aurora_pinyin.schema.yaml

# Rime schema
# encoding: utf-8

schema:
  schema_id: aurora_pinyin
  name: 极光拼音
  version: "0.1"
  author:
    - hosxy
  description: |
    为简体中文用户设计的拼音输入方案
    包括《通用规范汉字表》全部8105个汉字及部分未收录汉字/字符
  dependencies:
    - 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: ["gb2312","gbk","utf8"]                # 字符集
    reset: 1                                        # 默认为 GBK 字符集(常用字)
    states: [GB2312, 常用字(GBK), 生僻字(UTF-8)]

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
    - script_translator
  filters:
    - uniquifier
    - charset_filter@gb2312
    - charset_filter@gbk

speller:
  alphabet: zyxwvutsrqponmlkjihgfedcba
  delimiter: " '"
  algebra:
    - erase/^xx$/
    - fuzz/^([m-n]).+$/$1/ # 解决部分 单声母拟声字 候选词问题
    - abbrev/^([a-z]).+$/$1/
    - abbrev/^([zcs]h).+$/$1/
    - derive/^([nl])ve$/$1ue/
    - derive/^([jqxy])u/$1v/
    - derive/([dtngkhrzcs])o(u|ng)$/$1o/
    
    # 纠错
    - derive/un$/uen/ # uen > un
    - derive/ui$/uei/ # uei > ui
    - derive/iu$/iou/ # iou > ui
    - derive/([aeiou])ng$/$1gn/ # gn > ng
    - derive/ong$/on/ # on > ong
    - derive/ao$/oa/ # oa > ao
    - derive/([iu])a(o|ng?)$/a$1$2/ # aio > iao, aing > iang, aung > uang
    - derive/tie$/tei/ # tei > tie
    - derive/i$/ii/ # ii > i i 错按两次
    - derive/u$/uu/ # uu > u u 错按两次
    - derive/xia$/cia/ # cia > xia
    
translator:
  dictionary: aurora_pinyin
  preedit_format:
    - xform/([nl])v/$1ü/
    - xform/([nl])ue/$1üe/
    - xform/([jqxy])v/$1u/

reverse_lookup:
  dictionary: stroke
  enable_completion: true
  prefix: "`"
  suffix: "'"
  tips: 〔笔画〕
  preedit_format:
    - xlit/hspnz/一丨丿丶乙/
  comment_format:
    - xform/([nl])v/$1ü/

punctuator:
  import_preset: symbols
  full_shape:
    "\\" : "、"
    "#" : "#"
    "@" : "@"
  half_shape:
    "\\" : "、"
    "#" : "#"
    "@" : "@"
    
key_binder:
  import_preset: default

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