Add more conformant unicode handling

This update ensures emoji are correctly stripped (GH-4), and non-Latin
characters are not incorrectly lower-cased (GH-6), and adds a bunch of
(verified) test cases to ensure the future holds less broken builds.

Closes GH-4.
Closes GH-5.
Closes GH-8.

index.js

@@ -1,3 +1,5 @@
+var emoji = require('emoji-regex')
+
 module.exports = BananaSlug
 
 function BananaSlug () {
@@ -42,14 +44,19 @@ BananaSlug.prototype.reset = function () {
 
 var whitespace = /\s/g
 
+function lower (string) {
+  return string.toLowerCase()
+}
+
 function slugger (string) {
   var re = /[\u2000-\u206F\u2E00-\u2E7F\\'!"#$%&()*+,.\/:;<=>?@\[\]^`{|}~]/g
   var maintainCase = false
   var replacement = '-'
-  var result
 
   if (typeof string !== 'string') return ''
-  if (!maintainCase) string = string.toLowerCase()
-  result = string.trim().replace(re, '').replace(whitespace, replacement)
-  return result
+  if (!maintainCase) string = string.replace(/[A-Z]+/g, lower)
+  return string.trim()
+    .replace(re, '')
+    .replace(emoji(), '')
+    .replace(whitespace, replacement)
 }

package.json

@@ -3,9 +3,16 @@
   "description": "Generate a slug just like GitHub does for markdown headings.",
   "version": "1.1.0",
   "author": "Dan Flettre <[email protected]>",
+  "contributors": [
+    "Dan Flettre <[email protected]>",
+    "Titus Wormer <[email protected]> (http://wooorm.com)"
+  ],
   "bugs": {
     "url": "https://github.com/Flet/github-slugger/issues"
   },
+  "dependencies": {
+    "emoji-regex": "^6.0.0"
+  },
   "devDependencies": {
     "standard": "*",
     "tap-spec": "^4.0.2",

test/index.js

@@ -18,7 +18,7 @@ test('github test cases', function (t) {
   var slugger = new GithubSlugger()
 
   testCases.forEach(function (test) {
-    t.equals(test.slug, slugger.slug(test.text), test.mesg)
+    t.equals(slugger.slug(test.text), test.slug, test.mesg)
   })
   t.end()
 })
@@ -71,22 +71,114 @@ var testCases = [
   },
   {
     mesg: 'deals with duplicates correctly',
-    text: 'duplicate',
-    slug: 'duplicate'
+    text: 'duplicates',
+    slug: 'duplicates'
   },
   {
     mesg: 'deals with duplicates correctly-1',
-    text: 'duplicate',
-    slug: 'duplicate-1'
+    text: 'duplicates',
+    slug: 'duplicates-1'
   },
   {
     mesg: 'deals with duplicates correctly-2',
-    text: 'duplicate',
-    slug: 'duplicate-2'
+    text: 'duplicates',
+    slug: 'duplicates-2'
   },
   {
     mesg: 'deals with non-latin chars',
     text: 'Привет',
-    slug: 'привет'
+    slug: 'Привет'
+  },
+  // https://github.com/wooorm/gh-and-npm-slug-generation
+  {
+    mesg: 'gh-and-npm-slug-generation-1',
+    text: 'I ♥ unicode',
+    slug: 'i--unicode'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-2',
+    text: 'Dash-dash',
+    slug: 'dash-dash'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-3',
+    text: 'en–dash!',
+    slug: 'endash'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-4',
+    text: 'em–dash',
+    slug: 'emdash'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-5',
+    text: '😄 unicode emoji',
+    slug: '-unicode-emoji'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-6',
+    text: '😄-😄 unicode emoji',
+    slug: '--unicode-emoji'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-7',
+    text: '😄_😄 unicode emoji',
+    slug: '_-unicode-emoji'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-8',
+    text: '😄 - an emoji',
+    slug: '---an-emoji'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-9',
+    text: ':smile: - a gemoji',
+    slug: 'smile---a-gemoji'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-10',
+    text: '    Initial spaces',
+    slug: 'initial-spaces'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-11',
+    text: 'Final spaces   ',
+    slug: 'final-spaces'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-12',
+    text: 'duplicate',
+    slug: 'duplicate'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-13',
+    text: 'duplicate',
+    slug: 'duplicate-1'
+  },
+  {
+    mesg: 'gh-and-npm-slug-generation-14',
+    text: 'Привет non-latin 你好',
+    slug: 'Привет-non-latin-你好'
+  },
+  // https://github.com/chrisdickinson/emoji-slug-example
+  {
+    mesg: 'emoji-slug-example-1',
+    text: ':ok: No underscore',
+    slug: 'ok-no-underscore'
+  },
+  {
+    mesg: 'emoji-slug-example-2',
+    text: ':ok_hand: Single',
+    slug: 'ok_hand-single'
+  },
+  {
+    mesg: 'emoji-slug-example-3',
+    text: ':ok_hand::hatched_chick: Two in a row with no spaces',
+    slug: 'ok_handhatched_chick-two-in-a-row-with-no-spaces'
+  },
+  {
+    mesg: 'emoji-slug-example-4',
+    text: ':ok_hand: :hatched_chick: Two in a row',
+    slug: 'ok_hand-hatched_chick-two-in-a-row'
   }
 ]