icuSources/data/translit/Latin_NumericPinyin.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Latin_NumericPinyin.txt
   8 # Generated from CLDR
   9 #
  10
  11 # According to the pinyin definitions I've been able to find:
  12 # 'a', 'e' are the preferred bases
  13 # otherwise 'o'
  14 # otherwise last vowel
  15 # The trailing form of syllables are the following:
  16 #         "a", "ai", "ao", "an", "ang",
  17 #         "o", "ou", "ong",
  18 #         "e", "ei", "er", "en", "eng",
  19 #         "i", "ia", "iao", "ie", "iu", "ian", "in", "iang", "ing", "iong",
  20 #         "u", "ua", "uo", "uai", "ui", "uan", "un", "uang", "ueng",
  21 #         "ü", "üe", "üan", "ün"
  22 # so the letters the tone will 'hop' are:
  23 ::NFD (NFC);
  24 $tone = [\u0304\u0301\u030C\u0300\u0306] ;
  25 # Move the tone to the end of a syllable, and convert to number
  26 e {($tone) r} → r &Pinyin-NumericPinyin($1);
  27 ($tone) ( [i o n u {o n} {n g}]) → $2 &Pinyin-NumericPinyin($1);
  28 ($tone) → &Pinyin-NumericPinyin($1);
  29 # The following backs up until it finds the right vowel, then deposits the tone
  30 $vowel = [aAeEiIoOuU {u\u0308} {U\u0308} vV];
  31 $consonant = [[a-z A-Z] - [$vowel]];
  32 $digit = [1-5];
  33 $1 &NumericPinyin-Pinyin($3) $2 ← ([aAeE]) ($vowel* $consonant*) ($digit);
  34 $1 &NumericPinyin-Pinyin($3) $2 ← ([oO]) ([$vowel-[aeAE]]* $consonant*) ($digit);
  35 $1 &NumericPinyin-Pinyin($3) $2 ← ($vowel) ($consonant*) ($digit);
  36 &NumericPinyin-Pinyin($1) ← [:letter:] {($digit)};
  37 ::NFC (NFD);
  38