Re: [Boost-users] [regex] wildcard matching byte not character

Subject: Re: [Boost-users] [regex] wildcard matching byte not character
From: John Maddock (john_at_[hidden])
Date: 2010-03-01 05:22:07

> I'm having trouble with the behaviour of the wildcard character when
> using boost regex and unicode strings. I would expect a . to match a
> character, not a byte, but that's not the behaviour I'm seeing. I would
> have thought one wildcard would match any previous character, but for
> multi-byte characters in UTF-8 I have to use multiple wildcards to match
> them.
>
> I would appreciate it if someone could explain whether this is expected
> behaviour or not, or if there are flags that control this.
>
> What I'm trying to accomplish is to match a pattern (in UTF-8 ) against
> a string (in UTF-8). I'm creating icu UnicodeStrings since I'm having
> other problems with straight UTF-8 char*s and my platform doesn't
> support w_chars. I can show examples of the non-UnicodeString problems
> if desired.

You're constructing invalid UnicodeString's: the const char* constructor 
does not convert from UTF-8, if the strings are constructed as:

UnicodeString s(buf, "UTF8");

Then the output changes to

Success!
Failed

Which is what you expected.

HTH, John.