icuSources/data/translit/Han_Spacedhan.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2010, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Han_Spacedhan.txt
   8 # Generated from CLDR
   9 #
  10 :: [[㆒-㆟㈠-㉇㊀-㊰㋀-㋋㍘-㍰㍻-㍿㏠-㏾ 🈐-🈒🈔-🈺🉀-🉈🉐🉑][:ideographic:][:sc=han:]] nfkc;
  11 :: fullwidth-halfwidth;
  12 ｡ → '.';
  13 $terminalPunct = [\.\,\:\;\?\!．，：？！｡、；[:Pe:][:Pf:]];
  14 $initialPunct = [:Ps:][:Pi:];
  15 [[:Ideographic:] $terminalPunct] {} [:Letter:] → ' ' ;
  16 [:Letter:] [:Mark:]* {} [[:Ideographic:] $initialPunct] → ' ' ;
  17 ← [:Ideographic:] { ' ' } [:Letter:] ;
  18 ← [:Letter:] [:Mark:]* { ' ' } [:Ideographic:] ;