return _icu.character_name_from_code(code) or ''
except (TypeError, ValueError, KeyError):
return ''
def normalize(text, mode='NFC'):
# This is very slightly slower than using unicodedata.normalize, so stick with
# that unless you have very good reasons not too. Also, it's speed
# decreases on wide python builds, where conversion to/from ICU's string
# representation is slower.
return _icu.normalize(_nmodes[mode], unicode(text))
def contractions(col=None):
global _cmap
col = col or _collator
if col is None:
col = collator()
ans = _cmap.get(collator, None)
if ans is None:
ans = col.contractions()
ans = frozenset(filter(None, ans))
_cmap[col] = ans
return ans
################################################################################
if __name__ == '__main__':
from calibre.utils.icu_test import run
run(verbosity=4)
from collections import OrderedDict
items = sorted(items, key=lambda x:sort_key(key(x)), reverse=reverse)
ans = OrderedDict()
last_c, last_ordnum = ' ', 0
for item in items:
c = icu_upper(key(item) or ' ')
ordnum, ordlen = collation_order(c)
if last_ordnum != ordnum:
if not is_narrow_build:
ordlen = 1
last_c = c[0:ordlen]
last_ordnum = ordnum
try:
ans[last_c].append(item)
except KeyError:
ans[last_c] = [item]
return ans
# Return the number of unicode codepoints in a string
string_length = _icu.string_length if is_narrow_build else len
# Return the number of UTF-16 codepoints in a string
utf16_length = len if is_narrow_build else _icu.utf16_length
################################################################################
if __name__ == '__main__':
from calibre.utils.icu_test import run
run(verbosity=4)