def decode_article(self, size):
res = ""
while len(res) < size:
sym_idx = self._ltArticles.decode()
sym = self._article_symbols[sym_idx]
if sym <= 0x80:
if sym <= 0x3F:
start_pref_idx = self.bstr.read_bits(tools.bit_length(len(self.prefix)))
s = sym + 3
res += self.prefix[start_pref_idx:start_pref_idx + s]
else:
start_idx = self.bstr.read_bits(tools.bit_length(size))
s = sym - 0x3d
res += res[start_idx:start_idx + s]
else:
res += int2unichr(sym - 0x80)
return res
def decode_article(self, size):
res = ""
while len(res) < size:
sym_idx = self._ltArticles.decode()
sym = self._article_symbols[sym_idx]
if sym <= 0x80:
if sym <= 0x3F:
start_pref_idx = self.bstr.read_bits(
tools.bit_length(len(self.prefix)))
s = sym + 3
res += self.prefix[start_pref_idx:start_pref_idx + s]
else:
start_idx = self.bstr.read_bits(tools.bit_length(size))
s = sym - 0x3d
res += res[start_idx:start_idx + s]
else:
res += unichr(sym - 0x80)
return res
def read_reference(self, huffman_number):
reference = ""
code = self.bstr.read_bits(2)
if code == 3:
self.bstr.read_bits(32)
return reference
size = tools.bit_length(huffman_number)
assert(size >= 2)
return (code << (size - 2)) | self.bstr.read_bits(size - 2)
def read_reference(self, huffman_number):
reference = ""
code = self.bstr.read_bits(2)
if code == 3:
self.bstr.read_bits(32)
return reference
size = tools.bit_length(huffman_number)
assert (size >= 2)
return (code << (size - 2)) | self.bstr.read_bits(size - 2)
def decode_article(self, size):
"""
decode User and Abrv dict
"""
res = ""
while len(res) < size:
sym_idx = self._ltArticles.decode()
sym = self._article_symbols[sym_idx]
if sym >= 0x10000:
if sym >= 0x10040:
start_idx = self.bstr.read_bits(tools.bit_length(size))
s = sym - 0x1003d
res += res[start_idx:start_idx + s]
else:
prefix_idx = self.bstr.read_bits(tools.bit_length(len(self.prefix)))
s = sym - 0xfffd
res += self.prefix[prefix_idx:prefix_idx + s]
else:
res += int2unichr(sym)
return res
def decode_article(self, size):
"""
decode User and Abrv dict
"""
res = ""
while len(res) < size:
sym_idx = self._ltArticles.decode()
sym = self._article_symbols[sym_idx]
if sym >= 0x10000:
if sym >= 0x10040:
start_idx = self.bstr.read_bits(tools.bit_length(size))
s = sym - 0x1003d
res += res[start_idx:start_idx + s]
else:
prefix_idx = self.bstr.read_bits(
tools.bit_length(len(self.prefix)))
s = sym - 0xfffd
res += self.prefix[prefix_idx:prefix_idx + s]
else:
res += unichr(sym)
return res
def test_bit_length_0(self):
self.assertEqual(tools.bit_length(0), 1)
def test_bit_length_4(self):
self.assertEqual(tools.bit_length(4), 3)
def test_bit_length_2(self):
self.assertEqual(tools.bit_length(2), 2)
def test_bit_length_0(self):
self.assertEqual(tools.bit_length(0), 1)
def test_bit_length_4(self):
self.assertEqual(tools.bit_length(4), 3)
def test_bit_length_2(self):
self.assertEqual(tools.bit_length(2), 2)