def __store(self, s, bits):
"""Add string s to known data.
Seeks to avoid duplication, where possible.
For example, short-forms may be prefixes of long-forms.
"""
if not s:
return StringDataToken(0, 0, bits)
ucs2 = unicode2hex(s)
try:
index = self.text.index(s) - 1
matched = 0
while matched < len(ucs2):
index, matched = self.data.index(ucs2[0], index + 1), 1
if index + len(ucs2) >= len(self.data):
raise ValueError # not found after all !
while matched < len(ucs2) and self.data[
index + matched] == ucs2[matched]:
matched += 1
except ValueError:
index = len(self.data)
self.data += ucs2
self.text += s
assert index >= 0
try:
return StringDataToken(index, len(ucs2), bits)
except ValueError as e:
e.args += (self.name, s)
raise
def append(self, s):
s = s + '\0'
if s in self.hash:
return self.hash[s]
lst = unicode2hex(s)
index = len(self.data)
if index > 0xffff:
raise Error('Index ({}) outside the uint16 range !'.format(index))
self.hash[s] = index
self.data += lst
return index
def append(self, s):
if s in self.hash:
return self.hash[s]
lst = unicode2hex(s)
index = len(self.data)
if index > 0xffff:
raise Error('Data index {} is too big for uint16!'.format(index))
size = len(lst)
if size >= 0xffff:
raise Error('Data is too big ({}) for uint16 size!'.format(size))
token = None
try:
token = StringDataToken(index, size)
except Error as e:
e.message += '(on data "{}")'.format(s)
raise
self.hash[s] = token
self.data += lst
return token