Regex puzzle

re: Regex puzzle

haacked — Wed, 25 Feb 2004 18:20:00 GMT

Assuming that the allowed tags may not have attributes, something like this:

Search for this:

</?(p[^p>]+|b[^r>]+|br[^>]+)>

Replace with "".

If the allowed tags may have attributes, I'll get back to you.

(note, I haven't tested this).

re: Regex puzzle

haacked — Wed, 25 Feb 2004 18:31:00 GMT

Whoops. My solution was too simple. For example, it would not strip out this properly:

">

This is better (but much harder to understand.

</?(([^pb]|p[^p\s>]|b[^r\s>]|br[^\s>])\w+)("[^"]*"|'[^']*'|[^'">])*>

re: Regex puzzle

haacked — Wed, 25 Feb 2004 18:33:00 GMT

sorry again. I forget that you can't edit your posts on these things. Eric, if you could delete my repost, I should point out that the "\w+" portion should be "\w*".

re: Regex puzzle

Ricky Dhatt — Wed, 25 Feb 2004 19:00:00 GMT

Simple? -- there is no such thing. In my experience of doing this, there are too many things to deal with, like entities (&nbsp) and XHTML ( . I just used a full fledged parser now days. But I have the luxury of not worrying about the overhead.

re: Regex puzzle

Darren Neimke — Wed, 25 Feb 2004 19:16:00 GMT

Take a look at Html Agility Pack:

http://blogs.regexadvice.com/dneimke/archive/2004/02/11/500.aspx

There's a couple of examples at the bottom of that page which demonstrate the syntax for using it.

re: Regex puzzle

Jake — Wed, 25 Feb 2004 19:29:00 GMT

Would it be easier to bring in the HTML into an XML Reader and then parse it through that way? only problem is what would it do with the content (already know that it would handle the tags properly)

Jake

re: Regex puzzle

Paschal — Wed, 25 Feb 2004 20:07:00 GMT

Eric the user you talk about is not me by any chance :-) I asked the question few months ago and I still searching for a solution. This is really puzzled me.
I need it to clean some HTML documents but I want to keep the breaklines and paragraphs.
Most of the regex I saw stripped all the tags.

re: Regex puzzle

Raymond Chen — Wed, 25 Feb 2004 21:59:00 GMT

A wise man once said,

Some people, when confronted with a problem, think “I know, I’ll use regular expressions.” Now they have two problems.

I think this is one of the cases where trying a pure regex solution creates its own problems.

re: Regex puzzle

Eric TF Bat — Wed, 25 Feb 2004 23:31:00 GMT

Step one: look for all character #255s (if any) and double them: #255#255 (you'll see why in a sec).

Step two: look for all , , and (note typo in original question; is meaningless!) and replace with #255p, #255/p and #255br. Use regexps to allow for spaces between the br and the /, if you like.

Step three: convert all & to &

Step four: convert all < and > to < and >

Step five: convert all #255p to , #255/p to and #255br to (note space: required for old, dumb browsers)

Step six: convert all remaining #255#255 to #255, if you care.

Don't use regexps to handle HTML. Raymond is right; that way lies insanity.

Incidentally, it's impolite to randomly delete people's html, which is why I convert the < and > instead of rudely deleting it and giving them a nasty surprise. I just hope this blog's commenting system doesn't delete them, cos this message will be (even more) incomprehensible...

Take Outs: The Digital Doggy Bag of Blog Bits for 25 February 2004

Enjoy Every Sandwich — Thu, 26 Feb 2004 07:23:00 GMT

In the bag tonight: Less bitch'n and whin'n. Counts:Blogging: 8; Dev: 22; Otherwise: 8; SQL: 5; WILY: 8. Line of the night:

re: Regex puzzle

AvonWyss — Sun, 07 Mar 2004 14:03:00 GMT

What's wrong with this?

<(?!/?p|br)("[^"]*"|'[^']*'|[^>])+>

(I'd suggest to use this regex with the ExplicitCapture, IgnoreCase and SingleLine options enabled)

For some reason that I don't know, many neat features of the .NET regex engine (not exclusive to it, though) are rarely used, like assertion groups, backreferences, named groups, UNICODE char groups, and match evaluators for replacement patterns...

Anyways, to specifically clean out HTML code, I'd also rather use some HTML to XHTML converter (like HTML Tidy) and then use some code that works on the XML, or maybe just some XSLT, to get the wanted result.

re: Regex puzzle

Lost_In_JavaScript_Land — Tue, 13 Jul 2004 15:46:00 GMT

This works for me (in ASP/VBScript). It keeps , , and (upper and lower case) Also compensates for parameters.

Function FilterHTML(tempStr)
Dim re, tempStr2
Set re = New RegExp

re.IgnoreCase = True

re.Pattern = "<((?!P|BR).*).*>.*<\/\1>"
tempStr2 = re.Replace(tempStr, "")

re.Pattern = "<((?!P|\/P|BR).*)>"
FilterHTML = re.Replace(tempStr2, "")

Set re = Nothing
End Function

re: Regex puzzle

Lost_In_JavaScript_Land — Tue, 13 Jul 2004 15:52:00 GMT

Stupid me...just realized this is a C# blog...but converting the code shouldn't be too nasty ;)