Initial support for v2 API

[anaisbetts]

node-spellchecker has a number of deficiencies that this PR starts to fix up:

• Dictionary is hard-coded to en_US on Windows
• Dictionaries are hard-coded to the vendor directory, which fails on ASAR because Hunspell can't load it out of an ASAR archive
• These settings are globals, so having >1 language active is impossible (i.e. if you detect the user switched their keyboard layout to Spanish, you couldn't blow away the old Hunspell)

To fix this, this PR introduces a new class:

Spellchecker

• setDictionary(language, dictionaryPath) - This method must be called before any other method is called, it configures the language (en_US) and the path to the dictionary (optional, default is .)
• isMisspelled - Identical to v1 isMisspelled method, returns a bool marking whether a word is misspelled
• getCorrectionsForMisspelling - Identical to v1 getCorrectionsForMisspelling method, returns a list of corrections for a misspelled word.

This PR also maintains compatibility with the v1 API via creating a default Spellchecker when the original methods are called

[zcbenz]

This is awesome!

[cristianfraser]

I'm sorry, but how do I use the setDictionary() method? Doing c = require('spellchecker'); c.setDictionary returns undefined, I don't know if I'm missing something or not.

nvm, I realized they're in the Spellchecker class.

[anaisbetts]

@cristianfraser The methods directly exported are the old methods left in for compatibility, so existing apps still work

[cristianfraser]

@paulcbetts So, great tool! But I have one question, how does setDictionary(lang, path) works? I'm currently developing on OSX, and it doesn't seem possible to set a path for each language, it just uses the system's ones? Calling getAvailableDictionaries() returns many languages, none loaded by me so I guess it does. Also, it doesn't show en_US which is in the hunspell_dictionaries folders, though I can set it. Checking the code in spellchecker/src, it seems that for Mac and Windows the path argument isn't used?

I apologize if this isn't the right place to ask, but don't know if I should open an issue for this or simply ask here.

PS: This tidbit I found strange:

> y.setDictionary('en_US')
true
> y.setDictionary('en_USs')
true
> y.setDictionary('en_USsasdf')
true
> y.setDictionary('en_U')
true
> y.setDictionary('aen_U')
false

[anaisbetts]

Checking the code in spellchecker/src, it seems that for Mac and Windows the path argument isn't used?

This is correct.

PS: This tidbit I found strange:

OS X is weird.

[cristianfraser]

Ok, thanks.

So just to be sure, going with major languages like en and es should work on Mac and Windows without adding extra files?

[anaisbetts]

@cristianfraser Both >= Windows 8 and OS X actually ship with a ton of dictionaries in-box, I was surprised how complete it was. For Win7, you're on your own to ship Hunspell dictionaries (http://cgit.freedesktop.org/libreoffice/dictionaries/tree is where I was getting them)