def pinyin_indexes(content):
pinyin = find_pinyin(content)
# assert type(pinyin) == unicode
if not pinyin or pinyin == "_":
return ()
indexes = set()
# multiple pronunciations
for pinyin in re.split(r",|;", pinyin):
# find all pinyin ranges, use them to rip pinyin out
py = [
r._slice(pinyin)
for r in color.ranges_of_pinyin_in_string(pinyin)
]
# maybe no pinyin here
if not py:
return ()
# just pinyin, with diacritics, separated by whitespace
indexes.add("%s." % color.utf(" ".join(py)))
# pinyin with diacritics replaced by tone numbers
indexes.add("%s." % color.utf(" ".join(
["%s%d" % (
color.lowercase_string_by_removing_pinyin_tones(p),
color.determine_tone(p)) for p in py])))
return indexes
def pinyin_indexes(content):
pinyin = find_pinyin(content)
# assert type(pinyin) == unicode
if not pinyin or pinyin == '_':
return ()
indexes = set()
# multiple pronunciations
for pinyin in re.split(r',|;', pinyin):
# find all pinyin ranges, use them to rip pinyin out
py = [r._slice(pinyin) for r in color.ranges_of_pinyin_in_string(pinyin)]
# maybe no pinyin here
if not py:
return ()
# just pinyin, with diacritics, separated by whitespace
indexes.add('%s.' % color.utf(' '.join(py)))
# pinyin with diacritics replaced by tone numbers
indexes.add('%s.' % color.utf(' '.join(
['%s%d' % (
color.lowercase_string_by_removing_pinyin_tones(p),
color.determine_tone(p)) for p in py])))
return indexes
def pinyin_indexes(content):
pinyin = find_pinyin(content)
# assert type(pinyin) == unicode
if not pinyin or pinyin == '_':
return ()
indexes = set()
# multiple pronunciations
for pinyin in re.split(r',|;', pinyin):
# find all pinyin ranges, use them to rip pinyin out
py = [
r._slice(pinyin) for r in color.ranges_of_pinyin_in_string(pinyin)
]
# maybe no pinyin here
if not py:
return ()
# just pinyin, with diacritics, separated by whitespace
indexes.add('%s.' % color.utf(u' '.join(py)))
# pinyin with diacritics replaced by tone numbers
indexes.add('%s.' % color.utf(u' '.join(
map(
lambda p: '%s%d' %
(color.lowercase_string_by_removing_pinyin_tones(p),
color.determine_tone(p)),
# for each sub word
py))))
return indexes
def testFristTone(self):
self.assertEqual(1, colorize_pinyin.determine_tone('fāng'))
self.assertEqual(1, colorize_pinyin.determine_tone('yī'))
def testMixedPinyin(self):
self.assertEqual(3, colorize_pinyin.determine_tone('bǎiwén'))
def testNonPinyin(self):
self.assertEqual(0, colorize_pinyin.determine_tone('бурда'))
def testZeroTone(self):
self.assertEqual(0, colorize_pinyin.determine_tone('de'))
self.assertEqual(0, colorize_pinyin.determine_tone('ning'))
def testFourthTone(self, ):
self.assertEqual(4, colorize_pinyin.determine_tone('àn'))
self.assertEqual(4, colorize_pinyin.determine_tone('dìnggòu'))
def testThirdTone(self):
self.assertEqual(3, colorize_pinyin.determine_tone('fǎn'))
self.assertEqual(3, colorize_pinyin.determine_tone('lǚ'))
def testSecondTone(self):
self.assertEqual(2, colorize_pinyin.determine_tone('gán'))
self.assertEqual(2, colorize_pinyin.determine_tone('xún'))
def testFristTone(self):
self.assertEqual(1, colorize_pinyin.determine_tone('fāng'))
self.assertEqual(1, colorize_pinyin.determine_tone('yī'))
def testSecondTone(self):
self.assertEqual(2, colorize_pinyin.determine_tone('gán'))
self.assertEqual(2, colorize_pinyin.determine_tone('xún'))
def testMixedPinyin(self):
self.assertEqual(3, colorize_pinyin.determine_tone('bǎiwén'))
def testNonPinyin(self):
self.assertEqual(0, colorize_pinyin.determine_tone('бурда'))
def testZeroTone(self):
self.assertEqual(0, colorize_pinyin.determine_tone('de'))
self.assertEqual(0, colorize_pinyin.determine_tone('ning'))
def testFourthTone(self, ):
self.assertEqual(4, colorize_pinyin.determine_tone('àn'))
self.assertEqual(4, colorize_pinyin.determine_tone('dìnggòu'))
def testThirdTone(self):
self.assertEqual(3, colorize_pinyin.determine_tone('fǎn'))
self.assertEqual(3, colorize_pinyin.determine_tone('lǚ'))
