Binary cast don't works as expected

[alexander-akait]

Input:

$binary = (binary) $string;
$binary = b"binary string";

First parsed as cast node. Second as string node. I think it is invalid. Second should be cast node with option shortForm: true 👍

[ichiriac]

Hi @evilebottnawi,

It's not exactly the same thing. First is a cast, converting from a type to another, the second is a property of the text. You can get more details here : https://wiki.php.net/rfc/binary_string_deprecation

The b"text" is a deprecated construct, meant to handle unicode support on PHP 6.

If you encounter a string with a raw property starting with b, you probably should strip the b as it will certainly trigger an warning message. What I could do, is to expose the property unicode: true on the string node.

[alexander-akait]

/cc @ichiriac what about $a = (binary) $string, it is not deprecated and exists in documentation

[ichiriac]

Sorry that took so long to reply. Take a look at the documentation here :
http://php.net/manual/en/language.types.type-juggling.php

(binary) casting and b prefix forward support was added in PHP 5.2.1. Note that the (binary) cast is essential the same as (string), but it should not be relied upon.

Actually, if you cast as binary or as string it's strictly the same :

<?php 
$binary = (binary)$var;
$same = (string)$var;

I don't know for sure why this keyword has been introduced (maybe they planned to handle automatically the encoding into each string, so the cast was for disabling encoding and let strings manipulation work on binary contents) - but I'm 100% sure that everything was abandoned as the PHP tokenizer use exactly the same token on binary and string casts.

In order to have a clean code-base, you should avoid binary casting and treat them as string casting as it's more clear on how it will behave.

But as the parser may be closer to the source code, I've also introduced the binary cast mode + the unicode property on texts