def test_custom_style_with_call():
style_value = 'test_custom_style_with_call'
def func(pinyin, **kwargs):
return str(len(pinyin))
register(style_value, func=func)
hans = '北京'
origin_pinyin_s = pinyin(hans)
expected_pinyin_s = deepcopy(origin_pinyin_s)
for pinyin_s in expected_pinyin_s:
for index, py in enumerate(pinyin_s):
pinyin_s[index] = func(py)
assert pinyin(hans, style=style_value) == expected_pinyin_s
def test_custom_style_with_call():
style_value = 'test_custom_style_with_call'
def func(pinyin, **kwargs):
return str(len(pinyin))
register(style_value, func=func)
hans = '北京'
origin_pinyin_s = pinyin(hans)
expected_pinyin_s = deepcopy(origin_pinyin_s)
for pinyin_s in expected_pinyin_s:
for index, py in enumerate(pinyin_s):
pinyin_s[index] = func(py)
assert pinyin(hans, style=style_value) == expected_pinyin_s
)
CYRILLIC_TABLE = dict(zip(
u'abwgdEOrZiIklmnopRstufhqcCSHTMUevAV',
u'абвгдеёжзийклмнопрстуфхццчшщъьыэюяю'
))
class CyrillicfoConverter(object):
def to_cyrillic(self, pinyin, **kwargs):
pinyin = self._pre_convert(pinyin)
# 查表替换成注音
for find_re, replace in CYRILLIC_REPLACE:
pinyin = find_re.sub(replace, pinyin)
pinyin = ''.join(CYRILLIC_TABLE.get(x, x) for x in pinyin)
return pinyin
def to_cyrillic_first(self, pinyin, **kwargs):
pinyin = self.to_cyrillic(pinyin, **kwargs)
return pinyin[0]
def _pre_convert(self, pinyin):
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
# 将声调数字移动到最后
return RE_TONE3.sub(r'\1\3\2', pinyin)
converter = CyrillicfoConverter()
register(Style.CYRILLIC, func=converter.to_cyrillic)
register(Style.CYRILLIC_FIRST, func=converter.to_cyrillic_first)
from pypinyin.constants import Style
from pypinyin.style import register
from pypinyin.style._tone_convert import (to_finals, to_finals_tone,
to_finals_tone2, to_finals_tone3)
class FinalsConverter(object):
def to_finals(self, pinyin, **kwargs):
"""无声调韵母"""
return to_finals(pinyin, strict=kwargs.get('strict', True))
def to_finals_tone(self, pinyin, **kwargs):
"""声调在韵母头上"""
return to_finals_tone(pinyin, strict=kwargs.get('strict', True))
def to_finals_tone2(self, pinyin, **kwargs):
"""数字声调"""
return to_finals_tone2(pinyin, strict=kwargs.get('strict', True))
def to_finals_tone3(self, pinyin, **kwargs):
"""数字声调"""
return to_finals_tone3(pinyin, strict=kwargs.get('strict', True))
converter = FinalsConverter()
register(Style.FINALS, func=converter.to_finals)
register(Style.FINALS_TONE, func=converter.to_finals_tone)
register(Style.FINALS_TONE2, func=converter.to_finals_tone2)
register(Style.FINALS_TONE3, func=converter.to_finals_tone3)
)
BOPOMOFO_TABLE = dict(zip(
'bpmfdtnlgkhjqxZCSrzcsiuvaoeEAIOUMNKGR2340',
'ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒㄓㄔㄕㄖㄗㄘㄙㄧㄨㄩㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦˊˇˋ˙'
))
class BopomofoConverter(object):
def to_bopomofo(self, pinyin, **kwargs):
pinyin = self._pre_convert(pinyin)
# 查表替换成注音
for find_re, replace in BOPOMOFO_REPLACE:
pinyin = find_re.sub(replace, pinyin)
pinyin = ''.join(BOPOMOFO_TABLE.get(x, x) for x in pinyin)
return pinyin
def to_bopomofo_first(self, pinyin, **kwargs):
pinyin = self.to_bopomofo(pinyin, **kwargs)
return pinyin[0]
def _pre_convert(self, pinyin):
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
# 将声调数字移动到最后
return RE_TONE3.sub(r'\1\3\2', pinyin)
converter = BopomofoConverter()
register(Style.BOPOMOFO, func=converter.to_bopomofo)
register(Style.BOPOMOFO_FIRST, func=converter.to_bopomofo_first)
# -*- coding: utf-8 -*-
"""其他几个拼音风格实现:
Style.NORMAL
Style.FIRST_LETTER
"""
from __future__ import unicode_literals
from pypinyin.constants import Style
from pypinyin.style import register
from pypinyin.style._utils import replace_symbol_to_no_symbol
class OthersConverter(object):
def to_normal(self, pinyin, **kwargs):
pinyin = replace_symbol_to_no_symbol(pinyin)
return pinyin
def to_first_letter(self, pinyin, **kwargs):
# 用数字表示声调
pinyin = self.to_normal(pinyin)
return pinyin[0]
converter = OthersConverter()
register(Style.NORMAL, func=converter.to_normal)
register(Style.FIRST_LETTER, func=converter.to_first_letter)
Style.TONE2
Style.TONE3
"""
from __future__ import unicode_literals
from pypinyin.constants import Style
from pypinyin.style import register
from pypinyin.style._constants import RE_TONE3
from pypinyin.style._utils import replace_symbol_to_number
class ToneConverter(object):
def to_tone(self, pinyin, **kwargs):
return pinyin
def to_tone2(self, pinyin, **kwargs):
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
return pinyin
def to_tone3(self, pinyin, **kwargs):
pinyin = self.to_tone2(pinyin, **kwargs)
# 将声调数字移动到最后
return RE_TONE3.sub(r'\1\3\2', pinyin)
converter = ToneConverter()
register(Style.TONE, func=converter.to_tone)
register(Style.TONE2, func=converter.to_tone2)
register(Style.TONE3, func=converter.to_tone3)
whole_converted = _convert_whole(pinyin, _except_table)
if whole_converted != pinyin:
return _fixed_result(whole_converted)
whole_converted = _convert_whole(pinyin, _convert_table)
if whole_converted != pinyin:
return _fixed_result(whole_converted)
initials = get_initials(pinyin, strict=False)
tones = pinyin[len(initials):]
initials = _convert_whole(initials, _initial_table)
tones = _convert_whole(tones, _tone_table)
return _fixed_result('{}{}'.format(initials, tones))
def _fixed_result(pinyin):
return pinyin.replace('ü', 'v')
def _convert_whole(chars, table):
for pair in table:
f, r = pair
if f == chars:
return r
return chars
register(Style.WADEGILES, func=to_wade_glides)
(re.compile('^v(\d?$)'), 'vI'),
)
CYRILLIC_TABLE = dict(
zip(u'abwgdEOrZiIklmnopRstufhqcCSHTMUevAV',
u'абвгдеёжзийклмнопрстуфхццчшщъьыэюяю'))
class CyrillicfoConverter(object):
def to_cyrillic(self, pinyin, **kwargs):
pinyin = self._pre_convert(pinyin)
# 查表替换成注音
for find_re, replace in CYRILLIC_REPLACE:
pinyin = find_re.sub(replace, pinyin)
pinyin = ''.join(CYRILLIC_TABLE.get(x, x) for x in pinyin)
return pinyin
def to_cyrillic_first(self, pinyin, **kwargs):
pinyin = self.to_cyrillic(pinyin, **kwargs)
return pinyin[0]
def _pre_convert(self, pinyin):
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
# 将声调数字移动到最后
return RE_TONE3.sub(r'\1\3\2', pinyin)
converter = CyrillicfoConverter()
register(Style.CYRILLIC, func=converter.to_cyrillic)
register(Style.CYRILLIC_FIRST, func=converter.to_cyrillic_first)
(re.compile('1$'), ''),
)
BOPOMOFO_TABLE = dict(
zip('bpmfdtnlgkhjqxZCSrzcsiuvaoeEAIOUMNKGR2340',
'ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒㄓㄔㄕㄖㄗㄘㄙㄧㄨㄩㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦˊˇˋ˙'))
class BopomofoConverter(object):
def to_bopomofo(self, pinyin, **kwargs):
pinyin = self._pre_convert(pinyin)
# 查表替换成注音
for find_re, replace in BOPOMOFO_REPLACE:
pinyin = find_re.sub(replace, pinyin)
pinyin = ''.join(BOPOMOFO_TABLE.get(x, x) for x in pinyin)
return pinyin
def to_bopomofo_first(self, pinyin, **kwargs):
pinyin = self.to_bopomofo(pinyin, **kwargs)
return pinyin[0]
def _pre_convert(self, pinyin):
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
# 将声调数字移动到最后
return RE_TONE3.sub(r'\1\3\2', pinyin)
converter = BopomofoConverter()
register(Style.BOPOMOFO, func=converter.to_bopomofo)
register(Style.BOPOMOFO_FIRST, func=converter.to_bopomofo_first)
import pypinyin
from pypinyin.style import register
from .textnorm import textnorm
from .nonstd_pinyin import _nonstd_style
import jieba
jieba.initialize()
register('nonstd', _nonstd_style)
def _get_pinyin(text, std=True, pb=False):
'''
Params:
text: string, normalized sentences
std: boolean, standard pinyin stylc, default: standard pinyin style
pb: boolen, prosody label, default: False
Returns:
pinyin: string
'''
if std:
style = pypinyin.Style.TONE3
else:
style = 'nonstd'
text = text.strip()
punctuation = ', '
if text[-1:] in list(',.?!'):
punctuation = text[-1:] + ' '
text = text[:-1]
has_fi = has_finals(pinyin)
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
if not has_fi:
return pinyin
# 获取韵母部分
return get_finals(pinyin, strict=False)
def to_finals_tone3(self, pinyin, **kwargs):
if kwargs.get('strict'):
pinyin = convert_finals(pinyin)
has_fi = has_finals(pinyin)
# 用数字表示声调
pinyin = replace_symbol_to_number(pinyin)
# 将声调数字移动到最后
pinyin = RE_TONE3.sub(r'\1\3\2', pinyin)
if not has_fi:
return pinyin
# 获取韵母部分
return get_finals(pinyin, strict=False)
converter = FinalsConverter()
register(Style.FINALS, func=converter.to_finals)
register(Style.FINALS_TONE, func=converter.to_finals_tone)
register(Style.FINALS_TONE2, func=converter.to_finals_tone2)
register(Style.FINALS_TONE3, func=converter.to_finals_tone3)