3.1.0

Changed

• bind() now sets necessary attributes automatically (autocomplete, spellcheck etc)
• toKana() no longer converts zenkaku latin toKana now: ('ｉｍｉ imi' => 'ｉｍｉ いみ')
• toRomaji() extends vowels for katakana chōonpu, IE: ゲーム => geemu, toHiragana() converts to hyphen => ge-mu
• toKana() without IME mode converts lone 'n' => 'ん', 'nn' => 'んん' if no other chars present
• isJapanese() now returns false for latin numbers (201) (still true for zenkaku ２０１) - to retain prior behaviour use isJapanese(str, /[0-9]/)
• tokenize() splits into finer categories, view API documentation and tests for full details
• rewrite of conversion methods to allow custom mapping adapters

Added

• isJapanese and isRomaji now accept a second param, a regexp that will also pass the check
• global option romanization for toRomaji() (currently only 'hepburn' however)
• global option customKanaMapping for toKana()
• global option customRomajiMapping for toRomaji()

Fixed

• inserting text between already existing characters in a bind() IME input field now properly sets cursor to correct position after conversion
• toRomaji() little ヶヵ used in words like 一ヶ月 are no longer converted since they are used as symbols (like the kanji) and do not actually denote kana. Previous behaviour: 一ヶ月 => 一 ke 月
• toRomaji() no longer incorrectly duplicates non-glottal stops following っ. Previous behaviour: あっ、 -> a,,