Umlauts in regexp matching (via locale?)

问题:

I'm surprised that I'm not able to match a German umlaut in a regexp. I tried several approaches, most involving setting locales, but up to now to no avail.

locale.setlocale(locale.LC_ALL, 'de_DE.UTF-8')
re.findall(r'\w+', 'abc def g\xfci jkl', re.L)
re.findall(r'\w+', 'abc def g\xc3\xbci jkl', re.L)
re.findall(r'\w+', 'abc def güi jkl', re.L)
re.findall(r'\w+', u'abc def güi jkl', re.L)

None of these versions matches the umlaut-u (ü) correctly with \w+. Also removing the re.L flag or prefixing the pattern string with u (to make it unicode) did not help me.

Any ideas? How is the flag re.L used correctly?

回答1:

Have you tried to use the re.UNICODE flag, as described in the doc?

>>> re.findall(r'\w+', 'abc def güi jkl', re.UNICODE)
['abc', 'def', 'g\xc3\xbci', 'jkl']

A quick search points to this thread that gives some explanation:

re.LOCALE just passes the character to the underlying C library. It really only works on bytestrings which have 1 byte per character. UTF-8 encodes codepoints outside the ASCII range to multiple bytes per codepoint, and the re module will treat each of those bytes as a separate character.

回答2:

In my case \S gave me better results than \w, plus saving the file as utf-8, plus using re.UNICODE