mozillazg / python-pinyin / 502

pending completion

Build # 502

Build Type

push

travis-ci

Committed by

mozillazg

Commit Message

使用 `phrase-pinyin-data`_ v0.7.2 的词语拼音数据

* Fixes #112
* Fixes #117
* Fixes #122
* Fixes #131

Run Details

275 of 527 relevant lines covered (52.18%)

0.52 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

24.24

/pypinyin/style/_utils.py

# -*- coding: utf-8 -*-
from __future__ import unicode_literals

from pypinyin.standard import convert_finals
from pypinyin.style._constants import (
    _INITIALS, _INITIALS_NOT_STRICT,
    RE_PHONETIC_SYMBOL, PHONETIC_SYMBOL_DICT,
    RE_NUMBER
)


def get_initials(pinyin, strict):
    """获取单个拼音中的声母.

    :param pinyin: 单个拼音
    :type pinyin: unicode
    :param strict: 是否严格遵照《汉语拼音方案》来处理声母和韵母
    :return: 声母
    :rtype: unicode
    """
    if strict:
        _initials = _INITIALS
    else:
        _initials = _INITIALS_NOT_STRICT

    for i in _initials:
        if pinyin.startswith(i):
            return i
    return ''


def get_finals(pinyin, strict):
    """获取单个拼音中的韵母.

    :param pinyin: 单个拼音
    :type pinyin: unicode
    :param strict: 是否严格遵照《汉语拼音方案》来处理声母和韵母
    :return: 韵母
    :rtype: unicode
    """
    if strict:
        pinyin = convert_finals(pinyin)

    initials = get_initials(pinyin, strict=strict) or ''
    # 没有声母，整个都是韵母
    if not initials:
        return pinyin
    # 按声母分割，剩下的就是韵母
    return ''.join(pinyin.split(initials, 1))


def replace_symbol_to_number(pinyin):
    """把声调替换为数字"""
    def _replace(match):
        symbol = match.group(0)  # 带声调的字符
        # 返回使用数字标识声调的字符
        return PHONETIC_SYMBOL_DICT[symbol]

    # 替换拼音中的带声调字符
    return RE_PHONETIC_SYMBOL.sub(_replace, pinyin)


def replace_symbol_to_no_symbol(pinyin):
    """把带声调字符替换为没有声调的字符"""
    def _replace(match):
        symbol = match.group(0)  # 带声调的字符
        # 去掉声调: a1 -> a
        return RE_NUMBER.sub(r'', PHONETIC_SYMBOL_DICT[symbol])

    # 替换拼音中的带声调字符
    return RE_PHONETIC_SYMBOL.sub(_replace, pinyin)


def has_finals(pinyin):
    """判断是否有韵母"""
    # 鼻音: 'ḿ', 'ń', 'ň', 'ǹ ' 没有韵母
    for symbol in ['\u1e3f', '\u0144', '\u0148', '\u01f9']:
        if symbol in pinyin:
            return False

    return True

1	# -- coding: utf-8 --
2	from __future__ import unicode_literals
3
4	from pypinyin.standard import convert_finals
5	from pypinyin.style._constants import (
6	_INITIALS, _INITIALS_NOT_STRICT,
7	RE_PHONETIC_SYMBOL, PHONETIC_SYMBOL_DICT,
8	RE_NUMBER
9	)
10
11
12	def get_initials(pinyin, strict):
13	"""获取单个拼音中的声母.
14
15	:param pinyin: 单个拼音
16	:type pinyin: unicode
17	:param strict: 是否严格遵照《汉语拼音方案》来处理声母和韵母
18	:return: 声母
19	:rtype: unicode
20	"""
21	if strict:
22	_initials = _INITIALS
23	else:
24	_initials = _INITIALS_NOT_STRICT
25
26	for i in _initials:
27	if pinyin.startswith(i):
28	return i
29	return ''
30
31
32	def get_finals(pinyin, strict):
33	"""获取单个拼音中的韵母.
34
35	:param pinyin: 单个拼音
36	:type pinyin: unicode
37	:param strict: 是否严格遵照《汉语拼音方案》来处理声母和韵母
38	:return: 韵母
39	:rtype: unicode
40	"""
41	if strict:
42	pinyin = convert_finals(pinyin)
43
44	initials = get_initials(pinyin, strict=strict) or ''
45	# 没有声母，整个都是韵母
46	if not initials:
47	return pinyin
48	# 按声母分割，剩下的就是韵母
49	return ''.join(pinyin.split(initials, 1))
50
51
52	def replace_symbol_to_number(pinyin):
53	"""把声调替换为数字"""
54	def _replace(match):
55	symbol = match.group(0) # 带声调的字符
56	# 返回使用数字标识声调的字符
57	return PHONETIC_SYMBOL_DICT[symbol]
58
59	# 替换拼音中的带声调字符
60	return RE_PHONETIC_SYMBOL.sub(_replace, pinyin)
61
62
63	def replace_symbol_to_no_symbol(pinyin):
64	"""把带声调字符替换为没有声调的字符"""
65	def _replace(match):
66	symbol = match.group(0) # 带声调的字符
67	# 去掉声调: a1 -> a
68	return RE_NUMBER.sub(r'', PHONETIC_SYMBOL_DICT[symbol])
69
70	# 替换拼音中的带声调字符
71	return RE_PHONETIC_SYMBOL.sub(_replace, pinyin)
72
73
74	def has_finals(pinyin):
75	"""判断是否有韵母"""
76	# 鼻音: 'ḿ', 'ń', 'ň', 'ǹ ' 没有韵母
77	for symbol in ['\u1e3f', '\u0144', '\u0148', '\u01f9']:
78	if symbol in pinyin:
79	return False
80
81	return True

mozillazg / python-pinyin / 502

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous