icuSources/data/translit/Latin_NumericPinyin.txt

   1 #--------------------------------------------------------------------
   2 # Copyright (c) 1999-2004, International Business Machines
   3 # Corporation and others. All Rights Reserved.
   4 #--------------------------------------------------------------------
   5
   6 # According to the pinyin definitions I've been able to find:
   7 # 'a', 'e' are the preferred bases
   8 # otherwise 'o'
   9 # otherwise last vowel
  10
  11 # The trailing form of syllables are the following:
  12 #         "a", "ai", "ao", "an", "ang",
  13 #         "o", "ou", "ong",
  14 #         "e", "ei", "er", "en", "eng",
  15 #         "i", "ia", "iao", "ie", "iu", "ian", "in", "iang", "ing", "iong",
  16 #         "u", "ua", "uo", "uai", "ui", "uan", "un", "uang", "ueng",
  17 #         "ü", "üe", "üan", "ün"
  18 # so the letters the tone will 'hop' are:
  19
  20 ::NFD (NFC);
  21 $tone = [\u0304\u0301\u030C\u0300\u0306] ;
  22
  23 # Move the tone to the end of a syllable, and convert to number
  24 e {($tone) r} > r &tone-digit($1);
  25 ($tone) ( [i o n u {o n} {n g}]) > $2 &tone-digit($1);
  26 ($tone) > &tone-digit($1);
  27
  28 # The following backs up until it finds the right vowel, then deposits the tone
  29
  30 $vowel = [aAeEiIoOuUüÜ];
  31 $consonant = [[a-z A-Z] - [$vowel]];
  32 $digit = [1-5];
  33 $1 &digit-tone($3) $2 < ([aAeE]) ($vowel* $consonant*) ($digit);
  34 $1 &digit-tone($3) $2 < ([oO]) ([$vowel-[aeAE]]* $consonant*) ($digit);
  35 $1 &digit-tone($3) $2 < ($vowel) ($consonant*) ($digit);
  36 &digit-tone($1) < [:letter:] {($digit)};
  37
  38 ::NFC (NFD);
  39
  40
  41