Inproved default 2-characters tlds handle in flex links, closes markd…

…own-it#12

index.js

@@ -87,9 +87,16 @@ var defaultSchemas = {
   }
 };
 
+/*eslint-disable max-len*/
+
+// RE pattern for 2-character tlds (autogenerated by ./support/tlds_2char_gen.js)
+var tlds_2ch_src_re = 'a[cdefgilmnoqrstuwxz]|b[abdefghijmnorstvwyz]|c[acdfghiklmnoruvwxyz]|d[ejkmoz]|e[cegrstu]|f[ijkmor]|g[abdefghilmnpqrstuwy]|h[kmnrtu]|i[delmnoqrst]|j[emop]|k[eghimnprwyz]|l[abcikrstuvy]|m[acdeghklmnopqrstuvwxyz]|n[acefgilopruz]|om|p[aefghklmnrstwy]|qa|r[eosuw]|s[abcdeghijklmnortuvxyz]|t[cdfghjklmnortvwz]|u[agksyz]|v[aceginu]|w[fs]|y[et]|z[amw]';
+
 // DON'T try to make PRs with changes. Extend TLDs with LinkifyIt.tlds() instead
 var tlds_default = 'biz|com|edu|gov|net|org|pro|web|xxx|aero|asia|coop|info|museum|name|shop|рф'.split('|');
 
+/*eslint-enable max-len*/
+
 ////////////////////////////////////////////////////////////////////////////////
 
 function resetScanCache(self) {
@@ -125,7 +132,7 @@ function compile(self) {
   var tlds = self.__tlds__.slice();
 
   if (!self.__tlds_replaced__) {
-    tlds.push('[a-z]{2}');
+    tlds.push(tlds_2ch_src_re);
   }
   tlds.push(re.src_xn);
 

package.json

@@ -31,6 +31,6 @@
     "mocha": "*",
     "ndoc": "^3.1.0",
     "stylus": "~0.50.0",
-    "tlds": "^1.26.0"
+    "tlds": "^1.37.0"
   }
 }

support/tlds_2char_gen.js

@@ -0,0 +1,68 @@
+#!/usr/bin/env node
+
+// Generates list of 2-char english tlds.
+//
+// Code is dirty, i know, but it's needed only once
+//
+'use strict';
+
+
+/*eslint-disable no-console*/
+
+function toRanges(str) {
+  var ranges = [], i;
+
+  str = str.slice(1, -1);
+
+  while (str.length) {
+    for (i = 1; ; i++) {
+      if (str[i] !== String.fromCharCode(str[i - 1].charCodeAt(0) + 1)) {
+        if (i < 3) {
+          ranges.push(str.slice(0, i));
+        } else {
+          ranges.push(str[0] + '-' + str[i - 1]);
+        }
+        str = str.slice(i);
+        break;
+      }
+    }
+  }
+  return '[' + ranges.join('') + ']';
+}
+
+var tlds = require('tlds')
+              .filter(function (name) {
+                return /^[a-z]{2}$/.test(name);
+              })
+              .sort();
+
+//
+// group by first letter
+//
+
+var result = [];
+
+'abcdefghijklmnopqrstuvwxyz'.split('').forEach(function (letter) {
+  var list = tlds.filter(function (name) { return name[0] === letter; });
+
+  if (!list.length) { return; }
+
+  if (list.length < 2) {
+    result = result.concat(list);
+    return;
+  }
+
+  result.push(letter + '[' + list.map(function (n) { return n[1]; }).join('') + ']');
+});
+
+result = result.join('|');
+
+console.log(result);
+
+//
+// Compact ranges
+//
+
+result = result.replace(/\[[a-z]+\]/g, toRanges);
+
+// console.log(result);

test/fixtures/not_links.txt

@@ -19,6 +19,8 @@ _example.com
 http://example.com_
 @example.com
 
+node.js and io.js
+
 http://
 http://.
 http://..