icuSources/data/translit/Han_Latin_Names.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: Han_Latin_Names.txt
   5 # Generated from CLDR
   6 #
   7
   8 # This transform is primarily intended to produce readings for Chinese surnames, or for full
   9 # Chinese personal names - surname first - that occur at the beginning of a contiguous Han substring
  10 # (i.e. at the beginning of text, or immediately preceded by space or other non-Han characters).
  11 # Several Han characters have different readings in surnames, than the readings found in Han-Latin.
  12 # ----
  13 # Insert marker at start of each Han sequence (including Han after space).
  14 # Do this before ::Han-Spacedhan() to catch Han after space in original text,
  15 # and to apply before all other rules.
  16 $startOfHanMarker = \uFDD1;
  17 [:^script=Han:] { ([:script=Han:]) → $startOfHanMarker $1;
  18 # Need Spacedhan so the name transliterations get spaced properly
  19 ::Han-Spacedhan();
  20 # Convert special name readings that depend on next character
  21 令 } \u0020? 狐 →líng;
  22 万 } \u0020? 俟 →mò;
  23 澹 } \u0020? 台 →tán;
  24 # The following maps 长 to the standard Han-Latin reading zhǎng for this case,
  25 # to override the normal Han-Latin/Names reading 长→cháng further below
  26 $startOfHanMarker{ 长 } \u0020? 孙 →zhǎng;
  27 # Convert single characters with special name readings
  28 $startOfHanMarker{ 秘→bì;
  29 $startOfHanMarker{ 卜→bǔ;
  30 长→cháng;
  31 $startOfHanMarker{ 种→chóng;
  32 $startOfHanMarker{ 重→chóng;
  33 $startOfHanMarker{ 刀→diāo;
  34 干→gān;
  35 葛→gě;
  36 $startOfHanMarker{ 盖→gě;
  37 $startOfHanMarker{ 过→guō;
  38 $startOfHanMarker{ 华→huà;
  39 $startOfHanMarker{ 纪→jǐ;
  40 筠→jūn;
  41 $startOfHanMarker{ 阚→kàn;
  42 靓→liàng;
  43 $startOfHanMarker{ 牟→mù;
  44 $startOfHanMarker{ 粘→nián;
  45 $startOfHanMarker{ 区→ōu;
  46 $startOfHanMarker{ 繁→pó;
  47 $startOfHanMarker{ 覃→qín;
  48 仇→qiú;
  49 $startOfHanMarker{ 任→rén;
  50 $startOfHanMarker{ 单→shàn;
  51 $startOfHanMarker{ 召→shào;
  52 $startOfHanMarker{ 折→shé;
  53 $startOfHanMarker{ 舍→shè;
  54 $startOfHanMarker{ 沈→shěn;
  55 峙→shì;
  56 隗→wěi;
  57 $startOfHanMarker{ 解→xiè;
  58 莘→xīn;
  59 $startOfHanMarker{ 燕→yān;
  60 $startOfHanMarker{ 尉→yù;
  61 $startOfHanMarker{ 乐→yuè;
  62 $startOfHanMarker{ 员→yùn;
  63 $startOfHanMarker{ 查→zhā;
  64 翟→zhái;
  65 曾→zēng;
  66 # Convert $startOfHanMarkers to space, or to nothing if they are at the beginning of text.
  67 # Need to do this as a separate pass to get the spacing right.
  68 ::Null();
  69 [^$]{ $startOfHanMarker →\u0020;
  70 $startOfHanMarker →;
  71 # Then run the normal Han-Latin transform for the rest
  72 ::Han-Latin();
  73