URL Slug Generator - Clean Slugs from Any Title, One or Many

The title is split into words. Letters and digits stay. Spaces, punctuation, symbols and emoji end a word. Apostrophes are dropped without splitting, so 'Don't' becomes 'dont' and not 'don-t'. The words are lowercased and joined with a hyphen (or an underscore if you choose). Several separators in a row become one, and none is left at either end. Line breaks in a single title count as spaces.

Accents are removed with Unicode decomposition (NFKD): 'é' becomes 'e' plus a combining accent, and the accent is deleted. Some letters have no decomposition, so the tool replaces them from a fixed table, shown below: ß becomes ss, æ ae, œ oe, ø o, đ d, ð d, þ th, ł l, ħ h, ŧ t, ı i, ŋ ng and ĸ k. Compatibility characters are folded as well: a single-character ligature for 'fi' becomes the two letters, full-width letters become plain ones and a superscript 2 becomes 2. German spelling is a separate option that writes ä, ö and ü as ae, oe and ue instead of a, o and u (Müller becomes mueller).

Stop-word removal drops 31 common English words: a, an, and, are, as, at, be, but, by, for, if, in, into, is, it, of, on, or, such, that, the, their, then, there, these, they, this, to, was, will and with. The list is Apache Lucene's default English list without 'no' and 'not', because removing those can reverse the meaning of a title. A title made only of stop words keeps them. A length limit cuts the slug at a whole word, so the last word is never chopped, unless the first word alone is longer than the limit. Google recommends hyphens over underscores to separate words in a URL, and says URLs are case sensitive, so /APPLE and /apple are different addresses. One lowercase form avoids duplicates.

There is no transliteration. Sinhala, Tamil, Chinese, Arabic, Cyrillic and other scripts have no letter-for-letter English spelling here, so by default their letters are removed. 'Keep as typed' leaves them in the slug. A URL can carry such letters, but on the wire they travel as UTF-8 bytes written in percent-encoding (RFC 3986): the Sinhala letter ශ (U+0DC1) is the three bytes E0 B7 81, written %E0%B7%81, nine characters for one letter. Browsers usually show the readable form, yet email, chat and spreadsheets often show the long one, and Google says characters outside ASCII should be percent-encoded. 'Keep, percent-encoded' gives that form directly. Many sites use a short English slug that a person writes and keep Sinhala or Tamil in the page title.