[Summary view] [Print] [Text view]

   1  /**
   2   * @class jQuery.plugin.byteLength
   3   * @author Jan Paul Posma, 2011
   4   * @author Timo Tijhof, 2012
   5   * @author David Chan, 2013
   6   */
   7  
   8  /**
   9   * Calculate the byte length of a string (accounting for UTF-8).
  10   *
  11   * @static
  12   * @inheritable
  13   * @param {string} str
  14   * @return {string}
  15   */
  16  jQuery.byteLength = function ( str ) {
  17      // This basically figures out how many bytes a UTF-16 string (which is what js sees)
  18      // will take in UTF-8 by replacing a 2 byte character with 2 *'s, etc, and counting that.
  19      // Note, surrogate (\uD800-\uDFFF) characters are counted as 2 bytes, since there's two of them
  20      // and the actual character takes 4 bytes in UTF-8 (2*2=4). Might not work perfectly in
  21      // edge cases such as illegal sequences, but that should never happen.
  22  
  23      // https://en.wikipedia.org/wiki/UTF-8#Description
  24      // The mapping from UTF-16 code units to UTF-8 bytes is as follows:
  25      // > Range 0000-007F: codepoints that become 1 byte of UTF-8
  26      // > Range 0080-07FF: codepoints that become 2 bytes of UTF-8
  27      // > Range 0800-D7FF: codepoints that become 3 bytes of UTF-8
  28      // > Range D800-DFFF: Surrogates (each pair becomes 4 bytes of UTF-8)
  29      // > Range E000-FFFF: codepoints that become 3 bytes of UTF-8 (continued)
  30  
  31      return str
  32          .replace( /[\u0080-\u07FF\uD800-\uDFFF]/g, '**' )
  33          .replace( /[\u0800-\uD7FF\uE000-\uFFFF]/g, '***' )
  34          .length;
  35  };
  36  
  37  /**
  38   * @class jQuery
  39   * @mixins jQuery.plugin.byteLength
  40   */