What does $s[$i] return in PHP?

A string of one byte, not a number and not a multibyte character. PHP strings are plain byte sequences and strlen counts bytes, so the loops above treat each byte as one character. That is exact for the ASCII inputs here; ord($s[$i]) gives the byte as an integer if you want one.

How do I handle UTF-8 input such as "café☕é"?

Split it into characters first: $chars = mb_str_split($s); returns an array of UTF-8 characters (PHP 7.4 and later), and the same window runs over $chars with count($chars). The byte version returns 8 for "café☕é" because é is two bytes and ☕ three; the mb_str_split version returns 5.

Why unset the key instead of setting it to false?

Because isset only returns false for a missing key or a null value. After $window[$c] = false; the key still exists, isset($window[$c]) is still true, and the while loop would never stop shrinking. unset removes the key, so the next isset sees it gone.

Why do digit characters turn into integer keys?

PHP converts a string key that looks like a decimal integer into an int, so $seen["7"] is stored under the key 7. That is harmless here: each character still maps to exactly one key, and isset($seen["7"]) finds it again. It only matters if you later read the keys back and expect strings.

Is a fixed array with ord() faster?

A little. $last = array_fill(0, 256, -1); indexed by ord($c) avoids hashing string keys, and the check becomes if ($last[$o] >= $left) $left = $last[$o] + 1;. For 5,000 characters the difference is a fraction of a millisecond, so the readable string-keyed array is fine.

Longest Substring Without Repeating Characters in PHP: Brute Force and Sliding Window Solutions

length_of_longest_substring.php

PHP 8.5

Not run yet

+ 8 hidden

Input

s = "streetcar"

Expected

The double e breaks the word up. After it, "etcar" has five different characters.