def find_first_word(haystack):
match_pos, match_word = -1, None
for w in original_words:
idx = index_of(w, haystack, lang=lang)
if idx > -1 and (match_pos == -1 or match_pos > idx):
match_pos, match_word = idx, w
return match_pos, match_word
def test_break_iterator(self):
' Test the break iterator '
from calibre.spell.break_iterator import split_into_words as split, index_of, split_into_words_and_positions
for q in ('one two three', ' one two three', 'one\ntwo three ', ):
self.ae(split(unicode(q)), ['one', 'two', 'three'], 'Failed to split: %r' % q)
self.ae(split(u'I I\'m'), ['I', "I'm"])
self.ae(split(u'out-of-the-box'), ['out-of-the-box'])
self.ae(split(u'-one two-'), ['one', 'two'])
self.ae(split_into_words_and_positions('one \U0001f431 three'), [(0, 3), (7 if icu.is_narrow_build else 6, 5)])
for needle, haystack, pos in (
('word', 'a word b', 2),
('word', 'a word', 2),
('one-two', 'a one-two punch', 2),
('one-two', 'one-two punch', 0),
('one-two', 'one-two', 0),
('one', 'one-two one', 8),
('one-two', 'one-two-three one-two', 14),
('one', 'onet one', 5),
('two', 'one-two two', 8),
('i', 'i', 0),
('i', 'six i', 4),
('i', '', -1), ('', '', -1), ('', 'i', -1),
('i', 'six clicks', -1),
('i', '\U0001f431 i', (3 if icu.is_narrow_build else 2)),
):
fpos = index_of(needle, haystack)
self.ae(pos, fpos, 'Failed to find index of %r in %r (%d != %d)' % (needle, haystack, pos, fpos))
def find_first_word(haystack):
match_pos, match_word = -1, None
for w in original_words:
idx = index_of(w, haystack, lang=lang)
if idx > -1 and (match_pos == -1 or match_pos > idx):
match_pos, match_word = idx, w
return match_pos, match_word
def test_break_iterator(self):
' Test the break iterator '
from calibre.spell.break_iterator import split_into_words as split, index_of, split_into_words_and_positions, count_words
for q in (
'one two three',
' one two three',
'one\ntwo three ',
):
self.ae(split(str(q)), ['one', 'two', 'three'],
'Failed to split: %r' % q)
self.ae(split('I I\'m'), ['I', "I'm"])
self.ae(split('out-of-the-box'), ['out-of-the-box'])
self.ae(split('-one two-'), ['-one', 'two-'])
self.ae(split('-one a-b-c-d e'), ['-one', 'a-b-c-d', 'e'])
self.ae(split('-one -a-b-c-d- e'), ['-one', '-a-b-c-d-', 'e'])
self.ae(split_into_words_and_positions('one \U0001f431 three'),
[(0, 3), (6, 5)])
self.ae(count_words('a b c d e f'), 6)
for needle, haystack, pos in (
('word', 'a word b', 2),
('word', 'a word', 2),
('one-two', 'a one-two punch', 2),
('one-two', 'one-two punch', 0),
('one-two', 'one-two', 0),
('one', 'one-two one', 8),
('one-two', 'one-two-three one-two', 14),
('one', 'onet one', 5),
('two', 'one-two two', 8),
('two', 'two-one two', 8),
('-two', 'one-two -two', 8),
('-two', 'two', -1),
('i', 'i', 0),
('i', 'six i', 4),
('i', '', -1),
('', '', -1),
('', 'i', -1),
('i', 'six clicks', -1),
('i', '\U0001f431 i', 2),
('-a', 'b -a', 2),
('a-', 'a-b a- d', 4),
('-a-', 'b -a -a-', 5),
('-a-', '-a-', 0),
('-a-', 'a-', -1),
('-a-', '-a', -1),
('-a-', 'a', -1),
('a-', 'a-', 0),
('-a', '-a', 0),
('a-b-c-', 'a-b-c-d', -1),
('a-b-c-', 'a-b-c-.', 0),
('a-b-c-', 'a-b-c-d a-b-c- d', 8),
):
fpos = index_of(needle, haystack)
self.ae(
pos, fpos, 'Failed to find index of %r in %r (%d != %d)' %
(needle, haystack, pos, fpos))
def replace(text, original_word, new_word, lang):
indices = []
original_word, new_word, text = unicode_type(original_word), unicode_type(new_word), unicode_type(text)
q = text
offset = 0
while True:
idx = index_of(original_word, q, lang=lang)
if idx == -1:
break
indices.append(offset + idx)
offset += idx + len(original_word)
q = text[offset:]
for idx in reversed(indices):
text = text[:idx] + new_word + text[idx+len(original_word):]
return text, bool(indices)
def test_break_iterator(self):
' Test the break iterator '
from calibre.spell.break_iterator import split_into_words as split, index_of
for q in ('one two three', ' one two three', 'one\ntwo three ', 'one-two,three'):
self.ae(split(unicode(q)), ['one', 'two', 'three'], 'Failed to split: %r' % q)
self.ae(split(u'I I\'m'), ['I', "I'm"])
self.ae(0, index_of('i', 'i'))
self.ae(4, index_of('i', 'six i'))
self.ae(-1, index_of('i', ''))
self.ae(-1, index_of('', ''))
self.ae(-1, index_of('', 'i'))
self.ae(-1, index_of('i', 'six clicks'))
def test_break_iterator(self):
' Test the break iterator '
from calibre.spell.break_iterator import split_into_words as split, index_of
for q in ('one two three', ' one two three', 'one\ntwo three ', 'one-two,three'):
self.ae(split(unicode(q)), ['one', 'two', 'three'], 'Failed to split: %r' % q)
self.ae(split(u'I I\'m'), ['I', "I'm"])
self.ae(0, index_of('i', 'i'))
self.ae(4, index_of('i', 'six i'))
self.ae(-1, index_of('i', ''))
self.ae(-1, index_of('', ''))
self.ae(-1, index_of('', 'i'))
self.ae(-1, index_of('i', 'six clicks'))
def test_break_iterator(self):
' Test the break iterator '
from calibre.spell.break_iterator import split_into_words as split, index_of, split_into_words_and_positions
for q in ('one two three', ' one two three', 'one\ntwo three ', 'one-two,three'):
self.ae(split(unicode(q)), ['one', 'two', 'three'], 'Failed to split: %r' % q)
self.ae(split(u'I I\'m'), ['I', "I'm"])
self.ae(split_into_words_and_positions('one \U0001f431 three'), [(0, 3), (6 if sys.maxunicode >= 0x10ffff else 7, 5)])
self.ae(0, index_of('i', 'i'))
self.ae(4, index_of('i', 'six i'))
self.ae(-1, index_of('i', ''))
self.ae(-1, index_of('', ''))
self.ae(-1, index_of('', 'i'))
self.ae(-1, index_of('i', 'six clicks'))
