Re: [Exist-development] [Exist-open] matches() regex syntax

[Stefan M. <ste...@un...>]

Quite a while ago I wrote the message below. At that time I noticed that 
XQuery does not allow for look-aheads and look-behinds in regexes. 
Nonetheless, exist at that time used java regex syntax which allowed 
that. Obviously that has changed, as I get this error when migrating 
some patterns from 1.4 to current trunk.

a call to matches() with an offending regex yields the following message:

|Conversion from XPath2 to Java regular expression syntax failed: Error at character 2 in regular||  expression (\b)(test)(\b): invalid escape sequenc|


This is just a stupid example, I know, but at some points 
lookahead/behind can be very handy. I am just wondering if there is any 
chance to get java regex behaviour back? I fear it is for some reason 
not possible, like standards conformance, indexes?. I guess I could find 
some kind of work-around.

Anybody else stumbling over things like this?

cheers,
Stefan


On 19/02/09 14:47, Stefan Majewski wrote:
> I recently ran into the limitations of regexes in XSLT as according to
> the official specifications there is no lookahead or similar mechanisms
> allowed. This is very different to what would be allowed in Java's regex
> engine, where constructs like \b (?=  (?<= and the like are possible.
> Saxon does, even though it is implemented in Java, adhere very strictly
> to the standards and does consequently not allow these constructs.
>
> eXist nevertheless, does, even though this is not standard, allow for
> these extensions. I can't see a problem in this, as the regex engine
> still accepts everything that would be allowed by the XQuery/XPath specs
> for match(), tokenize() and replace().
>
> Nevertheless, now that I have moved portions of the logic outside of the
> XSL transformation, I wonder whether eXist's behaviour is intended to
> stay as it is or follow the saxon approach at one point and forbid
> lookahead patterns?
>
> It would be nice if anyone could comment on this, as if it weren't the
> case I would have to watch out and move it one day as a user-defined
> functioninto my private module.
>
> best,
>
> Stefan
>
>    


-- 
| Stefan Majewski  | Department of English, University of Vienna     |
| VOICE Corpus     | Spitalgasse 2-4, Universitätscampus AAKH, Hof 8 |
|                  | A-1090 Vienna                                   |
| Research Ass.(IT)| Phone: +43 1 4277 424 46                        |