At 22:09 27.05.2010, Delphi Inspiration wrote:
At 02:22 27.05.2010, Jim Bretti wrote:
I'm getting an access violation from DIPerlRegEx.Match when using the following subject / match pattern:
I could reduce this to the following:
Subject: für
Pattern: \p{Zs}*\R
The problem shows for the Umlaut letter 'ü' (but not 'ä') and requires both '/p{Zs}*' and '\R'. At a quick glance this looks like a PCRE problem. I will investigate further tomorrow.
Code analysis revealed that this is indeed a problem in PCRE. By mistake, it always tests Unicode Character Properties (the '\p{Zs}' part in your pattern) against UTF-8, even if coUtf8 is not set. The 'ü' Umlaut this therefore interpreted as the beginning of an UTF-8 sequence and results in an invalid character which finally leads to the access violation.
The workaround I suggested yesterday is safe and the recommended way to handle Umlaut characters with DIRegEx: Set the coUtf8 compile time option and encode both pattern and subject as UTF-8 before passing them to TDIRegEx methods.
I have forwarded the problem to the PCRE mailing list and will update DIRegEx when a fix becomes available.
Ralf
_______________________________________________
Delphi Inspiration mailing list
yunqa@xxxxxxxxxxxxx
//www.freelists.org/list/yunqa
_______________________________________________
Delphi Inspiration mailing list
yunqa@xxxxxxxxxxxxx
//www.freelists.org/list/yunqa
