Input method editor for the Japanese language

• The KKC of the ime is implemented in the separated file named kkc-engine installed at:
  • fcitx: /usr/lib/senn/fcitx/kkc-engine.
  • ibus: /usr/lib/senn/ibus/kkc-engine.
• When the ime is turned on, it starts kkc-engine and communcates with it using a pipe.
• kkc-engine can be replaced with any executable file as long as it satisfies the protocol described below to communication with the ime.

kkc-engine receives a request of a line from the standard input and sends a response of a line to the standard output. A line of the requests and responses is JSON encoded.

Each json form of requests is as follows:

{ "op": "<request op>", "args": "<request args>" }

The json form of a response depends on the request.

A Convert request is used to converts a sequence that the user inputs to a word sequence.

Request

• op: "CONVERT"
• args:
  • pron:
    • A string of the sequence (typically hiragana)

Response

• array of objects: // An array of objects that describes a word sequence
  • pron: // A substring of the input sequence for the word
  • candidates: array of objects:
    • form: // A string of kana and kanji for the word

A LIST_CANDIDATES request is used to list candidates for a pron.

Request

• op: "LIST_CANDIDATES"
• args:
  • pron:

Response

• array of objects: // An array of objects that describes candidates
  • form: // A string of kana and kanji