is_readable() not working on files inside phar

[petitj]

Description

Using the following code to create a basic phar file:

$stub = <<<STUB
<?php
    Phar::interceptFileFuncs();
    set_include_path('phar://' . __FILE__ . PATH_SEPARATOR . get_include_path());
    include "index.php";   
    
    __HALT_COMPILER();
STUB;

$index = <<<INDEX
<?php
    var_dump(file_exists('phar:///path/to/archive.phar/index.php')) . PHP_EOL;
    var_dump(is_readable('phar:///path/to/archive.phar/index.php')) . PHP_EOL;
    var_dump(file_exists('index.php')) . PHP_EOL;
    var_dump(is_readable('index.php')) . PHP_EOL;
INDEX;

$phar = new Phar('archive.phar');
$phar->addFromString('index.php', $index);
$phar->setStub($stub);

Then executing the following code:

php archive.phar

Resulted in this output:

bool(true)
bool(false)
bool(true)
bool(false)

But I expected this output instead:

bool(true)
bool(true)
bool(true)
bool(true)

Or executing the following code:

<?php
var_dump(is_readable('phar:///path/to/archive.phar/index.php'));

Resulted in this output:

bool(false)

But I expected this output instead:

bool(true)

PHP Version

PHP 8.0.12

Operating System

Ubuntu 20.04

[cmb69]

You need to use a phar wrapper:

var_dump(is_readable('phar://path/to/archive.phar/index.php'));

[petitj]

You need to use a phar wrapper:

var_dump(is_readable('phar://path/to/archive.phar/index.php'));

Sorry, that's what i was actually using. Error in the bug report that i just fixed. The bug is still there. Please look at the more detailed example and re-open it :)

[cmb69]

is_readable() with the proper phar URL returns false for me. This is on Windows, though. However, file_exists('index.php') should not return true for you, since there is no index.php involved (only within the phar).

[petitj]

is_readable() with the proper phar URL should return true. Not false. So it seems it is also a problem on windows.
file_exists('index.php') is executed in the context of the phar so true is the expected result.

[cmb69]

Sorry, my bad; it actually returned true for me, and I didn't really look at the modified code. Anyhow, I now did, and I get:

bool(true)
bool(true)
bool(true)
bool(true)

[petitj]

So this means it is an OS related bug?