What's a good way to remove script tags in body through the API?

[MarcC]

I have a client who had a previous site where they were just pasting analytics code (two types) into TinyMCE, so now all of their bodytext fields are kind of polluted with that stuff. What's a good way to remove this--assuming it'd be best just to do it using the PW API? These are script tags bracketed by HTML comments. 

[Macrura]

maybe regex when you run the search/replace with api..

i'm no regex guru, but i did find this...maybe a start

/<script.*>.*<script src='.+gaAddons.js' type='.+'><\/script>/s

or maybe html dom parser

http://simplehtmldom.sourceforge.net/

also - not sure how many pages there are, but in some cases i have used phpMyAdmin with inline mode and it's pretty fast to get through a lot of content

[diogo]

Find a regexp that matches comments and script tags and remove them with preg_replace() from all the bodytext fields. If you already imported everything to PW, you could do something like:

foreach($mypages as $p){
    $p->of(true);
    $p->body = preg_replace($pattern_for_script_tags,'', $p->body);
    $p->body = preg_replace($pattern_for_html_comments,'', $p->body);
    $p->save()
}

[ryan]

$body = preg_replace('{<script[^>]*>.*?</script>}is', '', $body);
$body = preg_replace('{<!--.*?-->}is', '', $body); 

The key here is to change the default "greedy" matching to be "lazy" matching using the .* followed by a question mark: .*?

That ensures that it will match only to the closest closing tag rather than the [default] furthest one. That way it won't wipe out legitimate copy. 

Also the "s" at the very end lets it traverse as many lines as needed to complete the match. Without that, it would only match opening and closing tags on the same line. 

[MarcC]

Awesome, thanks everybody. And thanks for the explanation, Ryan. I'm looking forward to improving my regexing