icuSources/test/testdata/test4.ucm

   1 # *******************************************************************************
   2 # * Copyright (C) 2001-2003, International Business Machines
   3 # * Corporation and others.  All Rights Reserved.
   4 # *******************************************************************************
   5 #
   6 # test4.ucm
   7 #
   8 # Test file for MBCS conversion with four-byte codepage data.
   9
  10 <code_set_name>     "test4"
  11 <mb_cur_max>        4
  12 <mb_cur_min>        1
  13 <uconv_class>       "MBCS"
  14
  15 # both subchars are single-byters, which does not make sense
  16 # but works - adding subchar1 for tests but don't want to
  17 # change old tests for a new subchar -- markus 20031028
  18 <subchar>           \xff
  19 <subchar1>          \xe1
  20 <icu:state>         0, 1:1, 5-9, e1, ff
  21 <icu:state>         2:2
  22 <icu:state>         3:3
  23 <icu:state>         a-f.p, ff
  24
  25 CHARMAP
  26
  27 # fromUnicode result is zero byte from other than U+0000
  28 <U20ac>     \x00 |0
  29
  30 # fallback from non-zero to zero possible with extension table
  31 <U20ad>     \x00 |1
  32
  33 # nothing special
  34 <U0005>     \x05 |0
  35
  36 # toUnicode result is fallback direct
  37 <U0006>     \x06 |3
  38
  39 # toUnicode result is direct non-BMP code point
  40 <U101234>   \x07 |0
  41 <Ufebcd>    \x08 |3
  42
  43 #unassigned \x09
  44
  45 # toUnicode result is surrogate pair: test real pair, single unit, unassigned
  46 <U23456>    \x01\x02\x03\x0a |0
  47 <U000b>     \x01\x02\x03\x0b |0
  48 #unassigned \x01\x02\x03\x0c
  49 <U34567>    \x01\x02\x03\x0d |3
  50 <U000e>     \x01\x02\x03\x0e |3
  51 #unassigned \x01\x02\x03\x0f
  52
  53 # <subchar1> non-mapping
  54 <U50005>    \xe1 |2
  55 # add a mapping that turns the above's Unicode side into a prefix
  56 <U50005><U60006> \x06 |1
  57
  58 # many bytes, and bytes per UChar
  59 <U30ab><U309a> \x01\x02\x03\x0a\x01\x02\x03\x0b\x01\x02\x03\x0c\x01\x02\x03\x0d\x01\x02\x03\x0e\x01\x02\x03\x0f\x01\x02\x03\x0a\x05\x06\x07 |0
  60
  61 # many UChars, and UChars per byte
  62 <U304b><U309a><U304d><U309a><U304f><U309a><U3051><U309a><U3053><U309a><U30ab><U309a><U30ad><U309a><U30af><U309a><U30b1><U309a><U0300> \x08\x09 |0
  63
  64 END CHARMAP