The anagram check normalizes both strings (lowercase, sort characters, rejoin) and compares the results; whitespace counts as a character so 'a gentleman' is NOT an anagram of 'elegant man' until you strip spaces explicitly. The duplicate-character finder uses a Map to count occurrences in one pass and returns just the keys with counts above 1. The literal-vs-object distinction is rare but real: new String('hi') is a boxed object that survives typeof as 'object', so legacy code that uses new String needs the dual check (typeof === 'string' OR instanceof String). In modern code, never use new String; just use string literals.

String.prototype.match returns an array of matches when the regex has the /g flag, or null if there are no matches; the ?? [] guard turns the null case into a clean zero. \d+ matches one-or-more digits and naturally splits a string into integer runs; map through Number to get real numeric values. The extension check uses an anchored regex ($) so notes.txt.bak correctly returns false. For a more general extension utility, splitting on . and taking the last part covers compound extensions like tar.gz only by returning gz; if you need tar.gz as a unit, walk the parts manually.

padStart is the built-in tool for left-padding to a target length: s.slice(-keepLast).padStart(s.length, '#') says "start with the last keepLast chars, then prepend hashes until the result matches the original length". The manual variant builds the prefix with String.repeat and concatenates; both produce the same output. Alphabet ranges work because letters have consecutive char codes within the same case ('a'..'z' is 97..122, 'A'..'Z' is 65..90). The empty-array return for reversed ranges is intentional; flip the bounds with Math.min/Math.max if you want to be permissive.

The adjacent-duplicate remover walks the string keeping a single prev character; whenever the current character differs, it joins the output. This is O(n) and uses constant extra state (other than the output). Notice it removes only RUNS, so aabbccddeeffaa becomes abcdefa (the last aa collapses to a single a but the earlier a survives because they are not adjacent). The selective-word reverse splits on spaces, conditionally reverses each word that exceeds the length threshold, then rejoins. Use String.prototype.split(/\s+/) and a final join with a single space if you want to normalize multiple-space input; the simple ' ' split preserves whatever spacing was there.