sal/textenc/generate/big5hkscs2001.pl

   1 #!/usr/bin/perl
   2 #*************************************************************************
   3 #
   4 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5 #
   6 # Copyright 2000, 2010 Oracle and/or its affiliates.
   7 #
   8 # OpenOffice.org - a multi-platform office productivity suite
   9 #
  10 # This file is part of OpenOffice.org.
  11 #
  12 # OpenOffice.org is free software: you can redistribute it and/or modify
  13 # it under the terms of the GNU Lesser General Public License version 3
  14 # only, as published by the Free Software Foundation.
  15 #
  16 # OpenOffice.org is distributed in the hope that it will be useful,
  17 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 # GNU Lesser General Public License version 3 for more details
  20 # (a copy is included in the LICENSE file that accompanied this code).
  21 #
  22 # You should have received a copy of the GNU Lesser General Public License
  23 # version 3 along with OpenOffice.org.  If not, see
  24 # <http://www.openoffice.org/license.html>
  25 # for a copy of the LGPLv3 License.
  26 #
  27 #*************************************************************************
  28
  29 # The following files must be available in a ./input subdir:
  30
  31 # <http://www.info.gov.hk/digital21/eng/hkscs/download/big5-iso.txt>
  32
  33 # <http://www.unicode.org/Public/MAPPINGS/OBSOLETE/EASTASIA/OTHER/BIG5.TXT>:
  34 #  "Unicode version: 1.1    Table version: 0.0d3    Date: 11 February 1994"
  35 #  Only used to track Unicode characters that are mapped from both Big5 and
  36 #  HKSCS.
  37
  38 # <http://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP950.TXT>:
  39 #  "Unicode version: 2.0    Table version: 2.01    Date: 1/7/2000"
  40 #  Only used to track Unicode characters that are mapped from both CP950 and
  41 #  HKSCS.
  42
  43 $surrogates = 0; # set to 1 to allow mappings to Unicode beyond Plane 0
  44
  45 $id = "Big5Hkscs2001";
  46
  47 sub isValidUtf32
  48 {
  49     my $utf32 = $_[0];
  50     return $utf32 >= 0 && $utf32 <= 0x10FFFF
  51            && !($utf32 >= 0xD800 && $utf32 <= 0xDFFF)
  52            && !($utf32 >= 0xFDD0 && $utf32 <= 0xFDEF)
  53            && ($utf32 & 0xFFFF) < 0xFFFE;
  54 }
  55
  56 sub printUtf32
  57 {
  58     my $utf32 = $_[0];
  59     return sprintf("U+%04X", $utf32);
  60 }
  61
  62 sub isValidBig5
  63 {
  64     my $big5 = $_[0];
  65     my $big5_row = $big5 >> 8;
  66     my $big5_column = $big5 & 0xFF;
  67     return $big5_row >= 0x81 && $big5_row <= 0xFE
  68            && ($big5_column >= 0x40 && $big5_column <= 0x7E
  69                || $big5_column >= 0xA1 && $big5_column <= 0xFE);
  70 }
  71
  72 sub printBig5
  73 {
  74     my $big5 = $_[0];
  75     return sprintf("%04X", $big5);
  76 }
  77
  78 sub printStats
  79 {
  80     my $used = $_[0];
  81     my $space = $_[1];
  82     return sprintf("%d/%d bytes (%.1f%%)",
  83                    $used,
  84                    $space,
  85                    $used * 100 / $space);
  86 }
  87
  88 sub printSpaces
  89 {
  90     my $column_width = $_[0];
  91     my $columns_per_line = $_[1];
  92     my $end = $_[2];
  93     $output = "";
  94     for ($i = int($end / $columns_per_line) * $columns_per_line;
  95          $i < $end;
  96          ++$i)
  97     {
  98         for ($j = 0; $j < $column_width; ++$j)
  99         {
 100             $output = $output . " ";
 101         }
 102     }
 103     return $output;
 104 }
 105
 106 sub addMapping
 107 {
 108     my $utf32 = $_[0];
 109     my $big5 = $_[1];
 110     my $comp = $_[2];
 111
 112     $uni_plane = $utf32 >> 16;
 113     $uni_page = ($utf32 >> 8) & 0xFF;
 114     $uni_index = $utf32 & 0xFF;
 115
 116     if (!defined($uni_plane_used[$uni_plane])
 117         || !defined($uni_page_used[$uni_plane][$uni_page])
 118         || !defined($uni_map[$uni_plane][$uni_page][$uni_index]))
 119     {
 120         $uni_map[$uni_plane][$uni_page][$uni_index] = $big5;
 121         $uni_plane_used[$uni_plane] = 1;
 122         $uni_page_used[$uni_plane][$uni_page] = 1;
 123         if ($comp != -1)
 124         {
 125             ++$compat[$comp];
 126         }
 127     }
 128     else
 129     {
 130         $big5_1 = $uni_map[$uni_plane][$uni_page][$uni_index];
 131         print "WARNING!  Mapping ", printUtf32($utf32), " to ",
 132               printBig5($big5_1), ", NOT ", ($comp ? "compat " : ""),
 133               printBig5($big5), "\n";
 134     }
 135 }
 136
 137 # Build mappings to track Unicode characters that are mapped from both Big5/
 138 # CP950 and HKSCS:
 139 {
 140     $filename = "BIG5.TXT";
 141     open IN, ("input/" . $filename) or die "Cannot read " . $filename;
 142     while (<IN>)
 143     {
 144         if (/(0x[0-9A-F][0-9A-F][0-9A-F][0-9A-F])[ \t]+(0x[0-9A-F]+)[ \t]+\#.*$/)
 145         {
 146             $big5 = oct($1);
 147             $utf32 = oct($2);
 148             isValidBig5($big5)
 149                 or die "Bad Big5 char " . printBig5($big5);
 150             isValidUtf32($utf32)
 151                 or die "Bad UTF32 char " . printUtf32($utf32);
 152             if ($utf32 != 0xFFFD)
 153             {
 154                 if (defined($underlying_big5[$utf32]))
 155                 {
 156                     print "WARNING!  In ", $filename, ", both ",
 157                           printBig5($underlying_big5[$utf32]), " and ",
 158                           printBig5($big5), " map to ", printUtf32($utf32),
 159                           "\n";
 160                 }
 161                 else
 162                 {
 163                     $underlying_big5[$utf32] = $big5;
 164                 }
 165             }
 166         }
 167     }
 168     close IN;
 169
 170     $filename = "CP950.TXT";
 171     open IN, ("input/" . $filename) or die "Cannot read " . $filename;
 172     while (<IN>)
 173     {
 174         if (/(0x[0-9A-F][0-9A-F][0-9A-F][0-9A-F])[ \t]+(0x[0-9A-F]+)[ \t]+\#.*$/)
 175         {
 176             $big5 = oct($1);
 177             $utf32 = oct($2);
 178             isValidBig5($big5)
 179                 or die "Bad Big5 char " . printBig5($big5);
 180             isValidUtf32($utf32)
 181                 or die "Bad UTF32 char " . printUtf32($utf32);
 182             if (defined($underlying_cp950[$utf32]))
 183             {
 184                 print "WARNING!  In ", $filename, ", both ",
 185                       printBig5($underlying_cp950[$utf32]), " and ",
 186                       printBig5($big5), " map to ", printUtf32($utf32), "\n";
 187             }
 188             else
 189             {
 190                 $underlying_cp950[$utf32] = $big5;
 191             }
 192         }
 193     }
 194     close IN;
 195 }
 196
 197 # The following are mapped by the underlying RTL_TEXTENCODING_BIG5 to some
 198 # nonstandard Unicode points, so they are explicitly mentioned here to map
 199 # to the standard Unicode PUA points.  (In the other direction, the unofficial
 200 # mappings from Unicode to RTL_TEXTENCODING_BIG5 C6A1--C7FE are harmless,
 201 # since all Unicode characters involved are already covered by the official
 202 # Big5-HKSCS mappings.)
 203 $big5_map[0xC6][0xCF] = 0xF6E0; addMapping(0xF6E0, 0xC6CF, -1);
 204 $big5_map[0xC6][0xD3] = 0xF6E4; addMapping(0xF6E4, 0xC6D3, -1);
 205 $big5_map[0xC6][0xD5] = 0xF6E6; addMapping(0xF6E6, 0xC6D5, -1);
 206 $big5_map[0xC6][0xD7] = 0xF6E8; addMapping(0xF6E8, 0xC6D7, -1);
 207 $big5_map[0xC6][0xDE] = 0xF6EF; addMapping(0xF6EF, 0xC6DE, -1);
 208 $big5_map[0xC6][0xDF] = 0xF6F0; addMapping(0xF6F0, 0xC6DF, -1);
 209
 210 # The following implements the mapping of Big5-HKSCS compatibility points
 211 # (GCCS characters unified with other HKSCS characters) to Unicode.  In the
 212 # other direction, characters from Unicode's PUA will map to these Big5-HKSCS
 213 # compatibility points.  (See the first list in <http://www.info.gov.hk/
 214 # digital21/eng/hkscs/download/big5cmp.txt>.)
 215 $big5_map[0x8E][0x69] = 0x7BB8;
 216 $big5_map[0x8E][0x6F] = 0x7C06;
 217 $big5_map[0x8E][0x7E] = 0x7CCE;
 218 $big5_map[0x8E][0xAB] = 0x7DD2;
 219 $big5_map[0x8E][0xB4] = 0x7E1D;
 220 $big5_map[0x8E][0xCD] = 0x8005;
 221 $big5_map[0x8E][0xD0] = 0x8028;
 222 $big5_map[0x8F][0x57] = 0x83C1;
 223 $big5_map[0x8F][0x69] = 0x84A8;
 224 $big5_map[0x8F][0x6E] = 0x840F;
 225 $big5_map[0x8F][0xCB] = 0x89A6;
 226 $big5_map[0x8F][0xCC] = 0x89A9;
 227 $big5_map[0x8F][0xFE] = 0x8D77;
 228 $big5_map[0x90][0x6D] = 0x90FD;
 229 $big5_map[0x90][0x7A] = 0x92B9;
 230 $big5_map[0x90][0xDC] = 0x975C;
 231 $big5_map[0x90][0xF1] = 0x97FF;
 232 $big5_map[0x91][0xBF] = 0x9F16;
 233 $big5_map[0x92][0x44] = 0x8503;
 234 $big5_map[0x92][0xAF] = 0x5159;
 235 $big5_map[0x92][0xB0] = 0x515B;
 236 $big5_map[0x92][0xB1] = 0x515D;
 237 $big5_map[0x92][0xB2] = 0x515E;
 238 $big5_map[0x92][0xC8] = 0x936E;
 239 $big5_map[0x92][0xD1] = 0x7479;
 240 $big5_map[0x94][0x47] = 0x6D67;
 241 $big5_map[0x94][0xCA] = 0x799B;
 242 $big5_map[0x95][0xD9] = 0x9097;
 243 $big5_map[0x96][0x44] = 0x975D;
 244 $big5_map[0x96][0xED] = 0x701E;
 245 $big5_map[0x96][0xFC] = 0x5B28;
 246 $big5_map[0x9B][0x76] = 0x7201;
 247 $big5_map[0x9B][0x78] = 0x77D7;
 248 $big5_map[0x9B][0x7B] = 0x7E87;
 249 $big5_map[0x9B][0xC6] = 0x99D6;
 250 $big5_map[0x9B][0xDE] = 0x91D4;
 251 $big5_map[0x9B][0xEC] = 0x60DE;
 252 $big5_map[0x9B][0xF6] = 0x6FB6;
 253 $big5_map[0x9C][0x42] = 0x8F36;
 254 $big5_map[0x9C][0x53] = 0x4FBB;
 255 $big5_map[0x9C][0x62] = 0x71DF;
 256 $big5_map[0x9C][0x68] = 0x9104;
 257 $big5_map[0x9C][0x6B] = 0x9DF0;
 258 $big5_map[0x9C][0x77] = 0x83CF;
 259 $big5_map[0x9C][0xBC] = 0x5C10;
 260 $big5_map[0x9C][0xBD] = 0x79E3;
 261 $big5_map[0x9C][0xD0] = 0x5A67;
 262 $big5_map[0x9D][0x57] = 0x8F0B;
 263 $big5_map[0x9D][0x5A] = 0x7B51;
 264 $big5_map[0x9D][0xC4] = 0x62D0;
 265 $big5_map[0x9E][0xA9] = 0x6062;
 266 $big5_map[0x9E][0xEF] = 0x75F9;
 267 $big5_map[0x9E][0xFD] = 0x6C4A;
 268 $big5_map[0x9F][0x60] = 0x9B2E;
 269 $big5_map[0x9F][0x66] = 0x9F17;
 270 $big5_map[0x9F][0xCB] = 0x50ED;
 271 $big5_map[0x9F][0xD8] = 0x5F0C;
 272 $big5_map[0xA0][0x63] = 0x880F;
 273 $big5_map[0xA0][0x77] = 0x62CE;
 274 $big5_map[0xA0][0xD5] = 0x7468;
 275 $big5_map[0xA0][0xDF] = 0x7162;
 276 $big5_map[0xA0][0xE4] = 0x7250;
 277 $big5_map[0xFA][0x5F] = 0x5029;
 278 $big5_map[0xFA][0x66] = 0x507D;
 279 $big5_map[0xFA][0xBD] = 0x5305;
 280 $big5_map[0xFA][0xC5] = 0x5344;
 281 $big5_map[0xFA][0xD5] = 0x537F;
 282 $big5_map[0xFB][0x48] = 0x5605;
 283 $big5_map[0xFB][0xB8] = 0x5A77;
 284 $big5_map[0xFB][0xF3] = 0x5E75;
 285 $big5_map[0xFB][0xF9] = 0x5ED0;
 286 $big5_map[0xFC][0x4F] = 0x5F58;
 287 $big5_map[0xFC][0x6C] = 0x60A4;
 288 $big5_map[0xFC][0xB9] = 0x6490;
 289 $big5_map[0xFC][0xE2] = 0x6674;
 290 $big5_map[0xFC][0xF1] = 0x675E;
 291 $big5_map[0xFD][0xB7] = 0x6C9C;
 292 $big5_map[0xFD][0xB8] = 0x6E1D;
 293 $big5_map[0xFD][0xBB] = 0x6E2F;
 294 $big5_map[0xFD][0xF1] = 0x716E;
 295 $big5_map[0xFE][0x52] = 0x732A;
 296 $big5_map[0xFE][0x6F] = 0x745C;
 297 $big5_map[0xFE][0xAA] = 0x74E9;
 298 $big5_map[0xFE][0xDD] = 0x7809;
 299
 300 $pua = 0;
 301 $compat[0] = 0; # 1993
 302 $compat[1] = 0; # 2000
 303 $compat[2] = 0; # 2001
 304
 305 $filename = "big5-iso.txt";
 306 open IN, ("input/" . $filename) or die "Cannot read " . $filename;
 307 while (<IN>)
 308 {
 309     if (/^([0-9A-F]+) +([0-9A-F]+) +([0-9A-F]+) +([0-9A-F]+)$/)
 310     {
 311         $big5 = oct("0x" . $1);
 312         $utf32_1993 = oct("0x" . $2);
 313         $utf32_2000 = oct("0x" . $3);
 314         $utf32_2001 = oct("0x" . $4);
 315         isValidBig5($big5)
 316             or die "Bad Big5 char " . printBig5($big5);
 317         isValidUtf32($utf32_1993)
 318             or die "Bad UTF32 char " . printUtf32($utf32_1993);
 319         isValidUtf32($utf32_2000)
 320             or die "Bad UTF32 char " . printUtf32($utf32_2000);
 321         isValidUtf32($utf32_2001)
 322             or die "Bad UTF32 char " . printUtf32($utf32_2001);
 323
 324         $utf32 = $surrogates ? $utf32_2001 : $utf32_2000;
 325
 326         if (defined($underlying_big5[$utf32])
 327             || defined($underlying_cp950[$utf32]))
 328         {
 329             if (defined($underlying_big5[$utf32])
 330                 && defined($underlying_cp950[$utf32])
 331                 && $underlying_big5[$utf32] == $underlying_cp950[$utf32]
 332                 && $underlying_big5[$utf32] == $big5
 333                 ||
 334                 defined($underlying_big5[$utf32])
 335                 && !defined($underlying_cp950[$utf32])
 336                 && $underlying_big5[$utf32] == $big5
 337                 ||
 338                 !defined($underlying_big5[$utf32])
 339                 && defined($underlying_cp950[$utf32])
 340                 && $underlying_cp950[$utf32] == $big5)
 341             {
 342                 # ignore
 343
 344                 # Depending on real underlying mapping (cf.
 345                 # ../convertbig5hkscs.tab), it would be possible to save some
 346                 # table space by dropping those HKSCS code points that are
 347                 # already covered by the underlying mapping.
 348             }
 349             else
 350             {
 351                 print "XXX mapping underlying";
 352                 if (defined($underlying_big5[$utf32])
 353                     && defined($underlying_cp950[$utf32])
 354                     && $underlying_big5[$utf32] == $underlying_cp950[$utf32])
 355                 {
 356                     print " Big5/CP950 ", printBig5($underlying_big5[$utf32]);
 357                 }
 358                 else
 359                 {
 360                     if (defined($underlying_big5[$utf32]))
 361                     {
 362                         print " Big5 ", printBig5($underlying_big5[$utf32]);
 363                     }
 364                     if (defined($underlying_cp950[$utf32]))
 365                     {
 366                         print " CP950 ", printBig5($underlying_cp950[$utf32]);
 367                     }
 368                 }
 369                 print " and HKSCS ", printBig5($big5), " to ",
 370                       printUtf32($utf32), "\n";
 371             }
 372         }
 373
 374         if ($utf32 >= 0xE000 && $utf32 <= 0xF8FF)
 375         {
 376             ++$pua;
 377         }
 378
 379         $big5_row = $big5 >> 8;
 380         $big5_column = $big5 & 0xFF;
 381         if (defined($big5_map[$big5_row][$big5_column]))
 382         {
 383             die "Bad Big5 mapping " . printBig5($big5);
 384         }
 385         $big5_map[$big5_row][$big5_column] = $utf32;
 386
 387         addMapping($utf32, $big5, -1);
 388
 389         if ($utf32_2001 != $utf32)
 390         {
 391             addMapping($utf32_2001, $big5, 2);
 392         }
 393         if ($utf32_2000 != $utf32 && $utf32_2000 != $utf32_2001)
 394         {
 395             addMapping($utf32_2000, $big5, 1);
 396         }
 397         if ($utf32_1993 != $utf32 && $utf32_1993 != $utf32_2000
 398             && $utf32_1993 != $utf32_2001)
 399         {
 400             addMapping($utf32_1993, $big5, 0);
 401         }
 402     }
 403 }
 404 close IN;
 405
 406 print $pua, " mappings to PUA\n";
 407 print $compat[0], " 1993 compatibility mappings\n" if ($compat[0] != 0);
 408 print $compat[1], " 2000 compatibility mappings\n" if ($compat[1] != 0);
 409 print $compat[2], " 2001 compatibility mappings\n" if ($compat[2] != 0);
 410
 411 if (defined($uni_plane_used[0]) && defined($uni_page_used[0][0]))
 412 {
 413     for ($utf32 = 0; $utf32 <= 0x7F; ++$utf32)
 414     {
 415         if (defined($uni_map[0][0][$uni_index]))
 416         {
 417             $big5 = $uni_map[0][0][$utf32];
 418             die "Mapping " . printUtf32($utf32) . " to " . printBig5($big5);
 419         }
 420     }
 421 }
 422
 423 $filename = lc($id) . ".tab";
 424 open OUT, ("> " . $filename) or die "Cannot write " . $filename;
 425
 426 {
 427     $filename = lc($id). ".pl";
 428     open IN, $filename or die "Cannot read ". $filename;
 429     $first = 1;
 430     while (<IN>)
 431     {
 432         if (/^\#!.*$/)
 433         {
 434         }
 435         elsif (/^\#(\*.*)$/)
 436         {
 437             if ($first == 1)
 438             {
 439                 print OUT "/", $1, "\n";
 440                 $first = 0;
 441             }
 442             else
 443             {
 444                 print OUT " ", substr($1, 0, length($1) - 1), "/\n";
 445             }
 446         }
 447         elsif (/^\# (.*)$/)
 448         {
 449             print OUT " *", $1, "\n";
 450         }
 451         elsif (/^\#(.*)$/)
 452         {
 453             print OUT " *", $1, "\n";
 454         }
 455         else
 456         {
 457             goto done;
 458         }
 459     }
 460   done:
 461 }
 462
 463 print OUT "\n",
 464           "#ifndef _SAL_TYPES_H_\n",
 465           "#include \"sal/types.h\"\n",
 466           "#endif\n",
 467           "\n";
 468
 469 print OUT "static sal_uInt16 const aImpl", $id, "ToUnicodeData[] = {\n";
 470 $big5_data_index = 0;
 471 $big5_rows = 0;
 472 $big5_chars = 0;
 473 for ($big5_row = 0; $big5_row <= 255; ++$big5_row)
 474 {
 475     $big5_row_first = -1;
 476     for ($big5_column = 0; $big5_column <= 255; ++$big5_column)
 477     {
 478         if (defined($big5_map[$big5_row][$big5_column]))
 479         {
 480             if ($big5_row_first == -1)
 481             {
 482                 $big5_row_first = $big5_column;
 483             }
 484             $big5_row_last = $big5_column;
 485         }
 486     }
 487     if ($big5_row_first != -1)
 488     {
 489         $big5_data_offsets[$big5_row] = $big5_data_index;
 490         ++$big5_rows;
 491         print OUT " /* row ", $big5_row, " */\n";
 492
 493         $big5_row_surrogates_first = -1;
 494         $big5_row_chars = 0;
 495         $big5_row_surrogates = 0;
 496
 497         print OUT "  ", $big5_row_first, " | (", $big5_row_last,
 498                   " << 8), /* first, last */\n";
 499         ++$big5_data_index;
 500
 501         print OUT "  ", printSpaces(7, 10, $big5_row_first);
 502         $bol = 0;
 503         for ($big5_column = $big5_row_first;
 504              $big5_column <= $big5_row_last;
 505              ++$big5_column)
 506         {
 507             if ($bol == 1)
 508             {
 509                 print OUT "  ";
 510                 $bol = 0;
 511             }
 512             if (defined($big5_map[$big5_row][$big5_column]))
 513             {
 514                 $utf32 = $big5_map[$big5_row][$big5_column];
 515                 ++$big5_row_chars;
 516                 if ($utf32 <= 0xFFFF)
 517                 {
 518                     printf OUT "0x%04X,", $utf32;
 519                 }
 520                 else
 521                 {
 522                     ++$big5_row_surrogates;
 523                     printf OUT "0x%04X,",
 524                                (0xD800 | (($utf32 - 0x10000) >> 10));
 525                     if ($big5_row_surrogates_first == -1)
 526                     {
 527                         $big5_row_surrogates_first = $big5_column;
 528                     }
 529                     $big5_row_surrogates_last = $big5_column;
 530                 }
 531             }
 532             else
 533             {
 534                 printf OUT "0xffff,";
 535             }
 536             ++$big5_data_index;
 537             if ($big5_column % 10 == 9)
 538             {
 539                 print OUT "\n";
 540                 $bol = 1;
 541             }
 542         }
 543         if ($bol == 0)
 544         {
 545             print OUT "\n";
 546         }
 547
 548         if ($big5_row_surrogates_first != -1)
 549         {
 550             print OUT "  ", $big5_row_surrogates_first,
 551                       ", /* first low-surrogate */\n";
 552             ++$big5_data_index;
 553
 554             print OUT "  ", printSpaces(7, 10, $big5_row_surrogates_first);
 555             $bol = 0;
 556             for ($big5_column = $big5_row_surrogates_first;
 557                  $big5_column <= $big5_row_surrogates_last;
 558                  ++$big5_column)
 559             {
 560                 if ($bol == 1)
 561                 {
 562                     print OUT "  ";
 563                     $bol = 0;
 564                 }
 565                 $utf32 = 0;
 566                 if (defined($big5_map[$big5_row][$big5_column]))
 567                 {
 568                     $utf32 = $big5_map[$big5_row][$big5_column];
 569                 }
 570                 if ($utf32 <= 0xFFFF)
 571                 {
 572                     printf OUT "     0,";
 573                 }
 574                 else
 575                 {
 576                     printf OUT "0x%04X,",
 577                                (0xDC00 | (($utf32 - 0x10000) & 0x3FF));
 578                 }
 579                 ++$big5_data_index;
 580                 if ($big5_column % 10 == 9)
 581                 {
 582                     print OUT "\n";
 583                     $bol = 1;
 584                 }
 585             }
 586             if ($bol == 0)
 587             {
 588                 print OUT "\n";
 589             }
 590         }
 591
 592         $big5_chars += $big5_row_chars;
 593         $big5_data_space[$big5_row]
 594             = ($big5_data_index - $big5_data_offsets[$big5_row]) * 2;
 595         $big5_data_used[$big5_row]
 596             = (1 + $big5_row_chars + ($big5_row_surrogates == 0 ?
 597                                           0 : 1 + $big5_row_surrogates))
 598                   * 2;
 599     }
 600     else
 601     {
 602         print OUT " /* row ", $big5_row, ": --- */\n";
 603         $big5_data_offsets[$big5_row] = -1;
 604     }
 605 }
 606 print OUT "};\n\n";
 607 print "big5 rows = ", $big5_rows, ", chars = ", $big5_chars, "\n";
 608
 609 print OUT "static sal_Int32 const aImpl", $id, "ToUnicodeRowOffsets[] = {\n";
 610 $big5_rowoffsets_used = 0;
 611 for ($big5_row = 0; $big5_row <= 255; ++$big5_row)
 612 {
 613     if ($big5_data_offsets[$big5_row] == -1)
 614     {
 615         print OUT "  -1, /* row ", $big5_row, " */\n";
 616     }
 617     else
 618     {
 619         print OUT "  ",
 620                   $big5_data_offsets[$big5_row],
 621                   ", /* row ",
 622                   $big5_row,
 623                   "; ",
 624                   printStats($big5_data_used[$big5_row],
 625                              $big5_data_space[$big5_row]),
 626                   " */\n";
 627         $big5_rowoffsets_used += 4;
 628     }
 629 }
 630 print OUT "};\n\n";
 631
 632 print OUT "static sal_uInt16 const aImplUnicodeTo", $id, "Data[] = {\n";
 633 $uni_data_index = 0;
 634 for ($uni_plane = 0; $uni_plane <= 16; ++$uni_plane)
 635 {
 636     if (defined($uni_plane_used[$uni_plane]))
 637     {
 638         for ($uni_page = 0; $uni_page <= 255; ++$uni_page)
 639         {
 640             if (defined($uni_page_used[$uni_plane][$uni_page]))
 641             {
 642                 $uni_data_offsets[$uni_plane][$uni_page] = $uni_data_index;
 643                 print OUT " /* plane ", $uni_plane, ", page ", $uni_page,
 644                           " */\n";
 645
 646                 $uni_page_first = -1;
 647                 for ($uni_index = 0; $uni_index <= 255; ++$uni_index)
 648                 {
 649                     if (defined($uni_map[$uni_plane][$uni_page][$uni_index]))
 650                     {
 651                         if ($uni_page_first == -1)
 652                         {
 653                             $uni_page_first = $uni_index;
 654                         }
 655                         $uni_page_last = $uni_index;
 656                     }
 657                 }
 658
 659                 $uni_data_used[$uni_plane][$uni_page] = 0;
 660
 661                 print OUT "  ", $uni_page_first, " | (", $uni_page_last,
 662                           " << 8), /* first, last */\n";
 663                 ++$uni_data_index;
 664                 $uni_data_used[$uni_plane][$uni_page] += 2;
 665
 666                 print OUT "  ", printSpaces(7, 10, $uni_page_first);
 667                 $bol = 0;
 668                 for ($uni_index = $uni_page_first;
 669                      $uni_index <= $uni_page_last;
 670                      ++$uni_index)
 671                 {
 672                     if ($bol == 1)
 673                     {
 674                         print OUT "  ";
 675                         $bol = 0;
 676                     }
 677                     if (defined($uni_map[$uni_plane][$uni_page][$uni_index]))
 678                     {
 679                         $big5 = $uni_map[$uni_plane][$uni_page][$uni_index];
 680                         printf OUT "0x%04X,", $big5;
 681                         $uni_data_used[$uni_plane][$uni_page] += 2;
 682                     }
 683                     else
 684                     {
 685                         print OUT "     0,";
 686                     }
 687                     ++$uni_data_index;
 688                     if ($uni_index % 10 == 9)
 689                     {
 690                         print OUT "\n";
 691                         $bol = 1;
 692                     }
 693                 }
 694                 if ($bol == 0)
 695                 {
 696                     print OUT "\n";
 697                 }
 698
 699                 $uni_data_space[$uni_plane][$uni_page]
 700                     = ($uni_data_index
 701                        - $uni_data_offsets[$uni_plane][$uni_page]) * 2;
 702             }
 703             else
 704             {
 705                 $uni_data_offsets[$uni_plane][$uni_page] = -1;
 706                 print OUT " /* plane ", $uni_plane, ", page ", $uni_page,
 707                           ": --- */\n";
 708             }
 709         }
 710     }
 711     else
 712     {
 713         print OUT " /* plane ", $uni_plane, ": --- */\n";
 714     }
 715 }
 716 print OUT "};\n\n";
 717
 718 print OUT "static sal_Int32 const aImplUnicodeTo", $id, "PageOffsets[] = {\n";
 719 for ($uni_plane = 0; $uni_plane <= 16; ++$uni_plane)
 720 {
 721     if (defined($uni_plane_used[$uni_plane]))
 722     {
 723         $uni_pageoffsets_used[$uni_plane] = 0;
 724         $uni_data_used_sum[$uni_plane] = 0;
 725         $uni_data_space_sum[$uni_plane] = 0;
 726         for ($uni_page = 0; $uni_page <= 255; ++$uni_page)
 727         {
 728             $offset = $uni_data_offsets[$uni_plane][$uni_page];
 729             if ($offset == -1)
 730             {
 731                 print OUT "  -1, /* plane ",
 732                           $uni_plane,
 733                           ", page ",
 734                           $uni_page,
 735                           " */\n";
 736             }
 737             else
 738             {
 739                 print OUT "  ",
 740                           $offset,
 741                           ", /* plane ",
 742                           $uni_plane,
 743                           ", page ",
 744                           $uni_page,
 745                           "; ",
 746                           printStats($uni_data_used[$uni_plane][$uni_page],
 747                                      $uni_data_space[$uni_plane][$uni_page]),
 748                           " */\n";
 749                 $uni_pageoffsets_used[$uni_plane] += 4;
 750                 $uni_data_used_sum[$uni_plane]
 751                     += $uni_data_used[$uni_plane][$uni_page];
 752                 $uni_data_space_sum[$uni_plane]
 753                     += $uni_data_space[$uni_plane][$uni_page];
 754             }
 755         }
 756     }
 757     else
 758     {
 759         print OUT "  /* plane ", $uni_plane, ": --- */\n";
 760     }
 761 }
 762 print OUT "};\n\n";
 763
 764 print OUT "static sal_Int32 const aImplUnicodeTo",
 765           $id,
 766           "PlaneOffsets[] = {\n";
 767 $uni_page_offset = 0;
 768 $uni_planeoffsets_used = 0;
 769 $uni_pageoffsets_used_sum = 0;
 770 $uni_pageoffsets_space_sum = 0;
 771 $uni_data_used_sum2 = 0;
 772 $uni_data_space_sum2 = 0;
 773 for ($uni_plane = 0; $uni_plane <= 16; ++$uni_plane)
 774 {
 775     if (defined ($uni_plane_used[$uni_plane]))
 776     {
 777         print OUT "  ",
 778                   $uni_page_offset++,
 779                   " * 256, /* plane ",
 780                   $uni_plane,
 781                   "; ",
 782                   printStats($uni_pageoffsets_used[$uni_plane], 256 * 4),
 783                   ", ",
 784                   printStats($uni_data_used_sum[$uni_plane],
 785                              $uni_data_space_sum[$uni_plane]),
 786                   " */\n";
 787         $uni_planeoffsets_used += 4;
 788         $uni_pageoffsets_used_sum += $uni_pageoffsets_used[$uni_plane];
 789         $uni_pageoffsets_space_sum += 256 * 4;
 790         $uni_data_used_sum2 += $uni_data_used_sum[$uni_plane];
 791         $uni_data_space_sum2 += $uni_data_space_sum[$uni_plane];
 792     }
 793     else
 794     {
 795         print OUT "  -1, /* plane ", $uni_plane, " */\n";
 796     }
 797 }
 798 print OUT " /* ",
 799           printStats($uni_planeoffsets_used, 17 * 4),
 800           ", ",
 801           printStats($uni_pageoffsets_used_sum, $uni_pageoffsets_space_sum),
 802           ", ",
 803           printStats($uni_data_used_sum2, $uni_data_space_sum2),
 804           " */\n};\n";
 805
 806 close OUT;