Re: EggRegex

[Marco Barisione <barisione gmail com>]

Matthias Clasen wrote:
> When I was last looking at regular expressions for GLib (which
> resulted in the current eggregex code), the first decision was to
> go for Perl regular expression, rather than posix. That naturally
> leads to PCRE. The main gripe with PCRE was (and is) that it
> had (and probably still has) relatively limited Unicode support.

The version of eggregex in libegg uses the three years old pcre 4.5. Now 
pcre 6.7 has a better support for Unicode.

Now PCRE:
- handles UTF-8
- knows that, doing a caseless match, à matches À
- has generic character types for non ASCII characters, so \p{Lt} 
matches a title case letter, \p{Sc} matches a currency symbol, and so on

Extended properties such as "Greek" or "InMusicalSymbols" are not supported.

> And it brings its own implementation of the necessary Unicode
> data, instead of using the GLib one.

Yes, but it shouldn't be too difficult to port pcre to use glib for 
Unicode. I can't do it because my knowledge of Unicode is very limited.

However this would mean that we should always use the internal PCRE 
instead of the system supplied one.

--
Marco Barisione
http://www.barisione.org/