replacements.js

// Simple typographical replacements
//
'use strict';

// TODO:
// - fractionals 1/2, 1/4, 3/4 -> ½, ¼, ¾
// - miltiplication 2 x 4 -> 2 × 4

var RARE_RE = /\+-|\.\.|\?\?\?\?|!!!!|,,|--/;

var SCOPED_ABBR_RE = /\((c|tm|r|p)\)/ig;
var SCOPED_ABBR = {
  'c': '©',
  'r': '®',
  'p': '§',
  'tm': '™'
};

function replaceScopedAbbr(str) {
  if (str.indexOf('(') < 0) { return str; }

  return str.replace(SCOPED_ABBR_RE, function(match, name) {
    return SCOPED_ABBR[name.toLowerCase()];
  });
}


module.exports = function replace(state) {
  var i, token, text, inlineTokens, blkIdx;

  if (!state.options.typographer) { return; }

  for (blkIdx = state.tokens.length - 1; blkIdx >= 0; blkIdx--) {

    if (state.tokens[blkIdx].type !== 'inline') { continue; }

    inlineTokens = state.tokens[blkIdx].children;

    for (i = inlineTokens.length - 1; i >= 0; i--) {
      token = inlineTokens[i];
      if (token.type === 'text') {
        text = token.content;

        text = replaceScopedAbbr(text);

        if (RARE_RE.test(text)) {
          text = text
            .replace(/\+-/g, '±')
            // .., ..., ....... -> …
            // but ?..... & !..... -> ?.. & !..
            .replace(/\.{2,}/g, '…').replace(/([?!])…/g, '$1..')
            .replace(/([?!]){4,}/g, '$1$1$1').replace(/,{2,}/g, ',')
            // em-dash
            .replace(/(^|[^-])---([^-]|$)/mg, '$1\u2014$2')
            // en-dash
            .replace(/(^|\s)--(\s|$)/mg, '$1\u2013$2')
            .replace(/(^|[^-\s])--([^-\s]|$)/mg, '$1\u2013$2');
        }

        token.content = text;
      }
    }
  }
};