New to regex but can't solve this problem

[jrw4]

If I have a string that looks like:

 

<content type="xhtml" xmlns:xhtml="http://www.w3.org/1999/xhtml">

 

or

 

<content type="xhtml" xmlns:x="http://www.w3.org/1999/xhtml">

 

I am trying to get the value that is between xmlns: and the equal sign.

 

<content type="xhtml" xmlns:[THIS IS WHAT IM LOOKING FOR]="http://www.w3.org/1999/xhtml">

 

So I have tried the following code:

 

$string = '<content type="xhtml" xmlns:xhtml="http://www.w3.org/1999/xhtml">';

preg_match("xmlns:[\w]+\=", $string, $matches);

var_dump($matches);

 

This returns null

 

So I am not sure how to find that.  What am I doing wrong and what should I do to find that part of the string?

[nrg_alpha]

The problem is that you're delimiters are lacking..., plus you are not isolating what you are looking for in your pattern.

Here is an example of how I would tackle it (throwing both url examples you listed as an array):

 

$html = array('<content type="xhtml" xmlns:x="http://www.w3.org/1999/xhtml">', '<content type="xhtml" xmlns:xhtml="http://www.w3.org/1999/xhtml">');
foreach($html as $val){
    preg_match('#xmlns:\K[^=]+#', $val, $match);
    echo $match[0] . "<br />\n";
}

 

Output:

x
xhtml

 

This way, $match[0] will only contain what is between xmlns and =.

 

EDIT - In our resources page, you read up about the 'Why delimiters?' thread, as well as delimiters in the php manual.

[cags]

Sorry to hijack the thread slightly, but it looks like nrg_alpha has solved it anyway, what does the \K modifer do, I tried looking it up and came up with 'Named Capturing Groups', but the references I found seemed to indicate \k was a .NET syntax.

 

EDIT: Nevermind I found it in one of the links you provided.

[nrg_alpha]

Sorry to hijack the thread slightly, but it looks like nrg_alpha has solved it anyway, what does the \K modifer do..

 

http://www.phpfreaks.com/blog/pcre-regex-spotlight-k

[cags]

Excellent, thanks for the link. I really should get around to checking out the articles here, I've only been around a few days and only focused on the forum.

[nrg_alpha]

No problem..take your time.. you're doing great

[jrw4]

Thanks for asking that as I had the same question on \K.

 

Now if I wanted to match something like:

 

<content[ANYTHING]>

 

My pattern would be:

 

$pattern = "/<content\K[\w]*>/";

[nrg_alpha]

Thanks for asking that as I had the same question on \K.

 

Now if I wanted to match something like:

 

<content[ANYTHING]>

 

My pattern would be:

 

$pattern = "/<content\K[\w]*>/";

 

Keep in mind that in that case, $0 (or if using preg_match, index[ 0 ] - either of which is the value that the entire pattern matched / captured is stored as) would be 'ANYTHING>' (I'm assuming that the [] brackets surrounding ANYTHING in the source string isn't there.. just displayed to surround ANYTHING for illustrative purposes...) Note that the > is included.. so chances are this is not what you would want.

 

In this case, you have a few options.. you can either put the > into a lookahead assertion like so:

$pattern = "/<content\s?\K\w*(?=>)/";

Since assertions don't consume any text, the > part is not included with the base match...

 

Or, depending on the string's circumstances (like let's assume that after '<content ' and the sequence of \w characters, it closes off with >, you might be able to even outright omit > completely:

$pattern = "/<content\s?\K\w*/";

 

This way, in either sample, the > character is not included in the base variable $0 (or if using preg_match, index[ 0 ]), which cleans things up a bit. Also note that I didn't use [\w], as \w is already a character class short hand in and of itself, which for all intents and purposes (without delving into the topic of locales, is understood as saying [a-zA-Z0-9_].. so if something like \w, or \d etc.. is the only thing being placed inside a character class, the character class is useless. And finally, if stuff follows <content, you probably don't want to include the initial space in your match, so I threw in the \s? just in case...

 

However, I don't think I would even use \w.. perhaps instead of that, I would use a character class to grab everything up to > like so:

 

$pattern = "/<content\s?\K[^>]*/";

[jrw4]

Well I was trying along the lines of:

 

$input = preg_replace("/<content\K[^\>]+/", "", $input);

 

Which just takes <content[ANYTHING]> and turns it into <content> which is alright but then I have to do a second line of code to remove that too:

 

$input = str_replace("<content>", "", $input);

 

I meant to ask how to do that in one line?

[nrg_alpha]

If I understand correctly, you are wiping out [ANYTHING] from <content[ANYTHING]> if applicable, but then you want to wipe out <content> itself?

I'll provide three samples which remove various levels of the <content> tags:

 

example:

$input = <<<EOF
Some text. <content class="whatever">Some content</content> Some more text yet again! <content>And yes, some more content!</content>
EOF;

# exmaple1: remove <content[ANYTHING]> only!
$input1 = preg_replace('#<content[^>]*>#i', '', $input);
echo $input1 . "<br />\n"; // Output: Some text. Some content</content> Some more text yet again! And yes, some more content!</content>

# example2: remove complete content tags
$input2 = preg_replace('#<content[^>]*>.*?</content>#is', '', $input);
echo $input2 . "<br />\n"; // Output: Some text.  Some more text yet again!

# example3: remove only the content tags (yet leave the text inside those tags in place)
$input3 = preg_replace('#<content[^>]*>(.*?)</content>#is', '$1', $input);
echo $input3 . "<br />\n"; // Output: Some text. Some content Some more text yet again! And yes, some more content!

[thebadbad]

Lol, I was about to post this, but then you beat me to it, haha:

 

Then you don't need \K at all, but just

 

$input = preg_replace('~<content[^>]*>~i', '', $input);

And my advance apologies goes to nrg, who is probably in the process of writing an elaborate answer (no offence - you do a great job explaining things in detail, while my answers just often aren't that long).

[nrg_alpha]

lol no harm, no foul