1
|
var deburrLetter = require('./_deburrLetter'),
|
2
|
toString = require('./toString');
|
3
|
|
4
|
/** Used to match Latin Unicode letters (excluding mathematical operators). */
|
5
|
var reLatin = /[\xc0-\xd6\xd8-\xf6\xf8-\xff\u0100-\u017f]/g;
|
6
|
|
7
|
/** Used to compose unicode character classes. */
|
8
|
var rsComboMarksRange = '\\u0300-\\u036f',
|
9
|
reComboHalfMarksRange = '\\ufe20-\\ufe2f',
|
10
|
rsComboSymbolsRange = '\\u20d0-\\u20ff',
|
11
|
rsComboRange = rsComboMarksRange + reComboHalfMarksRange + rsComboSymbolsRange;
|
12
|
|
13
|
/** Used to compose unicode capture groups. */
|
14
|
var rsCombo = '[' + rsComboRange + ']';
|
15
|
|
16
|
/**
|
17
|
* Used to match [combining diacritical marks](https://en.wikipedia.org/wiki/Combining_Diacritical_Marks) and
|
18
|
* [combining diacritical marks for symbols](https://en.wikipedia.org/wiki/Combining_Diacritical_Marks_for_Symbols).
|
19
|
*/
|
20
|
var reComboMark = RegExp(rsCombo, 'g');
|
21
|
|
22
|
/**
|
23
|
* Deburrs `string` by converting
|
24
|
* [Latin-1 Supplement](https://en.wikipedia.org/wiki/Latin-1_Supplement_(Unicode_block)#Character_table)
|
25
|
* and [Latin Extended-A](https://en.wikipedia.org/wiki/Latin_Extended-A)
|
26
|
* letters to basic Latin letters and removing
|
27
|
* [combining diacritical marks](https://en.wikipedia.org/wiki/Combining_Diacritical_Marks).
|
28
|
*
|
29
|
* @static
|
30
|
* @memberOf _
|
31
|
* @since 3.0.0
|
32
|
* @category String
|
33
|
* @param {string} [string=''] The string to deburr.
|
34
|
* @returns {string} Returns the deburred string.
|
35
|
* @example
|
36
|
*
|
37
|
* _.deburr('déjà vu');
|
38
|
* // => 'deja vu'
|
39
|
*/
|
40
|
function deburr(string) {
|
41
|
string = toString(string);
|
42
|
return string && string.replace(reLatin, deburrLetter).replace(reComboMark, '');
|
43
|
}
|
44
|
|
45
|
module.exports = deburr;
|