tungdzih.schema.yaml

# Rime schema
# encoding: utf-8

schema:
  schema_id: &schema_id tungdzih
  name: 通字羅馬字
  version: "2.2"
  author:
    - Baopaau <baopaau@gmail.com>
  description: |
    整理自趙元任《通字方案 (General Chinese)》
  dependencies:
    - terra_pinyin
    - jyut6ping3
    - stroke
    - cangjie5

switches:
  - name: ascii_mode
    reset: 0
    states: [ 中文, ASCII ]
  - name: full_shape
    states: [ 半寬, 全寬 ]
  - name: simplification
    states: [ 漢字, 汉字 ]
  - name: ascii_punct
    states: [ 。，, ．， ]

engine:
  processors:
    - ascii_composer
    - recognizer
    - key_binder
    - speller
    - punctuator
    - selector
    - navigator
    - express_editor
  segmentors:
    - ascii_segmentor
    - matcher
    - affix_segmentor@jyut6ping3
    - affix_segmentor@stroke
    - affix_segmentor@this_translator
    - affix_segmentor@terra_pinyin
    - affix_segmentor@cangjie5
    - abc_segmentor
    - punct_segmentor
    - fallback_segmentor
  translators:
    - punct_translator
    - script_translator
    - script_translator@jyut6ping3
    - table_translator@stroke
    - script_translator@this_translator
    - script_translator@terra_pinyin
    - table_translator@cangjie5
  filters:
    - simplifier
    - uniquifier
    - reverse_lookup_filter

speller:
  alphabet: zyxwvutsrqponmlkjihgfedcba
  delimiter: " '"
  algebra:
    # 忽略聲調
    - derive/h$//
    - derive/([nm])\1$/$1/
    - derive/([aoeui])\1(|[nm])$/$1$2/
    - derive/([aoeu])[ey]$/$1i/
    - derive/([aoe])[ow]$/$1u/
    - derive/([aoeui])[gq]$/$1ng/
    # 簡拼
    - abbrev/^([a-z]).+$/$1/
    - abbrev/^([bpmfvdtnlczsjckgqxhyw]?[hryvsz]y?).+$/$1/
    #- abbrev/^([a-z]{2}).+$/$1/

translator:
  dictionary: tungdzih
  prism: *schema_id
  spelling_hints: 7
  #initial_quality: 1
  comment_format: &comment

this_translator:
  tag: this_translator
  dictionary: tungdzih
  prism: *schema_id
  prefix: "~"
  tips: 〔通字〕
  comment_format: *comment
  preedit_format:
    - xform/^(.*)$/‹$1› 通字/

terra_pinyin:
  tag: terra_pinyin
  dictionary: terra_pinyin
  prism: terra_pinyin
  prefix: ":"
  tips: 〔普通話〕
  preedit_format:
    - 'xlit -/<\;,> 1234134'
    - xform/([iuvaoe](|ng?))vvvvv\>/${1}5/
    - xform/([iuvaoe](|ng?))vvvv\>/${1}4/
    - xform/([iuvaoe](|ng?))vvv\>/${1}3/
    - xform/([iuvaoe](|ng?))vv\>/${1}2/
    - xform/([iuvaoe](|ng?))v\>/${1}1/
    - xform/([nl])v/$1ü/
    - xform/([nl])ue/$1üe/
    - xform/([jqxy])v/$1u/
    - xlit/123456/¹²³⁴⁵⁶/
    - xform/^(.*)$/‹$1› 普通話/

jyut6ping3:
  tag: jyut6ping3
  dictionary: jyut6ping3
  prism: jyut6ping3
  prefix: ";"
  tips: 〔粵拼〕
  preedit_format:
    - xform/([aeioumngptk])vv/${1}4/
    - xform/([aeioumngptk])xx/${1}5/
    - xform/([aeioumngptk])qq/${1}6/
    - xform/([aeioumngptk])v/${1}1/
    - xform/([aeioumngptk])x/${1}2/
    - xform/([aeioumngptk])q/${1}3/
    - xlit/123456/¹²³⁴⁵⁶/
    - xform/^(.*)$/‹$1› 粵拼/

stroke:
  tag: stroke
  dictionary: stroke
  prefix: "`"
  tips: 〔筆畫〕
  preedit_format:
    - xlit/hspnz/一丨丿丶乙/
    - xform/^(.*)$/‹$1› 筆畫/

cangjie5:
  tag: cangjie5
  dictionary: cangjie5
  enable_user_dict: false
  prefix: "`c"
  tips: 〔倉頡〕
  preedit_format:
    - 'xform/^([a-z]*)$/$1：\U$1\E /'
    - "xlit|ABCDEFGHIJKLMNOPQRSTUVWXYZ|日月金木水火土竹戈十大中一弓人心手口尸廿山女田難卜符|"
    - xform/^(.*)$/‹$1› 倉頡/
  comment_format:
    - "xlit|abcdefghijklmnopqrstuvwxyz~|日月金木水火土竹戈十大中一弓人心手口尸廿山女田難卜符～|"
  disable_user_dict_for_patterns:
    - "^z.*$"
    - "^yyy.*$"

reverse_lookup:
  tags: [ terra_pinyin, jyut6ping3, stroke, cangjie5 ]
  overwrite_comment: true
  dictionary: tungdzih
  comment_format: *comment

abc_segmentor:
  extra_tags:
    #- jyut6ping3
    #- terra_pinyin
    #- cangjie5

punctuator:
  import_preset: default

key_binder:
  import_preset: default
  bindings:
    #- { when: has_menu, accept: space, send: Shift+Return }

recognizer:
  import_preset: default
  patterns:
    punct: "^/.*$"
    stroke: "^`.*$"
    jyut6ping3: "^;.*$"
    cangjie5: "^`c.*$"
    this_translator: "^~.*$"
    terra_pinyin: "^:.*$"