parse_url() returns false on colon(:) followed by number inside path

[RV7PR]

Description

The following code:

<?php
print_r(parse_url('https://test.com/a/b/c:354/'));
print_r(parse_url('test.com/a/b/c:354/'));
print_r(parse_url('/a/b/c:354/'));

Resulted in this output:

Array
(
    [scheme] => https
    [host] => test.com
    [path] => /a/b/c:354/
)
Array
(
    [host] => test.com
    [port] => 354
    [path] => /a/b/c:354/
)

But I expected this output instead:

Array
(
    [scheme] => https
    [host] => test.com
    [path] => /a/b/c:354/
)
Array
(
    [host] => test.com
    [path] => /a/b/c:354/
)
Array
(
    [path] => /a/b/c:354/
)

PHP Version

PHP 8.4.4

Operating System

No response

[cmb69]

Yeah: https://3v4l.org/hLssX

This is a know problem; IIRC, there is a shortcut in the "parser" to find the port, and that can screw up proper handling of relative URLs with colon-portlike in the path.

That is basically documented:

Partial and invalid URLs are also accepted, parse_url() tries its best to parse them correctly.

"Its best" is just not good enough here.

I don't think it makes sense to try to fix that (might break too many other edge cases), and since there is an RFC under discussion which would introduce proper URL/URI parsers anyway, it might make even less sense to fiddle with parse_url().

[iluuu1994]

This is pretty much an exact duplicate of #7890, so should probably be closed.