def test_pre_convert_style_return_value():
class A(DefaultConverter):
def pre_convert_style(self, han, orig_pinyin, style, strict, **kwargs):
return 'test'
han = '测试'
assert DefaultConverter().convert(han, Style.TONE2, False, 'ignore',
True) == [['ce4'], ['shi4']]
assert A().convert(han, Style.TONE2, False, 'ignore', True) == [['test'],
['test']]
def test_post_handle_nopinyin_return_value():
class A(DefaultConverter):
def post_handle_nopinyin(self, chars, style, heteronym, errors, strict,
pinyin, **kwargs):
return 'abc'
han = 'test'
assert DefaultConverter().convert(han, Style.TONE2, False, 'default',
True) == [['test']]
assert A().convert(han, Style.TONE2, False, 'default', True) == [['abc']]
def test_post_pinyin_return_value_phrase_pinyin():
class A(DefaultConverter):
def post_pinyin(self, han, heteronym, pinyin, **kwargs):
return {
'北': [['zhāo']],
'京': [['yáng']],
'北京': [['zhāo'], ['yáng']],
}[han]
han = '北京'
assert DefaultConverter().convert(han, Style.TONE3, False, 'ignore',
True) == [['bei3'], ['jing1']]
assert A().convert(han, Style.TONE3, False, 'ignore', True) == [['zhao1'],
['yang2']]
def test_post_pinyin_return_value_single_pinyin():
class A(DefaultConverter):
def post_pinyin(self, han, heteronym, pinyin, **kwargs):
return {
'测': [['zhāo']],
'试': [['yáng']],
'测试': [['zhāo'], ['yáng']],
}[han]
han = '测试'
assert DefaultConverter().convert(han, Style.TONE3, False, 'ignore',
True) == [['ce4'], ['shi4']]
assert A().convert(han, Style.TONE3, False, 'ignore', True) == [['zhao1'],
['yang2']]
def __init__(self, converter=None, **kwargs):
self._converter = converter or DefaultConverter()
"""对字符串进行分词后将调用 ``post_seg`` 方法对分词后的结果做处理。
默认原样返回传入的 ``seg_data``。
如果这个方法的返回值类型是 ``list``，表示对分词结果做了二次处理，此时，
``seg`` 方法将以这个返回的数据作为返回值。
:param hans: 分词前的字符串
:param seg_data: 分词后的结果
:type seg_data: list
:return: ``None`` or ``list``
"""
pass
_default_convert = DefaultConverter()
_default_pinyin = Pinyin(_default_convert)
def to_fixed(pinyin, style, strict=True):
# 用于向后兼容，TODO: 废弃
return _default_convert.convert_style(
'', pinyin, style=style, strict=strict, default=pinyin)
_to_fixed = to_fixed
def handle_nopinyin(chars, errors='default', heteronym=True):
# 用于向后兼容，TODO: 废弃
return _default_convert.handle_nopinyin(
