def read(self, lsd_decoder, bstr, known_prefix):
h = Heading()
prefix_len = lsd_decoder.decode_prefix_len()
postfix_len = lsd_decoder.decode_postfix_len()
h.text = known_prefix[:prefix_len]
h.text += lsd_decoder.decode_heading(postfix_len)
self.reference = lsd_decoder.read_reference2()
if bstr.read_bit():
# additional not visible formatting item in header
# join multisymbols item
ext_length = bstr.read_bits(8)
if ext_length != 0:
ext = ""
first_idx = prev_idx = 0
for i in range(ext_length):
idx = bstr.read_bits(8)
char = int2unichr(bstr.read_bits(16))
if ext == "":
ext += char
first_idx = prev_idx = idx
else:
if prev_idx + 1 == idx:
# join item with sequence idx
ext += char
prev_idx = idx
else:
# other item
h.extensions.append((first_idx, ext))
ext = char
first_idx = prev_idx = idx
# add last item
h.extensions.append((first_idx, ext))
self.headings.append(h)
return h.text
def read(self, lsd_decoder, bstr, known_prefix):
h = Heading()
prefix_len = lsd_decoder.decode_prefix_len()
postfix_len = lsd_decoder.decode_postfix_len()
h.text = known_prefix[:prefix_len]
h.text += lsd_decoder.decode_heading(postfix_len)
self.reference = lsd_decoder.read_reference2()
if bstr.read_bit():
# additional not visible formatting item in header
# join multisymbols item
ext_length = bstr.read_bits(8)
if ext_length != 0:
ext = ""
first_idx = prev_idx = 0
for i in range(ext_length):
idx = bstr.read_bits(8)
char = int2unichr(bstr.read_bits(16))
if ext == "":
ext += char
first_idx = prev_idx = idx
else:
if prev_idx + 1 == idx:
# join item with sequence idx
ext += char
prev_idx = idx
else:
# other item
h.extensions.append((first_idx, ext))
ext = char
first_idx = prev_idx = idx
# add last item
h.extensions.append((first_idx, ext))
self.headings.append(h)
return h.text
def decode_heading(self, size):
res = ""
for i in range(size):
sym_idx = self._ltHeadings.decode()
sym = self._heading_symbols[sym_idx]
assert(sym <= 0xffff) # LingvoEngine:2EAB84E8
res += int2unichr(sym)
return res
def decode_heading(self, size):
res = ""
for i in range(size):
sym_idx = self._ltHeadings.decode()
sym = self._heading_symbols[sym_idx]
assert (sym <= 0xffff) # LingvoEngine:2EAB84E8
res += int2unichr(sym)
return res
def read_unicode(self, size, big_endian=True):
res = ""
for i in range(size):
ch = self.read_some(2)
if not big_endian:
ch = reverse16(ch)
res += int2unichr(ch)
# res += unichr(self.read_some(2), big_endian))
return res
def read_unicode(self, size, big_endian=True):
res = ""
for i in range(size):
ch = self.read_some(2)
if not big_endian:
ch = reverse16(ch)
res += int2unichr(ch)
# res += unichr(self.read_some(2), big_endian))
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 += 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 += int2unichr(sym - 0x80)
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 += 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 += int2unichr(sym)
return res
def read_xored_prefix(self, size):
res = ""
for i in range(size):
res += int2unichr(self.bstr.read_bits(16) ^ 0x879A)
return res
def read_xored_prefix(self, size):
res = ""
for i in range(size):
res += int2unichr(self.bstr.read_bits(16) ^ 0x879A)
return res