querySelector and attribute's value

[dmitriy-krista]

When specifying the value for an attribute in querySelector() method like tag[attr='value'] , the method doesn't work as expected and returns null.

Example to reproduce the issue:

<?php

require "vendor/autoload.php";

$html = file_get_contents("https://github.com");
$document = new \Gt\Dom\HTMLDocument($html);

//will print "Enterprise"
echo $document->querySelector("nav > ul > li > a[data-ga-click]")->innerText . "\n";

//will throw PHP Notice: Trying to get property 'innerText' of non-object
echo $document->querySelector("nav > ul > li > a[data-ga-click='(Logged out) Header, go to Enterprise']")->innerText . "\n";

The expected behaviour is that the last line should print "Enterprise"

If I try to run the exact querySelector call in my browser, it works correctly and both lines print "Enterprise":

document.querySelector("nav > ul > li > a[data-ga-click]").innerText

document.querySelector("nav > ul > li > a[data-ga-click='(Logged out) Header, go to Enterprise']").innerText

[g105b]

Thanks for reporting this. I'll look into it shortly.

[g105b]

Hi @dmitriy-krista, it looks like the problem is with the use of the comma character, which isn't taking into account the quotes surrounding it. I've tested a simple attribute selector which works fine, so it looks like the way the attribute selector is parsed in PHP.Gt/CssXPath is to blame.

I'll transfer this issue there and find a fix soon.

[g105b]

I have isolated the bug to the naive use of explode on commas to separate queries:

CssXPath/src/Translator.php

Line 32 in 57e9a9e

$cssParts = explode(",", $this->cssSelector);

[g105b]

Here is a test case I'm working on before implementing: https://regex101.com/r/xB7rQ7/224

Notice the last example doesn't work. If anyone could help me understand backreferences in this example, we can squish the bug and close this issue.

[g105b]

Works a treat now. I'll release this to the next version which will be adopted by the next version of Dom.

Have fun!