Regex to select all image html tags conditionally on the src value

Question

I need a regex to do the following (unfortunately it has to be a regex, I can't code this because it's working within a purchased product):

I'd like to select all image tags in a chunk of html where either the image tag does not contain a class attribute, or, if it does contain a class attribute, that attribute does not contain a specific string at the beginning. Basically, I want to strip (by matching) all image tags from a chunk of html EXCEPT for images with a particular class applied to them.

This could be two separate regular expressions - I just want to match them - not extract any data.

So, for example, let's say the class I want to keep is called Pretty.

I'd like the regex to match:

<img src="xx"/>
<img border="x" src="xx"/>
<img whatever other attributes src="xx"/>
<img class="ugly" src="xx"/>
<img whatever other attributes class="fugly" src="xx"/>

but not match

<img class="Pretty" src="xx"/>
<img whatever other attributes class="Pretty" src="xx"/>
<img class="Pretty subpretty" src="xx"/>

If it's easier to do in one regex (one to match all image tags without class attribute, and one to match ones with class attributes that aren't 'pretty') that's totally fine too.

Answer 1

Use XPath instead, as that's what it's for:

//img[not(contains(@class,'Pretty'))]

This XPath expression looks for every img element whose class attribute does not contain the string 'Pretty'. I think it works for elements which are missing the class attribute.

Parsing XML and HTML with regular expressions is usually a very bad idea. Of course, XPath only works if the HTML in question is strict. If it's not a valid XML document then you might want to default back to something else, but even so regex isn't the right tool for the job.

Addendum: I was wrong about getting back to this in 30 minutes. Something came up and I don't have the time to sort it out. If it doesn't work for elements lacking the class attribute, use the following expression:

//img[(not(@class)) or (not(contains(@class,'Pretty')))]

Answer 2

A bit quick and dirty, but it works:

/(?!<img\b[^>]+\bclass="?[^>"]*\bPretty\b)<img\b[^>]*>/

How it works:

<img\b[^>]+\bclass="?[^>"]*\bPretty\b matches all "Pretty" images.

<img\b[^>]*> matches all images. So, put the "Pretty" image subpattern in a negative lookahead in front of the subpattern to match all images. This will then match all images, minus those that match the pretty subpattern.

Answer 3

<img(?:\s+(?:(?!class\b)\w+="[^"]*"|class="(?!Pretty)[^"]*"))*/>

That seems to answer your question, but there are many details you didn't address, like:

• Are the tag- and attribute names consistently lowercase?

• What if the class name starts with "pretty" (i.e., is it case sensitive)?

• Are attribute values always quoted, and always with double-quotes?

• Will there ever be extra whitespace, like around the "=" or before the final "/>"?

• Does your "purchased tool" support regexes with negative lookaheads?

Answer 4

Yes, to all those who suggested I would be better off using something other than regex you are of course right, but i guess you missed the first sentence in the question.

I ended up finding the solution, nick's and alan M's look the closest to that, thanks guys! Fortunately I can use negative lookahead's so it works perfectly :)