def is_traditional(text):
u"""
Determine whether a text is simplified Chinese
Returns True if written in Simplified, False otherwise.
Note: This assumes the text is known to be one or the other.
>>> is_traditional(u'Hello,這是麻煩啦')
True
"""
return hanzidentifier.identify(text) is TRADITIONAL
def is_simplified(text):
u"""
Determine whether a text is simplified Chinese
Returns True if written in Simplified, False otherwise.
Note: This assumes the text is known to be one or the other.
>>> is_simplified(u'这是麻烦啦')
True
"""
return hanzidentifier.identify(text) is SIMPLIFIED
def is_traditional(text):
u"""
Determine whether a text is simplified Chinese
Returns True if written in Simplified, False otherwise.
Note: This assumes the text is known to be one or the other.
>>> is_traditional(u'Hello,這是麻煩啦')
True
"""
return hanzidentifier.identify(text) is TRADITIONAL
def is_simplified(text):
u"""
Determine whether a text is simplified Chinese
Returns True if written in Simplified, False otherwise.
Note: This assumes the text is known to be one or the other.
>>> is_simplified(u'这是麻烦啦')
True
"""
return hanzidentifier.identify(text) is SIMPLIFIED
def identify(text):
u"""
Wrapper for hanzidentifier identify function
Returns:
None: if there are no recognizd Chinese characters.
EITHER: if the test is inconclusive.
TRADITIONAL: if the text is traditional.
SIMPLIFIED: if the text is simplified.
BOTH: the text has characters recognized as being solely traditional
and other characters recognized as being solely simplified.
NEITHER: (or None) It's neither simplified nor traditional Chinese text.
>>> identify(u'这是麻烦啦') is SIMPLIFIED
True
>>> identify(u'這是麻煩啦') is TRADITIONAL
True
>>> identify(u'这是麻烦啦! 這是麻煩啦') is BOTH
True
>>> identify(u'This is so mafan.') is NEITHER
True
"""
return hanzidentifier.identify(text)
def identify(text):
u"""
Wrapper for hanzidentifier identify function
Returns:
None: if there are no recognizd Chinese characters.
EITHER: if the test is inconclusive.
TRADITIONAL: if the text is traditional.
SIMPLIFIED: if the text is simplified.
BOTH: the text has characters recognized as being solely traditional
and other characters recognized as being solely simplified.
NEITHER: (or None) It's neither simplified nor traditional Chinese text.
>>> identify(u'这是麻烦啦') is SIMPLIFIED
True
>>> identify(u'這是麻煩啦') is TRADITIONAL
True
>>> identify(u'这是麻烦啦! 這是麻煩啦') is BOTH
True
>>> identify(u'This is so mafan.') is NEITHER
True
"""
return hanzidentifier.identify(text)