pinyin_to_mps.schema.yaml

# Rime schema
# encoding: utf-8

schema:
  schema_id: pinyin_to_mps
  name: 拼音-注音
  version: "0.1"
  author:
    - Baopaau <baopaau@gmail.com>
  description: |
    漢語拼音轉譯國語注音符號第一式
    空格鍵輸出上屏碼

    聲調輸入方法與地球拼音相同（`- / < \`，或以` ; `代一聲，以` , `代三聲，以` > `代四聲）
    另設單擊` v `鍵（按一次輸入第一聲，按兩次輸入第二聲，如此類推）
  dependencies:
    - terra_pinyin
    - cangjie5

switches:
  - name: ascii_mode
    reset: 0
    states: [ 中文, ASCII ]
  - name: full_shape
    states: [ 半寬, 全寬 ]
  - options: [ noop, simplification, trad_tw ]
    states: [ 原形, 陆简, 臺正 ]
    # 修改下行註釋，改變預設用字 [0, 1, 2 --> 對應上表]
    reset: 0
  - name: ascii_punct
    states: [ 。，, ．， ]
  - name: emoji_suggestion
    states: [ 無繪文字 , 帶繪文字 ]
    # 取消下行註釋，預設啓動 emoji
    reset: 1

engine:
  processors:
    - ascii_composer
    - recognizer
    - key_binder
    - speller
    - punctuator
    - selector
    - navigator
    - express_editor
  segmentors:
    - ascii_segmentor
    - matcher
    - abc_segmentor
    - punct_segmentor
    - fallback_segmentor
  translators:
    - punct_translator
    - reverse_lookup_translator
    - table_translator@custom_phrase
    - script_translator
  filters:
    - simplifier@trad_tw
    - simplifier
    - simplifier@emoji_suggestion
    #- reverse_lookup_filter
    - uniquifier

speller:
  alphabet: 'zyxwvutsrqponmlkjihgfedcba-;/<,>\'
  initials: zyxwvutsrqponmlkjihgfedcba
  delimiter: " '"
  algebra:
    - erase/^xx$/
    - xform/^r5$/er5/  # 〜兒
    - xform/^wong4$/weng4/ # 𥥈/𥦷

    - derive/^([bpmf])o/$1uo/ # 現代音ㄛ->ㄨㄛ
    - derive/^([bpmfw])eng/$1ong/ # 讀書音ㄥ->ㄨㄥ(ueng)
    
    - abbrev/^([a-z]).+$/$1/
    - abbrev/^([zcs]h).+$/$1/
    - derive/^([nl])ve/$1ue/
    - derive/^([jqxy])u/$1v/
    - derive/un([1-5])$/uen$1/
    - derive/ong/ung/
    - derive/ui/uei/
    - derive/iu/iou/
    - derive/ao/oa/
    - derive/ao/au/
    - derive/([aeiou])ng/$1gn/
    - derive/([iu])a(o|ng?)/a$1$2/
    - derive/^([a-z]+)[0-5]$/$1/
    - derive/([dtngkhrzcs])o(u|ng)$/$1o/
    - derive/ong$/on/

    - derive/5$/vvvvv/
    - derive/4$/vvvv/
    - derive/3$/vvv/
    - derive/2$/vv/
    - derive/1$/v/

    - 'erase/^.*5$/'
    - 'xlit 1234 -/<\'
    - 'derive/^(.*)-$/$1;/'
    - 'derive/^(.*)<$/$1,/'
    - 'derive/^(.*)\\$/$1>/'


translator:
  dictionary: terra_pinyin
  prism: pinyin_to_mps
  strict_spelling: true # fuzz項只在連詞拼寫中有效
  spelling_hints: 7
  comment_format: &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\>/${1}1/
    - xform/([iuvaoe](|ng?))\>/${1}1/ # 補全聲調

    # 開始轉譯
    - xform/e?r5$/er5/
    - xform/\<wong4\>/weng4/
    - xform/iu/iou/
    - xform/ui/uei/
    - xform/ong/ung/
    - xform/\<yi?/i/
    - xform/\<wu?/u/
    - xform/iu/v/
    - xform/\<([jqx])u/$1v/
    - xform/([iuv])n/$1en/
    - xform/zh/Z/
    - xform/ch/C/
    - xform/sh/S/
    - xform/ai/A/
    - xform/ei/I/
    - xform/ao/O/
    - xform/ou/U/
    - xform/ang/K/
    - xform/eng/G/
    - xform/an/M/
    - xform/en/N/
    - xform/er/R/
    - xform/eh/E/
    - xform/([iv])e/$1E/
    - xform/1//
    - 'xlit|bpmfdtnlgkhjqxZCSrzcsiuvaoeEAIOUMNKGR2345|ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒㄓㄔㄕㄖㄗㄘㄙㄧㄨㄩㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦˊˇˋ˙|'

  preedit_format: *format
    
custom_phrase:
  dictionary: ""
  user_dict: custom_phrase
  db_class: stabledb
  enable_completion: false
  enable_sentence: false
  initial_quality: 1
    
reverse_lookup:
  overwrite_comment: true
  dictionary: cangjie5
  prism: cangjie5
  prefix: "`"
  suffix: "'"
  tips: 〔倉頡〕
  comment_format: *format

trad_tw:
  option_name: trad_tw
  opencc_config: t2tw.json

emoji_suggestion:
  opencc_config: emoji.json
  option_name: emoji_suggestion
  tips: all
  
punctuator:
  import_preset: symbols

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

recognizer:
  import_preset: default
  patterns:
    punct: "^/.*$"
    reverse_lookup: "^`.*$"