Recent changes to bugs

Spurious *p++ constructs in code

John Byrd — Thu, 09 Feb 2012 00:23:10 -0000

On lines 510, 516, 522, 528, and 618 of trex.cc, there are several pointer increments of the form:

*p++

Where the value acquired by deferencing p is not used in any way. It is likely that the author meant:

p++

In all these places.

Incorrect matching behaviour

Maxim Zinal — Wed, 28 Dec 2011 12:54:42 -0000

T-Rex 1.3 does not correctly handle the following simple test case:

REGEXP STRING RESULT
^.*_C$ AB_C match (correct)
^.*_C$ A_B_C non-match (INCORRECT)
^.*_C$ _C non-match (INCORRECT)

Patch for C++ on MS Visual Studio 2005

Anonymous — Fri, 25 Feb 2011 17:32:21 -0000

MS Visual Studio 2005 doesn't seem too impressed with
extern "C" {
#include "trex.h"
}
and still compiles with C++ name mangling causing linker errors.

The fix is to place wrap the entire trex.h conditionally with extern "C" { } when __cplusplus is defined. Both should work, but MS is broken, so, apply the attached patch if having trouble.

unicode version not tested

Anonymous — Fri, 22 Aug 2008 10:35:26 -0000

The unicode version seems to not have been tested

#ifdef _UINCODE <==== _UNICODE
#define scisprint iswprint
#define scstrlen wcslen
#define scprintf wprintf
#define _SC(x) L(x) <==== L##x
#else
#define scisprint isprint
#define scstrlen strlen
#define scprintf printf
#define _SC(x) (x)
#endif

empty capture expression

ancuta_onofrei — Tue, 28 Aug 2007 11:03:06 -0000

Hello!

After downloading t-rex1.3 release i've discovered a bug: empty capture expression.

Using test.c i've tested some (pattern, test string) pairs.
Here it is:

For the pattern "abc([0-9]{2})(a)(b)" and the string to be tested "abc12ab".
The captures were:
[0]abc12ab
[1]12
[2]
[3]b
match! 4 sub matches.

For the pattern "abc([0-9]{1})(a)(b)" and test string "abc1ab" the captures were ok.

It surprised me to see that release 1.0 was displaying the correct output:
[0]abc12ab
[1]12
[2]a
[3]b
match! 4 sub matches