Typographer: enchanced plain rules

benchmark/profile.js

@@ -13,7 +13,7 @@ var md = new Remarkable({
   langPrefix: 'language-'
 });
 
-var data = fs.readFileSync(path.join(__dirname, '/samples/cdata.md'), 'utf8');
+var data = fs.readFileSync(path.join(__dirname, '/samples/lorem1.txt'), 'utf8');
 
 for (var i = 0; i < 20000; i++) {
   md.render(data);

lib/defaults.js

@@ -9,5 +9,6 @@ module.exports = {
   breaks: false,
   maxLevel: 20,
   langPrefix: 'language-',
+  typograph: false,
   highlight: function (/*str*/) { return ''; }
 };

lib/defaults_typographer.js

@@ -0,0 +1,18 @@
+// Default typograph options
+
+'use strict';
+
+
+module.exports = {
+  singleQuotes: '‘’',
+  doubleQuotes: '“”', // «» - russian, „“ - deutch
+  copyright:    true,
+  trademark:    true,
+  registered:   true,
+  plusminus:    true,
+  paragraph:    true,
+  ellipsis:     true,
+  dupes:        true,
+  emDashes:     true,
+  linkify:      true
+};

lib/typographer.js

@@ -12,6 +12,8 @@
 // - fractionals 1/2, 1/4, 3/4 -> ½, ¼, ¾
 // - miltiplication 2 x 4 -> 2 × 4
 
+var assign   = require('./common/utils').assign;
+var defaults = require('./defaults_typographer');
 var Ruler    = require('./ruler');
 
 
@@ -19,28 +21,46 @@ var rules = [];
 
 
 rules.push(function single(t, state) {
-  var i, token, text, tokens = state.tokens,
+  var i, token, text,
-      copyright    = t.copyright,
+      tokens = state.tokens,
-      trademark    = t.trademark,
+      options = t.options;
-      registered   = t.registered,
+
-      plusminus    = t.plusminus;
+  for (i = tokens.length - 1; i >= 0; i--) {
-
-  for (i = state.tokens.length - 1; i >= 0; i--) {
-    if (tokens[i].type === 'text') {
     token = tokens[i];
+    if (token.type === 'text') {
       text = token.content;
 
-      if (copyright && text.indexOf('(') >= 0) {
+      if (text.indexOf('(') >= 0) {
+        if (options.copyright) {
           text = text.replace(/\(c\)/gi, '©');
         }
-      if (trademark && text.indexOf('(') >= 0) {
+        if (options.trademark) {
           text = text.replace(/\(tm\)/gi, '™');
         }
-      if (registered && text.indexOf('(') >= 0) {
+        if (options.registered) {
           text = text.replace(/\(r\)/gi, '®');
         }
-      if (plusminus && text.indexOf('+/-') >= 0) {
+        if (options.paragraph) {
-        text = text.replace(/\+\/\-/g, '±');
+          text = text.replace(/\(p\)/gi, '§');
+        }
+      }
+
+      if (options.plusminus && text.indexOf('+-') >= 0) {
+        text = text.replace(/\+-/g, '±');
+      }
+      if (options.ellipsis && text.indexOf('..') >= 0) {
+        // .., ..., ....... -> …
+        // but ?..... & !..... -> ?.. & !..
+        text = text.replace(/\.{2,}/g, '…').replace(/([?!])…/g, '$1..');
+      }
+      if (options.dupes &&
+          (text.indexOf('????') >= 0 ||
+           text.indexOf('!!!!') >= 0 ||
+           text.indexOf(',,') >= 0)) {
+        text = text.replace(/([?!]){4,}/g, '$1$1$1').replace(/,{2,}/g, ',');
+      }
+      if (options.emDashes && text.indexOf('--') >= 0) {
+        text = text.replace(/(^|\s)--(\s|$)/mg, '$1—$2');
       }
 
       token.content = text;
@@ -50,12 +70,7 @@ rules.push(function single(t, state) {
 
 
 function Typographer() {
-  this.singleQuotes = '‘’';
+  this.options  = assign({}, defaults);
-  this.doubleQuotes = '“”'; // «» - russian, „“ - deutch
-  this.copyright    = true;
-  this.trademark    = true;
-  this.registered   = true;
-  this.plusminus    = true;
 
   this.ruler = new Ruler(this.rulesUpdate.bind(this));
 
@@ -70,6 +85,11 @@ Typographer.prototype.rulesUpdate = function () {
 };
 
 
+Typographer.prototype.set = function (options) {
+  assign(this.options, options);
+};
+
+
 Typographer.prototype.process = function (state) {
   var i, l, rules;
 

test/fixtures/remarkable/typographer.txt

@@ -22,9 +22,47 @@ trademark
 .
 
 
+paragraph
+.
+(p) (P)
+.
+<p>§ §</p>
+.
+
+
 plus-minus
 .
-+/-5
++-5
 .
 <p>±5</p>
 .
+
+
+ellipsis
+.
+test.. test... test..... test?..... test!....
+.
+<p>test… test… test… test?.. test!..</p>
+.
+
+
+dupes
+.
+!!!!!! ???? ,,
+.
+<p>!!! ??? ,</p>
+.
+
+
+em-dashes
+.
+-- remarkable -- super --
+.
+<p>— remarkable — super —</p>
+.
+
+
+
+TODO:
+
+- "Monitor 21""