Regular Expression – Where string contains ‘bar’ but is not preceded by ‘foo’

regex

I currently have a regular expression to match where a string contains 'bar' but is not preceded by 'foo':

(?<!foo)bar

I wish to modify the regex to match where a string contains 'bar' but is not preceded by 'foo' and (optionally) any additional character(s).

EG. 'footestbar' should not match as bar is preceded by 'foo' followed by 'test'.

I cannot find a way to achieve this as regular expressions do not permit the use of a quantifier within a negative lookbehind. Subsequently I cannot do the following:

(?<!foo.*)bar

Is there an alternative approach?

UPDATE:
The regular expression flavour is re2. I am currently reviewing the documentation, but on first glance have noticed the following:

(?<!re)     after text not matching re (NOT SUPPORTED)

Therefore the solution cannot use a negative lookbehind (if I'm interpreting the documentation correctly).

Best Answer

Amended Post

So it would seem that some (most?) Regex engines only work with a fixed length lookbehind. However, they seem fine to accept a variable length lookahead. I'm not sure what your use case for this is or how viable it would be, but one idea could be to reverse your string, and use:

rab(?!.*oof)

Of course, it looks ridiculous, and I don't know how your using it or how useful it will be. Most of my experience with Regex has been using .NET, and mostly not needed lookbehinds. Just an idea for you..!

Original Post - Only works with certain Regex engines (Tested using .NET)

Try the following regex:

(?<!foo(?:.*)?)bar

This adds an optional, non capturing group within the same negative look behind.

Using Javascript to match regular expressions

Maybe you want try this at chrome's console:

var p=/.*(regu).+?\ /gi; console.log( document.body.innerText.match(p) );

Just open console, copy and paste above to console and hit enter. You can test it here at this page.
This can be improved if it fits in.

Here we print to console match indexes and matched text. Here we try to match text that contains regu, 20 chars before (or less if start of line) and 10 chars after (or less if eol).

var p=/.{0,20}regu[^ \n]+[^\n]{0,10}/gi;
while (m = p.exec(document.body.innerText)) { 
    console.log( 'Index: '+m.index+' Match: '+m ); }

Also try this, it will paint background of all matches on page red, rexexp is not perfect but at least it should not mess with HTML tags:

var p=/(\>{1}[^\n\<]*?)([^\n\<]{0,30}regu[^\n\<]{0,10})/gi,b=document.body;
b.innerHTML=b.innerHTML.replace(p,'$1<span style="background-color:red;">$2</span>');

Bookmarking this:

Another way to use this is through javascript: protocol (same code as just above):

javascript:(function(){var p=/(\>{1}[^\n\<]*?)([^\n\<]{0,30}regu[^\n\<]{0,10})/gi,b=document.body;b.innerHTML=b.innerHTML.replace(p,'$1<span style="background-color:red;">$2</span>');})();

For example, using javascript: protocol one can insert a little search box to any web page for searching regexp's.
I think that you already know that simple regexp can also used to remove red matches from page.
If I continue to develop this for few more hours we may have search plugin that fit in bookmark :)

Is it possible to use NOT in a regular expression in TextMate

You want to use this bit of the syntax:

(?=subexp)         look-ahead
(?!subexp)         negative look-ahead
(?<=subexp)        look-behind
(?<!subexp)        negative look-behind

In your case, (?<!\.)foo(?!/)

Best Answer

Amended Post

Original Post - Only works with certain Regex engines (Tested using .NET)

Related Solutions

Chrome – How to search for regular expressions within webpages using Google Chrome or IE

Using Javascript to match regular expressions

Bookmarking this:

Is it possible to use NOT in a regular expression in TextMate

Related Question