Question

Generally I’d match HTML attributes with this regex

\w+=".*?"

but when the HTML contains PHP code it gets kind of dicy. Please consider the following tag:

<option value="<?php echo $img; ?>"<?php echo ($hpb[$i]['image_filename']==$img?' selected="selected"':''); ?>>
    <?php echo $img; ?>
</option>

the above regex will match the attribute selected="selected" which is determined inside PHP logic. Is there a way to match attributes which are not inside PHP tags while still matching the ones whose value may contain PHP logic? If not could I just remove the PHP code which isn’t part of an attribute value?

EDIT: Here’s what I have so far:

 \w+="(((.(?!<\?php))*?)|((.((?=<\?php).*?(?=\?>))*)*?))*"

Which basically means match a string which starts with a SPACE then greedily match alphanumeric characters followed by EQUALS sign followed by double quote and then match any of the following two while capturing as many characters as possible:

1. A sequence of characters which does not contain the string <?php
2. A sequence of characters containing the pattern <\?php.*?\?> or in other words greedily match the value part of the attribute with all of its PHP code
   All of that till a closing double quote is encountered…

Answer 1

/<\?php[\s\S]*?\?>|\s+(\w+)="([^"<]*(?:<\?php[\s\S]*?\?>[^<"]*)*)"/

This will match either a PHP code segment or a complete attribute="value" sequence in which the value may contain PHP code. After each match you can find out what you caught by checking the contents of the capturing groups. If it’s a pure PHP segment you matched, all but group[0] will be empty; otherwise, group[1] will contain the attribute name and group[2] will contain the value.

The regex assumes < will appear inside an attribute value only as the beginning of a <?php tag. Of course that’s not a syntactically valid assumption, but it’s probably safe anyway. I can make the regex more precise if you need me to, but it will be also be much less readable.