566 lines
28 KiB
Python
566 lines
28 KiB
Python
|
import audioop
|
||
|
import sys
|
||
|
import unittest
|
||
|
|
||
|
def pack(width, data):
|
||
|
return b''.join(v.to_bytes(width, sys.byteorder, signed=True) for v in data)
|
||
|
|
||
|
def unpack(width, data):
|
||
|
return [int.from_bytes(data[i: i + width], sys.byteorder, signed=True)
|
||
|
for i in range(0, len(data), width)]
|
||
|
|
||
|
packs = {w: (lambda *data, width=w: pack(width, data)) for w in (1, 2, 3, 4)}
|
||
|
maxvalues = {w: (1 << (8 * w - 1)) - 1 for w in (1, 2, 3, 4)}
|
||
|
minvalues = {w: -1 << (8 * w - 1) for w in (1, 2, 3, 4)}
|
||
|
|
||
|
datas = {
|
||
|
1: b'\x00\x12\x45\xbb\x7f\x80\xff',
|
||
|
2: packs[2](0, 0x1234, 0x4567, -0x4567, 0x7fff, -0x8000, -1),
|
||
|
3: packs[3](0, 0x123456, 0x456789, -0x456789, 0x7fffff, -0x800000, -1),
|
||
|
4: packs[4](0, 0x12345678, 0x456789ab, -0x456789ab,
|
||
|
0x7fffffff, -0x80000000, -1),
|
||
|
}
|
||
|
|
||
|
INVALID_DATA = [
|
||
|
(b'abc', 0),
|
||
|
(b'abc', 2),
|
||
|
(b'ab', 3),
|
||
|
(b'abc', 4),
|
||
|
]
|
||
|
|
||
|
|
||
|
class TestAudioop(unittest.TestCase):
|
||
|
|
||
|
def test_max(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.max(b'', w), 0)
|
||
|
self.assertEqual(audioop.max(bytearray(), w), 0)
|
||
|
self.assertEqual(audioop.max(memoryview(b''), w), 0)
|
||
|
p = packs[w]
|
||
|
self.assertEqual(audioop.max(p(5), w), 5)
|
||
|
self.assertEqual(audioop.max(p(5, -8, -1), w), 8)
|
||
|
self.assertEqual(audioop.max(p(maxvalues[w]), w), maxvalues[w])
|
||
|
self.assertEqual(audioop.max(p(minvalues[w]), w), -minvalues[w])
|
||
|
self.assertEqual(audioop.max(datas[w], w), -minvalues[w])
|
||
|
|
||
|
def test_minmax(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.minmax(b'', w),
|
||
|
(0x7fffffff, -0x80000000))
|
||
|
self.assertEqual(audioop.minmax(bytearray(), w),
|
||
|
(0x7fffffff, -0x80000000))
|
||
|
self.assertEqual(audioop.minmax(memoryview(b''), w),
|
||
|
(0x7fffffff, -0x80000000))
|
||
|
p = packs[w]
|
||
|
self.assertEqual(audioop.minmax(p(5), w), (5, 5))
|
||
|
self.assertEqual(audioop.minmax(p(5, -8, -1), w), (-8, 5))
|
||
|
self.assertEqual(audioop.minmax(p(maxvalues[w]), w),
|
||
|
(maxvalues[w], maxvalues[w]))
|
||
|
self.assertEqual(audioop.minmax(p(minvalues[w]), w),
|
||
|
(minvalues[w], minvalues[w]))
|
||
|
self.assertEqual(audioop.minmax(datas[w], w),
|
||
|
(minvalues[w], maxvalues[w]))
|
||
|
|
||
|
def test_maxpp(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.maxpp(b'', w), 0)
|
||
|
self.assertEqual(audioop.maxpp(bytearray(), w), 0)
|
||
|
self.assertEqual(audioop.maxpp(memoryview(b''), w), 0)
|
||
|
self.assertEqual(audioop.maxpp(packs[w](*range(100)), w), 0)
|
||
|
self.assertEqual(audioop.maxpp(packs[w](9, 10, 5, 5, 0, 1), w), 10)
|
||
|
self.assertEqual(audioop.maxpp(datas[w], w),
|
||
|
maxvalues[w] - minvalues[w])
|
||
|
|
||
|
def test_avg(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.avg(b'', w), 0)
|
||
|
self.assertEqual(audioop.avg(bytearray(), w), 0)
|
||
|
self.assertEqual(audioop.avg(memoryview(b''), w), 0)
|
||
|
p = packs[w]
|
||
|
self.assertEqual(audioop.avg(p(5), w), 5)
|
||
|
self .assertEqual(audioop.avg(p(5, 8), w), 6)
|
||
|
self.assertEqual(audioop.avg(p(5, -8), w), -2)
|
||
|
self.assertEqual(audioop.avg(p(maxvalues[w], maxvalues[w]), w),
|
||
|
maxvalues[w])
|
||
|
self.assertEqual(audioop.avg(p(minvalues[w], minvalues[w]), w),
|
||
|
minvalues[w])
|
||
|
self.assertEqual(audioop.avg(packs[4](0x50000000, 0x70000000), 4),
|
||
|
0x60000000)
|
||
|
self.assertEqual(audioop.avg(packs[4](-0x50000000, -0x70000000), 4),
|
||
|
-0x60000000)
|
||
|
|
||
|
def test_avgpp(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.avgpp(b'', w), 0)
|
||
|
self.assertEqual(audioop.avgpp(bytearray(), w), 0)
|
||
|
self.assertEqual(audioop.avgpp(memoryview(b''), w), 0)
|
||
|
self.assertEqual(audioop.avgpp(packs[w](*range(100)), w), 0)
|
||
|
self.assertEqual(audioop.avgpp(packs[w](9, 10, 5, 5, 0, 1), w), 10)
|
||
|
self.assertEqual(audioop.avgpp(datas[1], 1), 196)
|
||
|
self.assertEqual(audioop.avgpp(datas[2], 2), 50534)
|
||
|
self.assertEqual(audioop.avgpp(datas[3], 3), 12937096)
|
||
|
self.assertEqual(audioop.avgpp(datas[4], 4), 3311897002)
|
||
|
|
||
|
def test_rms(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.rms(b'', w), 0)
|
||
|
self.assertEqual(audioop.rms(bytearray(), w), 0)
|
||
|
self.assertEqual(audioop.rms(memoryview(b''), w), 0)
|
||
|
p = packs[w]
|
||
|
self.assertEqual(audioop.rms(p(*range(100)), w), 57)
|
||
|
self.assertAlmostEqual(audioop.rms(p(maxvalues[w]) * 5, w),
|
||
|
maxvalues[w], delta=1)
|
||
|
self.assertAlmostEqual(audioop.rms(p(minvalues[w]) * 5, w),
|
||
|
-minvalues[w], delta=1)
|
||
|
self.assertEqual(audioop.rms(datas[1], 1), 77)
|
||
|
self.assertEqual(audioop.rms(datas[2], 2), 20001)
|
||
|
self.assertEqual(audioop.rms(datas[3], 3), 5120523)
|
||
|
self.assertEqual(audioop.rms(datas[4], 4), 1310854152)
|
||
|
|
||
|
def test_cross(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.cross(b'', w), -1)
|
||
|
self.assertEqual(audioop.cross(bytearray(), w), -1)
|
||
|
self.assertEqual(audioop.cross(memoryview(b''), w), -1)
|
||
|
p = packs[w]
|
||
|
self.assertEqual(audioop.cross(p(0, 1, 2), w), 0)
|
||
|
self.assertEqual(audioop.cross(p(1, 2, -3, -4), w), 1)
|
||
|
self.assertEqual(audioop.cross(p(-1, -2, 3, 4), w), 1)
|
||
|
self.assertEqual(audioop.cross(p(0, minvalues[w]), w), 1)
|
||
|
self.assertEqual(audioop.cross(p(minvalues[w], maxvalues[w]), w), 1)
|
||
|
|
||
|
def test_add(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.add(b'', b'', w), b'')
|
||
|
self.assertEqual(audioop.add(bytearray(), bytearray(), w), b'')
|
||
|
self.assertEqual(audioop.add(memoryview(b''), memoryview(b''), w), b'')
|
||
|
self.assertEqual(audioop.add(datas[w], b'\0' * len(datas[w]), w),
|
||
|
datas[w])
|
||
|
self.assertEqual(audioop.add(datas[1], datas[1], 1),
|
||
|
b'\x00\x24\x7f\x80\x7f\x80\xfe')
|
||
|
self.assertEqual(audioop.add(datas[2], datas[2], 2),
|
||
|
packs[2](0, 0x2468, 0x7fff, -0x8000, 0x7fff, -0x8000, -2))
|
||
|
self.assertEqual(audioop.add(datas[3], datas[3], 3),
|
||
|
packs[3](0, 0x2468ac, 0x7fffff, -0x800000,
|
||
|
0x7fffff, -0x800000, -2))
|
||
|
self.assertEqual(audioop.add(datas[4], datas[4], 4),
|
||
|
packs[4](0, 0x2468acf0, 0x7fffffff, -0x80000000,
|
||
|
0x7fffffff, -0x80000000, -2))
|
||
|
|
||
|
def test_bias(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
for bias in 0, 1, -1, 127, -128, 0x7fffffff, -0x80000000:
|
||
|
self.assertEqual(audioop.bias(b'', w, bias), b'')
|
||
|
self.assertEqual(audioop.bias(bytearray(), w, bias), b'')
|
||
|
self.assertEqual(audioop.bias(memoryview(b''), w, bias), b'')
|
||
|
self.assertEqual(audioop.bias(datas[1], 1, 1),
|
||
|
b'\x01\x13\x46\xbc\x80\x81\x00')
|
||
|
self.assertEqual(audioop.bias(datas[1], 1, -1),
|
||
|
b'\xff\x11\x44\xba\x7e\x7f\xfe')
|
||
|
self.assertEqual(audioop.bias(datas[1], 1, 0x7fffffff),
|
||
|
b'\xff\x11\x44\xba\x7e\x7f\xfe')
|
||
|
self.assertEqual(audioop.bias(datas[1], 1, -0x80000000),
|
||
|
datas[1])
|
||
|
self.assertEqual(audioop.bias(datas[2], 2, 1),
|
||
|
packs[2](1, 0x1235, 0x4568, -0x4566, -0x8000, -0x7fff, 0))
|
||
|
self.assertEqual(audioop.bias(datas[2], 2, -1),
|
||
|
packs[2](-1, 0x1233, 0x4566, -0x4568, 0x7ffe, 0x7fff, -2))
|
||
|
self.assertEqual(audioop.bias(datas[2], 2, 0x7fffffff),
|
||
|
packs[2](-1, 0x1233, 0x4566, -0x4568, 0x7ffe, 0x7fff, -2))
|
||
|
self.assertEqual(audioop.bias(datas[2], 2, -0x80000000),
|
||
|
datas[2])
|
||
|
self.assertEqual(audioop.bias(datas[3], 3, 1),
|
||
|
packs[3](1, 0x123457, 0x45678a, -0x456788,
|
||
|
-0x800000, -0x7fffff, 0))
|
||
|
self.assertEqual(audioop.bias(datas[3], 3, -1),
|
||
|
packs[3](-1, 0x123455, 0x456788, -0x45678a,
|
||
|
0x7ffffe, 0x7fffff, -2))
|
||
|
self.assertEqual(audioop.bias(datas[3], 3, 0x7fffffff),
|
||
|
packs[3](-1, 0x123455, 0x456788, -0x45678a,
|
||
|
0x7ffffe, 0x7fffff, -2))
|
||
|
self.assertEqual(audioop.bias(datas[3], 3, -0x80000000),
|
||
|
datas[3])
|
||
|
self.assertEqual(audioop.bias(datas[4], 4, 1),
|
||
|
packs[4](1, 0x12345679, 0x456789ac, -0x456789aa,
|
||
|
-0x80000000, -0x7fffffff, 0))
|
||
|
self.assertEqual(audioop.bias(datas[4], 4, -1),
|
||
|
packs[4](-1, 0x12345677, 0x456789aa, -0x456789ac,
|
||
|
0x7ffffffe, 0x7fffffff, -2))
|
||
|
self.assertEqual(audioop.bias(datas[4], 4, 0x7fffffff),
|
||
|
packs[4](0x7fffffff, -0x6dcba989, -0x3a987656, 0x3a987654,
|
||
|
-2, -1, 0x7ffffffe))
|
||
|
self.assertEqual(audioop.bias(datas[4], 4, -0x80000000),
|
||
|
packs[4](-0x80000000, -0x6dcba988, -0x3a987655, 0x3a987655,
|
||
|
-1, 0, 0x7fffffff))
|
||
|
|
||
|
def test_lin2lin(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.lin2lin(datas[w], w, w), datas[w])
|
||
|
self.assertEqual(audioop.lin2lin(bytearray(datas[w]), w, w),
|
||
|
datas[w])
|
||
|
self.assertEqual(audioop.lin2lin(memoryview(datas[w]), w, w),
|
||
|
datas[w])
|
||
|
|
||
|
self.assertEqual(audioop.lin2lin(datas[1], 1, 2),
|
||
|
packs[2](0, 0x1200, 0x4500, -0x4500, 0x7f00, -0x8000, -0x100))
|
||
|
self.assertEqual(audioop.lin2lin(datas[1], 1, 3),
|
||
|
packs[3](0, 0x120000, 0x450000, -0x450000,
|
||
|
0x7f0000, -0x800000, -0x10000))
|
||
|
self.assertEqual(audioop.lin2lin(datas[1], 1, 4),
|
||
|
packs[4](0, 0x12000000, 0x45000000, -0x45000000,
|
||
|
0x7f000000, -0x80000000, -0x1000000))
|
||
|
self.assertEqual(audioop.lin2lin(datas[2], 2, 1),
|
||
|
b'\x00\x12\x45\xba\x7f\x80\xff')
|
||
|
self.assertEqual(audioop.lin2lin(datas[2], 2, 3),
|
||
|
packs[3](0, 0x123400, 0x456700, -0x456700,
|
||
|
0x7fff00, -0x800000, -0x100))
|
||
|
self.assertEqual(audioop.lin2lin(datas[2], 2, 4),
|
||
|
packs[4](0, 0x12340000, 0x45670000, -0x45670000,
|
||
|
0x7fff0000, -0x80000000, -0x10000))
|
||
|
self.assertEqual(audioop.lin2lin(datas[3], 3, 1),
|
||
|
b'\x00\x12\x45\xba\x7f\x80\xff')
|
||
|
self.assertEqual(audioop.lin2lin(datas[3], 3, 2),
|
||
|
packs[2](0, 0x1234, 0x4567, -0x4568, 0x7fff, -0x8000, -1))
|
||
|
self.assertEqual(audioop.lin2lin(datas[3], 3, 4),
|
||
|
packs[4](0, 0x12345600, 0x45678900, -0x45678900,
|
||
|
0x7fffff00, -0x80000000, -0x100))
|
||
|
self.assertEqual(audioop.lin2lin(datas[4], 4, 1),
|
||
|
b'\x00\x12\x45\xba\x7f\x80\xff')
|
||
|
self.assertEqual(audioop.lin2lin(datas[4], 4, 2),
|
||
|
packs[2](0, 0x1234, 0x4567, -0x4568, 0x7fff, -0x8000, -1))
|
||
|
self.assertEqual(audioop.lin2lin(datas[4], 4, 3),
|
||
|
packs[3](0, 0x123456, 0x456789, -0x45678a,
|
||
|
0x7fffff, -0x800000, -1))
|
||
|
|
||
|
def test_adpcm2lin(self):
|
||
|
self.assertEqual(audioop.adpcm2lin(b'\x07\x7f\x7f', 1, None),
|
||
|
(b'\x00\x00\x00\xff\x00\xff', (-179, 40)))
|
||
|
self.assertEqual(audioop.adpcm2lin(bytearray(b'\x07\x7f\x7f'), 1, None),
|
||
|
(b'\x00\x00\x00\xff\x00\xff', (-179, 40)))
|
||
|
self.assertEqual(audioop.adpcm2lin(memoryview(b'\x07\x7f\x7f'), 1, None),
|
||
|
(b'\x00\x00\x00\xff\x00\xff', (-179, 40)))
|
||
|
self.assertEqual(audioop.adpcm2lin(b'\x07\x7f\x7f', 2, None),
|
||
|
(packs[2](0, 0xb, 0x29, -0x16, 0x72, -0xb3), (-179, 40)))
|
||
|
self.assertEqual(audioop.adpcm2lin(b'\x07\x7f\x7f', 3, None),
|
||
|
(packs[3](0, 0xb00, 0x2900, -0x1600, 0x7200,
|
||
|
-0xb300), (-179, 40)))
|
||
|
self.assertEqual(audioop.adpcm2lin(b'\x07\x7f\x7f', 4, None),
|
||
|
(packs[4](0, 0xb0000, 0x290000, -0x160000, 0x720000,
|
||
|
-0xb30000), (-179, 40)))
|
||
|
|
||
|
# Very cursory test
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.adpcm2lin(b'\0' * 5, w, None),
|
||
|
(b'\0' * w * 10, (0, 0)))
|
||
|
|
||
|
def test_lin2adpcm(self):
|
||
|
self.assertEqual(audioop.lin2adpcm(datas[1], 1, None),
|
||
|
(b'\x07\x7f\x7f', (-221, 39)))
|
||
|
self.assertEqual(audioop.lin2adpcm(bytearray(datas[1]), 1, None),
|
||
|
(b'\x07\x7f\x7f', (-221, 39)))
|
||
|
self.assertEqual(audioop.lin2adpcm(memoryview(datas[1]), 1, None),
|
||
|
(b'\x07\x7f\x7f', (-221, 39)))
|
||
|
for w in 2, 3, 4:
|
||
|
self.assertEqual(audioop.lin2adpcm(datas[w], w, None),
|
||
|
(b'\x07\x7f\x7f', (31, 39)))
|
||
|
|
||
|
# Very cursory test
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.lin2adpcm(b'\0' * w * 10, w, None),
|
||
|
(b'\0' * 5, (0, 0)))
|
||
|
|
||
|
def test_invalid_adpcm_state(self):
|
||
|
# state must be a tuple or None, not an integer
|
||
|
self.assertRaises(TypeError, audioop.adpcm2lin, b'\0', 1, 555)
|
||
|
self.assertRaises(TypeError, audioop.lin2adpcm, b'\0', 1, 555)
|
||
|
# Issues #24456, #24457: index out of range
|
||
|
self.assertRaises(ValueError, audioop.adpcm2lin, b'\0', 1, (0, -1))
|
||
|
self.assertRaises(ValueError, audioop.adpcm2lin, b'\0', 1, (0, 89))
|
||
|
self.assertRaises(ValueError, audioop.lin2adpcm, b'\0', 1, (0, -1))
|
||
|
self.assertRaises(ValueError, audioop.lin2adpcm, b'\0', 1, (0, 89))
|
||
|
# value out of range
|
||
|
self.assertRaises(ValueError, audioop.adpcm2lin, b'\0', 1, (-0x8001, 0))
|
||
|
self.assertRaises(ValueError, audioop.adpcm2lin, b'\0', 1, (0x8000, 0))
|
||
|
self.assertRaises(ValueError, audioop.lin2adpcm, b'\0', 1, (-0x8001, 0))
|
||
|
self.assertRaises(ValueError, audioop.lin2adpcm, b'\0', 1, (0x8000, 0))
|
||
|
|
||
|
def test_lin2alaw(self):
|
||
|
self.assertEqual(audioop.lin2alaw(datas[1], 1),
|
||
|
b'\xd5\x87\xa4\x24\xaa\x2a\x5a')
|
||
|
self.assertEqual(audioop.lin2alaw(bytearray(datas[1]), 1),
|
||
|
b'\xd5\x87\xa4\x24\xaa\x2a\x5a')
|
||
|
self.assertEqual(audioop.lin2alaw(memoryview(datas[1]), 1),
|
||
|
b'\xd5\x87\xa4\x24\xaa\x2a\x5a')
|
||
|
for w in 2, 3, 4:
|
||
|
self.assertEqual(audioop.lin2alaw(datas[w], w),
|
||
|
b'\xd5\x87\xa4\x24\xaa\x2a\x55')
|
||
|
|
||
|
def test_alaw2lin(self):
|
||
|
encoded = b'\x00\x03\x24\x2a\x51\x54\x55\x58\x6b\x71\x7f'\
|
||
|
b'\x80\x83\xa4\xaa\xd1\xd4\xd5\xd8\xeb\xf1\xff'
|
||
|
src = [-688, -720, -2240, -4032, -9, -3, -1, -27, -244, -82, -106,
|
||
|
688, 720, 2240, 4032, 9, 3, 1, 27, 244, 82, 106]
|
||
|
for w in 1, 2, 3, 4:
|
||
|
decoded = packs[w](*(x << (w * 8) >> 13 for x in src))
|
||
|
self.assertEqual(audioop.alaw2lin(encoded, w), decoded)
|
||
|
self.assertEqual(audioop.alaw2lin(bytearray(encoded), w), decoded)
|
||
|
self.assertEqual(audioop.alaw2lin(memoryview(encoded), w), decoded)
|
||
|
|
||
|
encoded = bytes(range(256))
|
||
|
for w in 2, 3, 4:
|
||
|
decoded = audioop.alaw2lin(encoded, w)
|
||
|
self.assertEqual(audioop.lin2alaw(decoded, w), encoded)
|
||
|
|
||
|
def test_lin2ulaw(self):
|
||
|
self.assertEqual(audioop.lin2ulaw(datas[1], 1),
|
||
|
b'\xff\xad\x8e\x0e\x80\x00\x67')
|
||
|
self.assertEqual(audioop.lin2ulaw(bytearray(datas[1]), 1),
|
||
|
b'\xff\xad\x8e\x0e\x80\x00\x67')
|
||
|
self.assertEqual(audioop.lin2ulaw(memoryview(datas[1]), 1),
|
||
|
b'\xff\xad\x8e\x0e\x80\x00\x67')
|
||
|
for w in 2, 3, 4:
|
||
|
self.assertEqual(audioop.lin2ulaw(datas[w], w),
|
||
|
b'\xff\xad\x8e\x0e\x80\x00\x7e')
|
||
|
|
||
|
def test_ulaw2lin(self):
|
||
|
encoded = b'\x00\x0e\x28\x3f\x57\x6a\x76\x7c\x7e\x7f'\
|
||
|
b'\x80\x8e\xa8\xbf\xd7\xea\xf6\xfc\xfe\xff'
|
||
|
src = [-8031, -4447, -1471, -495, -163, -53, -18, -6, -2, 0,
|
||
|
8031, 4447, 1471, 495, 163, 53, 18, 6, 2, 0]
|
||
|
for w in 1, 2, 3, 4:
|
||
|
decoded = packs[w](*(x << (w * 8) >> 14 for x in src))
|
||
|
self.assertEqual(audioop.ulaw2lin(encoded, w), decoded)
|
||
|
self.assertEqual(audioop.ulaw2lin(bytearray(encoded), w), decoded)
|
||
|
self.assertEqual(audioop.ulaw2lin(memoryview(encoded), w), decoded)
|
||
|
|
||
|
# Current u-law implementation has two codes fo 0: 0x7f and 0xff.
|
||
|
encoded = bytes(range(127)) + bytes(range(128, 256))
|
||
|
for w in 2, 3, 4:
|
||
|
decoded = audioop.ulaw2lin(encoded, w)
|
||
|
self.assertEqual(audioop.lin2ulaw(decoded, w), encoded)
|
||
|
|
||
|
def test_mul(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.mul(b'', w, 2), b'')
|
||
|
self.assertEqual(audioop.mul(bytearray(), w, 2), b'')
|
||
|
self.assertEqual(audioop.mul(memoryview(b''), w, 2), b'')
|
||
|
self.assertEqual(audioop.mul(datas[w], w, 0),
|
||
|
b'\0' * len(datas[w]))
|
||
|
self.assertEqual(audioop.mul(datas[w], w, 1),
|
||
|
datas[w])
|
||
|
self.assertEqual(audioop.mul(datas[1], 1, 2),
|
||
|
b'\x00\x24\x7f\x80\x7f\x80\xfe')
|
||
|
self.assertEqual(audioop.mul(datas[2], 2, 2),
|
||
|
packs[2](0, 0x2468, 0x7fff, -0x8000, 0x7fff, -0x8000, -2))
|
||
|
self.assertEqual(audioop.mul(datas[3], 3, 2),
|
||
|
packs[3](0, 0x2468ac, 0x7fffff, -0x800000,
|
||
|
0x7fffff, -0x800000, -2))
|
||
|
self.assertEqual(audioop.mul(datas[4], 4, 2),
|
||
|
packs[4](0, 0x2468acf0, 0x7fffffff, -0x80000000,
|
||
|
0x7fffffff, -0x80000000, -2))
|
||
|
|
||
|
def test_ratecv(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.ratecv(b'', w, 1, 8000, 8000, None),
|
||
|
(b'', (-1, ((0, 0),))))
|
||
|
self.assertEqual(audioop.ratecv(bytearray(), w, 1, 8000, 8000, None),
|
||
|
(b'', (-1, ((0, 0),))))
|
||
|
self.assertEqual(audioop.ratecv(memoryview(b''), w, 1, 8000, 8000, None),
|
||
|
(b'', (-1, ((0, 0),))))
|
||
|
self.assertEqual(audioop.ratecv(b'', w, 5, 8000, 8000, None),
|
||
|
(b'', (-1, ((0, 0),) * 5)))
|
||
|
self.assertEqual(audioop.ratecv(b'', w, 1, 8000, 16000, None),
|
||
|
(b'', (-2, ((0, 0),))))
|
||
|
self.assertEqual(audioop.ratecv(datas[w], w, 1, 8000, 8000, None)[0],
|
||
|
datas[w])
|
||
|
self.assertEqual(audioop.ratecv(datas[w], w, 1, 8000, 8000, None, 1, 0)[0],
|
||
|
datas[w])
|
||
|
|
||
|
state = None
|
||
|
d1, state = audioop.ratecv(b'\x00\x01\x02', 1, 1, 8000, 16000, state)
|
||
|
d2, state = audioop.ratecv(b'\x00\x01\x02', 1, 1, 8000, 16000, state)
|
||
|
self.assertEqual(d1 + d2, b'\000\000\001\001\002\001\000\000\001\001\002')
|
||
|
|
||
|
for w in 1, 2, 3, 4:
|
||
|
d0, state0 = audioop.ratecv(datas[w], w, 1, 8000, 16000, None)
|
||
|
d, state = b'', None
|
||
|
for i in range(0, len(datas[w]), w):
|
||
|
d1, state = audioop.ratecv(datas[w][i:i + w], w, 1,
|
||
|
8000, 16000, state)
|
||
|
d += d1
|
||
|
self.assertEqual(d, d0)
|
||
|
self.assertEqual(state, state0)
|
||
|
|
||
|
expected = {
|
||
|
1: packs[1](0, 0x0d, 0x37, -0x26, 0x55, -0x4b, -0x14),
|
||
|
2: packs[2](0, 0x0da7, 0x3777, -0x2630, 0x5673, -0x4a64, -0x129a),
|
||
|
3: packs[3](0, 0x0da740, 0x377776, -0x262fca,
|
||
|
0x56740c, -0x4a62fd, -0x1298c0),
|
||
|
4: packs[4](0, 0x0da740da, 0x37777776, -0x262fc962,
|
||
|
0x56740da6, -0x4a62fc96, -0x1298bf26),
|
||
|
}
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.ratecv(datas[w], w, 1, 8000, 8000, None, 3, 1)[0],
|
||
|
expected[w])
|
||
|
self.assertEqual(audioop.ratecv(datas[w], w, 1, 8000, 8000, None, 30, 10)[0],
|
||
|
expected[w])
|
||
|
|
||
|
self.assertRaises(TypeError, audioop.ratecv, b'', 1, 1, 8000, 8000, 42)
|
||
|
self.assertRaises(TypeError, audioop.ratecv,
|
||
|
b'', 1, 1, 8000, 8000, (1, (42,)))
|
||
|
|
||
|
def test_reverse(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.reverse(b'', w), b'')
|
||
|
self.assertEqual(audioop.reverse(bytearray(), w), b'')
|
||
|
self.assertEqual(audioop.reverse(memoryview(b''), w), b'')
|
||
|
self.assertEqual(audioop.reverse(packs[w](0, 1, 2), w),
|
||
|
packs[w](2, 1, 0))
|
||
|
|
||
|
def test_tomono(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
data1 = datas[w]
|
||
|
data2 = bytearray(2 * len(data1))
|
||
|
for k in range(w):
|
||
|
data2[k::2*w] = data1[k::w]
|
||
|
self.assertEqual(audioop.tomono(data2, w, 1, 0), data1)
|
||
|
self.assertEqual(audioop.tomono(data2, w, 0, 1), b'\0' * len(data1))
|
||
|
for k in range(w):
|
||
|
data2[k+w::2*w] = data1[k::w]
|
||
|
self.assertEqual(audioop.tomono(data2, w, 0.5, 0.5), data1)
|
||
|
self.assertEqual(audioop.tomono(bytearray(data2), w, 0.5, 0.5),
|
||
|
data1)
|
||
|
self.assertEqual(audioop.tomono(memoryview(data2), w, 0.5, 0.5),
|
||
|
data1)
|
||
|
|
||
|
def test_tostereo(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
data1 = datas[w]
|
||
|
data2 = bytearray(2 * len(data1))
|
||
|
for k in range(w):
|
||
|
data2[k::2*w] = data1[k::w]
|
||
|
self.assertEqual(audioop.tostereo(data1, w, 1, 0), data2)
|
||
|
self.assertEqual(audioop.tostereo(data1, w, 0, 0), b'\0' * len(data2))
|
||
|
for k in range(w):
|
||
|
data2[k+w::2*w] = data1[k::w]
|
||
|
self.assertEqual(audioop.tostereo(data1, w, 1, 1), data2)
|
||
|
self.assertEqual(audioop.tostereo(bytearray(data1), w, 1, 1), data2)
|
||
|
self.assertEqual(audioop.tostereo(memoryview(data1), w, 1, 1),
|
||
|
data2)
|
||
|
|
||
|
def test_findfactor(self):
|
||
|
self.assertEqual(audioop.findfactor(datas[2], datas[2]), 1.0)
|
||
|
self.assertEqual(audioop.findfactor(bytearray(datas[2]),
|
||
|
bytearray(datas[2])), 1.0)
|
||
|
self.assertEqual(audioop.findfactor(memoryview(datas[2]),
|
||
|
memoryview(datas[2])), 1.0)
|
||
|
self.assertEqual(audioop.findfactor(b'\0' * len(datas[2]), datas[2]),
|
||
|
0.0)
|
||
|
|
||
|
def test_findfit(self):
|
||
|
self.assertEqual(audioop.findfit(datas[2], datas[2]), (0, 1.0))
|
||
|
self.assertEqual(audioop.findfit(bytearray(datas[2]),
|
||
|
bytearray(datas[2])), (0, 1.0))
|
||
|
self.assertEqual(audioop.findfit(memoryview(datas[2]),
|
||
|
memoryview(datas[2])), (0, 1.0))
|
||
|
self.assertEqual(audioop.findfit(datas[2], packs[2](1, 2, 0)),
|
||
|
(1, 8038.8))
|
||
|
self.assertEqual(audioop.findfit(datas[2][:-2] * 5 + datas[2], datas[2]),
|
||
|
(30, 1.0))
|
||
|
|
||
|
def test_findmax(self):
|
||
|
self.assertEqual(audioop.findmax(datas[2], 1), 5)
|
||
|
self.assertEqual(audioop.findmax(bytearray(datas[2]), 1), 5)
|
||
|
self.assertEqual(audioop.findmax(memoryview(datas[2]), 1), 5)
|
||
|
|
||
|
def test_getsample(self):
|
||
|
for w in 1, 2, 3, 4:
|
||
|
data = packs[w](0, 1, -1, maxvalues[w], minvalues[w])
|
||
|
self.assertEqual(audioop.getsample(data, w, 0), 0)
|
||
|
self.assertEqual(audioop.getsample(bytearray(data), w, 0), 0)
|
||
|
self.assertEqual(audioop.getsample(memoryview(data), w, 0), 0)
|
||
|
self.assertEqual(audioop.getsample(data, w, 1), 1)
|
||
|
self.assertEqual(audioop.getsample(data, w, 2), -1)
|
||
|
self.assertEqual(audioop.getsample(data, w, 3), maxvalues[w])
|
||
|
self.assertEqual(audioop.getsample(data, w, 4), minvalues[w])
|
||
|
|
||
|
def test_byteswap(self):
|
||
|
swapped_datas = {
|
||
|
1: datas[1],
|
||
|
2: packs[2](0, 0x3412, 0x6745, -0x6646, -0x81, 0x80, -1),
|
||
|
3: packs[3](0, 0x563412, -0x7698bb, 0x7798ba, -0x81, 0x80, -1),
|
||
|
4: packs[4](0, 0x78563412, -0x547698bb, 0x557698ba,
|
||
|
-0x81, 0x80, -1),
|
||
|
}
|
||
|
for w in 1, 2, 3, 4:
|
||
|
self.assertEqual(audioop.byteswap(b'', w), b'')
|
||
|
self.assertEqual(audioop.byteswap(datas[w], w), swapped_datas[w])
|
||
|
self.assertEqual(audioop.byteswap(swapped_datas[w], w), datas[w])
|
||
|
self.assertEqual(audioop.byteswap(bytearray(datas[w]), w),
|
||
|
swapped_datas[w])
|
||
|
self.assertEqual(audioop.byteswap(memoryview(datas[w]), w),
|
||
|
swapped_datas[w])
|
||
|
|
||
|
def test_negativelen(self):
|
||
|
# from issue 3306, previously it segfaulted
|
||
|
self.assertRaises(audioop.error,
|
||
|
audioop.findmax, bytes(range(256)), -2392392)
|
||
|
|
||
|
def test_issue7673(self):
|
||
|
state = None
|
||
|
for data, size in INVALID_DATA:
|
||
|
size2 = size
|
||
|
self.assertRaises(audioop.error, audioop.getsample, data, size, 0)
|
||
|
self.assertRaises(audioop.error, audioop.max, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.minmax, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.avg, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.rms, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.avgpp, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.maxpp, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.cross, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.mul, data, size, 1.0)
|
||
|
self.assertRaises(audioop.error, audioop.tomono, data, size, 0.5, 0.5)
|
||
|
self.assertRaises(audioop.error, audioop.tostereo, data, size, 0.5, 0.5)
|
||
|
self.assertRaises(audioop.error, audioop.add, data, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.bias, data, size, 0)
|
||
|
self.assertRaises(audioop.error, audioop.reverse, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.lin2lin, data, size, size2)
|
||
|
self.assertRaises(audioop.error, audioop.ratecv, data, size, 1, 1, 1, state)
|
||
|
self.assertRaises(audioop.error, audioop.lin2ulaw, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.lin2alaw, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.lin2adpcm, data, size, state)
|
||
|
|
||
|
def test_string(self):
|
||
|
data = 'abcd'
|
||
|
size = 2
|
||
|
self.assertRaises(TypeError, audioop.getsample, data, size, 0)
|
||
|
self.assertRaises(TypeError, audioop.max, data, size)
|
||
|
self.assertRaises(TypeError, audioop.minmax, data, size)
|
||
|
self.assertRaises(TypeError, audioop.avg, data, size)
|
||
|
self.assertRaises(TypeError, audioop.rms, data, size)
|
||
|
self.assertRaises(TypeError, audioop.avgpp, data, size)
|
||
|
self.assertRaises(TypeError, audioop.maxpp, data, size)
|
||
|
self.assertRaises(TypeError, audioop.cross, data, size)
|
||
|
self.assertRaises(TypeError, audioop.mul, data, size, 1.0)
|
||
|
self.assertRaises(TypeError, audioop.tomono, data, size, 0.5, 0.5)
|
||
|
self.assertRaises(TypeError, audioop.tostereo, data, size, 0.5, 0.5)
|
||
|
self.assertRaises(TypeError, audioop.add, data, data, size)
|
||
|
self.assertRaises(TypeError, audioop.bias, data, size, 0)
|
||
|
self.assertRaises(TypeError, audioop.reverse, data, size)
|
||
|
self.assertRaises(TypeError, audioop.lin2lin, data, size, size)
|
||
|
self.assertRaises(TypeError, audioop.ratecv, data, size, 1, 1, 1, None)
|
||
|
self.assertRaises(TypeError, audioop.lin2ulaw, data, size)
|
||
|
self.assertRaises(TypeError, audioop.lin2alaw, data, size)
|
||
|
self.assertRaises(TypeError, audioop.lin2adpcm, data, size, None)
|
||
|
|
||
|
def test_wrongsize(self):
|
||
|
data = b'abcdefgh'
|
||
|
state = None
|
||
|
for size in (-1, 0, 5, 1024):
|
||
|
self.assertRaises(audioop.error, audioop.ulaw2lin, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.alaw2lin, data, size)
|
||
|
self.assertRaises(audioop.error, audioop.adpcm2lin, data, size, state)
|
||
|
|
||
|
if __name__ == '__main__':
|
||
|
unittest.main()
|