pinyin_to_yale.schema.yaml

# Rime schema
# encoding: utf-8

__include: pinyin_to_mps.schema:/

schema:
  schema_id: pinyin_to_yale
  name: 拼音-耶魯
  version: "0.2"
  author:
    - Baopaau <baopaau@gmail.com>
  description: |
    漢語拼音轉譯漢語耶魯拼音
  dependencies:
    - terra_pinyin
    - cangjie5

translator:
  dictionary: terra_pinyin
  prism: pinyin_to_mps
  strict_spelling: true
  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}5/ # 補全聲調

    # 開始轉譯
    - xform/\<r5\>/er5/  # 〜兒
    - xform/\<wong4\>/weng4/ # 𥥈/𥦷

    - xform/\<([jqx])u/$1v/
    - xform/\<([bpmf])o/$1uo/

    - xform/\<([zcs])hi([1-5])\>/$1hr$2/
    - xform/\<ri([1-5])\>/r$1/
    - xform/\<zh/j/
    - xform/\<q/ch/
    - xform/\<x/sy/

    - xform/\<([zcs])i([1-5])\>/$1z$2/
    - xform/\<([zcs])ii([1-5])\>/$1i$2/ # 尖音字碼表用
    - xform/\<z([aeiouvz])/dz$1/
    - xform/\<c([aeiouvz])/ts$1/
    - xform/\<dzz([1-5])\>/dz$1/

    - xform/iu([1-5])\>/iou$1/
    - xform/v/iu/
    - xform/ui([1-5])\>/uei$1/
    - xform/ao([1-5])\>/au$1/
    - xform/([^yi])un([1-5])\>/$1uen$2/
    - xform/ong([1-5])\>/ung$1/

    - xform/iu([aeiou])/yw$1/
    - xform/i([aeiou])/y$1/
    - xform/u([aeiou])/w$1/
    - xform/\<syy/sy/

    # 正詞法（按需啓用）
    - xform/^(\w*[aeiourzng][1-5]) ([aeiourng]\w*?[1-5])$/$1-$2/
    - xform/([aeiou])-g/$1g/
    - xform/g-g/gg/
    - xform/^(\S+?) (\S+?)$/$1$2/

    # 轉換特殊字母
    - xform ([aeiou])(ng?|r)([1234]) $1$3$2
    - xform ([aeo])([iuo])([1234]) $1$3$2
    - xform a1 ā
    - xform a2 á
    - xform a3 ǎ
    - xform a4 à
    - xform e1 ē
    - xform e2 é
    - xform e3 ě
    - xform e4 è
    - xform o1 ō
    - xform o2 ó
    - xform o3 ǒ
    - xform o4 ò
    - xform i1 ī
    - xform i2 í
    - xform i3 ǐ
    - xform i4 ì
    - xform u1 ū
    - xform u2 ú
    - xform u3 ǔ
    - xform u4 ù
    - xform r1 r̄
    - xform r2 ŕ
    - xform r3 ř
    - xform r4 r̀
    - xform z1 z̄
    - xform z2 ź
    - xform z3 ž
    - xform z4 z̀
    - xform/eh[0-5]?/ê/
    - xform/([a-z]+)[0-5]/$1/

  preedit_format: *format