wubiluna.schema.yaml

# Rime schema settings
# encoding: utf-8

schema:
  schema_id: wubiluna
  name: "五笔朙月"
  version: "6"
  author:
    - Allen Choong
  description: |
    五筆字型86版修正中国字，加入emoticons。2.0 改编自仓颉检字法（Cangjie6）
  dependencies:
    - luna_pinyin
    - terra_pinyin
    - wubi86
    - stroke

switches:
  - name: ascii_mode
    reset: 0
    states: [ 中文, 西文 ]
  - name: full_shape
    states: [ 半角, 全角 ]
  - name: ascii_punct
    states: [ 。， ,  ．， ]
  - options: [ zh_simp, zh_tw, zh_trad ]
    # reset: 0 # Disable so that not reset to SC
    states:
      - 字型 → 汉字
      - 字型 → 漢字（臺）
      - 字型 → 漢字

engine:
  processors:
    - ascii_composer
    - recognizer
    - key_binder
    - speller
    - punctuator
    - selector
    - navigator
    - express_editor
  segmentors:
    - ascii_segmentor
    - matcher
    - abc_segmentor
    - affix_segmentor@pinyin
    - affix_segmentor@emoji
    - affix_segmentor@pinyin_lookup
    - affix_segmentor@stroke_lookup
    - affix_segmentor@stroke_reverse_lookup
    - affix_segmentor@reverse_lookup # cannot change to other name like wubi_lookup, don't know why
    - punct_segmentor
    - fallback_segmentor
  translators:
    - punct_translator
    - table_translator
    - script_translator@pinyin
    - script_translator@emoji
    - script_translator@pinyin_lookup
    - script_translator@stroke_lookup
    - script_translator@stroke_reverse_lookup
    - script_translator@reverse_lookup
  filters:
    - simplifier@zh_simp # So that by default, simplified pinyin is used
    - simplifier@zh_tw # Traditional based on Taiwan
    - uniquifier
    - reverse_lookup_filter@wubi_reverse_lookup
    - reverse_lookup_filter@pinyin_reverse_lookup
    - reverse_lookup_filter@stroke_reverse_lookup

zh_simp: # based on luna_pinyin_simp
  tag: zh_simp
  option_name: zh_simp

zh_tw:
  tag: zh_tw
  option_name: zh_tw
  opencc_config: t2tw.json

speller:
  alphabet: zyxwvutsrqponmlkjihgfedcba
  delimiter: " '"

translator: # Based on Cangjie6, wubi86, and wubi_pinyin
  dictionary: &dict
    wubiluna
  enable_user_dict: true
  enable_charset_filter: false # added to allows all words
  enable_sentence: true
  sentence_over_completion: true
  enable_encoder: true
  encode_commit_history: true
  max_phrase_length: 4
  initial_quality: 0.75

pinyin: # Based on Cangjie6 and luna_pinyin_simp
  tag: pinyin
  prefix: "~"
  suffix: ;
  tips: 【拼音】
  closing_tips: 【五笔】
  dictionary: luna_pinyin
  preedit_format: &pinyin
    - xform/([nl])v/$1ü/
    - xform/([nl])ue/$1üe/
    - xform/([jqxy])v/$1u/
  initial_quality: 0.2
#  user_dict: wubiluna # so that pinyin also share the same dictionary

emoji: # Based on emoji
  tag: emoji
  dictionary: emoji
  spelling_hints: 1
  comment_format:
    - "xform/;[a-z'`]*$//"
    - "xform/^/ /"
    - "xform/,/ (/"
    - xform/\((.*)$/\($1\)/
  initial_quality: 0.1

abc_segmentor:
  extra_tags: [pinyin, emoji]

reverse_lookup: # Based on Cangjie6, merge with pinyin_lookup, since there is only one reverse lookup
  tag: reverse_lookup
  prefix: "`"
  suffix: ;
  tips: 【拼音反查】
  closing_tips: 【五笔】
  dictionary: pinyin_simp # Based on wubi86 reverse lookup, so that will have simplified and traditional characters
  preedit_format: *pinyin
  initial_quality: 1
  # do not use extra tags, else it will overwrite the terra_pinyin code with wubi86

pinyin_lookup:
  tag: pinyin_lookup
  prefix: "`W" # don't know why must start from `, cannot start from W only
  suffix: ;
  dictionary: wubiluna
  tips: 【五笔反查】
  closing_tips: 【五笔】
  initial_quality: 1

wubi_reverse_lookup: # To show pinyin from wubi
  tags: [ pinyin_lookup ]
  dictionary: terra_pinyin
  overwrite_comment: true

pinyin_reverse_lookup: # To show the wubi code, based on Cangjie6
  tags: [ reverse_lookup ]
  dictionary: *dict
  overwrite_comment: true

stroke_lookup: # Based on Cangjie6, merge with pinyin_lookup, since there is only one reverse lookup
  tag: stroke_lookup
  prefix: "`S"
  suffix: ;
  tips: 【五笔画反查】
  closing_tips: 【五笔】
  dictionary: stroke # Based on wubi86 reverse lookup, so that will have simplified and traditional characters
  preedit_format:
    - xlit/hspnz/一丨丿丶乙/
  comment_format:
    - xform/~//
    - xlit/hspnz/一丨丿丶乙/
  initial_quality: 1

stroke_reverse_lookup: # will show the wubi when typing the stroke
  tags: [ stroke_lookup ]
  dictionary: *dict
  overwrite_comment: true

punctuator: # based on Cangjie6
  import_preset: symbols

key_binder: # based on luna_pinyin_simp
  import_preset: default
  bindings:
    - { when: always, accept: Control+Shift+4, toggle: zh_simp }
    - { when: always, accept: Control+Shift+dollar, toggle: zh_simp }

recognizer: # based on Cangjie6
  import_preset: default
  patterns:
    reverse_lookup: "`[a-z]*;?$" # pinyin to wubi
    pinyin_lookup: "`W[a-z]*;?$" # wubi to pinyin
    stroke_lookup: "`S[hspnz]*;?$" # stroke to wubi
    pinyin: "~[a-z']*;?$" # cannot use number to select the phrase, have to use cursor
    punct: "^/[a-z]*$"