PHP DOMXPath works with double quotes fails with single quotes

Question

I wrote a little script that extracts information from a web site using PHP's DOMXPath class.
I query for <div class="sku" /> and execute a substring-before on the result. The result contains text, non breaking spaces, a line break and more text.
So what I'm trying to do is cut before the   \r\n. It works fine when I use the following query:

$query = "substring-before(//div[@class='sku'],'\xC2\xA0\xC2\xA0\r\n')";

but fails as soon as I change the quotes (which shouldn't make any difference):

$query = 'substring-before(//div[@class="sku"],"\xC2\xA0\xC2\xA0\r\n")';

or

$query = 'substring-before(//div[@class=\'sku\'],\'\xC2\xA0\xC2\xA0\r\n\')';

How is this possible and how can I overcome this?

Live example here: http://codepad.viper-7.com/R1rCaj

Answer 1

The style of quotes makes a difference because when a string is enclosed in double-quotes PHP will interpret more escape sequences for special characters - including what you're using for non-breaking space \xC2\xA0, carriage return \r, and newline \n.

When you have these enclosed in single-quotes '\xC2\xA0\r\n', like in your second two queries, PHP treats them as those literal characters - backslash, x, C, 2... etc.

---

A little extra syntax highlighting may help show this off, escape sequences in orange:

---

If your string already has what would be escape sequences in it as literal characters, and there's no way to get that corrected^*, you're in the kinda dirty position of replacing them yourself.

This preg_replace_callback() will take care of the sort of sequences in your example, and it's trivial to extend to the rest of the escape sequences supported by double-quotes:

// Known good.
$query1 = "substring-before(//div[@class='sku'],'\xC2\xA0\xC2\xA0\r\n')";

// Known bad.
$query2 = 'substring-before(//div[@class=\'sku\'],\'\xC2\xA0\xC2\xA0\r\n\')';

$query2 = preg_replace_callback(
    '/\\\\(?:[rn]|(?:x[0-9A-Fa-f]{1,2}))/',
    function ($matches) {
        switch (substr($matches[0], 0, 2)) {
            case '\r':
                return "\r";
            case '\n':
                return "\n";
            case '\x':
                return hex2bin(substr($matches[0], 2));
            }
    },
    $query2
);

var_dump($query1 === $query2); // Now equal?

Output:

bool(true)

---

_{(*Really, you should get this fixed at the source.)}