icuSources/data/translit/Han_Latin_Names.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Han_Latin_Names.txt
   8 # Generated from CLDR
   9 #
  10
  11 # This transform is primarily intended to produce readings for Chinese surnames, or for full
  12 # Chinese personal names - surname first - that occur at the beginning of a contiguous Han substring
  13 # (i.e. at the beginning of text, or immediately preceded by space or other non-Han characters).
  14 # Several Han characters have different readings in surnames, than the readings found in Han-Latin.
  15 # ----
  16 # Insert marker at start of each Han sequence (including Han after space).
  17 # Do this before ::Han-Spacedhan() to catch Han after space in original text,
  18 # and to apply before all other rules.
  19 $startOfHanMarker = \uFDD1;
  20 [:^script=Han:] { ([:script=Han:]) → $startOfHanMarker $1;
  21 # Need Spacedhan so the name transliterations get spaced properly
  22 ::Han-Spacedhan();
  23 # Convert special name readings that depend on next character
  24 令 } \u0020? 狐 →líng;
  25 万 } \u0020? 俟 →mò;
  26 澹 } \u0020? 台 →tán;
  27 # The following maps 长 to the standard Han-Latin reading zhǎng for this case,
  28 # to override the normal Han-Latin/Names reading 长→cháng further below
  29 $startOfHanMarker{ 长 } \u0020? 孙 →zhǎng;
  30 # Convert single characters with special name readings
  31 $startOfHanMarker{ 秘→bì;
  32 $startOfHanMarker{ 卜→bǔ;
  33 长→cháng;
  34 $startOfHanMarker{ 种→chóng;
  35 $startOfHanMarker{ 重→chóng;
  36 $startOfHanMarker{ 刀→diāo;
  37 干→gān;
  38 葛→gě;
  39 $startOfHanMarker{ 盖→gě;
  40 $startOfHanMarker{ 过→guō;
  41 $startOfHanMarker{ 华→huà;
  42 $startOfHanMarker{ 纪→jǐ;
  43 筠→jūn;
  44 $startOfHanMarker{ 牟→mù;
  45 $startOfHanMarker{ 区→ōu;
  46 $startOfHanMarker{ 繁→pó;
  47 仇→qiú;
  48 $startOfHanMarker{ 任→rén;
  49 $startOfHanMarker{ 单→shàn;
  50 $startOfHanMarker{ 召→shào;
  51 $startOfHanMarker{ 折→shé;
  52 $startOfHanMarker{ 舍→shè;
  53 $startOfHanMarker{ 沈→shěn;
  54 峙→shì;
  55 隗→wěi;
  56 $startOfHanMarker{ 解→xiè;
  57 莘→xīn;
  58 $startOfHanMarker{ 燕→yān;
  59 $startOfHanMarker{ 尉→yù;
  60 $startOfHanMarker{ 乐→yuè;
  61 $startOfHanMarker{ 员→yùn;
  62 $startOfHanMarker{ 查→zhā;
  63 翟→zhái;
  64 曾→zēng;
  65 # Convert $startOfHanMarkers to space, or to nothing if they are at the beginning of text.
  66 # Need to do this as a separate pass to get the spacing right.
  67 ::Null();
  68 [^$]{ $startOfHanMarker →\u0020;
  69 $startOfHanMarker →;
  70 # Then run the normal Han-Latin transform for the rest
  71 ::Han-Latin();
  72