Why index the array with (unsigned char)?

Plain char is signed on most compilers, so a byte above 127 is a negative number and seen[s[i]] would read before the array. Casting to unsigned char maps every byte to 0..255, which is why the arrays have 256 entries.

Why not memset the int table to -1?

memset writes bytes, and -1 happens to be all one bits, so memset(last, -1, sizeof last) does give -1 in every int. It is a trick that only works for 0 and -1; a plain loop states the intent and works for any start value.

Does this handle UTF-8?

No. A char* is bytes, and é is two bytes in UTF-8, so the code finds the longest run of distinct bytes. The tests are ASCII, where one byte is one character. For UTF-8 you would decode code points first and key a hash table by them, since a 256-slot array no longer covers the alphabet.

Not here: an empty input arrives as "", a pointer to a single '\0', and every version returns 0 for it. In your own code a if (!s) return 0; guard costs nothing.

Why loop until s[right] is the NUL byte instead of calling strlen?

A C string carries no length, so strlen walks it once to count. The window loop already visits every character in order, so stopping at '\0' saves that extra pass; calling strlen once before the loop is fine too. Calling it in the loop condition is not, because it rescans the string on every step.

Longest Substring Without Repeating Characters in C: Sliding Window with Arrays

length_of_longest_substring.c

C 15.3

Not run yet

+ 8 hidden

Input

s = "streetcar"

Expected

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