jars/icu4j-4_4_2-src/main/tests/core/src/com/ibm/icu/dev/test/shaping/ArabicShapingTest.java

   1 /*\r
   2 *******************************************************************************\r
   3 *   Copyright (C) 2001-2009, International Business Machines\r
   4 *   Corporation and others.  All Rights Reserved.\r
   5 *******************************************************************************\r
   6 */\r
   7 \r
   8 package com.ibm.icu.dev.test.shaping;\r
   9 \r
  10 import com.ibm.icu.text.ArabicShaping;\r
  11 import com.ibm.icu.text.ArabicShapingException;\r
  12 \r
  13 /**\r
  14  * Interactive test for Arabic shaping.\r
  15  * Invoke from a command line passing args and strings.  Use '-help' to see description of arguments.\r
  16  */\r
  17 public class ArabicShapingTest{\r
  18     private static final int COPY = 0;\r
  19     private static final int INPLACE = 1;\r
  20     private static final int STRING = 2;\r
  21 \r
  22     public static final void main(String[] args) {\r
  23         int testtype = COPY;\r
  24         int options = 0;\r
  25         int ss = 0;\r
  26         int sl = -1;\r
  27         int ds = 0;\r
  28         int dl = -1;\r
  29         String text = "$22.4 test 123 \ufef6\u0644\u0622 456 \u0664\u0665\u0666!";\r
  30 \r
  31         for (int i = 0; i < args.length; ++i) {\r
  32             String arg = args[i];\r
  33             if (arg.charAt(0) == '-') {\r
  34                 String opt = arg.substring(1);\r
  35                 String val = opt;\r
  36                 int index = arg.indexOf(':');\r
  37                 if (index != -1) {\r
  38                     opt = opt.substring(0, Math.min(index, 3));\r
  39                     val = arg.substring(index + 1);\r
  40                 }\r
  41                 \r
  42                 if (opt.equalsIgnoreCase("len")) {\r
  43                     options &= ~ArabicShaping.LENGTH_MASK;\r
  44                     if (val.equalsIgnoreCase("gs")) {\r
  45                         options |= ArabicShaping.LENGTH_GROW_SHRINK;\r
  46                     } else if (val.equalsIgnoreCase("sn")) {\r
  47                         options |= ArabicShaping.LENGTH_FIXED_SPACES_NEAR;\r
  48                     } else if (val.equalsIgnoreCase("se")) {\r
  49                         options |= ArabicShaping.LENGTH_FIXED_SPACES_AT_END;\r
  50                     } else if (val.equalsIgnoreCase("sb")) {\r
  51                         options |= ArabicShaping.LENGTH_FIXED_SPACES_AT_BEGINNING;\r
  52                     } else {\r
  53                         throwValError(opt, val);\r
  54                     }\r
  55                 } else if (opt.equalsIgnoreCase("dir")) {\r
  56                     options &= ~ArabicShaping.TEXT_DIRECTION_MASK;\r
  57                     if (val.equalsIgnoreCase("log")) {\r
  58                         options |= ArabicShaping.TEXT_DIRECTION_LOGICAL;\r
  59                     } else if (val.equalsIgnoreCase("vis")) {\r
  60                         options |= ArabicShaping.TEXT_DIRECTION_VISUAL_LTR;\r
  61                     } else {\r
  62                         throwValError(opt, val);\r
  63                     }\r
  64                 } else if (opt.equalsIgnoreCase("let")) {\r
  65                     options &= ~ArabicShaping.LETTERS_MASK;\r
  66                     if (val.equalsIgnoreCase("no")) {\r
  67                         options |= ArabicShaping.LETTERS_NOOP;\r
  68                     } else if (val.equalsIgnoreCase("sh")) {\r
  69                         options |= ArabicShaping.LETTERS_SHAPE;\r
  70                     } else if (val.equalsIgnoreCase("un")) {\r
  71                         options |= ArabicShaping.LETTERS_UNSHAPE;\r
  72                     } else if (val.equalsIgnoreCase("ta")) {\r
  73                         options |= ArabicShaping.LETTERS_SHAPE_TASHKEEL_ISOLATED;\r
  74                     } else {\r
  75                         throwValError(opt, val);\r
  76                     }\r
  77                 } else if (opt.equalsIgnoreCase("dig")) {\r
  78                     options &= ~ArabicShaping.DIGITS_MASK;\r
  79                     if (val.equalsIgnoreCase("no")) {\r
  80                         options |= ArabicShaping.DIGITS_NOOP;\r
  81                     } else if (val.equalsIgnoreCase("ea")) {\r
  82                         options |= ArabicShaping.DIGITS_EN2AN;\r
  83                     } else if (val.equalsIgnoreCase("ae")) {\r
  84                         options |= ArabicShaping.DIGITS_AN2EN;\r
  85                     } else if (val.equalsIgnoreCase("lr")) {\r
  86                         options |= ArabicShaping.DIGITS_EN2AN_INIT_LR;\r
  87                     } else if (val.equalsIgnoreCase("al")) {\r
  88                         options |= ArabicShaping.DIGITS_EN2AN_INIT_AL;\r
  89                     } else {\r
  90                         throwValError(opt, val);\r
  91                     }\r
  92                 } else if (opt.equalsIgnoreCase("typ")) {\r
  93                     options &= ~ArabicShaping.DIGIT_TYPE_MASK;\r
  94                     if (val.equalsIgnoreCase("an")) {\r
  95                         options |= ArabicShaping.DIGIT_TYPE_AN;\r
  96                     } else if (val.equalsIgnoreCase("ex")) {\r
  97                         options |= ArabicShaping.DIGIT_TYPE_AN_EXTENDED;\r
  98                     } else {\r
  99                         throwValError(opt, val);\r
 100                     }\r
 101                 } else if (opt.equalsIgnoreCase("dst")) {\r
 102                     try {\r
 103                         ds = Integer.parseInt(val);\r
 104                     }\r
 105                     catch (Exception e) {\r
 106                         throwValError(opt, val);\r
 107                     }\r
 108                 } else if (opt.equalsIgnoreCase("dln")) {\r
 109                     try {\r
 110                         dl = Integer.parseInt(val);\r
 111                     }\r
 112                     catch (Exception e) {\r
 113                         throwValError(opt, val);\r
 114                     }\r
 115                 } else if (opt.equalsIgnoreCase("sst")) {\r
 116                     try {\r
 117                         ss = Integer.parseInt(val);\r
 118                     }\r
 119                     catch (Exception e) {\r
 120                         throwValError(opt, val);\r
 121                     }\r
 122                 } else if (opt.equalsIgnoreCase("sln")) {\r
 123                     try {\r
 124                         sl = Integer.parseInt(val);\r
 125                     }\r
 126                     catch (Exception e) {\r
 127                         throwValError(opt, val);\r
 128                     }\r
 129                 } else if (opt.equalsIgnoreCase("tes")) {\r
 130                     if (val.equalsIgnoreCase("cp")) {\r
 131                         testtype = COPY;\r
 132                     } else if (val.equalsIgnoreCase("ip")) {\r
 133                         testtype = INPLACE;\r
 134                     } else if (val.equalsIgnoreCase("st")) {\r
 135                         testtype = STRING;\r
 136                     } else {\r
 137                         throwValError(opt, val);\r
 138                     }\r
 139                 } else if (opt.equalsIgnoreCase("help")) {\r
 140                     System.out.println(usage);\r
 141                 } else {\r
 142                     throwOptError(opt);\r
 143                 }\r
 144             } else {\r
 145                 // assume text\r
 146                 text = parseText(arg);\r
 147             }\r
 148         }\r
 149 \r
 150         if (sl < 0) {\r
 151             sl = text.length() - ss;\r
 152             System.out.println("sl defaulting to " + sl);\r
 153         }\r
 154         if (dl < 0) {\r
 155             dl = 2 * sl;\r
 156             System.out.println("dl defaulting to " + dl);\r
 157         }\r
 158 \r
 159         ArabicShaping shaper = new ArabicShaping(options);\r
 160         System.out.println("shaper: " + shaper);\r
 161 \r
 162         char[] src = text.toCharArray();\r
 163         System.out.println(" input: '" + escapedText(src, ss, sl) + "'");\r
 164         if (testtype != STRING) {\r
 165             System.out.println("start: " + ss + " length: " + sl + " total length: " + src.length);\r
 166         }\r
 167 \r
 168         int result = -1;\r
 169         char[] dest = null;\r
 170 \r
 171         try {\r
 172             switch (testtype) {\r
 173             case COPY:\r
 174                 dest = new char[ds + dl];\r
 175                 result = shaper.shape(src, ss, sl, dest, ds, dl);\r
 176                 break;\r
 177 \r
 178             case INPLACE:\r
 179                 shaper.shape(src, ss, sl);\r
 180                 ds = ss;\r
 181                 result = sl;\r
 182                 dest = src;\r
 183                 break;\r
 184 \r
 185             case STRING:\r
 186                 dest = shaper.shape(text).toCharArray();\r
 187                 ds = 0;\r
 188                 result = dest.length;\r
 189                 break;\r
 190             }\r
 191 \r
 192             System.out.println("output: '" + escapedText(dest, ds, result) + "'");\r
 193             System.out.println("length: " + result);\r
 194             if (ds != 0 || result != dest.length) {\r
 195                 System.out.println("full output: '" + escapedText(dest, 0, dest.length) + "'");\r
 196             }\r
 197         }\r
 198         catch (ArabicShapingException e) {\r
 199             System.out.println("Caught ArabicShapingException");\r
 200             System.out.println(e);\r
 201         }\r
 202         catch (Exception e) {\r
 203             System.out.println("Caught Exception");\r
 204             System.out.println(e);\r
 205         }\r
 206     }\r
 207 \r
 208     private static void throwOptError(String opt) {\r
 209         throwUsageError("unknown option: " + opt);\r
 210     }\r
 211 \r
 212     private static void throwValError(String opt, String val) {\r
 213         throwUsageError("unknown value: " + val + " for option: " + opt);\r
 214     }\r
 215 \r
 216     private static void throwUsageError(String message) {\r
 217         StringBuffer buf = new StringBuffer("*** usage error ***\n");\r
 218         buf.append(message);\r
 219         buf.append("\n");\r
 220         buf.append(usage);\r
 221         throw new Error(buf.toString());\r
 222     }\r
 223 \r
 224     private static final String usage = \r
 225         "Usage: [option]* [text]\n" +\r
 226         "  where option is in the format '-opt[:val]'\n" +\r
 227         "  options are:\n" +\r
 228         "    -len:[gs|sn|se|sb]    (length: grow/shrink, spaces near, spaces end, spaces beginning)\n" +\r
 229         "    -dir:[log|vis]        (direction: logical, visual)\n" +\r
 230         "    -let:[no|sh|un|ta]    (letters: noop, shape, unshape, tashkeel)\n" +\r
 231         // "    -let:[no|sh|un]       (letters: noop, shape, unshape)\n" +\r
 232         "    -dig:[no|ea|ae|lr|al] (digits: noop, en2an, an2en, en2an_lr, en2an_al)\n" +\r
 233         "    -typ:[an|ex]          (digit type: arabic, arabic extended)\n" +\r
 234         "    -dst:#                (dest start: [integer])\n" +\r
 235         "    -dln:#                (dest length (max size): [integer])\n" +\r
 236         "    -sst:#                (source start: [integer])\n" +\r
 237         "    -sln:#                (source length: [integer])\n" +\r
 238         "    -tes:[cp|ip|st]       (test type: copy, in place, string)\n" +\r
 239         "    -help                 (print this help message)\n" +\r
 240         "  text can contain unicode escape values in the format '\\uXXXX' only\n";\r
 241         \r
 242     private static String escapedText(char[] text, int start, int length) {\r
 243         StringBuffer buf = new StringBuffer();\r
 244         for (int i = start, e = start + length; i < e; ++i) {\r
 245             char ch = text[i];\r
 246             if (ch < 0x20 || ch > 0x7e) {\r
 247                 buf.append("\\u");\r
 248                 if (ch < 0x1000) {\r
 249                     buf.append('0');\r
 250                 }\r
 251                 if (ch < 0x100) {\r
 252                     buf.append('0');\r
 253                 }\r
 254                 if (ch < 0x10) {\r
 255                     buf.append('0');\r
 256                 }\r
 257                 buf.append(Integer.toHexString(ch));\r
 258             } else {\r
 259                 buf.append(ch);\r
 260             }\r
 261         }\r
 262         return buf.toString();\r
 263     }\r
 264 \r
 265     private static String parseText(String text) {\r
 266         // process unicode escapes (only)\r
 267         StringBuffer buf = new StringBuffer();\r
 268         char[] chars = text.toCharArray();\r
 269         for (int i = 0; i < chars.length; ++i) {\r
 270             char ch = chars[i];\r
 271             if (ch == '\\') {\r
 272                 if ((i < chars.length - 1) &&\r
 273                     (chars[i+1] == 'u')) {\r
 274                     int val = Integer.parseInt(text.substring(i+2, i+6), 16);\r
 275                     buf.append((char)val);\r
 276                     i += 5;\r
 277                 } else {\r
 278                     buf.append('\\');\r
 279                 }\r
 280             } else {\r
 281                 buf.append(ch);\r
 282             }\r
 283         }\r
 284         return buf.toString();\r
 285     }\r
 286 }\r