Learn about our RFC process, Open RFC meetings & more.Join in the discussion! »


1.0.6 • Public • Published

Yet Another Diacritic Remover

But better, I think, read on...

Coveralls Status Build Status

The Problem

This is a non-broken diacritic remover. There are number of similar modules available but they're all based on the same code and are all deficient in some cases, which is why I wrote this one. This implementation is also much faster than most of the diacritic removers I've tested and ever-so-slightly faster than the fastest, which may be important to you, if you're grinding through a lot of strings. This is based on the same code as all the rest, original is here: http://web.archive.org/web/20121231230126/http://lehelk.com:80/2011/05/06/script-to-remove-diacritics/

As far as I can tell, the version referenced above and all of the derived examples I have found, suffer from not being able to strip all diacritics. Specifically, they fail when encountering words like "Rügen" it fails to remove the diaresis forming the umlaut for the letter "u". This because of the nature of the algorithm which prevents it from detecting and removing combining characters, in this case the Combining Diacritical Marks (0300–036F) unicode block. The letter in question in our example is actually a completely normal letter "u" which therefore does not get stripped (it doesn't need to be) followed by a combining diaresis which modifies the preceeding letter. The diaresis does not register as a letter and will be ignore by the standard algorithm.

To illustrate with our example word above:

// Word: Rügen
// Raw hex, before processing by V8: 0x52 0x75 0xCC 0x88 0x67 0x65 0x6E

Letter: R, code point: 0x52,   char code: 0x52
Letter: u, code point: 0x75,   char code: 0x75
Letter: ̈, code point:  0x0308, char code: 0x0308
Letter: g, code point: 0x67,   char code: 0x67
Letter: e, code point: 0x65,   char code: 0x65
Letter: n, code point: 0x6e,   char code: 0x6e

Notice how the the five letters in the word are unaccented. Also, notice how the diaresis appears on top of the comma. That's how a combining character works. This is why the normal strip routine fails, all it sees are five normal letters and one "letter" that isn't recongized as such.

I have modified the code below to correctly strip out combining characters. The combining characters are

  • Combining Diacritical Marks (0300–036F), since version 1.0, with modifications in subsequent versions down to 4.1
  • Combining Diacritical Marks Extended (1AB0–1AFF), version 7.0
  • Combining Diacritical Marks Supplement (1DC0–1DFF), versions 4.1 to 5.2
  • Combining Diacritical Marks for Symbols (20D0–20FF), since version 1.0, with modifications in subsequent versions down to 5.1
  • Combining Half Marks (FE20–FE2F), versions 1.0, with modifications in subsequent versions down to 8.0

Other Options

You can perform unicode normalization and avoid doing the combining character strip but that would actually involve more work. You can also check your string to see if it comprises only precomposed characters and can therefore benefit from using a simplified strip function. Of course, that test has a computational cost that probably makes that process not worthwhile. Note that you'll find several modules on npm that claim to perform normalizing of unicode strings. They are using the term in a misleading manner. All the ones I've seen do the same as this module: strip diacritics. A proper normalization function would take the combining character and the character it modifies and replace them with the precomposed version.


     strip = require( 'yads' ),
     testStr = `Rügen caractères spéciaux contrairement à la langue française`;
console.log( `${testStr} =>\n${strip.precomposed( testStr )}` );
// Rügen caractères spéciaux contrairement à la langue française =>
// Rügen caracteres speciaux contrairement a la langue francaise
console.log( `${testStr} =>\n${strip.combining( testStr )}` );
// Rügen caractères spéciaux contrairement à la langue française =>
// Rugen caracteres speciaux contrairement a la langue francaise

Note that precomposed() and combining() switches default functions:

console.log( `${testStr} =>\n${strip.remove_diacritics( testStr )}` );
// Rügen caractères spéciaux contrairement à la langue française =>
// Rugen caracteres speciaux contrairement a la langue francaise
console.log( `${testStr} =>\n${strip.remove_diacritics( testStr )}` );
// Rügen caractères spéciaux contrairement à la langue française =>
// Rügen caracteres speciaux contrairement a la langue francaise

In addition, some utility functions are included that might be useful for various searches, especially typeaheads.

const simpleStr = `42 caractères spéciaux!`;
console.log( `keep letters only: ${strip.letters_only( simpleStr )}` );
// caracteres speciaux
console.log( `keep letters and numbers only: ${strip.alphanum_only( simpleStr )}` );
// 42 caracteres speciaux
console.log( `packed letters only: ${strip.packed( simpleStr )}` );
// caracteresspeciaux
console.log( `packed letters and numbers only: ${strip.packed_alphanum( simpleStr )}` );
// 42caracteresspeciaux


npm i yads

DownloadsWeekly Downloads






Last publish


  • avatar