PRegEx, XML and you (me too)

[Fordi]

Ok, I have the following regex:
/<([^{\s])\s?(?🙁[=]+)=\"([\"])\"\s?)*\/>/}

Its purpose: to get all data from a sinlge XML tag that's expected to be in the following format:
<xmlTag xmlAtt1="xmlData1" xmlAtt2="xmlData2" ... />

My expected return in $regs (from a preg_match) is:
Array {
}
(I'd be able to determine the number of elements via (sizeof($regs)-2)/2, and get each token accordingly)

My problem is that it's only matching the last attibute (note the '*' after the body of the regex should be catching all occurences of foo="bar"), thus $regs is ALWAYS:

Array {
}

I've tried swapping greedy and ungreedy. Multiline is not an issue, as \r and \n are never to be part of the input stream. No literals, so case insensitive is not necessary. No matter what I try to do to it, it doesn't work right.

No, using the XML php module is not an option for me.

Any help on this would be greatly appreciated.

[ednark]

i think you are stuck with that... i don't know a way around it.. althought im by far not the foremost authority on regular expressions... your problem is also illustrated with this more simple regex....

maybe seeing a simpler example will help someone smarter reply....

<?php
$regex= '/(.)*/';
$xml = 'abcdefghijklmnop';

preg_match( $regex, $xml, $matches );

print_r($matches);
?>

the only think i can think of is trying a [man]preg_split[/man] on

/"\s.+="/

or something similar to get the data into a nice array form...

[Weedpacket]

There's preg_match_all, though. I'd go in two steps; a preg_match() to get the name of the tag, and then a preg_match_all() to get all the name=value pairs. uh...

([^\s]*)\s*=\s*"([^"]*)"

off the top of my head...

[ednark]

see... someone smarter

[marathoner]

As I understand it ereg can't do nongreedy matches so I must use preg_match.

I have a huge string with line breaks in it and I don't want to take them out before processing.

If I do the following code snippet:

$str="{tag}stuff1{/tag}{tag}stuff2{/tag}{tag}stuff3{/tag}";
$match = array("","","");
$a = preg_match("/{tag}(.*?){\/tag}/", $str, $match);

$match[0] contains what I want stuff1 (nongreedy) flanked by the tags, and $match[1] also contains what I want, which is stuff1 with the tags stripped off.

However if I change the string to

$str="{tag}stu
ff1{/tag}{tag}stuff2{/tag}{tag}stuff3{/tag}";

it fails to find stu\nff1 and instead finds stuff2. This is not what I want. I've tried PCRE_MULTILINE
and have been all over the web. How can I make this work?

Using php 4.3.0

[ednark]

try this... 4.3.0 with winXP apache 2.latest works fine with just the PCRE_DOTALL (?s) thingy in there

run this to see what i tested

<?php

header('Content-type: text/plain');

$strings = array(
    "{tag}stuff1{/tag}{tag}stuff2{/tag}{tag}stuff3{/tag}",
    "{tag}stu
ff1{/tag}{tag}stuff2{/tag}{tag}stuff3{/tag}"
);

$pregs = array(
    '/\{tag\}(.*?)\{\/tag\}/',
    '/(?s)\{tag\}(.*?)\{\/tag\}/',
    '/\{tag\}((.|\n)*?)\{\/tag\}/',
);

foreach ( $strings as $string ) {
    foreach ( $pregs as $preg ) {
        print "---------------------------------\n";
        print "PREG:\n    {$preg} \n\nTEST STRING:\n    \"{$string}\"\n\nResults:";
        preg_match_all( $preg, $string, $matches);
        print_r($matches[1]);
    }
}

?>

[marathoner]

thank you very much. I tried it in Linux. All the patterns work on the first string as expected. The second and third patterns also do what I want on the "broken" string. Now I have to go figure out what ?s and .| means. 🙂

[ednark]

everything you want to look up is here

http://www.php.net/manual/en/pcre.pattern.syntax.php

these are eqivalent:

edn(a|e|i|o|u)k
edn[aeiou]k

both are 'OR' statements

with [] you have to specify single characters
with ( | ) each section can be a whole regex in itself

so (.|ⁿ⁾ says, match anything, or a newline

be careful, this often seen expression is however not a legal perl compatible expression, but is something more sleezy

(|)

the (?s) is a tricky inline way of setting the PCRE_DOTALL