tools/tools/locale/tools/cldr2def.pl

   1 #!/usr/local/bin/perl -wC
   2
   3 # SPDX-License-Identifier: BSD-2-Clause-FreeBSD
   4 #
   5 # Copyright 2009 Edwin Groothuis <edwin@FreeBSD.org>
   6 # Copyright 2015 John Marino <draco@marino.st>
   7 #
   8 # Redistribution and use in source and binary forms, with or without
   9 # modification, are permitted provided that the following conditions
  10 # are met:
  11 # 1. Redistributions of source code must retain the above copyright
  12 #    notice, this list of conditions and the following disclaimer.
  13 # 2. Redistributions in binary form must reproduce the above copyright
  14 #    notice, this list of conditions and the following disclaimer in the
  15 #    documentation and/or other materials provided with the distribution.
  16 #
  17 # THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  18 # ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19 # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20 # ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  21 # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22 # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23 # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24 # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25 # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26 # OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27 # SUCH DAMAGE.
  28 #
  29 # $FreeBSD$
  30
  31 use strict;
  32 use File::Copy;
  33 use XML::Parser;
  34 use Tie::IxHash;
  35 use Text::Iconv;
  36 #use Data::Dumper;
  37 use Getopt::Long;
  38 use Digest::SHA qw(sha1_hex);
  39 require "charmaps.pm";
  40
  41
  42 if ($#ARGV < 2) {
  43         print "Usage: $0 --unidir=<unidir> --etc=<etcdir> --type=<type>\n";
  44         exit(1);
  45 }
  46
  47 my $DEFENCODING = "UTF-8";
  48
  49 my $UNIDIR = undef;
  50 my $ETCDIR = undef;
  51 my $TYPE = undef;
  52
  53 my $result = GetOptions (
  54                 "unidir=s"      => \$UNIDIR,
  55                 "etc=s"         => \$ETCDIR,
  56                 "type=s"        => \$TYPE,
  57             );
  58
  59 my %convertors = ();
  60
  61 my %ucd = ();
  62 my %values = ();
  63 my %hashtable = ();
  64 my %languages = ();
  65 my %translations = ();
  66 my %encodings = ();
  67 my %alternativemonths = ();
  68 get_languages();
  69
  70 my %utf8map = ();
  71 my %utf8aliases = ();
  72 get_unidata($UNIDIR);
  73 get_utf8map("$UNIDIR/posix/$DEFENCODING.cm");
  74 get_encodings("$ETCDIR/charmaps");
  75
  76 my %keys = ();
  77 tie(%keys, "Tie::IxHash");
  78 tie(%hashtable, "Tie::IxHash");
  79
  80 my %FILESNAMES = (
  81         "monetdef"      => "LC_MONETARY",
  82         "timedef"       => "LC_TIME",
  83         "msgdef"        => "LC_MESSAGES",
  84         "numericdef"    => "LC_NUMERIC",
  85         "colldef"       => "LC_COLLATE",
  86         "ctypedef"      => "LC_CTYPE"
  87 );
  88
  89 my %callback = (
  90         mdorder => \&callback_mdorder,
  91         altmon => \&callback_altmon,
  92         cformat => \&callback_cformat,
  93         dformat => \&callback_dformat,
  94         dtformat => \&callback_dtformat,
  95         cbabmon => \&callback_abmon,
  96         cbampm => \&callback_ampm,
  97         data => undef,
  98 );
  99
 100 my %DESC = (
 101
 102         # numericdef
 103         "decimal_point" => "decimal_point",
 104         "thousands_sep" => "thousands_sep",
 105         "grouping"      => "grouping",
 106
 107         # monetdef
 108         "int_curr_symbol"       => "int_curr_symbol (last character always " .
 109                                    "SPACE)",
 110         "currency_symbol"       => "currency_symbol",
 111         "mon_decimal_point"     => "mon_decimal_point",
 112         "mon_thousands_sep"     => "mon_thousands_sep",
 113         "mon_grouping"          => "mon_grouping",
 114         "positive_sign"         => "positive_sign",
 115         "negative_sign"         => "negative_sign",
 116         "int_frac_digits"       => "int_frac_digits",
 117         "frac_digits"           => "frac_digits",
 118         "p_cs_precedes"         => "p_cs_precedes",
 119         "p_sep_by_space"        => "p_sep_by_space",
 120         "n_cs_precedes"         => "n_cs_precedes",
 121         "n_sep_by_space"        => "n_sep_by_space",
 122         "p_sign_posn"           => "p_sign_posn",
 123         "n_sign_posn"           => "n_sign_posn",
 124
 125         # msgdef
 126         "yesexpr"       => "yesexpr",
 127         "noexpr"        => "noexpr",
 128         "yesstr"        => "yesstr",
 129         "nostr"         => "nostr",
 130
 131         # timedef
 132         "abmon"         => "Short month names",
 133         "mon"           => "Long month names (as in a date)",
 134         "abday"         => "Short weekday names",
 135         "day"           => "Long weekday names",
 136         "t_fmt"         => "X_fmt",
 137         "d_fmt"         => "x_fmt",
 138         "c_fmt"         => "c_fmt",
 139         "am_pm"         => "AM/PM",
 140         "d_t_fmt"       => "date_fmt",
 141         "altmon"        => "Long month names (without case ending)",
 142         "md_order"      => "md_order",
 143         "t_fmt_ampm"    => "ampm_fmt",
 144 );
 145
 146 if ($TYPE eq "colldef") {
 147         transform_collation();
 148         make_makefile();
 149 }
 150
 151 if ($TYPE eq "ctypedef") {
 152         transform_ctypes();
 153         make_makefile();
 154 }
 155
 156 if ($TYPE eq "numericdef") {
 157         %keys = (
 158             "decimal_point"     => "s",
 159             "thousands_sep"     => "s",
 160             "grouping"          => "ai",
 161         );
 162         get_fields();
 163         print_fields();
 164         make_makefile();
 165 }
 166
 167 if ($TYPE eq "monetdef") {
 168         %keys = (
 169             "int_curr_symbol"   => "s",
 170             "currency_symbol"   => "s",
 171             "mon_decimal_point" => "s",
 172             "mon_thousands_sep" => "s",
 173             "mon_grouping"      => "ai",
 174             "positive_sign"     => "s",
 175             "negative_sign"     => "s",
 176             "int_frac_digits"   => "i",
 177             "frac_digits"       => "i",
 178             "p_cs_precedes"     => "i",
 179             "p_sep_by_space"    => "i",
 180             "n_cs_precedes"     => "i",
 181             "n_sep_by_space"    => "i",
 182             "p_sign_posn"       => "i",
 183             "n_sign_posn"       => "i"
 184         );
 185         get_fields();
 186         print_fields();
 187         make_makefile();
 188 }
 189
 190 if ($TYPE eq "msgdef") {
 191         %keys = (
 192             "yesexpr"           => "s",
 193             "noexpr"            => "s",
 194             "yesstr"            => "s",
 195             "nostr"             => "s"
 196         );
 197         get_fields();
 198         print_fields();
 199         make_makefile();
 200 }
 201
 202 if ($TYPE eq "timedef") {
 203         %keys = (
 204             "abmon"             => "<cbabmon<abmon<as",
 205             "mon"               => "as",
 206             "abday"             => "as",
 207             "day"               => "as",
 208             "t_fmt"             => "s",
 209             "d_fmt"             => "<dformat<d_fmt<s",
 210             "c_fmt"             => "<cformat<d_t_fmt<s",
 211             "am_pm"             => "<cbampm<am_pm<as",
 212             "d_t_fmt"           => "<dtformat<d_t_fmt<s",
 213             "altmon"            => "<altmon<mon<as",
 214             "md_order"          => "<mdorder<d_fmt<s",
 215             "t_fmt_ampm"        => "s",
 216         );
 217         get_fields();
 218         print_fields();
 219         make_makefile();
 220 }
 221
 222 sub callback_ampm {
 223         my $s = shift;
 224         my $nl = $callback{data}{l} . "_" . $callback{data}{c};
 225         my $enc = $callback{data}{e};
 226
 227         if ($nl eq 'ru_RU') {
 228                 if ($enc eq 'UTF-8') {
 229                         $s = 'дп;пп';
 230                 } else {
 231                         my  $converter = Text::Iconv->new("utf-8", "$enc");
 232                         $s = $converter->convert("дп;пп");
 233                 }
 234         }
 235         return $s;
 236 }
 237
 238 sub callback_cformat {
 239         my $s = shift;
 240         my $nl = $callback{data}{l} . "_" . $callback{data}{c};
 241
 242         if ($nl eq 'ko_KR') {
 243                 $s =~ s/(> )(%p)/$1%A $2/;
 244         }
 245         $s =~ s/\.,/\./;
 246         $s =~ s/ %Z//;
 247         $s =~ s/ %z//;
 248         $s =~ s/^"%e\./%A %e/;
 249         $s =~ s/^"(%B %e, )/"%A, $1/;
 250         $s =~ s/^"(%e %B )/"%A $1/;
 251         return $s;
 252 };
 253
 254 sub callback_dformat {
 255         my $s = shift;
 256
 257         $s =~ s/(%m(<SOLIDUS>|[-.]))%e/$1%d/;
 258         $s =~ s/%e((<SOLIDUS>|[-.])%m)/%d$1/;
 259         return $s;
 260 };
 261
 262 sub callback_dtformat {
 263         my $s = shift;
 264         my $nl = $callback{data}{l} . "_" . $callback{data}{c};
 265
 266         if ($nl eq 'ja_JP') {
 267                 $s =~ s/(> )(%H)/$1%A $2/;
 268         } elsif ($nl eq 'ko_KR' || $nl eq 'zh_CN' || $nl eq 'zh_TW') {
 269                 if ($nl ne 'ko_KR') {
 270                         $s =~ s/%m/%_m/;
 271                 }
 272                 $s =~ s/(> )(%p)/$1%A $2/;
 273         }
 274         $s =~ s/\.,/\./;
 275         $s =~ s/^"%e\./%A %e/;
 276         $s =~ s/^"(%B %e, )/"%A, $1/;
 277         $s =~ s/^"(%e %B )/"%A $1/;
 278         return $s;
 279 };
 280
 281 sub callback_mdorder {
 282         my $s = shift;
 283         return undef if (!defined $s);
 284         $s =~ s/[^dem]//g;
 285         $s =~ s/e/d/g;
 286         return $s;
 287 };
 288
 289 sub callback_altmon {
 290         # if the language/country is known in %alternative months then
 291         # return that, otherwise repeat mon
 292         my $s = shift;
 293
 294         if (defined $alternativemonths{$callback{data}{l}}{$callback{data}{c}}) {
 295                 my @altnames = split(";",$alternativemonths{$callback{data}{l}}{$callback{data}{c}});
 296                 my @cleaned;
 297                 foreach (@altnames)
 298                 {
 299                         $_ =~ s/^\s+//;
 300                         $_ =~ s/\s+$//;
 301                         push @cleaned, $_;
 302                 }
 303                 return join(";",@cleaned);
 304         }
 305
 306         return $s;
 307 }
 308
 309 sub callback_abmon {
 310         # for specified CJK locales, pad result with a space to enable
 311         # columns to line up (style established in FreeBSD in 2001)
 312         my $s = shift;
 313         my $nl = $callback{data}{l} . "_" . $callback{data}{c};
 314
 315         if ($nl eq 'ja_JP' || $nl eq 'ko_KR' || $nl eq 'zh_CN' ||
 316             $nl eq 'zh_HK' || $nl eq 'zh_TW') {
 317                 my @monthnames = split(";", $s);
 318                 my @cleaned;
 319                 foreach (@monthnames)
 320                 {
 321                         if ($_ =~ /^"<(two|three|four|five|six|seven|eight|nine)>/ ||
 322                            ($_ =~ /^"<one>/ && $_ !~ /^"<one>(<zero>|<one>|<two>)/))
 323                         {
 324                                 $_ =~ s/^"/"<space>/;
 325                         }
 326                         push @cleaned, $_;
 327                 }
 328                 return join(";",@cleaned);
 329         }
 330         return $s;
 331 }
 332
 333 ############################
 334
 335 sub get_unidata {
 336         my $directory = shift;
 337
 338         open(FIN, "$directory/UnicodeData.txt")
 339             or die("Cannot open $directory/UnicodeData.txt");;
 340         my @lines = <FIN>;
 341         chomp(@lines);
 342         close(FIN);
 343
 344         foreach my $l (@lines) {
 345                 my @a = split(/;/, $l);
 346
 347                 $ucd{code2name}{"$a[0]"} = $a[1];       # Unicode name
 348                 $ucd{name2code}{"$a[1]"} = $a[0];       # Unicode code
 349         }
 350 }
 351
 352 sub get_utf8map {
 353         my $file = shift;
 354
 355         open(FIN, $file);
 356         my @lines = <FIN>;
 357         close(FIN);
 358         chomp(@lines);
 359
 360         my $prev_k = undef;
 361         my $prev_v = "";
 362         my $incharmap = 0;
 363         foreach my $l (@lines) {
 364                 $l =~ s/\r//;
 365                 next if ($l =~ /^\#/);
 366                 next if ($l eq "");
 367
 368                 if ($l eq "CHARMAP") {
 369                         $incharmap = 1;
 370                         next;
 371                 }
 372
 373                 next if (!$incharmap);
 374                 last if ($l eq "END CHARMAP");
 375
 376                 $l =~ /^<([^\s]+)>\s+(.*)/;
 377                 my $k = $1;
 378                 my $v = $2;
 379                 $k =~ s/_/ /g;          # unicode char string
 380                 $v =~ s/\\x//g;         # UTF-8 char code
 381                 $utf8map{$k} = $v;
 382
 383                 $utf8aliases{$k} = $prev_k if ($prev_v eq $v);
 384
 385                 $prev_v = $v;
 386                 $prev_k = $k;
 387         }
 388 }
 389
 390 sub get_encodings {
 391         my $dir = shift;
 392         foreach my $e (sort(keys(%encodings))) {
 393                 if (!open(FIN, "$dir/$e.TXT")) {
 394                         print "Cannot open charmap for $e\n";
 395                         next;
 396
 397                 }
 398                 $encodings{$e} = 1;
 399                 my @lines = <FIN>;
 400                 close(FIN);
 401                 chomp(@lines);
 402                 foreach my $l (@lines) {
 403                         $l =~ s/\r//;
 404                         next if ($l =~ /^\#/);
 405                         next if ($l eq "");
 406
 407                         my @a = split(" ", $l);
 408                         next if ($#a < 1);
 409                         $a[0] =~ s/^0[xX]//;    # local char code
 410                         $a[1] =~ s/^0[xX]//;    # unicode char code
 411                         $convertors{$e}{uc($a[1])} = uc($a[0]);
 412                 }
 413         }
 414 }
 415
 416 sub get_languages {
 417         my %data = get_xmldata($ETCDIR);
 418         %languages = %{$data{L}};
 419         %translations = %{$data{T}};
 420         %alternativemonths = %{$data{AM}};
 421         %encodings = %{$data{E}};
 422 }
 423
 424 sub transform_ctypes {
 425         # Add the C.UTF-8
 426         $languages{"C"}{"x"}{data}{"x"}{$DEFENCODING} = undef;
 427
 428         foreach my $l (sort keys(%languages)) {
 429         foreach my $f (sort keys(%{$languages{$l}})) {
 430         foreach my $c (sort keys(%{$languages{$l}{$f}{data}})) {
 431                 next if (defined $languages{$l}{$f}{definitions}
 432                     && $languages{$l}{$f}{definitions} !~ /$TYPE/);
 433                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 0; # unread
 434                 my $file = $l;
 435                 $file .= "_" . $f if ($f ne "x");
 436                 $file .= "_" . $c if ($c ne "x");
 437                 my $actfile = $file;
 438
 439                 my $filename = "$UNIDIR/posix/xx_Comm_C.UTF-8.src";
 440                 if (! -f $filename) {
 441                         print STDERR "Cannot open $filename\n";
 442                         next;
 443                 }
 444                 open(FIN, "$filename");
 445                 print "Reading from $filename for ${l}_${f}_${c}\n";
 446                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 1; # read
 447                 my @lines;
 448                 my $shex;
 449                 my $uhex;
 450                 while (<FIN>) {
 451                         push @lines, $_;
 452                 }
 453                 close(FIN);
 454                 $shex = sha1_hex(join("\n", @lines));
 455                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = $shex;
 456                 $hashtable{$shex}{"${l}_${f}_${c}.$DEFENCODING"} = 1;
 457                 open(FOUT, ">$TYPE.draft/$actfile.$DEFENCODING.src");
 458                 print FOUT @lines;
 459                 close(FOUT);
 460                 foreach my $enc (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
 461                         next if ($enc eq $DEFENCODING);
 462                         $filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
 463                         if (! -f $filename) {
 464                                 print STDERR "Cannot open $filename\n";
 465                                 next;
 466                         }
 467                         @lines = ();
 468                         open(FIN, "$filename");
 469                         while (<FIN>) {
 470                                 if ((/^comment_char\s/) || (/^escape_char\s/)){
 471                                         push @lines, $_;
 472                                 }
 473                                 if (/^LC_CTYPE/../^END LC_CTYPE/) {
 474                                         push @lines, $_;
 475                                 }
 476                         }
 477                         close(FIN);
 478                         $uhex = sha1_hex(join("\n", @lines) . $enc);
 479                         $languages{$l}{$f}{data}{$c}{$enc} = $uhex;
 480                         $hashtable{$uhex}{"${l}_${f}_${c}.$enc"} = 1;
 481                         open(FOUT, ">$TYPE.draft/$actfile.$enc.src");
 482                         print FOUT <<EOF;
 483 # Warning: Do not edit. This file is automatically extracted from the
 484 # tools in /usr/src/tools/tools/locale. The data is obtained from the
 485 # CLDR project, obtained from http://cldr.unicode.org/
 486 # -----------------------------------------------------------------------------
 487 EOF
 488                         print FOUT @lines;
 489                         close(FOUT);
 490                 }
 491         }
 492         }
 493         }
 494 }
 495
 496
 497 sub transform_collation {
 498         foreach my $l (sort keys(%languages)) {
 499         foreach my $f (sort keys(%{$languages{$l}})) {
 500         foreach my $c (sort keys(%{$languages{$l}{$f}{data}})) {
 501                 next if (defined $languages{$l}{$f}{definitions}
 502                     && $languages{$l}{$f}{definitions} !~ /$TYPE/);
 503                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 0; # unread
 504                 my $file;
 505                 $file = $l . "_";
 506                 $file .= $f . "_" if ($f ne "x");
 507                 $file .= $c;
 508                 my $actfile = $file;
 509
 510                 my $filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
 511                 $filename = "$ETCDIR/$file.$DEFENCODING.src"
 512                     if (! -f $filename);
 513                 if (! -f $filename
 514                  && defined $languages{$l}{$f}{fallback}) {
 515                         $file = $languages{$l}{$f}{fallback};
 516                         $filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
 517                 }
 518                 $filename = "$UNIDIR/posix/$file.$DEFENCODING.src"
 519                     if (! -f $filename);
 520                 if (! -f $filename) {
 521                         print STDERR
 522                             "Cannot open $file.$DEFENCODING.src or fallback\n";
 523                         next;
 524                 }
 525                 open(FIN, "$filename");
 526                 print "Reading from $filename for ${l}_${f}_${c}\n";
 527                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 1; # read
 528                 my @lines;
 529                 my $shex;
 530                 while (<FIN>) {
 531                         if ((/^comment_char\s/) || (/^escape_char\s/)){
 532                                 push @lines, $_;
 533                         }
 534                         if (/^LC_COLLATE/../^END LC_COLLATE/) {
 535                                 $_ =~ s/[ ]+/ /g;
 536                                 push @lines, $_;
 537                         }
 538                 }
 539                 close(FIN);
 540                 $shex = sha1_hex(join("\n", @lines));
 541                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = $shex;
 542                 $hashtable{$shex}{"${l}_${f}_${c}.$DEFENCODING"} = 1;
 543                 open(FOUT, ">$TYPE.draft/$actfile.$DEFENCODING.src");
 544                 print FOUT <<EOF;
 545 # Warning: Do not edit. This file is automatically extracted from the
 546 # tools in /usr/src/tools/tools/locale. The data is obtained from the
 547 # CLDR project, obtained from http://cldr.unicode.org/
 548 # -----------------------------------------------------------------------------
 549 EOF
 550                 print FOUT @lines;
 551                 close(FOUT);
 552
 553                 foreach my $enc (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
 554                         next if ($enc eq $DEFENCODING);
 555                         copy ("$TYPE.draft/$actfile.$DEFENCODING.src",
 556                               "$TYPE.draft/$actfile.$enc.src");
 557                         $languages{$l}{$f}{data}{$c}{$enc} = $shex;
 558                         $hashtable{$shex}{"${l}_${f}_${c}.$enc"} = 1;
 559                 }
 560         }
 561         }
 562         }
 563 }
 564
 565 sub get_fields {
 566         foreach my $l (sort keys(%languages)) {
 567         foreach my $f (sort keys(%{$languages{$l}})) {
 568         foreach my $c (sort keys(%{$languages{$l}{$f}{data}})) {
 569                 next if (defined $languages{$l}{$f}{definitions}
 570                     && $languages{$l}{$f}{definitions} !~ /$TYPE/);
 571
 572                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 0; # unread
 573                 my $file;
 574                 $file = $l . "_";
 575                 $file .= $f . "_" if ($f ne "x");
 576                 $file .= $c;
 577
 578                 my $filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
 579                 $filename = "$ETCDIR/$file.$DEFENCODING.src"
 580                     if (! -f $filename);
 581                 if (! -f $filename
 582                  && defined $languages{$l}{$f}{fallback}) {
 583                         $file = $languages{$l}{$f}{fallback};
 584                         $filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
 585                 }
 586                 $filename = "$UNIDIR/posix/$file.$DEFENCODING.src"
 587                     if (! -f $filename);
 588                 if (! -f $filename) {
 589                         print STDERR
 590                             "Cannot open $file.$DEFENCODING.src or fallback\n";
 591                         next;
 592                 }
 593                 open(FIN, "$filename");
 594                 print "Reading from $filename for ${l}_${f}_${c}\n";
 595                 $languages{$l}{$f}{data}{$c}{$DEFENCODING} = 1; # read
 596                 my @lines = <FIN>;
 597                 chomp(@lines);
 598                 close(FIN);
 599                 my $continue = 0;
 600                 foreach my $k (keys(%keys)) {
 601                         foreach my $line (@lines) {
 602                                 $line =~ s/\r//;
 603                                 next if (!$continue && $line !~ /^$k\s/);
 604                                 if ($continue) {
 605                                         $line =~ s/^\s+//;
 606                                 } else {
 607                                         $line =~ s/^$k\s+//;
 608                                 }
 609
 610                                 $values{$l}{$f}{$c}{$k} = ""
 611                                         if (!defined $values{$l}{$f}{$c}{$k});
 612
 613                                 $continue = ($line =~ /\/$/);
 614                                 $line =~ s/\/$// if ($continue);
 615
 616                                 while ($line =~ /_/) {
 617                                         $line =~
 618                                             s/\<([^>_]+)_([^>]+)\>/<$1 $2>/;
 619                                 }
 620                                 die "_ in data - $line" if ($line =~ /_/);
 621                                 $values{$l}{$f}{$c}{$k} .= $line;
 622
 623                                 last if (!$continue);
 624                         }
 625                 }
 626         }
 627         }
 628         }
 629 }
 630
 631 sub decodecldr {
 632         my $e = shift;
 633         my $s = shift;
 634
 635         my $v = undef;
 636
 637         if ($e eq "UTF-8") {
 638                 #
 639                 # Conversion to UTF-8 can be done from the Unicode name to
 640                 # the UTF-8 character code.
 641                 #
 642                 $v = $utf8map{$s};
 643                 die "Cannot convert $s in $e (charmap)" if (!defined $v);
 644         } else {
 645                 #
 646                 # Conversion to these encodings can be done from the Unicode
 647                 # name to Unicode code to the encodings code.
 648                 #
 649                 my $ucc = undef;
 650                 $ucc = $ucd{name2code}{$s} if (defined $ucd{name2code}{$s});
 651                 $ucc = $ucd{name2code}{$utf8aliases{$s}}
 652                         if (!defined $ucc
 653                          && $utf8aliases{$s}
 654                          && defined $ucd{name2code}{$utf8aliases{$s}});
 655
 656                 if (!defined $ucc) {
 657                         if (defined $translations{$e}{$s}{hex}) {
 658                                 $v = $translations{$e}{$s}{hex};
 659                                 $ucc = 0;
 660                         } elsif (defined $translations{$e}{$s}{ucc}) {
 661                                 $ucc = $translations{$e}{$s}{ucc};
 662                         }
 663                 }
 664
 665                 die "Cannot convert $s in $e (ucd string)" if (!defined $ucc);
 666                 $v = $convertors{$e}{$ucc} if (!defined $v);
 667
 668                 $v = $translations{$e}{$s}{hex}
 669                         if (!defined $v && defined $translations{$e}{$s}{hex});
 670
 671                 if (!defined $v && defined $translations{$e}{$s}{unicode}) {
 672                         my $ucn = $translations{$e}{$s}{unicode};
 673                         $ucc = $ucd{name2code}{$ucn}
 674                                 if (defined $ucd{name2code}{$ucn});
 675                         $ucc = $ucd{name2code}{$utf8aliases{$ucn}}
 676                                 if (!defined $ucc
 677                                  && defined $ucd{name2code}{$utf8aliases{$ucn}});
 678                         $v = $convertors{$e}{$ucc};
 679                 }
 680
 681                 die "Cannot convert $s in $e (charmap)" if (!defined $v);
 682         }
 683
 684         return pack("C", hex($v)) if (length($v) == 2);
 685         return pack("CC", hex(substr($v, 0, 2)), hex(substr($v, 2, 2)))
 686                 if (length($v) == 4);
 687         return pack("CCC", hex(substr($v, 0, 2)), hex(substr($v, 2, 2)),
 688             hex(substr($v, 4, 2))) if (length($v) == 6);
 689         print STDERR "Cannot convert $e $s\n";
 690         return "length = " . length($v);
 691
 692 }
 693
 694 sub translate {
 695         my $enc = shift;
 696         my $v = shift;
 697
 698         return $translations{$enc}{$v} if (defined $translations{$enc}{$v});
 699         return undef;
 700 }
 701
 702 sub print_fields {
 703         foreach my $l (sort keys(%languages)) {
 704         foreach my $f (sort keys(%{$languages{$l}})) {
 705         foreach my $c (sort keys(%{$languages{$l}{$f}{data}})) {
 706                 next if (defined $languages{$l}{$f}{definitions}
 707                     && $languages{$l}{$f}{definitions} !~ /$TYPE/);
 708                 foreach my $enc (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
 709                         if ($languages{$l}{$f}{data}{$c}{$DEFENCODING} eq "0") {
 710                                 print "Skipping ${l}_" .
 711                                     ($f eq "x" ? "" : "${f}_") .
 712                                     "${c} - not read\n";
 713                                 next;
 714                         }
 715                         my $file = $l;
 716                         $file .= "_" . $f if ($f ne "x");
 717                         $file .= "_" . $c;
 718                         print "Writing to $file in $enc\n";
 719
 720                         if ($enc ne $DEFENCODING &&
 721                             !defined $convertors{$enc}) {
 722                                 print "Failed! Cannot convert to $enc.\n";
 723                                 next;
 724                         };
 725
 726                         open(FOUT, ">$TYPE.draft/$file.$enc.new");
 727                         my $okay = 1;
 728                         my $output = "";
 729                         print FOUT <<EOF;
 730 # Warning: Do not edit. This file is automatically generated from the
 731 # tools in /usr/src/tools/tools/locale. The data is obtained from the
 732 # CLDR project, obtained from http://cldr.unicode.org/
 733 # -----------------------------------------------------------------------------
 734 EOF
 735                         foreach my $k (keys(%keys)) {
 736                                 my $g = $keys{$k};
 737
 738                                 die("Unknown $k in \%DESC")
 739                                         if (!defined $DESC{$k});
 740
 741                                 $output .= "#\n# $DESC{$k}\n";
 742
 743                                 # Replace one row with another
 744                                 if ($g =~ /^>/) {
 745                                         $k = substr($g, 1);
 746                                         $g = $keys{$k};
 747                                 }
 748
 749                                 # Callback function
 750                                 if ($g =~ /^\</) {
 751                                         $callback{data}{c} = $c;
 752                                         $callback{data}{k} = $k;
 753                                         $callback{data}{f} = $f;
 754                                         $callback{data}{l} = $l;
 755                                         $callback{data}{e} = $enc;
 756                                         my @a = split(/\</, substr($g, 1));
 757                                         my $rv =
 758                                             &{$callback{$a[0]}}($values{$l}{$f}{$c}{$a[1]});
 759                                         $values{$l}{$f}{$c}{$k} = $rv;
 760                                         $g = $a[2];
 761                                         $callback{data} = ();
 762                                 }
 763
 764                                 my $v = $values{$l}{$f}{$c}{$k};
 765                                 $v = "undef" if (!defined $v);
 766
 767                                 if ($g eq "i") {
 768                                         $output .= "$v\n";
 769                                         next;
 770                                 }
 771                                 if ($g eq "ai") {
 772                                         $output .= "$v\n";
 773                                         next;
 774                                 }
 775                                 if ($g eq "s") {
 776                                         $v =~ s/^"//;
 777                                         $v =~ s/"$//;
 778                                         my $cm = "";
 779                                         while ($v =~ /^(.*?)<(.*?)>(.*)/) {
 780                                                 my $p1 = $1;
 781                                                 $cm = $2;
 782                                                 my $p3 = $3;
 783
 784                                                 my $rv = decodecldr($enc, $cm);
 785 #                                               $rv = translate($enc, $cm)
 786 #                                                       if (!defined $rv);
 787                                                 if (!defined $rv) {
 788                                                         print STDERR
 789 "Could not convert $k ($cm) from $DEFENCODING to $enc\n";
 790                                                         $okay = 0;
 791                                                         next;
 792                                                 }
 793
 794                                                 $v = $p1 . $rv . $p3;
 795                                         }
 796                                         $output .= "$v\n";
 797                                         next;
 798                                 }
 799                                 if ($g eq "as") {
 800                                         foreach my $v (split(/;/, $v)) {
 801                                                 $v =~ s/^"//;
 802                                                 $v =~ s/"$//;
 803                                                 my $cm = "";
 804                                                 while ($v =~ /^(.*?)<(.*?)>(.*)/) {
 805                                                         my $p1 = $1;
 806                                                         $cm = $2;
 807                                                         my $p3 = $3;
 808
 809                                                         my $rv =
 810                                                             decodecldr($enc,
 811                                                                 $cm);
 812 #                                                       $rv = translate($enc,
 813 #                                                           $cm)
 814 #                                                           if (!defined $rv);
 815                                                         if (!defined $rv) {
 816                                                                 print STDERR
 817 "Could not convert $k ($cm) from $DEFENCODING to $enc\n";
 818                                                                 $okay = 0;
 819                                                                 next;
 820                                                         }
 821
 822                                                         $v = $1 . $rv . $3;
 823                                                 }
 824                                                 $output .= "$v\n";
 825                                         }
 826                                         next;
 827                                 }
 828
 829                                 die("$k is '$g'");
 830
 831                         }
 832
 833                         $languages{$l}{$f}{data}{$c}{$enc} = sha1_hex($output);
 834                         $hashtable{sha1_hex($output)}{"${l}_${f}_${c}.$enc"} = 1;
 835                         print FOUT "$output# EOF\n";
 836                         close(FOUT);
 837
 838                         if ($okay) {
 839                                 rename("$TYPE.draft/$file.$enc.new",
 840                                     "$TYPE.draft/$file.$enc.src");
 841                         } else {
 842                                 rename("$TYPE.draft/$file.$enc.new",
 843                                     "$TYPE.draft/$file.$enc.failed");
 844                         }
 845                 }
 846         }
 847         }
 848         }
 849 }
 850
 851 sub make_makefile {
 852         print "Creating Makefile for $TYPE\n";
 853         my $SRCOUT;
 854         my $SRCOUT2;
 855         my $SRCOUT3 = "";
 856         my $SRCOUT4 = "";
 857         my $MAPLOC;
 858         if ($TYPE eq "colldef") {
 859                 $SRCOUT = "localedef \${LOCALEDEF_ENDIAN} -D -U " .
 860                         "-i \${.IMPSRC} \\\n" .
 861                         "\t-f \${MAPLOC}/map.\${.TARGET:T:R:E:C/@.*//} " .
 862                         "\${.OBJDIR}/\${.IMPSRC:T:R}";
 863                 $MAPLOC = "MAPLOC=\t\t\${.CURDIR}/../../tools/tools/" .
 864                                 "locale/etc/final-maps\n";
 865                 $SRCOUT2 = "LC_COLLATE";
 866                 $SRCOUT3 = "" .
 867                         ".for f t in \${LOCALES_MAPPED}\n" .
 868                         "FILES+=\t\$t.LC_COLLATE\n" .
 869                         "FILESDIR_\$t.LC_COLLATE=\t\${LOCALEDIR}/\$t\n" .
 870                         "\$t.LC_COLLATE: \${.CURDIR}/\$f.src\n" .
 871                         "\tlocaledef \${LOCALEDEF_ENDIAN} -D -U " .
 872                         "-i \${.ALLSRC} \\\n" .
 873                         "\t\t-f \${MAPLOC}/map.\${.TARGET:T:R:E:C/@.*//} \\\n" .
 874                         "\t\t\${.OBJDIR}/\${.TARGET:T:R}\n" .
 875                         ".endfor\n\n";
 876                 $SRCOUT4 = "## LOCALES_MAPPED\n";
 877         }
 878         elsif ($TYPE eq "ctypedef") {
 879                 $SRCOUT = "localedef \${LOCALEDEF_ENDIAN} -D -U -c " .
 880                         "-w \${MAPLOC}/widths.txt \\\n" .
 881                         "\t-f \${MAPLOC}/map.\${.IMPSRC:T:R:E} " .
 882                         "\\\n\t-i \${.IMPSRC} \${.OBJDIR}/\${.IMPSRC:T:R} " .
 883                         " || true";
 884                 $SRCOUT2 = "LC_CTYPE";
 885                 $MAPLOC = "MAPLOC=\t\t\${.CURDIR}/../../tools/tools/" .
 886                                 "locale/etc/final-maps\n";
 887                 $SRCOUT3 = "## SYMPAIRS\n\n" .
 888                         ".for s t in \${SYMPAIRS}\n" .
 889                         "\${t:S/src\$/LC_CTYPE/}: " .
 890                         "\$s\n" .
 891                         "\tlocaledef \${LOCALEDEF_ENDIAN} -D -U -c " .
 892                         "-w \${MAPLOC}/widths.txt \\\n" .
 893                         "\t-f \${MAPLOC}/map.\${.TARGET:T:R:C/^.*\\.//} " .
 894                         "\\\n\t-i \${.ALLSRC} \${.OBJDIR}/\${.TARGET:T:R} " .
 895                         " || true\n" .
 896                         ".endfor\n\n";
 897         }
 898         else {
 899                 $SRCOUT = "grep -v -E '^(\#\$\$|\#[ ])' < \${.IMPSRC} > \${.TARGET}";
 900                 $SRCOUT2 = "out";
 901                 $MAPLOC = "";
 902         }
 903         open(FOUT, ">$TYPE.draft/Makefile");
 904         print FOUT <<EOF;
 905 # \$FreeBSD\$
 906 # Warning: Do not edit. This file is automatically generated from the
 907 # tools in /usr/src/tools/tools/locale.
 908
 909 LOCALEDIR=      \${SHAREDIR}/locale
 910 FILESNAME=      $FILESNAMES{$TYPE}
 911 .SUFFIXES:      .src .${SRCOUT2}
 912 ${MAPLOC}
 913 EOF
 914
 915         if ($TYPE eq "colldef" || $TYPE eq "ctypedef") {
 916                 print FOUT <<EOF;
 917 .include <bsd.endian.mk>
 918
 919 EOF
 920         }
 921
 922         print FOUT <<EOF;
 923 .src.${SRCOUT2}:
 924         $SRCOUT
 925
 926 ## PLACEHOLDER
 927
 928 ${SRCOUT4}
 929
 930 EOF
 931
 932         foreach my $hash (keys(%hashtable)) {
 933                 # For colldef, weight LOCALES to UTF-8
 934                 #     Sort as upper-case and reverse to achieve it
 935                 #     Make en_US, ru_RU, and ca_AD preferred
 936                 my @files;
 937                 if ($TYPE eq "colldef") {
 938                         @files = sort {
 939                                 if ($a eq 'en_x_US.UTF-8' ||
 940                                     $a eq 'ru_x_RU.UTF-8' ||
 941                                     $a eq 'ca_x_AD.UTF-8') { return -1; }
 942                                 elsif ($b eq 'en_x_US.UTF-8' ||
 943                                        $b eq 'ru_x_RU.UTF-8' ||
 944                                        $b eq 'ca_x_AD.UTF-8') { return 1; }
 945                                 else { return uc($b) cmp uc($a); }
 946                                 } keys(%{$hashtable{$hash}});
 947                 } elsif ($TYPE eq "ctypedef") {
 948                         @files = sort {
 949                                 if ($a eq 'C_x_x.UTF-8') { return -1; }
 950                                 elsif ($b eq 'C_x_x.UTF-8') { return 1; }
 951                                 if ($a =~ /^en_x_US/) { return -1; }
 952                                 elsif ($b =~ /^en_x_US/) { return 1; }
 953
 954                                 if ($a =~ /^en_x_GB.ISO8859-15/ ||
 955                                     $a =~ /^ru_x_RU/) { return -1; }
 956                                 elsif ($b =~ /^en_x_GB.ISO8859-15/ ||
 957                                        $b =~ /ru_x_RU/) { return 1; }
 958                                 else { return uc($b) cmp uc($a); }
 959
 960                                 } keys(%{$hashtable{$hash}});
 961                 } else {
 962                         @files = sort {
 963                                 if ($a =~ /_Comm_/ ||
 964                                     $b eq 'en_x_US.UTF-8') { return 1; }
 965                                 elsif ($b =~ /_Comm_/ ||
 966                                        $a eq 'en_x_US.UTF-8') { return -1; }
 967                                 else { return uc($b) cmp uc($a); }
 968                                 } keys(%{$hashtable{$hash}});
 969                 }
 970                 if ($#files > 0) {
 971                         my $link = shift(@files);
 972                         $link =~ s/_x_x//;      # special case for C
 973                         $link =~ s/_x_/_/;      # strip family if none there
 974                         foreach my $file (@files) {
 975                                 my @a = split(/_/, $file);
 976                                 my @b = split(/\./, $a[-1]);
 977                                 $file =~ s/_x_/_/;
 978                                 print FOUT "SAME+=\t\t$link $file\n";
 979                                 undef($languages{$a[0]}{$a[1]}{data}{$b[0]}{$b[1]});
 980                         }
 981                 }
 982         }
 983
 984         foreach my $l (sort keys(%languages)) {
 985         foreach my $f (sort keys(%{$languages{$l}})) {
 986         foreach my $c (sort keys(%{$languages{$l}{$f}{data}})) {
 987                 next if (defined $languages{$l}{$f}{definitions}
 988                     && $languages{$l}{$f}{definitions} !~ /$TYPE/);
 989                 if (defined $languages{$l}{$f}{data}{$c}{$DEFENCODING}
 990                  && $languages{$l}{$f}{data}{$c}{$DEFENCODING} eq "0") {
 991                         print "Skipping ${l}_" . ($f eq "x" ? "" : "${f}_") .
 992                             "${c} - not read\n";
 993                         next;
 994                 }
 995                 foreach my $e (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
 996                         my $file = $l;
 997                         $file .= "_" . $f if ($f ne "x");
 998                         $file .= "_" . $c if ($c ne "x");
 999                         next if (!defined $languages{$l}{$f}{data}{$c}{$e});
1000                         print FOUT "LOCALES+=\t$file.$e\n";
1001                 }
1002
1003                 if (defined $languages{$l}{$f}{nc_link}) {
1004                         foreach my $e (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
1005                                 my $file = $l . "_";
1006                                 $file .= $f . "_" if ($f ne "x");
1007                                 $file .= $c;
1008                                 print FOUT "SAME+=\t\t$file.$e $languages{$l}{$f}{nc_link}.$e\t# legacy (lang/country change)\n";
1009                         }
1010                 }
1011
1012                 if (defined $languages{$l}{$f}{e_link}) {
1013                         foreach my $el (split(" ", $languages{$l}{$f}{e_link})) {
1014                                 my @a = split(/:/, $el);
1015                                 my $file = $l . "_";
1016                                 $file .= $f . "_" if ($f ne "x");
1017                                 $file .= $c;
1018                                 print FOUT "SAME+=\t\t$file.$a[0] $file.$a[1]\t# legacy (same charset)\n";
1019                         }
1020                 }
1021
1022         }
1023         }
1024         }
1025
1026         print FOUT <<EOF;
1027
1028 FILES=          \${LOCALES:S/\$/.${SRCOUT2}/}
1029 CLEANFILES=     \${FILES}
1030
1031 .for f t in \${SAME}
1032 SYMLINKS+=      ../\$f/\${FILESNAME} \\
1033     \${LOCALEDIR}/\$t/\${FILESNAME}
1034 .endfor
1035
1036 .for f in \${LOCALES}
1037 FILESDIR_\${f}.${SRCOUT2}= \${LOCALEDIR}/\${f}
1038 .endfor
1039
1040 ${SRCOUT3}.include <bsd.prog.mk>
1041 EOF
1042
1043         close(FOUT);
1044 }