def test_setgetword():
for dtype in [numpy.ubyte, numpy.int32, numpy.float64]:
arr = numpy.array(list(range(-256,256)), dtype=dtype)
arr2 = numpy.zeros(arr.shape, dtype=arr.dtype)
for i in range (arr.nbytes):
word, next = bittools.getword (arr,i*8,8)
bittools.setword (arr2,i*8,8,word)
assert (arr==arr2).all(),repr((arr,arr2))
print('ok')
def test_wordbits ():
for width in range (1,64+1):
arr = numpy.array([131,17,235], dtype=numpy.int64)
arr2 = numpy.zeros((1+width//8), dtype=arr.dtype)
bstr = tobinary(arr)
word, next = bittools.getword(arr, 0, width)
if width > 7:
assert word==arr[0],repr((word, arr[0], bstr))
bittools.setword (arr2, 0, width, word)
assert bittools.getword (arr2, 0, width)[0]==word
assert tobinary(arr2)[:width] == bstr[:width],repr((tobinary(arr2)[:width], bstr[:width]))
def encode_bittools(seq, max_bits=12):
""" Compress sequence using Lempel-Ziv-Welch algorithm for TIFF.
Parameters
----------
seq : {str, numpy.ndarray}
max_bits : int
Specify maximum bits for encoding table.
Returns
-------
bseq : numpy.ndarray
See also
--------
decode_bittools
"""
if isinstance(seq, numpy.ndarray):
nbytes = seq.nbytes * 2
seq = seq.tostring()
else:
nbytes = len(seq) * 2
r = numpy.zeros((nbytes, ), dtype=numpy.ubyte)
init_table = [(chr(code), code) for code in range(256)]
table = {}
table_get = table.get
table_clear = table.clear
table_update = table.update
sup_code2 = (1 << max_bits) - 2
next_code = CODESTART
bits = 9
max_code = (1 << bits)
s = ''
table_update(init_table)
index = setword(r, 0, bits, CODECLEAR, 1)
for c in seq:
s1 = s + c
if s1 in table:
s = s1
else:
index = setword(r, index, bits, table_get(s), 1)
table[s1] = next_code
next_code += 1
s = c
if next_code == sup_code2:
index = setword(r, index, bits, table_get(s), 1)
index = setword(r, index, bits, CODECLEAR, 1)
s = ''
table_clear()
table_update(init_table)
next_code = CODESTART
bits = 9
max_code = (1 << bits)
elif next_code == max_code:
bits += 1
max_code = (1 << bits)
if s:
index = setword(r, index, bits, table_get(s), 1)
index = setword(r, index, bits, CODEEOI)
bytes = index // 8
if 8 * bytes < index:
bytes += 1
return r[:bytes]
def encode_bittools(seq, max_bits=12):
""" Compress sequence using Lempel-Ziv-Welch algorithm for TIFF.
Parameters
----------
seq : {str, numpy.ndarray}
max_bits : int
Specify maximum bits for encoding table.
Returns
-------
bseq : numpy.ndarray
See also
--------
decode_bittools
"""
if isinstance (seq, numpy.ndarray):
nbytes = seq.nbytes*2
seq = seq.tostring()
else:
nbytes = len(seq)*2
r = numpy.zeros((nbytes,), dtype=numpy.ubyte)
init_table = [(chr(code),code) for code in range (256)]
table = {}
table_get = table.get
table_clear = table.clear
table_update = table.update
sup_code2 = (1<<max_bits) - 2
next_code = CODESTART
bits = 9
max_code = (1<<bits)
s = ''
table_update(init_table)
index = setword(r, 0, bits, CODECLEAR, 1)
for c in seq:
s1 = s + c
if s1 in table:
s = s1
else:
index = setword(r, index, bits, table_get(s), 1)
table[s1] = next_code
next_code += 1
s = c
if next_code==sup_code2:
index = setword(r, index, bits, table_get(s), 1)
index = setword(r, index, bits, CODECLEAR, 1)
s = ''
table_clear()
table_update(init_table)
next_code = CODESTART
bits = 9
max_code = (1<<bits)
elif next_code==max_code:
bits += 1
max_code = (1<<bits)
if s:
index = setword(r, index, bits, table_get(s), 1)
index = setword(r, index, bits, CODEEOI)
bytes = index//8
if 8*bytes < index:
bytes += 1
return r[:bytes]
