Why does PHP substr() change ASCII carriage return byte?

Question

I was going to use a long string to manipulate a large number of bit flags, keeping the result string in Redis. However, stumbled upon a php bug (?). A byte that contains bits 00001101 read using substr() returns an unexpected value:

$bin = 0b00001101;  // 13 - ASCII Carriage return
$c = substr($bin, 0, 1);    // read this character
printf("Expectation: 00001101, reality: %08b\n", $c); // 00000001

Ideone

Is the assumption that substr() is binary-safe wrong? Also tried mb_substr(), setting encoding to 8bit with exactly the same result.

Hygrostat

Answer 1

You're setting $bin to an integer 13

Using substr() against $bin is casting $bin to a string ("13")

You're reading the first character of that string ("1")

Using printf() with %b, you're explicitly casting that string back to an integer 1

the argument is treated as an integer, and presented as a binary number.

EDIT

This code should give the result that you're expecting

$bin = 0b00001101;  // 13 - ASCII Carriage return
$c = substr(chr($bin), 0, 1);    // read this character
printf("Expectation: 00001101, reality: %08b\n", ord($c)); // 00001101

Triplicate

Thanks! Probably I created a bad example/isolation of the problem. Originally I have a longer binary string, reading a particular byte from it. Will rewrite the example. – Hygrostat

True, now I get it. Thanks again! – Hygrostat