Merge pull request #2 from atom/master

Catch up with atom/node-spellchecker.

Author: goddessfreya

.travis.yml

@@ -32,3 +32,4 @@ env:
   matrix:
     - NODE_VERSION=8.9.3
     - NODE_VERSION=10
+    - NODE_VERSION=12

README.md

@@ -54,3 +54,42 @@ When using Hunspell, this will not modify the .dic file; new words must be added
 `word` - String word to add.
 
 Returns nothing.
+
+### new Spellchecker()
+
+In addition to the above functions that are used on a default instance, a new instance of SpellChecker can be instantiated with the use of the `new` operator. The same methods are available with the instance but the dictionary and underlying API can be changed independently from the default instance.
+
+```javascript
+const checker = new SpellChecker.Spellchecker()
+```
+
+#### SpellChecker.Spellchecker.setSpellcheckerType(type)
+
+Overrides the library selection for checking. Without this, the checker will use [Hunspell](http://hunspell.github.io/) on Linux, the [Spell Checking API](https://docs.microsoft.com/en-us/windows/desktop/intl/spell-checker-api) for Windows, and [NSSpellChecker](https://developer.apple.com/documentation/appkit/nsspellchecker) on Macs.
+
+If the environment variable `SPELLCHECKER_PREFER_HUNSPELL` is set to any value, the library will fallback to always using the Hunspell implementation.
+
+This is the same behavior as calling `setSpellcheckerType` with the `USE_SYSTEM_DEFAULTS` constant:
+
+```coffeescript
+checker = new SpellChecker.Spellchecker
+checker.setSpellcheckerType SpellChecker.USE_SYSTEM_DEFAULTS
+```
+
+To always use the system API and not fallback to Hunspell regardless of the environment variable, use the `ALWAYS_USE_SYSTEM` constant:
+
+```coffeescript
+checker = new SpellChecker.Spellchecker
+checker.setSpellcheckerType SpellChecker.ALWAYS_USE_SYSTEM
+```
+
+Likewise, Hunspell can be forced with the `ALWAYS_USE_HUNSPELL` constant.
+
+```javascript
+const checker = new SpellChecker.Spellchecker();
+checker.setSpellcheckerType(SpellChecker.ALWAYS_USE_SYSTEM);
+```
+
+On Linux, Hunspell is always used regardless of the setting. This method must also be called before any spelling is done otherwise it will throw an error.
+
+This returns nothing.

appveyor.yml

@@ -4,7 +4,6 @@ environment:
     - nodejs_version: '10'
       SPELLCHECKER_PREFER_HUNSPELL: true
     - nodejs_version: '8'
-    - nodejs_version: '6'
 
 install:
   - ps: Install-Product node $env:nodejs_version

lib/spellchecker.js

@@ -103,5 +103,8 @@ module.exports = {
   getAvailableDictionaries: getAvailableDictionaries,
   getCorrectionsForMisspelling: getCorrectionsForMisspelling,
   getDictionaryPath: getDictionaryPath,
-  Spellchecker: Spellchecker
+  Spellchecker: Spellchecker,
+  USE_SYSTEM_DEFAULTS: 0,
+  ALWAYS_USE_SYSTEM: 1,
+  ALWAYS_USE_HUNSPELL: 2,
 };

package-lock.json

@@ -2,7 +2,7 @@
   "main": "./lib/spellchecker.js",
   "name": "spellchecker",
   "description": "Bindings to native spellchecker",
-  "version": "3.5.3",
+  "version": "3.7.0",
   "license": "MIT",
   "repository": {
     "type": "git",
@@ -16,10 +16,10 @@
     "test": "jasmine-focused --captureExceptions --coffee spec/"
   },
   "devDependencies": {
-    "jasmine-focused": "1.x"
+    "jasmine-focused": "^1.0.7"
   },
   "dependencies": {
     "any-promise": "^1.3.0",
-    "nan": "^2.10.0"
+    "nan": "^2.14.0"
   }
 }