URL slug generator

How is a URL slug made from a title?

The title is split into words: runs of letters and digits, with spaces, punctuation and symbols such as & and ! treated as gaps. The words are lowercased and joined with hyphens, so there is never a hyphen at either end or two in a row. Apostrophes are dropped rather than turned into hyphens, so “Founder’s” becomes founders. Hyphens and underscores are both unreserved characters in RFC 3986, so neither needs encoding in a URL, but Google recommends hyphens.

Accented Latin letters become plain ones: the text is decomposed with Unicode NFKD, which splits é into e plus a combining accent, and the accents are removed. NFKD can't split a few letters that are separate letters in their own right, so a table maps them: ß → ss, æ → ae, ø → o, đ → d, ł → l, œ → oe, þ → th, ð → d and ı → i.

Words in other scripts, such as Hindi, Tamil or Arabic, have no plain-Latin spelling that a rule can produce, and their vowel signs are combining marks too, so stripping marks would break them. This tool keeps them by default: Google's URL guidance says to use words in your audience's language and to percent-encode non-ASCII characters in links, which the result shows. Choose ASCII only to remove them instead; the result then says which script was removed rather than dropping it silently.

Small words are removed only when you ask, from this list: a, an, the, and, or, but, nor, of, to, in, on, at, by, for, from, with, as, into, is, are, was, were, be, it, its. If every word is on the list, none are removed. A length cap drops whole words from the end until the slug fits.

title → NFKD, remove accents (é → e), map ß æ ø đ ł œ þ ð ı
      → lowercase → drop apostrophes
      → words = runs of letters, digits and marks
      → [remove small words] → [cut whole words to fit]
slug = words joined with "-"