How do I return the substring itself instead of its length?

Remember where the best window starts. In the dictionary version, replace the max line with if right - left + 1 > best: start, best = left, right - left + 1 and return s[start:start + best]. That returns 'etcar' for "streetcar" and ' world' for "hello world", leading space included. With > instead of >= you get the first of several equally long answers, so "noon" gives 'no'.

Does it work with Unicode, spaces and punctuation?

Yes. A Python str is a sequence of Unicode code points, and any of them can be a dictionary key, so "café☕é" gives 'café☕', five characters long. One catch: a letter built from a base plus a combining accent, or an emoji with a skin-tone modifier, is several code points, so it counts as several characters.

What about the longest substring with at most k distinct characters?

Same window, different rule. Keep a dictionary of counts for the characters in the window, add s[right], and while len(counts) > k decrease the count of s[left], delete it when it reaches zero, and move left right. For "mississippi" with k = 2 the answer is 7, from "ississi". The variant where each character may appear up to k times works the same way with a check on the count of the new character.

What does it return for an empty string?

All three versions return 0, because the loop body never runs and best keeps its starting value. No special case is needed.

Why does the dictionary version check last_seen[c] >= left?

Because the dictionary never forgets old positions. A character may have been seen before but sit to the left of the current window, and then it is not a repeat. Without the check, "noon" returns 3 instead of 2, because the left edge jumps backwards to a stale index. left = max(left, last_seen[c] + 1) is an equivalent way to write it.

Longest Substring Without Repeating Characters (Python)

length_of_longest_substring.py

Python 3.14

Not run yet

+ 8 hidden

Input

s = "streetcar"

Expected

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