(
amine@207:             0.2,
amine@207:             5,
amine@207:             0.2,
amine@207:             True,
amine@207:             False,
amine@207:             {"eth": 50},
amine@207:             [(2, 14), (17, 29), (34, 76)],
amine@207:         ),
amine@264:         drop_trailing_silence_2=(1.5, 5, 0.2, True, False, {"eth": 50}, [(34, 76)]),
amine@207:         strict_min_dur=(
amine@207:             0.3,
amine@207:             2,
amine@207:             0.2,
amine@207:             False,
amine@207:             True,
amine@207:             {"eth": 50},
amine@207:             [(2, 16), (17, 31), (34, 54), (54, 74)],
amine@207:         ),
amine@207:     )
amine@207:     def test_split_params(
amine@207:         self,
amine@207:         min_dur,
amine@207:         max_dur,
amine@207:         max_silence,
amine@207:         drop_trailing_silence,
amine@207:         strict_min_dur,
amine@207:         kwargs,
amine@207:         expected,
amine@207:     ):
amine@207:         with open("tests/data/test_split_10HZ_mono.raw", "rb") as fp:
amine@207:             data = fp.read()
amine@207: 
amine@207:         regions = split(
amine@207:             data,
amine@207:             min_dur,
amine@207:             max_dur,
amine@207:             max_silence,
amine@207:             drop_trailing_silence,
amine@207:             strict_min_dur,
amine@207:             analysis_window=0.1,
amine@207:             sr=10,
amine@207:             sw=2,
amine@207:             ch=1,
amine@207:             **kwargs
amine@207:         )
amine@255: 
amine@255:         region = AudioRegion(data, 10, 2, 1)
amine@255:         regions_ar = region.split(
amine@255:             min_dur,
amine@255:             max_dur,
amine@255:             max_silence,
amine@255:             drop_trailing_silence,
amine@255:             strict_min_dur,
amine@255:             analysis_window=0.1,
amine@255:             **kwargs
amine@255:         )
amine@255: 
amine@207:         regions = list(regions)
amine@255:         regions_ar = list(regions_ar)
amine@207:         err_msg = "Wrong number of regions after split, expected: "
amine@210:         err_msg += "{}, found: {}".format(len(expected), len(regions))
amine@207:         self.assertEqual(len(regions), len(expected), err_msg)
amine@255:         err_msg = "Wrong number of regions after AudioRegion.split, expected: "
amine@255:         err_msg += "{}, found: {}".format(len(expected), len(regions_ar))
amine@255:         self.assertEqual(len(regions_ar), len(expected), err_msg)
amine@207: 
amine@207:         sample_width = 2
amine@255:         for reg, reg_ar, exp in zip(regions, regions_ar, expected):
amine@207:             onset, offset = exp
amine@207:             exp_data = data[onset * sample_width : offset * sample_width]
amine@207:             self.assertEqual(bytes(reg), exp_data)
amine@255:             self.assertEqual(reg, reg_ar)
amine@264:   
amine@211:     @genty_dataset(
amine@241:         stereo_all_default=(2, {}, [(2, 32), (34, 76)]),
amine@213:         mono_max_read=(1, {"max_read": 5}, [(2, 16), (17, 31), (34, 50)]),
amine@213:         mono_max_read_short_name=(1, {"mr": 5}, [(2, 16), (17, 31), (34, 50)]),
amine@211:         mono_use_channel_1=(
amine@211:             1,
amine@241:             {"eth": 50, "use_channel": 0},
amine@211:             [(2, 16), (17, 31), (34, 76)],
amine@211:         ),
amine@211:         mono_uc_1=(1, {"eth": 50, "uc": 1}, [(2, 16), (17, 31), (34, 76)]),
amine@211:         mono_use_channel_None=(
amine@211:             1,
amine@211:             {"eth": 50, "use_channel": None},
amine@211:             [(2, 16), (17, 31), (34, 76)],
amine@211:         ),
amine@211:         stereo_use_channel_1=(
amine@211:             2,
amine@241:             {"eth": 50, "use_channel": 0},
amine@211:             [(2, 16), (17, 31), (34, 76)],
amine@211:         ),
amine@264:         stereo_use_channel_no_use_channel_given=(2, {"eth": 50}, [(2, 32), (34, 76)]),
amine@211:         stereo_use_channel_minus_2=(
amine@211:             2,
amine@211:             {"eth": 50, "use_channel": -2},
amine@211:             [(2, 16), (17, 31), (34, 76)],
amine@211:         ),
amine@241:         stereo_uc_2=(2, {"eth": 50, "uc": 1}, [(10, 32), (36, 76)]),
amine@211:         stereo_uc_minus_1=(2, {"eth": 50, "uc": -1}, [(10, 32), (36, 76)]),
amine@264:         mono_uc_mix=(1, {"eth": 50, "uc": "mix"}, [(2, 16), (17, 31), (34, 76)]),
amine@213:         stereo_use_channel_mix=(
amine@213:             2,
amine@213:             {"energy_threshold": 53.5, "use_channel": "mix"},
amine@213:             [(54, 76)],
amine@213:         ),
amine@213:         stereo_uc_mix=(2, {"eth": 52, "uc": "mix"}, [(17, 26), (54, 76)]),
amine@264:         stereo_uc_mix_default_eth=(2, {"uc": "mix"}, [(10, 16), (17, 31), (36, 76)]),
amine@211:     )
amine@211:     def test_split_kwargs(self, channels, kwargs, expected):
amine@211: 
amine@211:         mono_or_stereo = "mono" if channels == 1 else "stereo"
amine@211:         filename = "tests/data/test_split_10HZ_{}.raw".format(mono_or_stereo)
amine@211:         with open(filename, "rb") as fp:
amine@211:             data = fp.read()
amine@211: 
amine@211:         regions = split(
amine@211:             data,
amine@211:             min_dur=0.2,
amine@211:             max_dur=5,
amine@211:             max_silence=0.2,
amine@211:             drop_trailing_silence=False,
amine@211:             strict_min_dur=False,
amine@211:             analysis_window=0.1,
amine@211:             sr=10,
amine@211:             sw=2,
amine@211:             ch=channels,
amine@211:             **kwargs
amine@211:         )
amine@255: 
amine@255:         region = AudioRegion(data, 10, 2, channels)
amine@255:         regions_ar = region.split(
amine@255:             min_dur=0.2,
amine@255:             max_dur=5,
amine@255:             max_silence=0.2,
amine@255:             drop_trailing_silence=False,
amine@255:             strict_min_dur=False,
amine@255:             analysis_window=0.1,
amine@255:             **kwargs
amine@255:         )
amine@255: 
amine@212:         regions = list(regions)
amine@255:         regions_ar = list(regions_ar)
amine@211:         err_msg = "Wrong number of regions after split, expected: "
amine@241:         err_msg += "{}, found: {}".format(len(expected), len(regions))
amine@211:         self.assertEqual(len(regions), len(expected), err_msg)
amine@255:         err_msg = "Wrong number of regions after AudioRegion.split, expected: "
amine@255:         err_msg += "{}, found: {}".format(len(expected), len(regions_ar))
amine@255:         self.assertEqual(len(regions_ar), len(expected), err_msg)
amine@255: 
amine@255:         sample_width = 2
amine@241:         sample_size_bytes = sample_width * channels
amine@255:         for reg, reg_ar, exp in zip(regions, regions_ar, expected):
amine@212:             onset, offset = exp
amine@264:             exp_data = data[onset * sample_size_bytes : offset * sample_size_bytes]
amine@241:             self.assertEqual(len(bytes(reg)), len(exp_data))
amine@255:             self.assertEqual(reg, reg_ar)
amine@211: 
amine@212:     @genty_dataset(
amine@264:         mono_aw_0_2_max_silence_0_2=(0.2, 5, 0.2, 1, {"aw": 0.2}, [(2, 30), (34, 76)]),
amine@264:         mono_aw_0_2_max_silence_0_3=(0.2, 5, 0.3, 1, {"aw": 0.2}, [(2, 30), (34, 76)]),
amine@264:         mono_aw_0_2_max_silence_0_4=(0.2, 5, 0.4, 1, {"aw": 0.2}, [(2, 32), (34, 76)]),
amine@231:         mono_aw_0_2_max_silence_0=(
amine@231:             0.2,
amine@231:             5,
amine@231:             0,
amine@231:             1,
amine@241:             {"aw": 0.2},
amine@231:             [(2, 14), (16, 24), (26, 28), (34, 76)],
amine@231:         ),
amine@241:         mono_aw_0_2=(0.2, 5, 0.2, 1, {"aw": 0.2}, [(2, 30), (34, 76)]),
amine@231:         mono_aw_0_3_max_silence_0=(
amine@231:             0.3,
amine@231:             5,
amine@231:             0,
amine@231:             1,
amine@241:             {"aw": 0.3},
amine@231:             [(3, 12), (15, 24), (36, 76)],
amine@231:         ),
amine@264:         mono_aw_0_3_max_silence_0_3=(0.3, 5, 0.3, 1, {"aw": 0.3}, [(3, 27), (36, 76)]),
amine@264:         mono_aw_0_3_max_silence_0_5=(0.3, 5, 0.5, 1, {"aw": 0.3}, [(3, 27), (36, 76)]),
amine@264:         mono_aw_0_3_max_silence_0_6=(0.3, 5, 0.6, 1, {"aw": 0.3}, [(3, 30), (36, 76)]),
amine@231:         mono_aw_0_4_max_silence_0=(
amine@231:             0.2,
amine@231:             5,
amine@232:             0,
amine@231:             1,
amine@241:             {"aw": 0.4},
amine@231:             [(4, 12), (16, 24), (36, 76)],
amine@231:         ),
amine@231:         mono_aw_0_4_max_silence_0_3=(
amine@231:             0.2,
amine@231:             5,
amine@231:             0.3,
amine@231:             1,
amine@241:             {"aw": 0.4},
amine@231:             [(4, 12), (16, 24), (36, 76)],
amine@231:         ),
amine@264:         mono_aw_0_4_max_silence_0_4=(0.2, 5, 0.4, 1, {"aw": 0.4}, [(4, 28), (36, 76)]),
amine@241:         stereo_uc_0_analysis_window_0_2=(
amine@241:             0.2,
amine@241:             5,
amine@241:             0.2,
amine@241:             2,
amine@241:             {"uc": 0, "analysis_window": 0.2},
amine@241:             [(2, 30), (34, 76)],
amine@241:         ),
amine@220:         stereo_uc_1_analysis_window_0_2=(
amine@220:             0.2,
amine@220:             5,
amine@220:             0.2,
amine@220:             2,
amine@220:             {"uc": 1, "analysis_window": 0.2},
amine@231:             [(10, 32), (36, 76)],
amine@231:         ),
amine@233:         stereo_uc_mix_aw_0_1_max_silence_0=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.1},
amine@233:             [(10, 14), (17, 24), (26, 29), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_1_max_silence_0_1=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.1,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.1},
amine@233:             [(10, 15), (17, 25), (26, 30), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_1_max_silence_0_2=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.2,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.1},
amine@233:             [(10, 16), (17, 31), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_1_max_silence_0_3=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.3,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.1},
amine@233:             [(10, 32), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_min_dur_0_3=(
amine@233:             0.3,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 14), (16, 24), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_min_dur_0_41=(
amine@233:             0.41,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(16, 24), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_1=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.1,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 14), (16, 24), (26, 28), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_2=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.2,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 30), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_4=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.4,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 32), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_5=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.5,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 32), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_2_max_silence_0_6=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.2},
amine@233:             [(10, 34), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 24), (27, 30), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_min_dur_0_3=(
amine@233:             0.4,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 24), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_6=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 57), (57, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_6_max_dur_5_1=(
amine@233:             0.2,
amine@233:             5.1,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 60), (60, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_6_max_dur_5_2=(
amine@233:             0.2,
amine@233:             5.2,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 60), (60, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_6_max_dur_5_3=(
amine@233:             0.2,
amine@233:             5.3,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 60), (60, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_3_max_silence_0_6_max_dur_5_4=(
amine@233:             0.2,
amine@233:             5.4,
amine@233:             0.6,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.3},
amine@233:             [(9, 63), (63, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_4_max_silence_0=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.4},
amine@233:             [(16, 24), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_4_max_silence_0_3=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.3,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.4},
amine@233:             [(16, 24), (36, 76)],
amine@233:         ),
amine@233:         stereo_uc_mix_aw_0_4_max_silence_0_4=(
amine@233:             0.2,
amine@233:             5,
amine@233:             0.4,
amine@233:             2,
amine@233:             {"uc": "mix", "analysis_window": 0.4},
amine@233:             [(16, 28), (36, 76)],
amine@233:         ),
amine@220:     )
amine@220:     def test_split_analysis_window(
amine@220:         self, min_dur, max_dur, max_silence, channels, kwargs, expected
amine@220:     ):
amine@220: 
amine@220:         mono_or_stereo = "mono" if channels == 1 else "stereo"
amine@220:         filename = "tests/data/test_split_10HZ_{}.raw".format(mono_or_stereo)
amine@220:         with open(filename, "rb") as fp:
amine@220:             data = fp.read()
amine@220: 
amine@220:         regions = split(
amine@220:             data,
amine@220:             min_dur=min_dur,
amine@220:             max_dur=max_dur,
amine@220:             max_silence=max_silence,
amine@220:             drop_trailing_silence=False,
amine@220:             strict_min_dur=False,
amine@220:             sr=10,
amine@220:             sw=2,
amine@220:             ch=channels,
amine@220:             **kwargs
amine@220:         )
amine@255: 
amine@255:         region = AudioRegion(data, 10, 2, channels)
amine@255:         regions_ar = region.split(
amine@255:             min_dur=min_dur,
amine@255:             max_dur=max_dur,
amine@255:             max_silence=max_silence,
amine@255:             drop_trailing_silence=False,
amine@255:             strict_min_dur=False,
amine@255:             **kwargs
amine@255:         )
amine@255: 
amine@220:         regions = list(regions)
amine@255:         regions_ar = list(regions_ar)
amine@255:         err_msg = "Wrong number of regions after split, expected: "
amine@255:         err_msg += "{}, found: {}".format(len(expected), len(regions))
amine@255:         self.assertEqual(len(regions), len(expected), err_msg)
amine@255:         err_msg = "Wrong number of regions after AudioRegion.split, expected: "
amine@255:         err_msg += "{}, found: {}".format(len(expected), len(regions_ar))
amine@255:         self.assertEqual(len(regions_ar), len(expected), err_msg)
amine@255: 
amine@220:         sample_width = 2
amine@255:         sample_size_bytes = sample_width * channels
amine@255:         for reg, reg_ar, exp in zip(regions, regions_ar, expected):
amine@220:             onset, offset = exp
amine@264:             exp_data = data[onset * sample_size_bytes : offset * sample_size_bytes]
amine@220:             self.assertEqual(bytes(reg), exp_data)
amine@255:             self.assertEqual(reg, reg_ar)
amine@220: 
amine@220:     @genty_dataset(
amine@212:         filename_audio_format=(
amine@212:             "tests/data/test_split_10HZ_stereo.raw",
amine@212:             {"audio_format": "raw", "sr": 10, "sw": 2, "ch": 2},
amine@212:         ),
amine@212:         filename_audio_format_short_name=(
amine@212:             "tests/data/test_split_10HZ_stereo.raw",
amine@212:             {"fmt": "raw", "sr": 10, "sw": 2, "ch": 2},
amine@212:         ),
amine@212:         filename_no_audio_format=(
amine@212:             "tests/data/test_split_10HZ_stereo.raw",
amine@212:             {"sr": 10, "sw": 2, "ch": 2},
amine@212:         ),
amine@212:         filename_no_long_audio_params=(
amine@212:             "tests/data/test_split_10HZ_stereo.raw",
amine@212:             {"sampling_rate": 10, "sample_width": 2, "channels": 2},
amine@212:         ),
amine@212:         bytes_=(
amine@212:             open("tests/data/test_split_10HZ_stereo.raw", "rb").read(),
amine@212:             {"sr": 10, "sw": 2, "ch": 2},
amine@212:         ),
amine@212:         audio_reader=(
amine@212:             AudioDataSource(
amine@212:                 "tests/data/test_split_10HZ_stereo.raw",
amine@212:                 sr=10,
amine@212:                 sw=2,
amine@212:                 ch=2,
amine@212:                 block_dur=0.1,
amine@212:             ),
amine@212:             {},
amine@212:         ),
amine@212:         audio_region=(
amine@212:             AudioRegion(
amine@264:                 open("tests/data/test_split_10HZ_stereo.raw", "rb").read(), 10, 2, 2
amine@212:             ),
amine@212:             {},
amine@212:         ),
amine@212:         audio_source=(
amine@212:             get_audio_source(
amine@212:                 "tests/data/test_split_10HZ_stereo.raw", sr=10, sw=2, ch=2
amine@212:             ),
amine@212:             {},
amine@212:         ),
amine@212:     )
amine@212:     def test_split_input_type(self, input, kwargs):
amine@212: 
amine@241:         with open("tests/data/test_split_10HZ_stereo.raw", "rb") as fp:
amine@212:             data = fp.read()
amine@212: 
amine@212:         regions = split(
amine@212:             input,
amine@212:             min_dur=0.2,
amine@212:             max_dur=5,
amine@212:             max_silence=0.2,
amine@212:             drop_trailing_silence=False,
amine@212:             strict_min_dur=False,
amine@212:             analysis_window=0.1,
amine@212:             **kwargs
amine@212:         )
amine@212:         regions = list(regions)
amine@241:         expected = [(2, 32), (34, 76)]
amine@212:         sample_width = 2
amine@212:         err_msg = "Wrong number of regions after split, expected: "
amine@212:         err_msg += "{}, found: {}".format(expected, regions)
amine@212:         self.assertEqual(len(regions), len(expected), err_msg)
amine@211:         for reg, exp in zip(regions, expected):
amine@211:             onset, offset = exp
amine@264:             exp_data = data[onset * sample_width * 2 : offset * sample_width * 2]
amine@211:             self.assertEqual(bytes(reg), exp_data)
amine@211: 
amine@223:     @genty_dataset(
amine@223:         min_dur_greater_than_max_dur=(0.5, 0.4, 0.1),
amine@223:         durations_OK_but_wrong_number_of_analysis_windows=(0.44, 0.49, 0.1),
amine@223:     )
amine@223:     def test_split_wrong_min_max_dur(self, min_dur, max_dur, analysis_window):
amine@223: 
amine@223:         with self.assertRaises(ValueError) as val_err:
amine@223:             split(
amine@223:                 b"0" * 16,
amine@223:                 min_dur=min_dur,
amine@223:                 max_dur=max_dur,
amine@223:                 max_silence=0.2,
amine@223:                 sr=16000,
amine@223:                 sw=1,
amine@223:                 ch=1,
amine@223:                 analysis_window=analysis_window,
amine@223:             )
amine@223: 
amine@223:         err_msg = "'min_dur' ({0} sec.) results in {1} analysis "
amine@223:         err_msg += "window(s) ({1} == ceil({0} / {2})) which is "
amine@223:         err_msg += "higher than the number of analysis window(s) for "
amine@223:         err_msg += "'max_dur' ({3} == floor({4} / {2}))"
amine@223: 
amine@223:         err_msg = err_msg.format(
amine@223:             min_dur,
amine@223:             math.ceil(min_dur / analysis_window),
amine@223:             analysis_window,
amine@223:             math.floor(max_dur / analysis_window),
amine@223:             max_dur,
amine@223:         )
amine@223:         self.assertEqual(err_msg, str(val_err.exception))
amine@223: 
amine@224:     @genty_dataset(
amine@224:         max_silence_equals_max_dur=(0.5, 0.5, 0.1),
amine@224:         max_silence_greater_than_max_dur=(0.5, 0.4, 0.1),
amine@224:         durations_OK_but_wrong_number_of_analysis_windows=(0.44, 0.49, 0.1),
amine@224:     )
amine@224:     def test_split_wrong_max_silence_max_dur(
amine@224:         self, max_silence, max_dur, analysis_window
amine@224:     ):
amine@224: 
amine@224:         with self.assertRaises(ValueError) as val_err:
amine@224:             split(
amine@224:                 b"0" * 16,
amine@224:                 min_dur=0.2,
amine@224:                 max_dur=max_dur,
amine@224:                 max_silence=max_silence,
amine@224:                 sr=16000,
amine@224:                 sw=1,
amine@224:                 ch=1,
amine@224:                 analysis_window=analysis_window,
amine@224:             )
amine@224: 
amine@224:         err_msg = "'max_silence' ({0} sec.) results in {1} analysis "
amine@224:         err_msg += "window(s) ({1} == floor({0} / {2})) which is "
amine@224:         err_msg += "higher or equal to the number of analysis window(s) for "
amine@224:         err_msg += "'max_dur' ({3} == floor({4} / {2}))"
amine@224: 
amine@224:         err_msg = err_msg.format(
amine@224:             max_silence,
amine@224:             math.floor(max_silence / analysis_window),
amine@224:             analysis_window,
amine@224:             math.floor(max_dur / analysis_window),
amine@224:             max_dur,
amine@224:         )
amine@224:         self.assertEqual(err_msg, str(val_err.exception))
amine@224: 
amine@226:     @genty_dataset(
amine@226:         negative_min_dur=({"min_dur": -1},),
amine@226:         zero_min_dur=({"min_dur": 0},),
amine@226:         negative_max_dur=({"max_dur": -1},),
amine@226:         zero_max_dur=({"max_dur": 0},),
amine@226:         negative_max_silence=({"max_silence": -1},),
amine@237:         zero_analysis_window=({"analysis_window": 0},),
amine@237:         negative_analysis_window=({"analysis_window": -1},),
amine@226:     )
amine@226:     def test_split_negative_temporal_params(self, wrong_param):
amine@226: 
amine@237:         params = {
amine@237:             "min_dur": 0.2,
amine@237:             "max_dur": 0.5,
amine@237:             "max_silence": 0.1,
amine@237:             "analysis_window": 0.1,
amine@237:         }
amine@226:         params.update(wrong_param)
amine@226:         with self.assertRaises(ValueError) as val_err:
amine@226:             split(None, **params)
amine@226: 
amine@226:         name = set(wrong_param).pop()
amine@226:         value = wrong_param[name]
amine@226:         err_msg = "'{}' ({}) must be >{} 0".format(
amine@226:             name, value, "=" if name == "max_silence" else ""
amine@226:         )
amine@226:         self.assertEqual(err_msg, str(val_err.exception))
amine@226: 
amine@236:     def test_split_too_small_analysis_window(self):
amine@236:         with self.assertRaises(ValueError) as val_err:
amine@236:             split(b"", sr=10, sw=1, ch=1, analysis_window=0.09)
amine@236:         err_msg = "Too small 'analysis_windows' (0.09) for sampling rate (10)."
amine@236:         err_msg += " Analysis windows should at least be 1/10 to cover one "
amine@236:         err_msg += "single data sample"
amine@236:         self.assertEqual(err_msg, str(val_err.exception))
amine@264:     
amine@264:     def test_split_and_plot(self):
amine@236: 
amine@264:         with open("tests/data/test_split_10HZ_mono.raw", "rb") as fp:
amine@264:             data = fp.read()
amine@264: 
amine@264:         region = AudioRegion(data, 10, 2, 1)
amine@264:         with patch("auditok.core.plot_detections") as patch_fn:
amine@264:             regions = region.split_and_plot(
amine@264:                 min_dur=0.2,
amine@264:                 max_dur=5,
amine@264:                 max_silence=0.2,
amine@264:                 drop_trailing_silence=False,
amine@264:                 strict_min_dur=False,
amine@264:                 analysis_window=0.1,
amine@264:                 sr=10,
amine@264:                 sw=2,
amine@264:                 ch=1,
amine@264:                 eth=50,
amine@264:             )
amine@264:         self.assertTrue(patch_fn.called)
amine@264:         expected = [(2, 16), (17, 31), (34, 76)]
amine@264:         sample_width = 2
amine@264:         expected_regions = []
amine@264:         for (onset, offset) in expected:
amine@264:             onset *= sample_width
amine@264:             offset *= sample_width
amine@264:             expected_regions.append(AudioRegion(data[onset:offset], 10, 2, 1))
amine@264:         self.assertEqual(regions, expected_regions)
amine@207: 
amine@207: @genty
amine@207: class TestAudioRegion(TestCase):
amine@86:     @genty_dataset(
amine@86:         simple=(b"\0" * 8000, 0, 8000, 1, 1, 1, 1, 1000),
amine@264:         one_ms_less_than_1_sec=(b"\0" * 7992, 0, 8000, 1, 1, 0.999, 0.999, 999),
amine@86:         tree_quarter_ms_less_than_1_sec=(
amine@86:             b"\0" * 7994,
amine@86:             0,
amine@86:             8000,
amine@86:             1,
amine@86:             1,
amine@86:             0.99925,
amine@86:             0.99925,
amine@86:             999,
amine@86:         ),
amine@264:         half_ms_less_than_1_sec=(b"\0" * 7996, 0, 8000, 1, 1, 0.9995, 0.9995, 1000),
amine@86:         quarter_ms_less_than_1_sec=(
amine@86:             b"\0" * 7998,
amine@86:             0,
amine@86:             8000,
amine@86:             1,
amine@86:             1,
amine@86:             0.99975,
amine@86:             0.99975,
amine@86:             1000,
amine@86:         ),
amine@86:         simple_sample_width_2=(b"\0" * 8000 * 2, 0, 8000, 2, 1, 1, 1, 1000),
amine@86:         simple_stereo=(b"\0" * 8000 * 2, 0, 8000, 1, 2, 1, 1, 1000),
amine@86:         simple_multichannel=(b"\0" * 8000 * 5, 0, 8000, 1, 5, 1, 1, 1000),
amine@86:         simple_sample_width_2_multichannel=(
amine@86:             b"\0" * 8000 * 2 * 5,
amine@86:             0,
amine@86:             8000,
amine@86:             2,
amine@86:             5,
amine@86:             1,
amine@86:             1,
amine@86:             1000,
amine@86:         ),
amine@86:         one_ms_less_than_1s_sw_2_multichannel=(
amine@86:             b"\0" * 7992 * 2 * 5,
amine@86:             0,
amine@86:             8000,
amine@86:             2,
amine@86:             5,
amine@86:             0.999,
amine@86:             0.999,
amine@86:             999,
amine@86:         ),
amine@86:         tree_qrt_ms_lt_1_s_sw_2_multichannel=(
amine@86:             b"\0" * 7994 * 2 * 5,
amine@86:             0,
amine@86:             8000,
amine@86:             2,
amine@86:             5,
amine@86:             0.99925,
amine@86:             0.99925,
amine@86:             999,
amine@86:         ),
amine@86:         half_ms_lt_1s_sw_2_multichannel=(
amine@86:             b"\0" * 7996 * 2 * 5,
amine@86:             0,
amine@86:             8000,
amine@86:             2,
amine@86:             5,
amine@86:             0.9995,
amine@86:             0.9995,
amine@86:             1000,
amine@86:         ),
amine@86:         quarter_ms_lt_1s_sw_2_multichannel=(
amine@86:             b"\0" * 7998 * 2 * 5,
amine@86:             0,
amine@86:             8000,
amine@86:             2,
amine@86:             5,
amine@86:             0.99975,
amine@86:             0.99975,
amine@86:             1000,
amine@86:         ),
amine@86:         arbitrary_length_1=(
amine@86:             b"\0" * int(8000 * 1.33),
amine@86:             2.7,
amine@86:             8000,
amine@86:             1,
amine@86:             1,
amine@86:             4.03,
amine@86:             1.33,
amine@86:             1330,
amine@86:         ),
amine@86:         arbitrary_length_2=(
amine@86:             b"\0" * int(8000 * 0.476),
amine@86:             11.568,
amine@86:             8000,
amine@86:             1,
amine@86:             1,
amine@86:             12.044,
amine@86:             0.476,
amine@86:             476,
amine@86:         ),
amine@86:         arbitrary_length_sw_2_multichannel=(
amine@86:             b"\0" * int(8000 * 1.711) * 2 * 3,
amine@86:             9.415,
amine@86:             8000,
amine@86:             2,
amine@86:             3,
amine@86:             11.126,
amine@86:             1.711,
amine@86:             1711,
amine@86:         ),
amine@86:         arbitrary_samplig_rate=(
amine@86:             b"\0" * int(3172 * 1.318),
amine@86:             17.236,
amine@86:             3172,
amine@86:             1,
amine@86:             1,
amine@86:             17.236 + int(3172 * 1.318) / 3172,
amine@86:             int(3172 * 1.318) / 3172,
amine@86:             1318,
amine@86:         ),
amine@86:         arbitrary_sr_sw_2_multichannel=(
amine@86:             b"\0" * int(11317 * 0.716) * 2 * 3,
amine@86:             18.811,
amine@86:             11317,
amine@86:             2,
amine@86:             3,
amine@86:             18.811 + int(11317 * 0.716) / 11317,
amine@86:             int(11317 * 0.716) / 11317,
amine@86:             716,
amine@86:         ),
amine@86:     )
amine@86:     def test_creation(
amine@86:         self,
amine@86:         data,
amine@86:         start,
amine@86:         sampling_rate,
amine@86:         sample_width,
amine@86:         channels,
amine@86:         expected_end,
amine@86:         expected_duration_s,
amine@86:         expected_duration_ms,
amine@86:     ):
amine@244:         meta = {"start": start, "end": expected_end}
amine@244:         region = AudioRegion(data, sampling_rate, sample_width, channels, meta)
amine@86:         self.assertEqual(region.sampling_rate, sampling_rate)
amine@86:         self.assertEqual(region.sr, sampling_rate)
amine@86:         self.assertEqual(region.sample_width, sample_width)
amine@86:         self.assertEqual(region.sw, sample_width)
amine@86:         self.assertEqual(region.channels, channels)
amine@86:         self.assertEqual(region.ch, channels)
amine@244:         self.assertEqual(region.meta.start, start)
amine@244:         self.assertEqual(region.meta.end, expected_end)
amine@86:         self.assertEqual(region.duration, expected_duration_s)
amine@245:         self.assertEqual(len(region.ms), expected_duration_ms)
amine@86:         self.assertEqual(bytes(region), data)
amine@88: 
amine@97:     def test_creation_invalid_data_exception(self):
amine@97:         with self.assertRaises(AudioParameterError) as audio_param_err:
amine@97:             _ = AudioRegion(
amine@244:                 data=b"ABCDEFGHI", sampling_rate=8, sample_width=2, channels=1
amine@97:             )
amine@97:         self.assertEqual(
amine@97:             "The length of audio data must be an integer "
amine@97:             "multiple of `sample_width * channels`",
amine@97:             str(audio_param_err.exception),
amine@97:         )
amine@97: 
amine@88:     @genty_dataset(
amine@192:         simple=("output.wav", 1.230, "output.wav"),
amine@244:         start=("output_{meta.start:g}.wav", 1.230, "output_1.23.wav"),
amine@244:         start_2=("output_{meta.start}.wav", 1.233712, "output_1.233712.wav"),
amine@244:         start_3=("output_{meta.start:.2f}.wav", 1.2300001, "output_1.23.wav"),
amine@244:         start_4=("output_{meta.start:.3f}.wav", 1.233712, "output_1.234.wav"),
amine@264:         start_5=("output_{meta.start:.8f}.wav", 1.233712, "output_1.23371200.wav"),
amine@192:         start_end_duration=(
amine@244:             "output_{meta.start}_{meta.end}_{duration}.wav",
amine@192:             1.455,
amine@192:             "output_1.455_2.455_1.0.wav",
amine@192:         ),
amine@192:         start_end_duration_2=(
amine@244:             "output_{meta.start}_{meta.end}_{duration}.wav",
amine@192:             1.455321,
amine@192:             "output_1.455321_2.455321_1.0.wav",
amine@192:         ),
amine@192:     )
amine@192:     def test_save(self, format, start, expected):
amine@192:         with TemporaryDirectory() as tmpdir:
amine@244:             region = AudioRegion(b"0" * 160, 160, 1, 1)
amine@244:             meta = {"start": start, "end": start + region.duration}
amine@244:             region.meta = meta
amine@192:             format = os.path.join(tmpdir, format)
amine@192:             filename = region.save(format)[len(tmpdir) + 1 :]
amine@192:             self.assertEqual(filename, expected)
amine@192: 
amine@193:     def test_save_file_exists_exception(self):
amine@193:         with TemporaryDirectory() as tmpdir:
amine@193:             filename = os.path.join(tmpdir, "output.wav")
amine@193:             open(filename, "w").close()
amine@244:             region = AudioRegion(b"0" * 160, 160, 1, 1)
amine@193:             with self.assertRaises(FileExistsError):
amine@193:                 region.save(filename, exists_ok=False)
amine@193: 
amine@192:     @genty_dataset(
amine@194:         first_half=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@194:             slice(0, 500),
amine@244:             b"a" * 80,
amine@244:         ),
amine@244:         second_half=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(500, None),
amine@244:             b"b" * 80,
amine@244:         ),
amine@244:         second_half_negative=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-500, None),
amine@244:             b"b" * 80,
amine@244:         ),
amine@244:         middle=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(200, 750),
amine@244:             b"a" * 48 + b"b" * 40,
amine@244:         ),
amine@244:         middle_negative=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-800, -250),
amine@244:             b"a" * 48 + b"b" * 40,
amine@244:         ),
amine@244:         middle_sw2=(
amine@244:             AudioRegion(b"a" * 160 + b"b" * 160, 160, 2, 1),
amine@244:             slice(200, 750),
amine@244:             b"a" * 96 + b"b" * 80,
amine@244:         ),
amine@244:         middle_ch2=(
amine@244:             AudioRegion(b"a" * 160 + b"b" * 160, 160, 1, 2),
amine@244:             slice(200, 750),
amine@244:             b"a" * 96 + b"b" * 80,
amine@244:         ),
amine@244:         middle_sw2_ch2=(
amine@244:             AudioRegion(b"a" * 320 + b"b" * 320, 160, 2, 2),
amine@244:             slice(200, 750),
amine@244:             b"a" * 192 + b"b" * 160,
amine@244:         ),
amine@244:         but_first_sample=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(1, None),
amine@244:             b"a" * (4000 - 8) + b"b" * 4000,
amine@244:         ),
amine@244:         but_first_sample_negative=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(-999, None),
amine@244:             b"a" * (4000 - 8) + b"b" * 4000,
amine@244:         ),
amine@244:         but_last_sample=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(0, 999),
amine@244:             b"a" * 4000 + b"b" * (4000 - 8),
amine@244:         ),
amine@244:         but_last_sample_negative=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(0, -1),
amine@244:             b"a" * 4000 + b"b" * (4000 - 8),
amine@244:         ),
amine@244:         big_negative_start=(
amine@244:             AudioRegion(b"a" * 160, 160, 1, 1),
amine@244:             slice(-5000, None),
amine@244:             b"a" * 160,
amine@244:         ),
amine@264:         big_negative_stop=(AudioRegion(b"a" * 160, 160, 1, 1), slice(None, -1500), b""),
amine@264:         empty=(AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1), slice(0, 0), b""),
amine@244:         empty_start_stop_reversed=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(200, 100),
amine@244:             b"",
amine@244:         ),
amine@244:         empty_big_positive_start=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(2000, 3000),
amine@244:             b"",
amine@244:         ),
amine@244:         empty_negative_reversed=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-100, -200),
amine@244:             b"",
amine@244:         ),
amine@244:         empty_big_negative_stop=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(0, -2000),
amine@244:             b"",
amine@244:         ),
amine@244:         arbitrary_sampling_rate=(
amine@244:             AudioRegion(b"a" * 124 + b"b" * 376, 1234, 1, 1),
amine@244:             slice(100, 200),
amine@244:             b"a" + b"b" * 123,
amine@244:         ),
amine@244:     )
amine@244:     def test_region_temporal_slicing(self, region, slice_, expected_data):
amine@244:         sub_region = region.millis[slice_]
amine@244:         self.assertEqual(bytes(sub_region), expected_data)
amine@244:         start_sec = slice_.start / 1000 if slice_.start is not None else None
amine@244:         stop_sec = slice_.stop / 1000 if slice_.stop is not None else None
amine@244:         sub_region = region.sec[start_sec:stop_sec]
amine@244:         self.assertEqual(bytes(sub_region), expected_data)
amine@244: 
amine@244:     @genty_dataset(
amine@244:         first_half=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(0, 80),
amine@194:             0,
amine@194:             b"a" * 80,
amine@194:         ),
amine@194:         second_half=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(80, None),
amine@194:             0.5,
amine@194:             b"b" * 80,
amine@194:         ),
amine@194:         second_half_negative=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-80, None),
amine@194:             0.5,
amine@194:             b"b" * 80,
amine@194:         ),
amine@194:         middle=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(160 // 5, 160 // 4 * 3),
amine@194:             0.2,
amine@194:             b"a" * 48 + b"b" * 40,
amine@194:         ),
amine@194:         middle_negative=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-160 // 5 * 4, -160 // 4),
amine@194:             0.2,
amine@194:             b"a" * 48 + b"b" * 40,
amine@194:         ),
amine@194:         middle_sw2=(
amine@244:             AudioRegion(b"a" * 160 + b"b" * 160, 160, 2, 1),
amine@244:             slice(160 // 5, 160 // 4 * 3),
amine@194:             0.2,
amine@194:             b"a" * 96 + b"b" * 80,
amine@194:         ),
amine@194:         middle_ch2=(
amine@244:             AudioRegion(b"a" * 160 + b"b" * 160, 160, 1, 2),
amine@244:             slice(160 // 5, 160 // 4 * 3),
amine@194:             0.2,
amine@194:             b"a" * 96 + b"b" * 80,
amine@194:         ),
amine@194:         middle_sw2_ch2=(
amine@244:             AudioRegion(b"a" * 320 + b"b" * 320, 160, 2, 2),
amine@244:             slice(160 // 5, 160 // 4 * 3),
amine@194:             0.2,
amine@194:             b"a" * 192 + b"b" * 160,
amine@194:         ),
amine@194:         but_first_sample=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@194:             slice(1, None),
amine@244:             1 / 8000,
amine@244:             b"a" * (4000 - 1) + b"b" * 4000,
amine@194:         ),
amine@194:         but_first_sample_negative=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(-7999, None),
amine@244:             1 / 8000,
amine@244:             b"a" * (4000 - 1) + b"b" * 4000,
amine@194:         ),
amine@194:         but_last_sample=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@244:             slice(0, 7999),
amine@194:             0,
amine@244:             b"a" * 4000 + b"b" * (4000 - 1),
amine@194:         ),
amine@194:         but_last_sample_negative=(
amine@244:             AudioRegion(b"a" * 4000 + b"b" * 4000, 8000, 1, 1),
amine@194:             slice(0, -1),
amine@194:             0,
amine@244:             b"a" * 4000 + b"b" * (4000 - 1),
amine@194:         ),
amine@194:         big_negative_start=(
amine@244:             AudioRegion(b"a" * 160, 160, 1, 1),
amine@244:             slice(-1600, None),
amine@194:             0,
amine@194:             b"a" * 160,
amine@194:         ),
amine@194:         big_negative_stop=(
amine@244:             AudioRegion(b"a" * 160, 160, 1, 1),
amine@244:             slice(None, -1600),
amine@194:             0,
amine@194:             b"",
amine@194:         ),
amine@264:         empty=(AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1), slice(0, 0), 0, b""),
amine@194:         empty_start_stop_reversed=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(80, 40),
amine@244:             0.5,
amine@194:             b"",
amine@194:         ),
amine@194:         empty_big_positive_start=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(1600, 3000),
amine@244:             10,
amine@194:             b"",
amine@194:         ),
amine@194:         empty_negative_reversed=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@244:             slice(-16, -32),
amine@194:             0.9,
amine@194:             b"",
amine@194:         ),
amine@194:         empty_big_negative_stop=(
amine@244:             AudioRegion(b"a" * 80 + b"b" * 80, 160, 1, 1),
amine@194:             slice(0, -2000),
amine@194:             0,
amine@194:             b"",
amine@194:         ),
amine@230:         arbitrary_sampling_rate=(
amine@244:             AudioRegion(b"a" * 124 + b"b" * 376, 1235, 1, 1),
amine@231:             slice(100, 200),
amine@231:             100 / 1235,
amine@231:             b"a" * 24 + b"b" * 76,
amine@231:         ),
amine@231:         arbitrary_sampling_rate_middle_sw2_ch2=(
amine@244:             AudioRegion(b"a" * 124 + b"b" * 376, 1235, 2, 2),
amine@231:             slice(25, 50),
amine@231:             25 / 1235,
amine@231:             b"a" * 24 + b"b" * 76,
amine@231:         ),
amine@231:     )
amine@264:     def test_region_sample_slicing(self, region, slice_, time_shift, expected_data):
amine@231:         sub_region = region[slice_]
amine@231:         self.assertEqual(bytes(sub_region), expected_data)
amine@231: 
amine@231:     @genty_dataset(
amine@88:         simple=(8000, 1, 1),
amine@88:         stereo_sw_2=(8000, 2, 2),
amine@229:         arbitrary_sr_multichannel=(5413, 2, 3),
amine@88:     )
amine@88:     def test_concatenation(self, sampling_rate, sample_width, channels):
amine@88: 
amine@88:         region_1, region_2 = _make_random_length_regions(
amine@88:             [b"a", b"b"], sampling_rate, sample_width, channels
amine@88:         )
amine@88:         expected_duration = region_1.duration + region_2.duration
amine@88:         expected_data = bytes(region_1) + bytes(region_2)
amine@88:         concat_region = region_1 + region_2
amine@264:         self.assertAlmostEqual(concat_region.duration, expected_duration, places=6)
amine@88:         self.assertEqual(bytes(concat_region), expected_data)
amine@88: 
amine@88:     @genty_dataset(
amine@88:         simple=(8000, 1, 1),
amine@88:         stereo_sw_2=(8000, 2, 2),
amine@229:         arbitrary_sr_multichannel=(5413, 2, 3),
amine@88:     )
amine@88:     def test_concatenation_many(self, sampling_rate, sample_width, channels):
amine@88: 
amine@88:         regions = _make_random_length_regions(
amine@88:             [b"a", b"b", b"c"], sampling_rate, sample_width, channels
amine@88:         )
amine@88:         expected_duration = sum(r.duration for r in regions)
amine@88:         expected_data = b"".join(bytes(r) for r in regions)
amine@88:         concat_region = sum(regions)
amine@88: 
amine@264:         self.assertAlmostEqual(concat_region.duration, expected_duration, places=6)
amine@88:         self.assertEqual(bytes(concat_region), expected_data)
amine@88: 
amine@88:     def test_concatenation_different_sampling_rate_error(self):
amine@88: 
amine@244:         region_1 = AudioRegion(b"a" * 100, 8000, 1, 1)
amine@244:         region_2 = AudioRegion(b"b" * 100, 3000, 1, 1)
amine@88: 
amine@88:         with self.assertRaises(ValueError) as val_err:
amine@88:             region_1 + region_2
amine@88:         self.assertEqual(
amine@88:             "Can only concatenate AudioRegions of the same "
amine@88:             "sampling rate (8000 != 3000)",
amine@88:             str(val_err.exception),
amine@88:         )
amine@88: 
amine@88:     def test_concatenation_different_sample_width_error(self):
amine@88: 
amine@244:         region_1 = AudioRegion(b"a" * 100, 8000, 2, 1)
amine@244:         region_2 = AudioRegion(b"b" * 100, 8000, 4, 1)
amine@88: 
amine@88:         with self.assertRaises(ValueError) as val_err:
amine@88:             region_1 + region_2
amine@88:         self.assertEqual(
amine@264:             "Can only concatenate AudioRegions of the same " "sample width (2 != 4)",
amine@88:             str(val_err.exception),
amine@88:         )
amine@88: 
amine@88:     def test_concatenation_different_number_of_channels_error(self):
amine@88: 
amine@244:         region_1 = AudioRegion(b"a" * 100, 8000, 1, 1)
amine@244:         region_2 = AudioRegion(b"b" * 100, 8000, 1, 2)
amine@88: 
amine@88:         with self.assertRaises(ValueError) as val_err:
amine@88:             region_1 + region_2
amine@88:         self.assertEqual(
amine@88:             "Can only concatenate AudioRegions of the same "
amine@88:             "number of channels (1 != 2)",
amine@88:             str(val_err.exception),
amine@88:         )
amine@196: 
amine@196:     @genty_dataset(
amine@245:         simple=(0.01, 0.03, 240, 30),
amine@245:         rounded_len_floor=(0.00575, 0.01725, 138, 17),
amine@245:         rounded_len_ceil=(0.00625, 0.01875, 150, 19),
amine@196:     )
amine@196:     def test_multiplication(
amine@245:         self, duration, expected_duration, expected_len, expected_len_ms
amine@196:     ):
amine@196:         sw = 2
amine@196:         data = b"0" * int(duration * 8000 * sw)
amine@244:         region = AudioRegion(data, 8000, sw, 1)
amine@196:         m_region = 1 * region * 3
amine@196:         self.assertEqual(bytes(m_region), data * 3)
amine@196:         self.assertEqual(m_region.sr, 8000)
amine@196:         self.assertEqual(m_region.sw, 2)
amine@196:         self.assertEqual(m_region.ch, 1)
amine@196:         self.assertEqual(m_region.duration, expected_duration)
amine@245:         self.assertEqual(len(m_region), expected_len)
amine@245:         self.assertEqual(m_region.len, expected_len)
amine@245:         self.assertEqual(m_region.s.len, expected_duration)
amine@245:         self.assertEqual(len(m_region.ms), expected_len_ms)
amine@245:         self.assertEqual(m_region.ms.len, expected_len_ms)
amine@197: 
amine@198:     @genty_dataset(_str=("x", "str"), _float=(1.4, "float"))
amine@197:     def test_multiplication_non_int(self, factor, _type):
amine@197:         with self.assertRaises(TypeError) as type_err:
amine@244:             AudioRegion(b"0" * 80, 8000, 1, 1) * factor
amine@197:             err_msg = "Can't multiply AudioRegion by a non-int of type '{}'"
amine@197:             self.assertEqual(err_msg.format(_type), str(type_err.exception))
amine@254: 
amine@254:     @genty_dataset(
amine@254:         simple=([b"a" * 80, b"b" * 80],),
amine@254:         extra_samples_1=([b"a" * 31, b"b" * 31, b"c" * 30],),
amine@254:         extra_samples_2=([b"a" * 31, b"b" * 30, b"c" * 30],),
amine@254:         extra_samples_3=([b"a" * 11, b"b" * 11, b"c" * 10, b"c" * 10],),
amine@254:     )
amine@252:     def test_truediv(self, data):
amine@254: 
amine@252:         region = AudioRegion(b"".join(data), 80, 1, 1)
amine@252: 
amine@252:         sub_regions = region / len(data)
amine@252:         for data_i, region in zip(data, sub_regions):
amine@252:             self.assertEqual(len(data_i), len(bytes(region)))
amine@254: 
amine@254:     @genty_dataset(
amine@254:         mono_sw_1=(b"a" * 10, 1, 1, "b", [97] * 10),
amine@254:         mono_sw_2=(b"a" * 10, 2, 1, "h", [24929] * 5),
amine@254:         mono_sw_4=(b"a" * 8, 4, 1, "i", [1633771873] * 2),
amine@254:         stereo_sw_1=(b"ab" * 5, 1, 2, "b", [[97] * 5, [98] * 5]),
amine@254:     )
amine@254:     def test_samples(self, data, sample_width, channels, fmt, expected):
amine@254: 
amine@254:         region = AudioRegion(data, 10, sample_width, channels)
amine@254:         if isinstance(expected[0], list):
amine@254:             expected = [array_(fmt, exp) for exp in expected]
amine@254:         else:
amine@254:             expected = array_(fmt, expected)
amine@254:         samples = region.samples
amine@254:         equal = samples == expected
amine@254:         try:
amine@254:             # for numpy
amine@254:             equal = equal.all()
amine@254:         except:
amine@254:             pass
amine@254:         self.assertTrue(equal)