Regexp problem

Oersoep

Problem:

I'm writing a little script that translates pages to morse. The plan is to isolate any HTML, CSS and scripts, translate the texts and than put it all back together.
It's working quite well, but at some pages the regexps start acting funny.

The following piece of code is a small exerpt of what happens.
The first piece of code works like it should, but the same doesn't work when $script contains a full webpage.

not working example

working example

the same function working very well on a tiny page

Seems like some kind of maximum string length for handling regexps. I searched, but I can't find it in any docs.
Does anyone know a solution? Does anyone know the exact maximum?
Should I cut the page into pieces? What if I cut a HTML tag? (I want to keep those intact)

Thanks!!

<?
// this works

$string = '<script LANGUAGE="JavaScript">
  blah!
</script>  etcetera...' ;

if( eregi( '^.*</script[^>]*>', $string, $res ) ){
	echo $res[0];
}else{
	// geen match
	echo "Why hast thou forsaken me?!?";
}


// this doesn't

$array = file( 'http://www.zend.com/manual/', 'r');
$string = implode( '', $array );

if( eregi( '^.*</script[^>]*>', $string, $res ) ){
	echo $res[0];
}else{
	// geen match
	echo "Why hast thou forsaken me?!?";
}


?>

Oersoep

Is really no one able to help?

zmola

Hi. I can try to help, but could you explain what's not working please? I see that the "working" and "not working" examples have a lot of morse code. Maybe I'm just too tired to see straight, especially with all those dots and dashes. 😉

Oersoep

Well, if you see morse, the script is working.

The problem was that regexps didn't work when the 'victim' string got too long. So with some pages it works, and with some it doesn't. (i.e. the script I gave.)

I worked around the problem by working with chunk_splitting the 'victim'. Not very pretty, but it works.

Still the problem persists and I still haven't found any documentation about this maximum string length.

Even stranger: I tried the script on a different machine, and the maximum was different.

zmola

Hi. When I look at the "not working" example, I see morse code from top to bottom, along with broken image tags. I did find this regarding maximum string length, but who knows if "worked with" means regexp.

Oersoep

As I said, I fixed the working examples by cutting the string in pieces...

When it failes, the string itself stays intact... The regexps just don't match anything.

The broken image tags are caused by missing full urls on the original pages. Has anyone got an advice on that one?

zmola

Perhaps a possible solution...

<?
$url = "http://www.zend.com/manual/";
$img = "<IMG src=\"/images/zend_logo.gif\">";

$url = parse_url($url);
$url = $url[scheme]."://".$url[host];

$img = str_replace("src=\"","src=\"$url",$img);

echo $img;
?>

Cheers.

ctimmer

1st a suggestion, change:
if( eregi( '^{.</script[^>]>',} $string, $res ) ){
to
if( eregi( '</script[^>]*>', $string, $res ) ){

You don't need to skip 'any characters' at the beginning and this can cause problems.

In the nonworking example, I suspect that your search is terminating at the first linefeed. Using pregi, there is an option (PCRE_MULTILINE) to search multiple lines.

Curt