gen-unicode-tables.pl

   1 #! /usr/bin/perl -w
   2
   3 #    Copyright (C) 1998, 1999 Tom Tromey
   4 #    Copyright (C) 2001 Red Hat Software
   5
   6 #    This program is free software; you can redistribute it and/or modify
   7 #    it under the terms of the GNU General Public License as published by
   8 #    the Free Software Foundation; either version 2, or (at your option)
   9 #    any later version.
  10
  11 #    This program is distributed in the hope that it will be useful,
  12 #    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 #    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 #    GNU General Public License for more details.
  15
  16 #    You should have received a copy of the GNU General Public License
  17 #    along with this program; if not, write to the Free Software
  18 #    Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  19 #    02111-1307, USA.
  20
  21 # Contributer(s):
  22 #   Andrew Taylor <andrew.taylor@montage.ca>
  23
  24 # gen-unicode-tables.pl - Generate tables for libunicode from Unicode data.
  25 # See http://www.unicode.org/Public/UNIDATA/UnicodeCharacterDatabase.html
  26 # Usage: gen-unicode-tables.pl [-decomp | -both] UNICODE-VERSION UnicodeData.txt LineBreak.txt SpecialCasing.txt CaseFolding.txt
  27 # I consider the output of this program to be unrestricted.  Use it as
  28 # you will.
  29
  30 # FIXME:
  31 # * For decomp table it might make sense to use a shift count other
  32 #   than 8.  We could easily compute the perfect shift count.
  33
  34 use vars qw($CODE $NAME $CATEGORY $COMBINING_CLASSES $BIDI_CATEGORY $DECOMPOSITION $DECIMAL_VALUE $DIGIT_VALUE $NUMERIC_VALUE $MIRRORED $OLD_NAME $COMMENT $UPPER $LOWER $TITLE $BREAK_CODE $BREAK_CATEGORY $BREAK_NAME $CASE_CODE $CASE_LOWER $CASE_TITLE $CASE_UPPER $CASE_CONDITION);
  35
  36 # Names of fields in Unicode data table.
  37 $CODE = 0;
  38 $NAME = 1;
  39 $CATEGORY = 2;
  40 $COMBINING_CLASSES = 3;
  41 $BIDI_CATEGORY = 4;
  42 $DECOMPOSITION = 5;
  43 $DECIMAL_VALUE = 6;
  44 $DIGIT_VALUE = 7;
  45 $NUMERIC_VALUE = 8;
  46 $MIRRORED = 9;
  47 $OLD_NAME = 10;
  48 $COMMENT = 11;
  49 $UPPER = 12;
  50 $LOWER = 13;
  51 $TITLE = 14;
  52
  53 # Names of fields in the line break table
  54 $BREAK_CODE = 0;
  55 $BREAK_PROPERTY = 1;
  56
  57 # Names of fields in the SpecialCasing table
  58 $CASE_CODE = 0;
  59 $CASE_LOWER = 1;
  60 $CASE_TITLE = 2;
  61 $CASE_UPPER = 3;
  62 $CASE_CONDITION = 4;
  63
  64 # Names of fields in the CaseFolding table
  65 $FOLDING_CODE = 0;
  66 $FOLDING_STATUS = 1;
  67 $FOLDING_MAPPING = 2;
  68
  69 # Map general category code onto symbolic name.
  70 %mappings =
  71     (
  72      # Normative.
  73      'Lu' => "G_UNICODE_UPPERCASE_LETTER",
  74      'Ll' => "G_UNICODE_LOWERCASE_LETTER",
  75      'Lt' => "G_UNICODE_TITLECASE_LETTER",
  76      'Mn' => "G_UNICODE_NON_SPACING_MARK",
  77      'Mc' => "G_UNICODE_COMBINING_MARK",
  78      'Me' => "G_UNICODE_ENCLOSING_MARK",
  79      'Nd' => "G_UNICODE_DECIMAL_NUMBER",
  80      'Nl' => "G_UNICODE_LETTER_NUMBER",
  81      'No' => "G_UNICODE_OTHER_NUMBER",
  82      'Zs' => "G_UNICODE_SPACE_SEPARATOR",
  83      'Zl' => "G_UNICODE_LINE_SEPARATOR",
  84      'Zp' => "G_UNICODE_PARAGRAPH_SEPARATOR",
  85      'Cc' => "G_UNICODE_CONTROL",
  86      'Cf' => "G_UNICODE_FORMAT",
  87      'Cs' => "G_UNICODE_SURROGATE",
  88      'Co' => "G_UNICODE_PRIVATE_USE",
  89      'Cn' => "G_UNICODE_UNASSIGNED",
  90
  91      # Informative.
  92      'Lm' => "G_UNICODE_MODIFIER_LETTER",
  93      'Lo' => "G_UNICODE_OTHER_LETTER",
  94      'Pc' => "G_UNICODE_CONNECT_PUNCTUATION",
  95      'Pd' => "G_UNICODE_DASH_PUNCTUATION",
  96      'Ps' => "G_UNICODE_OPEN_PUNCTUATION",
  97      'Pe' => "G_UNICODE_CLOSE_PUNCTUATION",
  98      'Pi' => "G_UNICODE_INITIAL_PUNCTUATION",
  99      'Pf' => "G_UNICODE_FINAL_PUNCTUATION",
 100      'Po' => "G_UNICODE_OTHER_PUNCTUATION",
 101      'Sm' => "G_UNICODE_MATH_SYMBOL",
 102      'Sc' => "G_UNICODE_CURRENCY_SYMBOL",
 103      'Sk' => "G_UNICODE_MODIFIER_SYMBOL",
 104      'So' => "G_UNICODE_OTHER_SYMBOL"
 105      );
 106
 107 %break_mappings =
 108     (
 109      'BK' => "G_UNICODE_BREAK_MANDATORY",
 110      'CR' => "G_UNICODE_BREAK_CARRIAGE_RETURN",
 111      'LF' => "G_UNICODE_BREAK_LINE_FEED",
 112      'CM' => "G_UNICODE_BREAK_COMBINING_MARK",
 113      'SG' => "G_UNICODE_BREAK_SURROGATE",
 114      'ZW' => "G_UNICODE_BREAK_ZERO_WIDTH_SPACE",
 115      'IN' => "G_UNICODE_BREAK_INSEPARABLE",
 116      'GL' => "G_UNICODE_BREAK_NON_BREAKING_GLUE",
 117      'CB' => "G_UNICODE_BREAK_CONTINGENT",
 118      'SP' => "G_UNICODE_BREAK_SPACE",
 119      'BA' => "G_UNICODE_BREAK_AFTER",
 120      'BB' => "G_UNICODE_BREAK_BEFORE",
 121      'B2' => "G_UNICODE_BREAK_BEFORE_AND_AFTER",
 122      'HY' => "G_UNICODE_BREAK_HYPHEN",
 123      'NS' => "G_UNICODE_BREAK_NON_STARTER",
 124      'OP' => "G_UNICODE_BREAK_OPEN_PUNCTUATION",
 125      'CL' => "G_UNICODE_BREAK_CLOSE_PUNCTUATION",
 126      'QU' => "G_UNICODE_BREAK_QUOTATION",
 127      'EX' => "G_UNICODE_BREAK_EXCLAMATION",
 128      'ID' => "G_UNICODE_BREAK_IDEOGRAPHIC",
 129      'NU' => "G_UNICODE_BREAK_NUMERIC",
 130      'IS' => "G_UNICODE_BREAK_INFIX_SEPARATOR",
 131      'SY' => "G_UNICODE_BREAK_SYMBOL",
 132      'AL' => "G_UNICODE_BREAK_ALPHABETIC",
 133      'PR' => "G_UNICODE_BREAK_PREFIX",
 134      'PO' => "G_UNICODE_BREAK_POSTFIX",
 135      'SA' => "G_UNICODE_BREAK_COMPLEX_CONTEXT",
 136      'AI' => "G_UNICODE_BREAK_AMBIGUOUS",
 137      'XX' => "G_UNICODE_BREAK_UNKNOWN"
 138      );
 139
 140 # Title case mappings.
 141 %title_to_lower = ();
 142 %title_to_upper = ();
 143
 144 # Maximum length of special-case strings
 145
 146 my $special_case_len = 0;
 147 my @special_cases;
 148
 149 $do_decomp = 0;
 150 $do_props = 1;
 151 if (@ARGV && $ARGV[0] eq '-decomp')
 152 {
 153     $do_decomp = 1;
 154     $do_props = 0;
 155     shift @ARGV;
 156 }
 157 elsif (@ARGV && $ARGV[0] eq '-both')
 158 {
 159     $do_decomp = 1;
 160     shift @ARGV;
 161 }
 162
 163 if (@ARGV != 6) {
 164     $0 =~ s@.*/@@;
 165     die "Usage: $0 [-decomp | -both] UNICODE-VERSION UnicodeData.txt LineBreak.txt SpecialCasing.txt CaseFolding.txt CompositionExclusions.txt\n";
 166 }
 167
 168 print "Creating decomp table\n" if ($do_decomp);
 169 print "Creating property table\n" if ($do_props);
 170
 171 print "Composition exlusions from $ARGV[5]\n";
 172
 173 open (INPUT, "< $ARGV[5]") || exit 1;
 174
 175 while (<INPUT>) {
 176
 177     chop;
 178
 179     next if /^#/;
 180     next if /^\s*$/;
 181
 182     s/\s*#.*//;
 183
 184     s/^\s*//;
 185     s/\s*$//;
 186
 187     $composition_exclusions{hex($_)} = 1;
 188 }
 189
 190 close INPUT;
 191
 192 print "Unicode data from $ARGV[1]\n";
 193
 194 open (INPUT, "< $ARGV[1]") || exit 1;
 195
 196 $last_code = -1;
 197 while (<INPUT>)
 198 {
 199     chop;
 200     @fields = split (';', $_, 30);
 201     if ($#fields != 14)
 202     {
 203         printf STDERR ("Entry for $fields[$CODE] has wrong number of fields (%d)\n", $#fields);
 204     }
 205
 206     $code = hex ($fields[$CODE]);
 207
 208     last if ($code > 0xFFFF); # ignore characters out of the basic plane
 209
 210     if ($code > $last_code + 1)
 211     {
 212         # Found a gap.
 213         if ($fields[$NAME] =~ /Last>/)
 214         {
 215             # Fill the gap with the last character read,
 216             # since this was a range specified in the char database
 217             @gfields = @fields;
 218         }
 219         else
 220         {
 221             # The gap represents undefined characters.  Only the type
 222             # matters.
 223             @gfields = ('', '', 'Cn', '0', '', '', '', '', '', '', '',
 224                         '', '', '', '');
 225         }
 226         for (++$last_code; $last_code < $code; ++$last_code)
 227         {
 228             $gfields{$CODE} = sprintf ("%04x", $last_code);
 229             &process_one ($last_code, @gfields);
 230         }
 231     }
 232     &process_one ($code, @fields);
 233     $last_code = $code;
 234 }
 235
 236 close INPUT;
 237
 238 @gfields = ('', '', 'Cn', '0', '', '', '', '', '', '', '',
 239             '', '', '', '');
 240 for (++$last_code; $last_code < 0x10000; ++$last_code)
 241 {
 242     $gfields{$CODE} = sprintf ("%04x", $last_code);
 243     &process_one ($last_code, @gfields);
 244 }
 245 --$last_code;                   # Want last to be 0xFFFF.
 246
 247 print "Creating line break table\n";
 248
 249 print "Line break data from $ARGV[2]\n";
 250
 251 open (INPUT, "< $ARGV[2]") || exit 1;
 252
 253 $last_code = -1;
 254 while (<INPUT>)
 255 {
 256     my ($start_code, $end_code);
 257
 258     chop;
 259
 260     next if /^#/;
 261
 262     s/\s*#.*//;
 263
 264     @fields = split (';', $_, 30);
 265     if ($#fields != 1)
 266     {
 267         printf STDERR ("Entry for $fields[$CODE] has wrong number of fields (%d)\n", $#fields);
 268         next;
 269     }
 270
 271     if ($fields[$CODE] =~ /([A-F0-9]{4})..([A-F0-9]{4})/)
 272     {
 273         $start_code = hex ($1);
 274         $end_code = hex ($2);
 275     } else {
 276         $start_code = $end_code = hex ($fields[$CODE]);
 277
 278     }
 279
 280     last if ($start_code > 0xFFFF); # FIXME ignore characters out of the basic plane
 281
 282     if ($start_code > $last_code + 1)
 283     {
 284         # The gap represents undefined characters. If assigned,
 285         # they are AL, if not assigned, XX
 286         for (++$last_code; $last_code < $start_code; ++$last_code)
 287         {
 288             if ($type[$last_code] eq 'Cn')
 289             {
 290                 $break_props[$last_code] = 'XX';
 291             }
 292             else
 293             {
 294                 $break_props[$last_code] = 'AL';
 295             }
 296         }
 297     }
 298
 299     for ($last_code = $start_code; $last_code <= $end_code; $last_code++)
 300     {
 301         $break_props[$last_code] = $fields[$BREAK_PROPERTY];
 302     }
 303
 304     $last_code = $end_code;
 305 }
 306
 307 close INPUT;
 308
 309 for (++$last_code; $last_code < 0x10000; ++$last_code)
 310 {
 311   if ($type[$last_code] eq 'Cn')
 312     {
 313       $break_props[$last_code] = 'XX';
 314     }
 315   else
 316     {
 317       $break_props[$last_code] = 'AL';
 318     }
 319 }
 320 --$last_code;                   # Want last to be 0xFFFF.
 321
 322 print STDERR "Last code is not 0xFFFF" if ($last_code != 0xFFFF);
 323
 324 print "Reading special-casing table for case conversion\n";
 325
 326 open (INPUT, "< $ARGV[3]") || exit 1;
 327
 328 while (<INPUT>)
 329 {
 330     my $code;
 331
 332     chop;
 333
 334     next if /^#/;
 335     next if /^\s*$/;
 336
 337     s/\s*#.*//;
 338
 339     @fields = split ('\s*;\s*', $_, 30);
 340
 341     $raw_code = $fields[$CASE_CODE];
 342     $code = hex ($raw_code);
 343
 344     if ($#fields != 4 && $#fields != 5)
 345     {
 346         printf STDERR ("Entry for $raw_code has wrong number of fields (%d)\n", $#fields);
 347         next;
 348     }
 349
 350     if (!defined $type[$code])
 351     {
 352         printf STDERR "Special case for code point: $code, which has no defined type\n";
 353         next;
 354     }
 355
 356     if (defined $fields[5]) {
 357         # Ignore conditional special cases - we'll handle them in code
 358         next;
 359     }
 360
 361     if ($type[$code] eq 'Lu')
 362     {
 363         (hex $fields[$CASE_UPPER] == $code) || die "$raw_code is Lu and UCD_Upper($raw_code) != $raw_code";
 364
 365         &add_special_case ($code, $value[$code],$fields[$CASE_LOWER], $fields[$CASE_TITLE]);
 366
 367     } elsif ($type[$code] eq 'Lt')
 368     {
 369         (hex $fields[$CASE_TITLE] == $code) || die "$raw_code is Lt and UCD_Title($raw_code) != $raw_code";
 370
 371         &add_special_case ($code, undef,$fields[$CASE_LOWER], $fields[$CASE_UPPER]);
 372     } elsif ($type[$code] eq 'Ll')
 373     {
 374         (hex $fields[$CASE_LOWER] == $code) || die "$raw_code is Ll and UCD_Lower($raw_code) != $raw_code";
 375
 376         &add_special_case ($code, $value[$code],$fields[$CASE_UPPER], $fields[$CASE_TITLE]);
 377     } else {
 378         printf STDERR "Special case for non-alphabetic code point: $raw_code\n";
 379         next;
 380     }
 381 }
 382
 383 close INPUT;
 384
 385 open (INPUT, "< $ARGV[4]") || exit 1;
 386
 387 my $casefoldlen = 0;
 388 my @casefold;
 389
 390 while (<INPUT>)
 391 {
 392     my $code;
 393
 394     chop;
 395
 396     next if /^#/;
 397     next if /^\s*$/;
 398
 399     s/\s*#.*//;
 400
 401     @fields = split ('\s*;\s*', $_, 30);
 402
 403     $raw_code = $fields[$FOLDING_CODE];
 404     $code = hex ($raw_code);
 405
 406     next if $code > 0xffff;     # FIXME!
 407
 408     if ($#fields != 3)
 409     {
 410         printf STDERR ("Entry for $raw_code has wrong number of fields (%d)\n", $#fields);
 411         next;
 412     }
 413
 414     next if ($fields[$FOLDING_STATUS] eq 'S');
 415
 416     @values = map { hex ($_) } split /\s+/, $fields[$FOLDING_MAPPING];
 417
 418     # Check simple case
 419
 420     if (@values == 1 &&
 421         !(defined $value[$code] && $value[$code] >= 0xd800 && $value[$code] < 0xdc00) &&
 422         defined $type[$code]) {
 423
 424         my $lower;
 425         if ($type[$code] eq 'Ll')
 426         {
 427             $lower = $code;
 428         } elsif ($type[$code] eq 'Lt')
 429         {
 430             $lower = $title_to_lower{$code};
 431         } elsif ($type[$code] eq 'Lu')
 432         {
 433             $lower = $value[$code];
 434         } else {
 435             $lower = $code;
 436         }
 437
 438         if ($lower == $values[0]) {
 439             next;
 440         }
 441     }
 442
 443     my $string = pack ("U*", @values);
 444     if (1 + length $string > $casefoldlen) {
 445         $casefoldlen = 1 + length $string;
 446     }
 447
 448     push @casefold, [ $code, $string ];
 449 }
 450
 451 close INPUT;
 452
 453 if ($do_props) {
 454     &print_tables ($last_code)
 455 }
 456 if ($do_decomp) {
 457     &print_decomp ($last_code);
 458     &output_composition_table;
 459 }
 460
 461 &print_line_break ($last_code);
 462
 463 exit 0;
 464
 465 # Process a single character.
 466 sub process_one
 467 {
 468     my ($code, @fields) = @_;
 469
 470     $type[$code] = $fields[$CATEGORY];
 471     if ($type[$code] eq 'Nd')
 472     {
 473         $value[$code] = int ($fields[$DECIMAL_VALUE]);
 474     }
 475     elsif ($type[$code] eq 'Ll')
 476     {
 477         $value[$code] = hex ($fields[$UPPER]);
 478     }
 479     elsif ($type[$code] eq 'Lu')
 480     {
 481         $value[$code] = hex ($fields[$LOWER]);
 482     }
 483
 484     if ($type[$code] eq 'Lt')
 485     {
 486         $title_to_lower{$code} = hex ($fields[$LOWER]);
 487         $title_to_upper{$code} = hex ($fields[$UPPER]);
 488     }
 489
 490     $cclass[$code] = $fields[$COMBINING_CLASSES];
 491
 492     # Handle decompositions.
 493     if ($fields[$DECOMPOSITION] ne '')
 494     {
 495         if ($fields[$DECOMPOSITION] =~ s/\<.*\>\s*//) {
 496            $decompose_compat[$code] = 1;
 497         } else {
 498            $decompose_compat[$code] = 0;
 499
 500            if (!exists $composition_exclusions{$code}) {
 501                $compositions{$code} = $fields[$DECOMPOSITION];
 502            }
 503         }
 504         $decompositions[$code] = $fields[$DECOMPOSITION];
 505     }
 506 }
 507
 508 sub print_tables
 509 {
 510     my ($last) = @_;
 511     my ($outfile) = "gunichartables.h";
 512
 513     local ($bytes_out) = 0;
 514
 515     print "Writing $outfile...\n";
 516
 517     open (OUT, "> $outfile");
 518
 519     print OUT "/* This file is automatically generated.  DO NOT EDIT!\n";
 520     print OUT "   Instead, edit gen-unicode-tables.pl and re-run.  */\n\n";
 521
 522     print OUT "#ifndef CHARTABLES_H\n";
 523     print OUT "#define CHARTABLES_H\n\n";
 524
 525     print OUT "#define G_UNICODE_DATA_VERSION \"$ARGV[0]\"\n\n";
 526
 527     printf OUT "#define G_UNICODE_LAST_CHAR 0x%04x\n\n", $last;
 528
 529     printf OUT "#define G_UNICODE_MAX_TABLE_INDEX 1000\n\n";
 530
 531     $table_index = 0;
 532     printf OUT "static const char type_data[][256] = {\n";
 533     for ($count = 0; $count <= $last; $count += 256)
 534     {
 535         $row[$count / 256] = &print_row ($count, 1, \&fetch_type);
 536     }
 537     printf OUT "\n};\n\n";
 538
 539     print OUT "static const short type_table[256] = {\n";
 540     for ($count = 0; $count <= $last; $count += 256)
 541     {
 542         print OUT ",\n" if $count > 0;
 543         print OUT "  ", $row[$count / 256];
 544         $bytes_out += 2;
 545     }
 546     print OUT "\n};\n\n";
 547
 548
 549     #
 550     # Now print attribute table.
 551     #
 552
 553     $table_index = 0;
 554     printf OUT "static const unsigned short attr_data[][256] = {\n";
 555     for ($count = 0; $count <= $last; $count += 256)
 556     {
 557         $row[$count / 256] = &print_row ($count, 2, \&fetch_attr);
 558     }
 559     printf OUT "\n};\n\n";
 560
 561     print OUT "static const short attr_table[256] = {\n";
 562     for ($count = 0; $count <= $last; $count += 256)
 563     {
 564         print OUT ",\n" if $count > 0;
 565         print OUT "  ", $row[$count / 256];
 566         $bytes_out += 2;
 567     }
 568     print OUT "\n};\n\n";
 569
 570     #
 571     # print title case table
 572     #
 573
 574     # FIXME: type.
 575     print OUT "static const unsigned short title_table[][3] = {\n";
 576     my ($item);
 577     my ($first) = 1;
 578     foreach $item (sort keys %title_to_lower)
 579     {
 580         print OUT ",\n"
 581             unless $first;
 582         $first = 0;
 583         printf OUT "  { 0x%04x, 0x%04x, 0x%04x }", $item, $title_to_upper{$item}, $title_to_lower{$item};
 584         $bytes_out += 6;
 585     }
 586     print OUT "\n};\n\n";
 587
 588     #
 589     # And special case conversion table -- conversions that change length
 590     #
 591     &output_special_case_table (\*OUT);
 592     &output_casefold_table (\*OUT);
 593
 594     print OUT "#endif /* CHARTABLES_H */\n";
 595
 596     close (OUT);
 597
 598     printf STDERR "Generated %d bytes in tables\n", $bytes_out;
 599 }
 600
 601 # A fetch function for the type table.
 602 sub fetch_type
 603 {
 604     my ($index) = @_;
 605     return $mappings{$type[$index]};
 606 }
 607
 608 # A fetch function for the attribute table.
 609 sub fetch_attr
 610 {
 611     my ($index) = @_;
 612     if (defined $value[$index])
 613       {
 614         return sprintf ("0x%04x", $value[$index]);
 615       }
 616     else
 617       {
 618         return "0x0000";
 619       }
 620 }
 621
 622 sub print_row
 623 {
 624     my ($start, $typsize, $fetcher) = @_;
 625
 626     my ($i);
 627     my (@values);
 628     my ($flag) = 1;
 629     my ($off);
 630
 631     for ($off = 0; $off < 256; ++$off)
 632     {
 633         $values[$off] = $fetcher->($off + $start);
 634         if ($values[$off] ne $values[0])
 635         {
 636             $flag = 0;
 637         }
 638     }
 639     if ($flag)
 640     {
 641         return $values[0] . " + G_UNICODE_MAX_TABLE_INDEX";
 642     }
 643
 644     printf OUT ",\n" if ($table_index != 0);
 645     printf OUT "  { /* page %d, index %d */\n    ", $start / 256, $table_index;
 646     my ($column) = 4;
 647     for ($i = $start; $i < $start + 256; ++$i)
 648     {
 649         print OUT ", "
 650             if $i > $start;
 651         my ($text) = $values[$i - $start];
 652         if (length ($text) + $column + 2 > 78)
 653         {
 654             print OUT "\n    ";
 655             $column = 4;
 656         }
 657         print OUT $text;
 658         $column += length ($text) + 2;
 659     }
 660     print OUT "\n  }";
 661
 662     $bytes_out += 256 * $typsize;
 663
 664     return sprintf "%d /* page %d */", $table_index++, $start / 256;
 665 }
 666
 667 # Generate the character decomposition header.
 668 sub print_decomp
 669 {
 670     my ($last) = @_;
 671     my ($outfile) = "gunidecomp.h";
 672
 673     local ($bytes_out) = 0;
 674
 675     print "Writing $outfile...\n";
 676
 677     open (OUT, "> $outfile") || exit 1;
 678
 679     print OUT "/* This file is automatically generated.  DO NOT EDIT! */\n\n";
 680     print OUT "#ifndef DECOMP_H\n";
 681     print OUT "#define DECOMP_H\n\n";
 682
 683     printf OUT "#define G_UNICODE_LAST_CHAR 0x%04x\n\n", $last;
 684
 685     printf OUT "#define G_UNICODE_MAX_TABLE_INDEX 1000\n\n";
 686
 687     my ($count, @row);
 688     $table_index = 0;
 689     printf OUT "static const unsigned char cclass_data[][256] = {\n";
 690     for ($count = 0; $count <= $last; $count += 256)
 691     {
 692         $row[$count / 256] = &print_row ($count, 1, \&fetch_cclass);
 693     }
 694     printf OUT "\n};\n\n";
 695
 696     print OUT "static const short combining_class_table[256] = {\n";
 697     for ($count = 0; $count <= $last; $count += 256)
 698     {
 699         print OUT ",\n" if $count > 0;
 700         print OUT "  ", $row[$count / 256];
 701         $bytes_out += 2;
 702     }
 703     print OUT "\n};\n\n";
 704
 705     print OUT "typedef struct\n{\n";
 706     # FIXME: type.
 707     print OUT "  unsigned short ch;\n";
 708     print OUT "  unsigned char canon_offset;\n";
 709     print OUT "  unsigned char compat_offset;\n";
 710     print OUT "  unsigned short expansion_offset;\n";
 711     print OUT "} decomposition;\n\n";
 712
 713     print OUT "static const decomposition decomp_table[] =\n{\n";
 714     my ($iter);
 715     my ($first) = 1;
 716     my ($decomp_string) = "";
 717     my ($decomp_string_offset) = 0;
 718     for ($count = 0; $count <= $last; ++$count)
 719     {
 720         if (defined $decompositions[$count])
 721         {
 722             print OUT ",\n"
 723                 if ! $first;
 724             $first = 0;
 725
 726             my $canon_decomp;
 727             my $compat_decomp;
 728
 729             if (!$decompose_compat[$count]) {
 730                 $canon_decomp = make_decomp ($count, 0);
 731             }
 732             $compat_decomp = make_decomp ($count, 1);
 733
 734             if (defined $canon_decomp && $compat_decomp eq $canon_decomp) {
 735                 undef $compat_decomp;
 736             }
 737
 738             my $string = "";
 739             my $canon_offset = 0xff;
 740             my $compat_offset = 0xff;
 741
 742             if (defined $canon_decomp) {
 743                 $canon_offset = 0;
 744                 $string .= $canon_decomp;
 745             }
 746             if (defined $compat_decomp) {
 747                 if (defined $canon_decomp) {
 748                     $string .= "\\x00\\x00";
 749                 }
 750                 $compat_offset = (length $string) / 4;
 751                 $string .= $compat_decomp;
 752             }
 753
 754             if (!defined($decomp_offsets{$string})) {
 755                 $decomp_offsets{$string} = $decomp_string_offset;
 756                 $decomp_string .= "\n  \"".$string."\\0\\0\" /* offset ".
 757                     $decomp_string_offset." */";
 758                 $decomp_string_offset += ((length $string) / 4) + 2;
 759
 760                 $bytes_out += (length $string) / 4 + 2; # "\x20"
 761             }
 762
 763             printf OUT qq(  { 0x%04x, %u, %u, %d }),
 764                 $count, $canon_offset, $compat_offset, $decomp_offsets{$string};
 765             $bytes_out += 6;
 766
 767         }
 768     }
 769     print OUT "\n};\n\n";
 770
 771     printf OUT "static const unsigned char decomp_expansion_string[] = %s;\n\n", $decomp_string;
 772
 773     print OUT "#endif /* DECOMP_H */\n";
 774
 775     printf STDERR "Generated %d bytes in decomp tables\n", $bytes_out;
 776 }
 777
 778 sub print_line_break
 779 {
 780     my ($last) = @_;
 781     my ($outfile) = "gunibreak.h";
 782
 783     local ($bytes_out) = 0;
 784
 785     print "Writing $outfile...\n";
 786
 787     open (OUT, "> $outfile");
 788
 789     print OUT "/* This file is automatically generated.  DO NOT EDIT!\n";
 790     print OUT "   Instead, edit gen-unicode-tables.pl and re-run.  */\n\n";
 791
 792     print OUT "#ifndef BREAKTABLES_H\n";
 793     print OUT "#define BREAKTABLES_H\n\n";
 794
 795     print OUT "#define G_UNICODE_DATA_VERSION \"$ARGV[0]\"\n\n";
 796
 797     printf OUT "#define G_UNICODE_LAST_CHAR 0x%04x\n\n", $last;
 798
 799     printf OUT "#define G_UNICODE_MAX_TABLE_INDEX 1000\n\n";
 800
 801     $table_index = 0;
 802     printf OUT "static const char break_property_data[][256] = {\n";
 803     for ($count = 0; $count <= $last; $count += 256)
 804     {
 805         $row[$count / 256] = &print_row ($count, 1, \&fetch_break_type);
 806     }
 807     printf OUT "\n};\n\n";
 808
 809     print OUT "static const short break_property_table[256] = {\n";
 810     for ($count = 0; $count <= $last; $count += 256)
 811     {
 812         print OUT ",\n" if $count > 0;
 813         print OUT "  ", $row[$count / 256];
 814         $bytes_out += 2;
 815     }
 816     print OUT "\n};\n\n";
 817
 818     print OUT "#endif /* BREAKTABLES_H */\n";
 819
 820     close (OUT);
 821
 822     printf STDERR "Generated %d bytes in break tables\n", $bytes_out;
 823 }
 824
 825
 826 # A fetch function for the break properties table.
 827 sub fetch_break_type
 828 {
 829     my ($index) = @_;
 830     return $break_mappings{$break_props[$index]};
 831 }
 832
 833 # Fetcher for combining class.
 834 sub fetch_cclass
 835 {
 836     my ($i) = @_;
 837     return $cclass[$i];
 838 }
 839
 840 # Expand a character decomposition recursively.
 841 sub expand_decomp
 842 {
 843     my ($code, $compat) = @_;
 844
 845     my ($iter, $val);
 846     my (@result) = ();
 847     foreach $iter (split (' ', $decompositions[$code]))
 848     {
 849         $val = hex ($iter);
 850         if (defined $decompositions[$val] &&
 851             ($compat || !$decompose_compat[$val]))
 852         {
 853             push (@result, &expand_decomp ($val, $compat));
 854         }
 855         else
 856         {
 857             push (@result, $val);
 858         }
 859     }
 860
 861     return @result;
 862 }
 863
 864 sub make_decomp
 865 {
 866     my ($code, $compat) = @_;
 867
 868     my $result = "";
 869     foreach $iter (&expand_decomp ($code, $compat))
 870     {
 871         $result .= sprintf "\\x%02x\\x%02x", $iter / 256, $iter & 0xff;
 872     }
 873
 874     $result;
 875 }
 876 # Generate special case data string from two fields
 877 sub add_special_case
 878 {
 879     my ($code, $single, $field1, $field2) = @_;
 880
 881     @values = (defined $single ? $single : (),
 882                (map { hex ($_) } split /\s+/, $field1),
 883                0,
 884                (map { hex ($_) } split /\s+/, $field2));
 885     $result = "";
 886
 887
 888     for $value (@values) {
 889         $result .= sprintf ("\\x%02x\\x%02x", $value / 256, $value & 0xff);
 890     }
 891
 892     $result .= "\\0";
 893
 894     if (2 * @values + 2 > $special_case_len) {
 895         $special_case_len = 2 * @values + 2;
 896     }
 897
 898     push @special_cases, $result;
 899
 900     #
 901     # We encode special cases in the surrogate pair space
 902     #
 903     $value[$code] = 0xD800 + scalar(@special_cases) - 1;
 904 }
 905
 906 sub output_special_case_table
 907 {
 908     my $out = shift;
 909
 910     print $out <<EOT;
 911
 912 /* Table of special cases for case conversion; each record contains
 913  * First, the best single character mapping to lowercase if Lu,
 914  * and to uppercase if Ll, followed by the output mapping for the two cases
 915  * other than the case of the codepoint, in the order [Ll],[Lu],[Lt],
 916  * separated and terminated by a double NUL.
 917  */
 918 static const unsigned char special_case_table[][$special_case_len] = {
 919 EOT
 920
 921     for $case (@special_cases) {
 922         print $out qq( "$case",\n);
 923     }
 924
 925     print $out <<EOT;
 926 };
 927
 928 EOT
 929
 930     print STDERR "Generated ", ($special_case_len * scalar @special_cases), " bytes in special case table\n";
 931 }
 932
 933 sub enumerate_ordered
 934 {
 935     my ($array) = @_;
 936
 937     my $n = 0;
 938     for my $code (sort { $a <=> $b } keys %$array) {
 939         if ($array->{$code} == 1) {
 940             delete $array->{$code};
 941             next;
 942         }
 943         $array->{$code} = $n++;
 944     }
 945
 946     return $n;
 947 }
 948
 949 sub output_composition_table
 950 {
 951     print STDERR "Generating composition table\n";
 952
 953     local ($bytes_out) = 0;
 954
 955     my %first;
 956     my %second;
 957
 958     # First we need to go through and remove decompositions
 959     # starting with a non-starter, and single-character
 960     # decompositions. At the same time, record
 961     # the first and second character of each decomposition
 962
 963     for $code (keys %compositions) {
 964         @values = map { hex ($_) } split /\s+/, $compositions{$code};
 965         if ($cclass[$values[0]]) {
 966             delete $compositions{$code};
 967             next;
 968         }
 969         if (@values == 1) {
 970             delete $compositions{$code};
 971             next;
 972         }
 973         if (@values != 2) {
 974             die "$code has more than two elements in its decomposition!\n";
 975         }
 976
 977         if (exists $first{$values[0]}) {
 978             $first{$values[0]}++;
 979         } else {
 980             $first{$values[0]} = 1;
 981         }
 982     }
 983
 984     # Assign integer indicices, removing singletons
 985     my $n_first = enumerate_ordered (\%first);
 986
 987     # Now record the second character if each (non-singleton) decomposition
 988     for $code (keys %compositions) {
 989         @values = map { hex ($_) } split /\s+/, $compositions{$code};
 990
 991         if (exists $first{$values[0]}) {
 992             if (exists $second{$values[1]}) {
 993                 $second{$values[1]}++;
 994             } else {
 995                 $second{$values[1]} = 1;
 996             }
 997         }
 998     }
 999
1000     # Assign integer indices, removing duplicate
1001     my $n_second = enumerate_ordered (\%second);
1002
1003     # Build reverse table
1004
1005     my @first_singletons;
1006     my @second_singletons;
1007     my %reverse;
1008     for $code (keys %compositions) {
1009         @values = map { hex ($_) } split /\s+/, $compositions{$code};
1010
1011         my $first = $first{$values[0]};
1012         my $second = $second{$values[1]};
1013
1014         if (defined $first && defined $second) {
1015             $reverse{"$first|$second"} = $code;
1016         } elsif (!defined $first) {
1017             push @first_singletons, [ $values[0], $values[1], $code ];
1018         } else {
1019             push @second_singletons, [ $values[1], $values[0], $code ];
1020         }
1021     }
1022
1023     @first_singletons = sort { $a->[0] <=> $b->[0] } @first_singletons;
1024     @second_singletons = sort { $a->[0] <=> $b->[0] } @second_singletons;
1025
1026     my %vals;
1027
1028     open OUT, ">gunicomp.h" or die "Cannot open gunicomp.h: $!\n";
1029
1030     # Assign values in lookup table for all code points involved
1031
1032     my $total = 1;
1033     my $last = 0;
1034     printf OUT "#define COMPOSE_FIRST_START %d\n", $total;
1035     for $code (keys %first) {
1036         $vals{$code} = $first{$code} + $total;
1037         $last = $code if $code > $last;
1038     }
1039     $total += $n_first;
1040     $i = 0;
1041     printf OUT "#define COMPOSE_FIRST_SINGLE_START %d\n", $total;
1042     for $record (@first_singletons) {
1043         my $code = $record->[0];
1044         $vals{$code} = $i++ + $total;
1045         $last = $code if $code > $last;
1046     }
1047     $total += @first_singletons;
1048     printf OUT "#define COMPOSE_SECOND_START %d\n", $total;
1049     for $code (keys %second) {
1050         $vals{$code} = $second{$code} + $total;
1051         $last = $code if $code > $last;
1052     }
1053     $total += $n_second;
1054     $i = 0;
1055     printf OUT "#define COMPOSE_SECOND_SINGLE_START %d\n\n", $total;
1056     for $record (@second_singletons) {
1057         my $code = $record->[0];
1058         $vals{$code} = $i++ + $total;
1059         $last = $code if $code > $last;
1060     }
1061
1062     # Output lookup table
1063
1064     my @row;
1065     $table_index = 0;
1066     printf OUT "static const int compose_data[][256] = {\n";
1067     for (my $count = 0; $count <= $last; $count += 256)
1068     {
1069         $row[$count / 256] = &print_row ($count, 2, sub { exists $vals{$_[0]} ? $vals{$_[0]} : 0; });
1070     }
1071     printf OUT "\n};\n\n";
1072
1073     print OUT "static const short compose_table[256] = {\n";
1074     for (my $count = 0; $count <= $last; $count += 256)
1075     {
1076         print OUT ",\n" if $count > 0;
1077         print OUT "  ", $row[$count / 256];
1078         $bytes_out += 4;
1079     }
1080     print OUT "\n};\n\n";
1081
1082     # Output first singletons
1083
1084     print OUT "static const int compose_first_single[][2] = {\n";
1085     $i = 0;
1086     for $record (@first_singletons) {
1087         print OUT ",\n" if $i++ > 0;
1088         printf OUT " { %#06x, %#06x }", $record->[1], $record->[2];
1089     }
1090     print OUT "\n};\n";
1091
1092     $bytes_out += @first_singletons * 4;
1093
1094     # Output second singletons
1095
1096     print OUT "static const int compose_second_single[][2] = {\n";
1097     $i = 0;
1098     for $record (@second_singletons) {
1099         print OUT ",\n" if $i++ > 0;
1100         printf OUT " { %#06x, %#06x }", $record->[1], $record->[2];
1101     }
1102     print OUT "\n};\n";
1103
1104     $bytes_out += @second_singletons * 4;
1105
1106     # Output array of composition pairs
1107
1108     print OUT <<EOT;
1109 static const int compose_array[$n_first][$n_second] = {
1110 EOT
1111
1112     for (my $i = 0; $i < $n_first; $i++) {
1113         print OUT ",\n" if $i;
1114         print OUT " { ";
1115         for (my $j = 0; $j < $n_second; $j++) {
1116             print OUT ", " if $j;
1117             if (exists $reverse{"$i|$j"}) {
1118                 printf OUT "%#06x", $reverse{"$i|$j"};
1119             } else {
1120                 print OUT "     0";
1121             }
1122         }
1123         print OUT " }";
1124     }
1125     print OUT "\n";
1126
1127     print OUT <<EOT;
1128 };
1129 EOT
1130
1131     $bytes_out += $n_first * $n_second * 2;
1132
1133     printf STDERR "Generated %d bytes in compose tables\n", $bytes_out;
1134 }
1135
1136 sub output_casefold_table
1137 {
1138     my $out = shift;
1139
1140     print $out <<EOT;
1141
1142 /* Table of casefolding cases that can't be derived by lowercasing
1143  */
1144 static const struct {
1145   guint16 ch;
1146   gchar data[$casefoldlen];
1147 } casefold_table[] = {
1148 EOT
1149
1150    @casefold = sort { $a->[0] <=> $b->[0] } @casefold;
1151
1152    for $case (@casefold) {
1153        $code = $case->[0];
1154        $string = $case->[1];
1155        print $out sprintf(qq({ %#04x, "$string" },\n), $code);
1156
1157    }
1158
1159     print $out <<EOT;
1160 };
1161
1162 EOT
1163
1164    my $recordlen = (2+$casefoldlen+1) & ~1;
1165    printf "Generated %d bytes for casefold table\n", $recordlen * @casefold;
1166 }
1167
1168
1169