How it works

The text is decomposed to Unicode NFD, combining marks (U+0300 to U+036F) are removed, and letters that Unicode does not decompose (ß ø ł æ œ đ þ ð ı ħ ŧ ŋ) are replaced from a table. The result is recomposed to NFC.

Notes on stripping accents

Accents come off with Unicode normalisation rather than with a find and replace list. NFD splits a letter such as é into a plain e followed by a combining acute accent, every combining mark is deleted, and what is left is recomposed as NFC. The same three steps work for every language, including Vietnamese letters that carry two marks at once.

Some letters are not accented letters at all. The stroke in ø, the bar in ł, the ligature in æ and the shape of ß are part of the letter itself, so Unicode has nothing to decompose. Those come from a small table: ß becomes ss, ø becomes o, ł becomes l, æ becomes ae, þ becomes th. Switch the option off when you want those letters left alone, for example while cleaning up a name.

Slug mode is for URLs, filenames and search keys. It lowercases the stripped text, turns every run of spaces and punctuation into a single hyphen and trims both ends, so Crème Brûlée becomes creme-brulee. Stripping does change meaning in some languages: Turkish ı and i are separate letters, and Polish ł, ó and ż are separate letters from l, o and z. Keep the stripped text for machine readable keys and keep the original for anything a person will read.