Can't figure out how to invert this regex

Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

Monks,

I have a regex dilemna. The following regex will replace html entities with the word REPLACED:

s/&.{2,10}?;/REPLACED/g;
[download]

However, what I want to do is invert that, so that it replaces & symbols NOT in an html entity with the word REPLACED:

s/(?<!.{2,10}?;)&/REPLACED/g;
[download]

returns an error

/(?<!.{2,10}?;)&/: variable length lookbehind not implemented

Where did I derail?

Comment on Can't figure out how to invert this regex Select or Download Code

Replies are listed 'Best First'.
Re: Can't figure out how to invert this regex by merlyn (Sage) on Nov 25, 2005 at 14:45 UTC
If I recall, an entity can be only alphanums, so this should do it: `s/&(?![A-Za-z0-9]+;)/REPLACED/g;` [download] Literally: an ampersand not followed by (1 or more alphanums followed by a semicolon). If that's not the correct definition of an entity, season to taste. Not sure why you were looking to the left to determine ampersandy-ness. The interesting part is to the right. -- Randal L. Schwartz, Perl hacker Be sure to read my standard disclaimer if this is a reply. update: Ahh yes, the pound form. And the #x form. OK, try this: `s/&(?!(?:[A-Za-z0-9]+\|#\d+\|#x[0-9A-Fa-f]+);)/REPLACED/g;` [download]	[reply] [d/l] [select]
Re^2: Can't figure out how to invert this regex by Anonymous Monk on Nov 25, 2005 at 15:05 UTC
actually, an entity can also have the form { so there would be a need to check for a pound as well as alpha-num.	[reply]
Re^2: Can't figure out how to invert this regex by Anonymous Monk on Nov 25, 2005 at 15:03 UTC
"Not sure why you were looking to the left to determine ampersandy-ness." Just a mistake out of my own inexperience with regex's. I didn't really understand where it was looking. Now I get it.	[reply]
Re: Can't figure out how to invert this regex by anniyan (Monk) on Nov 25, 2005 at 14:41 UTC
You are using variable length in negative look behind, as per the perl regex tutorial perlre, we cannot use variable length in negative look behind condition. Rather we can use variable length in positive and negative look ahead. updated:Thanks sauoq and apologises for the mistake. Regards, Anniyan (CREATED in HELL by DEVIL to s\|EVILS\|GOODS\|g in WORLD)	[reply]
Re^2: Can't figure out how to invert this regex by sauoq (Abbot) on Nov 25, 2005 at 22:42 UTC
Rather we can use variable length in positive look behind No, that's incorrect. Variable length patterns are supported in neither negative nor positive look-behind. -sauoq "My two cents aren't worth a dime.";	[reply]
Re^2: Can't figure out how to invert this regex by Anonymous Monk on Nov 25, 2005 at 14:51 UTC
Ah... so: `s/&(?!.{2,10}?;)/REPLACED/g;` [download] Gives me what i need. Thanks.	[reply] [d/l]