icuSources/tools/genrb/genrbjar.pl

   1 #!/usr/bin/perl
   2 #  ********************************************************************
   3 #  * COPYRIGHT:
   4 #  * Copyright (c) 2002-2004, International Business Machines Corporation and
   5 #  * others. All Rights Reserved.
   6 #  ********************************************************************
   7
   8 # Script to generate the ICULocaleData.jar file.  This file is
   9 # part of icu4j.  It is checked into CVS.  It is generated from
  10 # locale data in the icu4c project.  See usage() notes (below)
  11 # for more information.
  12
  13 # This script requires perl.  For Win32, I recommend www.activestate.com.
  14
  15 # Alan Liu
  16
  17 use strict;
  18 use warnings;
  19 use File::Path;
  20 use File::Copy;
  21
  22 my $isMSWin32 = ($^O eq 'MSWin32');
  23 usage() unless (@ARGV >= ($isMSWin32 ? 3:2));
  24 my $ICU_ROOT = shift;
  25 my $ICU4J_ROOT = shift;
  26 my $prefix = ''; # e.g. "LD_LIBRARY_PATH=..."
  27 my $flavor = ''; # e.g. "Debug/"
  28 if ($isMSWin32) {
  29     $flavor = shift() . '/';
  30 } else {
  31     my $isDarwin = ($^O eq 'darwin');
  32     my $ldVar = ($isDarwin ? 'DYLD_LIBRARY_PATH' : 'LD_LIBRARY_PATH');
  33     $prefix = ($isMSWin32 ? '' : "$ldVar=$ICU_ROOT/source/common:$ICU_ROOT/source/i18n:$ICU_ROOT/source/tools/toolutil:$ICU_ROOT/source/data/out:$ICU_ROOT/source/data: ");
  34 }
  35 checkPlatform();
  36
  37 # Step 1.  Run genrb.
  38 print "\n[Step 1: Run genrb]\n";
  39 my $genrb = "$ICU_ROOT/source/tools/genrb/${flavor}genrb";
  40 my $dataDir = "$ICU_ROOT/source/data/locales";
  41 my $javaRootDir = "$dataDir/java";
  42 my $pkg = "com/ibm/icu/impl/data";
  43 my $javaDir = "$javaRootDir/$pkg";
  44 chdir($dataDir);
  45 mkpath($javaDir);
  46 my $op = "$prefix$genrb -s. -d$javaDir -j -p com.ibm.icu.impl.data -b LocaleElements ";
  47 print "{Command: $op*.txt}\n";
  48 print "Directory: $dataDir\n";
  49 my @list;
  50 if (@ARGV) {
  51     @list = @ARGV;
  52     foreach (@list) { $_ .= ".txt" unless (/\.txt$/i); }
  53 } else {
  54     @list = glob("*.txt");
  55 }
  56 my $count = 0;
  57 my $errCount = 0;
  58 foreach (sort @list) {
  59     cmd("$op $_", " $_ ");
  60     ++$count;
  61 }
  62
  63 print "\nProcessed $count locale file(s)\n";
  64
  65 # Step 2.  Create LocaleElements_index.java.
  66 print "\n[Step 2: Create LocaleElements_index.java]\n";
  67 chdir("$ICU_ROOT/source/data/out/build");
  68 cmd("$op res_index.txt");
  69 chdir($javaDir);
  70 my $f = "LocaleElements_index.java";
  71 unlink $f if (-e $f);
  72 rename "LocaleElements_res_index.java", $f;
  73 patchIndex("LocaleElements_index.java");
  74
  75 # Step 3.  Find %%ALIAS tags.
  76 # Assume that it looks like this:
  77 #    public LocaleElements_no_NO_NY  () {
  78 #          contents = new Object[][] {
  79 #                {
  80 #                    "%%ALIAS",
  81 #                    "nn_NO",
  82 #                },
  83 #            };
  84 #        }
  85 print "\n[Step 3: Scan for %%ALIAS tags]\n";
  86 print "Directory: $javaDir\n";
  87 chdir($javaDir);
  88 @list = glob("LocaleElements*.java");
  89 my %aliases;
  90 foreach my $file (sort @list) {
  91     my $aliasOf = '';
  92     open(IN, $file) or die;
  93     while (<IN>) {
  94         if (/^\s*\"\%\%ALIAS\"/) {
  95             # This is an alias of the locale on the next line
  96             $aliasOf = <IN>;
  97             die "Can't parse $aliasOf" unless
  98                 ($aliasOf =~ s/^\s*\"(.+?)\",\s*$/$1/);
  99             last;
 100         }
 101     }
 102     close(IN);
 103
 104     if ($aliasOf) {
 105         my $me = $file;
 106         $me =~ s/^LocaleElements_(.+)\.java$/$1/i;
 107         $aliases{$me} = $aliasOf;
 108         print " $me is an alias of $aliasOf\n";
 109     }
 110 }
 111
 112 # Step 4.  Fix %%ALIAS tags.
 113 print "\n[Step 4: Fix %%ALIAS tags]\n";
 114 my %patched; # Record any locales that we patch
 115 foreach my $loc (sort keys %aliases) {
 116     # $loc is an alias of $aliases{$loc}
 117     # Make $loc point to package private static _contents of $aliases{$loc}
 118     my $aliasee = $aliases{$loc};
 119     if (!exists($patched{$aliasee})) {
 120         # Patch the alias
 121         #patchAliasee($aliasee);
 122         $patched{$aliasee} = 1;
 123     }
 124     patchAlias($loc, $aliasee);
 125 }
 126
 127 # Step 5.  Patch transliteration resources.
 128 # ICU resources have TransliterateLATIN but ICU4J resources expect Transliterate_LATIN
 129 print "\n[Step 5: Patch transliteration resources]\n";
 130 foreach my $file (sort @list) {
 131     my $hasTrans = 0;
 132     open(IN, $file) or die;
 133     while (<IN>) {
 134         # Ignore files that are already patched
 135         if (/^\s*\"Transliterate[^_].*\"/) {
 136             $hasTrans = 1;
 137             last;
 138         }
 139     }
 140     close(IN);
 141
 142     patchTrans($file) if ($hasTrans);
 143 }
 144
 145 # Step 6.  Compile .java files
 146 print "\n[Step 6: Compile .java files]\n";
 147 my $cmd = "javac -classpath $ICU4J_ROOT/classes:$javaRootDir:%CLASSPATH% $pkg/*.java";
 148 chdir($javaRootDir);
 149 print "Directory: $javaRootDir\n";
 150 cmd($cmd);
 151
 152 # Step 7.  Update .jar file.  Do a "jar u" to update the existing file.
 153 print "\n[Step 7: Update .jar file]\n";
 154 my $jarFile = "$ICU4J_ROOT/src/$pkg/ICULocaleData.jar";
 155 my $filesToBePackaged= "$pkg/*.class $pkg/*.col $pkg/*.brk $pkg/*.utf8";
 156 $cmd = "jar uf $jarFile $filesToBePackaged";
 157 # Do jar command
 158 print "Directory: $javaRootDir\n";
 159 chdir($javaRootDir);
 160 if(-e "$jarFile"){
 161    if (! -e "$jarFile.orig") {
 162        copy("$jarFile","$jarFile.orig");
 163    }
 164 }else{
 165    $jarFile ="$ICU_ROOT/source/data/locales/java/ICULocaleData.jar";
 166    $cmd = "jar cvf $jarFile $filesToBePackaged";
 167 }
 168 cmd($cmd);
 169 print " $jarFile updated\n";
 170
 171 # Done!
 172 print "\n[All done]\n";
 173 checkPlatform();
 174 exit(0);
 175
 176 #-----------------------------------------------------------------------
 177 # Execute a command
 178 # Param: Command
 179 # Param: Display line, or '' to display command
 180 sub cmd {
 181     my $cmd = shift;
 182     my $prompt = shift;
 183     if ($prompt) {
 184         print $prompt;
 185     } else {
 186         print "{Command: $cmd}..";
 187     }
 188     my_system($cmd);
 189     my $exit_value  = $? >> 8;
 190     #my $signal_num  = $? & 127;
 191     #my $dumped_core = $? & 128;
 192     if ($exit_value == 0) {
 193         print "ok\n" unless ($prompt);
 194     } else {
 195         ++$errCount;
 196         print "ERROR ($exit_value)\n";
 197         exit(1);
 198     }
 199 }
 200
 201 # A system()-like sub that does NOT ignore SIGINT
 202 sub my_system {
 203     my $pid = fork;
 204     if (! defined $pid) {
 205         return -1;
 206     } elsif ($pid) {
 207         return waitpid($pid, 0);
 208     } else {
 209         exec(@_) or exit $!;
 210     }
 211 }
 212
 213 #-----------------------------------------------------------------------
 214 # Patch the file that an %%ALIAS tag points to
 215 sub patchAliasee {
 216     my $loc = shift;
 217     my $file = "LocaleElements_$loc.java";
 218     my $omitNextBrace = 0;
 219     open(IN, $file) or die;
 220     open(OUT, ">$file.new") or die;
 221     while (<IN>) {
 222         #if (/^\s*data\s*=\s*new\s+Object/) {
 223          #   print OUT "        super.contents = data;\n";
 224        #     print OUT "    };\n";
 225        #     print OUT '    static final Object[][] data =', "\n";
 226        #     s/^\s*contents\s*=\s*/        /;
 227        #     print OUT;
 228        # } elsif (/^\s*\}\s*;/) {
 229        #     # Omit the "}" after this
 230        #     print OUT;
 231        #     $omitNextBrace = 1;
 232        # } elsif ($omitNextBrace && /^\s*\}\s*$/) {
 233        #     # Omit it
 234        #     $omitNextBrace = 0;
 235        # } else {
 236             print OUT;
 237        # }
 238     }
 239     close(IN);
 240     close(OUT);
 241     unlink($file);
 242     rename("$file.new", $file);
 243     print " $file patched (aliasee)\n";
 244 }
 245
 246 #-----------------------------------------------------------------------
 247 # Patch the file that contains the %%ALIAS tag
 248 sub patchAlias {
 249     my $loc = shift;
 250     my $aliasee = shift;
 251     my $file = "LocaleElements_$loc.java";
 252     open(IN, $file) or die;
 253     open(OUT, ">$file.new") or die;
 254     my $var = "static final Object";
 255     while (<IN>) {
 256         if(/$var/){
 257             # Output our new data
 258             print OUT "          static final Object[][] data  = LocaleElements_$aliasee.data;\n";
 259            #consume the next 3 lines
 260             <IN>;
 261             <IN>;
 262             <IN>;
 263             <IN>;
 264             <IN>;
 265         } else {
 266             print OUT;
 267         }
 268     }
 269     close(IN);
 270     close(OUT);
 271     unlink($file);
 272     rename("$file.new", $file);
 273     print " $file patched (alias)\n";
 274 }
 275
 276 #-----------------------------------------------------------------------
 277 # Patch a file with a transliteration resource.
 278 sub patchTrans {
 279     my $file = shift;
 280     open(IN, $file) or die;
 281     open(OUT, ">$file.new") or die;
 282     while (<IN>) {
 283         # This should look like "TransliterateFOO" but if underscores
 284         # have crept in, ignore them.
 285         s/^(\s*\"Transliterate)_*(.+?\")/$1_$2/;
 286         print OUT;
 287     }
 288     close(IN);
 289     close(OUT);
 290     unlink($file);
 291     rename("$file.new", $file);
 292     print " $file patched (trans)\n";
 293 }
 294
 295 #-----------------------------------------------------------------------
 296 # Patch the index file, renaming res_index to index
 297 sub patchIndex {
 298     my $file = shift;
 299     open(IN, $file) or die;
 300     open(OUT, ">$file.new") or die;
 301     while (<IN>) {
 302         s/res_(index)/$1/;
 303         print OUT;
 304     }
 305     close(IN);
 306     close(OUT);
 307     unlink($file);
 308     rename("$file.new", $file);
 309     print " $file patched (index)\n";
 310 }
 311
 312 #-----------------------------------------------------------------------
 313 sub checkPlatform {
 314     my $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/;
 315     if (!$is_big_endian) {
 316         print "*******\n";
 317         print "WARNING: You are running on a LITTLE ENDIAN machine.\n";
 318         print "WARNING: You cannot use the resulting ICULocaleData.jar\n";
 319         print "WARNING: *.col files will have incorrect byte order.\n";
 320         print "*******\n";
 321     }
 322 }
 323
 324 #-----------------------------------------------------------------------
 325 sub usage {
 326     print << "END";
 327 Usage: genrbjar.pl <icu_root_dir> <icu4j_root_dir> [<locale>+]
 328        genrbjar.pl <icu_root_dir> <icu4j_root_dir> ('Debug' | 'Release') [<locale>+]
 329
 330 'Debug' or 'Release' is required on MSWin32, and absent on UNIX.
 331
 332 genrbjar creates the ICULocaleData.jar file in the icu4j project.  It
 333 uses locale data files in the icu4c directory and processes them with
 334 genrb to generate Java source.  It makes necessary edits to the Java
 335 source, then compiles the Java to .class files, then creates a .jar
 336 file.  The ICULocaleData.jar file is created in its correct location
 337 within the icu4j directory structure.
 338
 339 Optionally, one or more locales may be specified on the command line.
 340 If this is done, only those locales will be processed.  If no locales
 341 are listed, all locales are processed.
 342
 343 Before running this tool, a JDK must be installed and the javac and
 344 jar binaries for that JDK must be on the system path.
 345 Examples:
 346  i) on Linux:  ./genrbjar.pl ~/icu ~/icu4j
 347  ii) on Win32: perl genrbjar.pl C:\\icu C:\\icu4j Debug
 348
 349 NOTE: You CANNOT use the ICULocaleData.jar created on little endian
 350 machines (e.g. Win32) because the *.col files will have the wrong byte
 351 order.  However, you can use the *.class files and look at the *.java
 352 files.
 353 END
 354   exit(0);
 355 }
 356
 357 __END__
 358 :endofperl