Question

In text i want to find structures like every thing till some text, but not match between some word.

Example in text:

Templates : You can add custom templates for your theme. Updated on 2010 look[124] end
Media RSS feed : Add the Cooliris Effect to your gallery Updated on 2011 look[124]
Role settings : Each gallery has a author Updated at 2010 ...  look[124] end
AJAX based thumbnail generator : No more server Updated on 2010 look[124] end limitation during the batch process Copy/Move : Copy or move images between Updated on 2010 this look[124] galleries Sortable Albums : Create your own sets of images Updated on 2010 this look[124] end
Upload or pictures via a zip-file (Not in Safe-mode)
Watermark function : You can add a watermark image or text 
...

I need to find “Updated .*[124] end” every match must start this “Update” and ends with “[number]” and word “end“. But some text looks very similar, but not ends with word “end“. This text must not mach. How to make it work?

I try to write

/Updated(.*?)\[.*?\]\send/msi

or

Updated(.*?)\[.*?\](?!Updated)\send

But this takes strings like:

Updated on 2011 look[124] Role settings : Each gallery has a author Updated at 2010 ...  look[124] end
Updated on 2010 this look[124] galleries Sortable Albums : Create your own sets of images Updated on 2010 this look[124] end

How to write regex witch skips bad matches?

http://regexr.com?2vh1j

Thanks for your opinion.

Answer 1

I think this is what you were trying for with your second regex:

Updated\s++(?>(?!Updated\b|end\b)\S+\s+)*+end\b

In other words, match Updated and look for the corresponding end. If you find another Updated first, you know you started at the wrong place, so abandon that match. I excluded end as well because that lets me match the words possessively (i.e., with *+); the regex never has to backtrack to find or (more importantly) eliminate a match.

If you really have to specify the look[nnn] part, this should do the trick:

Updated\s++(?>(?!Updated\b|end\b|look\[\d+\])\S+\s+)*+look\[\d+\]\s+end\b

Add the i flag for a case-insensitive match if you need to, but you don’t need the m or s flags. If this seems overly complicated, it’s because I don’t know your data as well as you do. There’s a good chance this is all you really need:

Updated(?:(?!Updated).)*\send