ansaurus

Question

How can I remove and transform selective HTML tags?

Answer 1

+2 A:

Here's the Perl for removing the non-p tags--it won't work across lines though

perl -pe 's/<\/?(?>[^p]|p\w+)[^>]*>//ig'

That will print it out to standard out and you can redirect it from there.

If you have only one link, you could do this:

perl -pe 's/<embed\s+src="(.*?\.swf)"\/?>/<a href="$1">Click here for external description<\/a>/i;s/<\/?(?>[^ap]|[ap]\w+)[^>]*>//ig'

Axeman 2008-11-07 15:06:29

beware that literal '>' is allowed within attribute values, so your regexp would fail for obscure-but-valid html.

kch 2008-11-07 15:09:04

I didn't know that was the case. Most of the time, I've gotten away with just that expression.

Axeman 2008-11-07 15:21:19

This is a pretty easy job for HTML::Parser, and that handles everything correctly. :)

brian d foy 2008-11-07 20:30:13

Answer 2

+1 A:

I came up with this, for sed:

sed -e 's/[<][/][^Pp][^>]*[>]//g' -e 's/[<][^/Pp][^>]*[>]//g' file

Steve Baker 2008-11-07 15:14:10

Good point, I should make mine case sensitive as well.

Axeman 2008-11-07 15:25:29

Answer 3

A:

Use xmlstarlet to filter out the tags you want; as it knows XML syntax (which is considerably more complex than can be correctly captured in a regex), it isn't prone to failure in corner cases like nested tags, and can decode & and kin within elements correctly. The select subcommand will take XPath, so a trivial expression to find A elements will do what you want.

The format subcommand has an option to take HTML as input, so you'll want to do an initial processing stage with that if your content isn't valid XHTML.

Charles Duffy 2008-11-07 15:41:24

Answer 4

A:

If you want to do this in Perl, there are many HTML utility modules that can help you transform and filter HTML. Other languages probably have similar libraries. See, for instance, HTML::Parser. You define handlers for the parts you want to affect. Once you get control of that part, you can transform it and decide how to output it.

If you have very regular output that is generated by another program, you might be able to get away with a regex. However, whenever I've done that the regex starts to get pretty nasty as I find the special cases and exceptions.

It sounds like you may be able to turn this problem around though. Are you really just trying to extract URLs and link to them? Instead of messing with the parsers, can you just pull out the URLs and write a completely new document without caring about the old one? Or maybe just completely replace a region with the larger document?

brian d foy 2008-11-07 20:35:57

ansaurus

tags:

views:

answers:

How can I remove and transform selective HTML tags?

related questions