markdown-it/lib/rules_core/replacements.js

// Simple typographic replacements
//
// (c) (C) → ©
// (tm) (TM) → ™
// (r) (R) → ®
// +- → ±
// (p) (P) -> §
// ... → … (also ?.... → ?.., !.... → !..)
// ???????? → ???, !!!!! → !!!, `,,` → `,`
// -- → –, --- → —
//
'use strict';

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

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

var DASHES_RE = /--/;

// Workaround for phantomjs - need regex without /g flag,
// or root check will fail every second time
var SCOPED_ABBR_TEST_RE = /\((c|tm|r|p)\)/i;

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

function replaceFn(match, name) {
  return SCOPED_ABBR[name.toLowerCase()];
}

function replace_scoped(inlineTokens) {
  var i, token, inside_autolink = 0;

  for (i = inlineTokens.length - 1; i >= 0; i--) {
    token = inlineTokens[i];

    if (token.type === 'text' && !inside_autolink) {
      token.content = token.content.replace(SCOPED_ABBR_RE, replaceFn);
    }

    if (token.type === 'link_open' && token.info === 'auto') {
      inside_autolink--;
    }

    if (token.type === 'link_close' && token.info === 'auto') {
      inside_autolink++;
    }
  }
}

function replace_helper(inlineTokens, re, replacer) {
  var i, token, inside_autolink = 0;

  for (i = inlineTokens.length - 1; i >= 0; i--) {
    token = inlineTokens[i];

    if (token.type === 'text' && !inside_autolink) {
      if (re.test(token.content)) {
        token.content = replacer(token.content);
      }
    }

    if (token.type === 'link_open' && token.info === 'auto') {
      inside_autolink--;
    }

    if (token.type === 'link_close' && token.info === 'auto') {
      inside_autolink++;
    }
  }
}

function replace_rare(inlineTokens) {
  replace_helper(inlineTokens, RARE_RE, function (content) {
    return content
      .replace(/\+-/g, '±')
      // .., ..., ....... -> …
      // but ?..... & !..... -> ?.. & !..
      .replace(/\.{2,}/g, '…').replace(/([?!])…/g, '$1..')
      .replace(/([?!]){4,}/g, '$1$1$1').replace(/,{2,}/g, ',');
  });
}

function replace_dashes(inlineTokens) {
  replace_helper(inlineTokens, DASHES_RE, function (content) {
    return content
      // em-dash
      .replace(/(^|[^-])---(?=[^-]|$)/mg, '$1\u2014')
      // en-dash
      .replace(/(^|\s)--(?=\s|$)/mg, '$1\u2013')
      .replace(/(^|[^-\s])--(?=[^-\s]|$)/mg, '$1\u2013');
  });
}

module.exports = function replace(state) {
  var blkIdx;
  var typographerOption = state.md.options.typographer;

  if (!typographerOption) { return; }

  var optionIsObject = typeof typographerOption === 'object';
  var replaceRare = !optionIsObject || typographerOption.rare;
  var replaceIP = !optionIsObject || typographerOption.intellectualProperty;
  var replaceDashes = !optionIsObject || typographerOption.dashes;

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

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

    if (replaceIP && SCOPED_ABBR_TEST_RE.test(state.tokens[blkIdx].content)) {
      replace_scoped(state.tokens[blkIdx].children);
    }

    if (replaceDashes && DASHES_RE.test(state.tokens[blkIdx].content)) {
      replace_dashes(state.tokens[blkIdx].children);
    }

    if (replaceRare && RARE_RE.test(state.tokens[blkIdx].content)) {
      replace_rare(state.tokens[blkIdx].children);
    }

  }
};