sg1pb3.schema.yaml

# Rime schema
# encoding: utf-8

schema:
  schema_id: sg1pb3
  name: "SGPB 雙拼"
  version: '1.0'
  author:
    - MISTER_CHAN <mister.chan@qq.com>
  description: |
    粵語雙拼
    採用香港語言學學會粵語拼音方案
  dependencies:
    - cangjie5
    - wubi86

switches:
  - name: ascii_mode
    reset: 0
    states: [ 中文, 西文 ]
  - name: full_shape
    states: [ 半角, 全角 ]
  - name: ascii_punct
    states: [ "。，", "．，" ]
  - name: simplification
    states: [ 漢字, 汉字 ]
  - name: emoji_suggestion
    reset: 1

engine:
  processors:
    - ascii_composer
    - recognizer
    - key_binder
    - speller
    - punctuator
    - selector
    - navigator
    - express_editor
  segmentors:
    - ascii_segmentor
    - matcher
    - affix_segmentor@cangjie5
    - affix_segmentor@wubi86
    - abc_segmentor
    - punct_segmentor
    - fallback_segmentor
  translators:
    - punct_translator
    - script_translator
    - table_translator@cangjie5
    - table_translator@wubi86
  filters:
    - simplifier
    - simplifier@emoji_suggestion
    - uniquifier
    - charset_filter@gbk
    - reverse_lookup_filter@jyut6ping3_reverse_lookup

speller:
  alphabet: abcdefghijklmnopqrstuvwxyz123456789
  initials: abcdefghijklmnopqrstuvwxyz
  delimiter: " '"
  algebra:
    - xform/(?<=[ptk])1/7/
    - xform/(?<=[ptk])3/8/
    - xform/(?<=[ptk])6/9/
    - xform/aau/Q/
    - xform/^w|aai/W/
    - xform/(?<!o)e(?=\d)/E/
    - xform/k(?=u\d|ui|un\d|ut)|kw|(?<![aeou])e?i(?=\d)/I/
    - xform/g(?=u\d|ui|un\d|ut)|gw|ui|eoi/R/
    - xform/^t|(?<!^)[ei]u/T/
    - xform/(?:yu|(?<!^)(?<![csz])oe)(?=\d)/Y/
    - xform/(?<![aeio])o?u(?=\d)/U/
    - xform/o(?=\d)/O/
    - xform/^p|oi/P/
    - xform/aa(?=\d)/A/
    - xform/s(?!oe\d)|au/S/
    - xform/d|ai/D/
    - xform/f|aa[mp]/F/
    - xform/^g|(?<!^)oe(?:ng|k)/G/
    - xform/h|y?u[nt](?=\d)/H/
    - xform/j|e?o[nt](?=\d)/J/
    - xform/^k|i[nt](?=\d)/K/
    - xform/l|(?<!^)[ei][mp]/L/
    - xform/z(?!oe\d)|aa[nt](?=\d)/Z/
    - xform/[sz]|(?<!a)a[nt](?=\d)/X/
    - xform/c|aa?(?:ng|k)/C/
    - xform/^ng|(?<!a)a[mp]/V/
    - xform/b|(?:i|(?<!o)e)(?:ng|k)/B/
    - xform/^n|u(?:ng|k)/N/
    - xform/^m|o(?:ng|k)/M/
    - erase/^.*[a-z].*$/
    - xlit/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/
    - derive/\d//
    - abbrev/^..+$/./

translator:
  dictionary: jyut6ping3
  spelling_hints: 0
  prism: sg1pb3

cangjie5:
  tag: cangjie5
  dictionary: cangjie5
  enable_user_dict: false
  prefix: "`"
  suffix: "'"
  tips: 〔倉頡五代〕
  preedit_format:
    - 'xform/^([a-z]*)$/$1\t（\U$1\E）/'
    - "xlit|ABCDEFGHIJKLMNOPQRSTUVWXYZ|日月金木水火土竹戈十大中一弓人心手口尸廿山女田難卜符|"
  comment_format:
    - "xlit|abcdefghijklmnopqrstuvwxyz~|日月金木水火土竹戈十大中一弓人心手口尸廿山女田難卜符～|"
  disable_user_dict_for_patterns:
    - "^z.*$"
    - "^yyy.*$"

wubi86:
  tag: wubi86
  dictionary: wubi86
  enable_user_dict: false
  prefix: "y"
  suffix: "'"
  tips: 〔五笔〕
  
jyut6ping3_reverse_lookup:
  tags: [ cangjie5, wubi86 ]
  overwrite_comment: false
  dictionary: jyut6ping3

emoji_suggestion:
  opencc_config: emoji.json
  option_name: emoji_suggestion
  tips: 

punctuator:
  import_preset: symbols_cantonese

key_binder:
  import_preset: default

recognizer:
  import_preset: default
  patterns:
    punct: "^/(?:[0-9A-Za-z]+)$"
    cangjie5: "^`[a-z]*'?$"
    wubi86: "^y[a-z]*'?$"