1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/os/textandloc/charconvfw/charconvplugins/tools/cnvtool.pl Fri Jun 15 03:10:57 2012 +0200
1.3 @@ -0,0 +1,1533 @@
1.4 +#
1.5 +# Copyright (c) 1997-2009 Nokia Corporation and/or its subsidiary(-ies).
1.6 +# All rights reserved.
1.7 +# This component and the accompanying materials are made available
1.8 +# under the terms of "Eclipse Public License v1.0"
1.9 +# which accompanies this distribution, and is available
1.10 +# at the URL "http://www.eclipse.org/legal/epl-v10.html".
1.11 +#
1.12 +# Initial Contributors:
1.13 +# Nokia Corporation - initial contribution.
1.14 +#
1.15 +# Contributors:
1.16 +#
1.17 +# Description:
1.18 +#
1.19 +
1.20 +use strict;
1.21 +use integer;
1.22 +
1.23 +BEGIN
1.24 + {
1.25 + my $perlScriptPath=$0;
1.26 + my $os = $^O; #get the OS type
1.27 + #check OS type
1.28 + if($os=~/MSWin32/) #Windows OS
1.29 + {
1.30 + $perlScriptPath=~s/\//\\/g; # replace any forward-slashes with back-slashes
1.31 + $perlScriptPath=~s/(\\?)[^\\]+$/$1/; # get rid of this Perl-script's file-name
1.32 + }
1.33 + else #Unix OS
1.34 + {
1.35 + $perlScriptPath=~s/\\/\//g; # replace any back-slashes with forward-slashes
1.36 + $perlScriptPath=~s/(\/?)[^\/]+$/$1/; # get rid of this Perl-script's file-name
1.37 + }
1.38 + unshift(@INC, $perlScriptPath); # can't do "use lib $perlScriptPath" here as "use lib" only seems to work with *hard-coded* directory names
1.39 + }
1.40 +use PARSER;
1.41 +use WRITER;
1.42 +
1.43 +$|=1; # ensures that any progress information sent to the screen is displayed immediately and not buffered
1.44 +if ((@ARGV==0) || ($ARGV[0]=~/\?/i) || ($ARGV[0]=~/-h/i) || ($ARGV[0]=~/help/i))
1.45 + {
1.46 + die("\nVersion 021\n\nCharacter-set conversion-table generating tool\nCopyright (c) 2000-2009 Nokia Corporation and/or its subsidiary(-ies).\n\nUsage:\n\n\tcnvtool <control-file> <source-file> <output-file> [options]\n\nwhere the following options are available (each has a short form and a long form which are shown below separated by a '|'):\n\n\t-s | -generateSourceCode\n\t-c | -columns(<a>: <b>, <c>)\n\t-r | -omitReplacementForUnconvertibleUnicodeCharacters\n\t-p | -cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed\n\t-u | -sourceFilesToSubtract(<a>, <b>, ...)\n\n");
1.47 + }
1.48 +my $generateSourceCode=0;
1.49 +my @columns=(2, 1, 2);
1.50 +my $omitReplacementForUnconvertibleUnicodeCharacters=0;
1.51 +my $cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed=0;
1.52 +my @sourceFilesToSubtract=();
1.53 +my $flattenHashAndSave=0; # this flag is not published for use outside of the CHARCONV component
1.54 +&extractCommandLineFlags(\$generateSourceCode, \@columns, \$omitReplacementForUnconvertibleUnicodeCharacters, \$cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed, \@sourceFilesToSubtract, \$flattenHashAndSave);
1.55 +(!$omitReplacementForUnconvertibleUnicodeCharacters || $generateSourceCode) or die("Error: bad combination of flags\n");
1.56 +my $controlFile=shift;
1.57 +my $sourceFile=shift;
1.58 +my $outputFile=shift;
1.59 +print("Generating $outputFile...\n");
1.60 +my $uid=0;
1.61 +my $endiannessAsText='';
1.62 +my $endianness=0;
1.63 +my $replacementForUnconvertibleUnicodeCharacters='';
1.64 +my @foreignVariableByteData=();
1.65 +my @foreignToUnicodeData=();
1.66 +my @unicodeToForeignData=();
1.67 +my %foreignCharacterCodes=();
1.68 +my %unicodeCharacterCodes=();
1.69 +my %preferredForeignCharacterCodesForConflictResolution=();
1.70 +my %preferredUnicodeCharacterCodesForConflictResolution=();
1.71 +my %additionalSubsetTables=();
1.72 +my %privateUseUnicodeCharacterSlotsUsed=();
1.73 +
1.74 +print(" reading $controlFile...\n");
1.75 +open(CONTROL_FILE, "< $controlFile") or die("Error: could not open \"$controlFile\" for reading\n");
1.76 +&readHeaderFromControlFile(\*CONTROL_FILE, $controlFile, $generateSourceCode, \$uid, \$endiannessAsText, \$endianness, \$replacementForUnconvertibleUnicodeCharacters, $flattenHashAndSave);
1.77 +&readForeignVariableByteDataFromControlFile(\*CONTROL_FILE, $controlFile, \@foreignVariableByteData);
1.78 +&readOneDirectionDataFromControlFile(\*CONTROL_FILE, $controlFile, \@foreignToUnicodeData, \%preferredUnicodeCharacterCodesForConflictResolution, \%additionalSubsetTables, 1);
1.79 +&readOneDirectionDataFromControlFile(\*CONTROL_FILE, $controlFile, \@unicodeToForeignData, \%preferredForeignCharacterCodesForConflictResolution, \%additionalSubsetTables, 0);
1.80 +close(CONTROL_FILE) or die("Error: could not close \"$controlFile\"\n");
1.81 +
1.82 +print(" reading $sourceFile...\n");
1.83 +open(SOURCE_FILE, "< $sourceFile") or die("Error: could not open \"$sourceFile\" for reading\n");
1.84 +&readSourceFile(\*SOURCE_FILE, $sourceFile, \%foreignCharacterCodes, \%unicodeCharacterCodes, \@columns, $cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed, \%privateUseUnicodeCharacterSlotsUsed, 0);
1.85 +close(SOURCE_FILE) or die("Error: could not close \"$sourceFile\"\n");
1.86 +
1.87 +my $sourceFileToSubtract;
1.88 +foreach $sourceFileToSubtract (@sourceFilesToSubtract)
1.89 + {
1.90 + print(" subtracting from $sourceFileToSubtract...\n");
1.91 + open(SOURCE_FILE_TO_SUBTRACT, "< $sourceFileToSubtract") or die("Error: could not open \"$sourceFileToSubtract\" for reading\n");
1.92 + &readSourceFile(\*SOURCE_FILE_TO_SUBTRACT, $sourceFileToSubtract, \%foreignCharacterCodes, \%unicodeCharacterCodes, \@columns, $cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed, \%privateUseUnicodeCharacterSlotsUsed, 1);
1.93 + close(SOURCE_FILE_TO_SUBTRACT) or die("Error: could not close \"$sourceFileToSubtract\"\n");
1.94 + }
1.95 +
1.96 +&warnIfAnyPrivateUseUnicodeCharacterSlotsBeingUsed(\%privateUseUnicodeCharacterSlotsUsed);
1.97 +&resolveConflictsAndFlattenArraysToScalars(\%foreignCharacterCodes, \%preferredForeignCharacterCodesForConflictResolution, 'Unicode', 'foreign');
1.98 +&resolveConflictsAndFlattenArraysToScalars(\%unicodeCharacterCodes, \%preferredUnicodeCharacterCodesForConflictResolution, 'foreign', 'Unicode');
1.99 +&checkForeignVariableByteData($endianness, \@foreignVariableByteData, \@foreignToUnicodeData);
1.100 +
1.101 +print(" writing $outputFile...\n");
1.102 +open(OUTPUT_FILE, "> $outputFile") or die("Error: could not open \"$outputFile\" for writing\n");
1.103 +if ($generateSourceCode)
1.104 + {
1.105 + my @sourceCodeOfForeignToUnicodeIndexedTables16=();
1.106 + my @sourceCodeOfForeignToUnicodeKeyedTables1616=();
1.107 + my @sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_indexedEntries=();
1.108 + my @sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_keyedEntries=();
1.109 +
1.110 + my @sourceCodeOfUnicodeToForeignIndexedTables16=();
1.111 + my @sourceCodeOfUnicodeToForeignKeyedTables1616=();
1.112 + my @sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_indexedEntries=();
1.113 + my @sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_keyedEntries=();
1.114 +
1.115 + # new for 32 bit encoding begin
1.116 + my @sourceCodeOfForeignToUnicodeIndexedTables32=();
1.117 + my @sourceCodeOfForeignToUnicodeKeyedTables3232=();
1.118 + my @sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_indexedEntries=();
1.119 + my @sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_keyedEntries=();
1.120 +
1.121 + my @sourceCodeOfUnicodeToForeignIndexedTables32=();
1.122 + my @sourceCodeOfUnicodeToForeignKeyedTables3232=();
1.123 + my @sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_indexedEntries=();
1.124 + my @sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_keyedEntries=();
1.125 + # new for 32 bit endcoding end
1.126 +
1.127 + my @sourceCodeOfTopLevelStructures=();
1.128 +
1.129 + &writeSourceCodeHeader(\*OUTPUT_FILE, $outputFile, $replacementForUnconvertibleUnicodeCharacters);
1.130 + &writeSourceCodeForeignVariableByteData(\@sourceCodeOfTopLevelStructures, \@foreignVariableByteData);
1.131 + &writeSourceCodeOneDirectionData(\@sourceCodeOfTopLevelStructures,
1.132 + \@sourceCodeOfForeignToUnicodeIndexedTables16, \@sourceCodeOfForeignToUnicodeKeyedTables1616, \@sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_indexedEntries, \@sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_keyedEntries,
1.133 + \@sourceCodeOfForeignToUnicodeIndexedTables32, \@sourceCodeOfForeignToUnicodeKeyedTables3232, \@sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_indexedEntries, \@sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_keyedEntries,
1.134 + \@foreignToUnicodeData, \%unicodeCharacterCodes, 1);
1.135 + &writeSourceCodeOneDirectionData(\@sourceCodeOfTopLevelStructures,
1.136 + \@sourceCodeOfUnicodeToForeignIndexedTables16, \@sourceCodeOfUnicodeToForeignKeyedTables1616, \@sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_indexedEntries, \@sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_keyedEntries,
1.137 + \@sourceCodeOfUnicodeToForeignIndexedTables32, \@sourceCodeOfUnicodeToForeignKeyedTables3232, \@sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_indexedEntries, \@sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_keyedEntries,
1.138 + \@unicodeToForeignData, \%foreignCharacterCodes, 0);
1.139 + &writeSourceCodeFinalStuff(\*OUTPUT_FILE,
1.140 + \@sourceCodeOfForeignToUnicodeIndexedTables16, \@sourceCodeOfForeignToUnicodeKeyedTables1616, \@sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_indexedEntries, \@sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_keyedEntries,
1.141 + \@sourceCodeOfUnicodeToForeignIndexedTables16, \@sourceCodeOfUnicodeToForeignKeyedTables1616, \@sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_indexedEntries, \@sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_keyedEntries,
1.142 + \@sourceCodeOfForeignToUnicodeIndexedTables32, \@sourceCodeOfForeignToUnicodeKeyedTables3232, \@sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_indexedEntries, \@sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_keyedEntries,
1.143 + \@sourceCodeOfUnicodeToForeignIndexedTables32, \@sourceCodeOfUnicodeToForeignKeyedTables3232, \@sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_indexedEntries, \@sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_keyedEntries,
1.144 + \@sourceCodeOfTopLevelStructures, $endiannessAsText, \%additionalSubsetTables);
1.145 + }
1.146 +elsif ($flattenHashAndSave)
1.147 + {
1.148 + binmode OUTPUT_FILE;
1.149 + #instead of calling the writeBinaryHeader, just write the data I need followed by
1.150 + # writeBinaryForeignVariableByteData...
1.151 + &write8(\*OUTPUT_FILE, $endianness);
1.152 + &write8(\*OUTPUT_FILE, length($replacementForUnconvertibleUnicodeCharacters));
1.153 + &writeString(\*OUTPUT_FILE, $replacementForUnconvertibleUnicodeCharacters);
1.154 + &writeBinaryForeignVariableByteData(\*OUTPUT_FILE, \@foreignVariableByteData);
1.155 + #choose %unicodeCharacterCodes and write the data as keypair
1.156 + my $key;
1.157 + my $rangekey;
1.158 + my $limit;
1.159 + foreach $key (keys(%unicodeCharacterCodes))
1.160 + {
1.161 + &write16(\*OUTPUT_FILE,$key);
1.162 + &write16(\*OUTPUT_FILE,$unicodeCharacterCodes{$key});
1.163 + }
1.164 + }
1.165 +else
1.166 + {
1.167 + binmode OUTPUT_FILE;
1.168 + &writeBinaryHeader(\*OUTPUT_FILE, $uid, $endianness, $replacementForUnconvertibleUnicodeCharacters);
1.169 + &writeBinaryForeignVariableByteData(\*OUTPUT_FILE, \@foreignVariableByteData);
1.170 + &writeBinaryOneDirectionData(\*OUTPUT_FILE, \@foreignToUnicodeData, \%unicodeCharacterCodes, 1);
1.171 + &writeBinaryOneDirectionData(\*OUTPUT_FILE, \@unicodeToForeignData, \%foreignCharacterCodes, 0);
1.172 + }
1.173 +close(OUTPUT_FILE) or die("Error: could not close \"$outputFile\"\n");
1.174 +print("complete\n\n");
1.175 +
1.176 +sub extractCommandLineFlags()
1.177 + {
1.178 + my $generateSourceCode=shift;
1.179 + my $columns=shift;
1.180 + my $omitReplacementForUnconvertibleUnicodeCharacters=shift;
1.181 + my $cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed=shift;
1.182 + my $sourceFilesToSubtract=shift;
1.183 + my $flattenHashAndSave=shift;
1.184 + my $i;
1.185 + for ($i=0; $i<=$#ARGV;) # (i) not cache-ing $#ARGV into a variable as @ARGV may change length in this loop (ii) iterate forwards as some parameters may occupy more than one element in @ARGV
1.186 + {
1.187 + if (($ARGV[$i]=~/^-s$/i) || ($ARGV[$i]=~/^-generateSourceCode$/i))
1.188 + {
1.189 + if ($$flattenHashAndSave==1)
1.190 + {
1.191 + die ("Error: Cannot have -s and -b flags set at the same time");
1.192 + }
1.193 + else
1.194 + {
1.195 + splice(@ARGV, $i, 1);
1.196 + $$generateSourceCode=1;
1.197 + }
1.198 + }
1.199 + elsif (($ARGV[$i]=~/^-c\b(.*)$/i) || ($ARGV[$i]=~/^-columns\b(.*)$/i))
1.200 + {
1.201 + my $columnsData=$1;
1.202 + splice(@ARGV, $i, 1);
1.203 + for (;;)
1.204 + {
1.205 + if ($columnsData=~/^\s*\(\s*(\d+)\s*:\s*(\d+)\s*\,?\s*(\d+)\s*\)\s*$/)
1.206 + {
1.207 + @$columns=($1, $2, $3);
1.208 + last;
1.209 + }
1.210 + ($#ARGV>=$i) or die("Error: bad \"-columns\" format\n");
1.211 + $columnsData.=(splice(@ARGV, $i, 1))[0];
1.212 + }
1.213 + }
1.214 + elsif (($ARGV[$i]=~/^-r$/i) || ($ARGV[$i]=~/^-omitReplacementForUnconvertibleUnicodeCharacters$/i))
1.215 + {
1.216 + splice(@ARGV, $i, 1);
1.217 + $$omitReplacementForUnconvertibleUnicodeCharacters=1;
1.218 + }
1.219 + elsif (($ARGV[$i]=~/^-p$/i) || ($ARGV[$i]=~/^-cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed$/i))
1.220 + {
1.221 + splice(@ARGV, $i, 1);
1.222 + $$cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed=1;
1.223 + }
1.224 + elsif (($ARGV[$i]=~/^-u\b(.*)$/i) || ($ARGV[$i]=~/^-sourceFilesToSubtract\b(.*)$/i))
1.225 + {
1.226 + my $sourceFilesData=$1;
1.227 + splice(@ARGV, $i, 1);
1.228 + for (;;)
1.229 + {
1.230 + if ($sourceFilesData=~/^\s*\(\s*(.+)\)\s*$/)
1.231 + {
1.232 + my $sourceFilesData=$1;
1.233 + @$sourceFilesToSubtract=split(/,/, $sourceFilesData, -1);
1.234 + my $j;
1.235 + for ($j=$#$sourceFilesToSubtract; $j>=0; --$j)
1.236 + {
1.237 + $sourceFilesToSubtract->[$j]=~s/^\s+//;
1.238 + $sourceFilesToSubtract->[$j]=~s/\s+$//;
1.239 + ($sourceFilesToSubtract->[$j] ne '') or die("Error: bad \"-sourceFilesToSubtract\" format (1)\n");
1.240 + }
1.241 + last;
1.242 + }
1.243 + ($#ARGV>=$i) or die("Error: bad \"-sourceFilesToSubtract\" format (2)\n");
1.244 + $sourceFilesData.=(splice(@ARGV, $i, 1))[0];
1.245 + }
1.246 + }
1.247 + elsif (($ARGV[$i]=~/^-f$/i) || ($ARGV[$i]=~/^-flattenHashAndSave$/i))
1.248 + {
1.249 + if ($$generateSourceCode==1)
1.250 + {
1.251 + die ("Error: Cannot have -s and -b flags set at the same time");
1.252 + }
1.253 + else
1.254 + {
1.255 + splice(@ARGV, $i, 1);
1.256 + $$flattenHashAndSave=1;
1.257 + }
1.258 + }
1.259 + else
1.260 + {
1.261 + ++$i;
1.262 + }
1.263 + }
1.264 + }
1.265 +
1.266 +sub algorithm
1.267 + {
1.268 + my $algorithmAsText=shift;
1.269 + if ($algorithmAsText=~/^Direct$/i)
1.270 + {
1.271 + return 0;
1.272 + }
1.273 + elsif ($algorithmAsText=~/^Offset$/i)
1.274 + {
1.275 + return 1;
1.276 + }
1.277 + elsif ($algorithmAsText=~/^IndexedTable16$/i)
1.278 + {
1.279 + return 2;
1.280 + }
1.281 + elsif ($algorithmAsText=~/^KeyedTable1616$/i)
1.282 + {
1.283 + return 3;
1.284 + }
1.285 + elsif ($algorithmAsText=~/^KeyedTable16OfIndexedTables16$/i)
1.286 + {
1.287 + return 4;
1.288 + }
1.289 + elsif ($algorithmAsText=~/^IndexedTable32$/i)
1.290 + {
1.291 + return 5;
1.292 + }
1.293 + elsif ($algorithmAsText=~/^KeyedTable3232$/i)
1.294 + {
1.295 + return 6;
1.296 + }
1.297 + elsif ($algorithmAsText=~/^KeyedTable32OfIndexedTables32$/i)
1.298 + {
1.299 + return 7;
1.300 + }
1.301 + else
1.302 + {
1.303 + return -1;
1.304 + }
1.305 + }
1.306 +
1.307 +sub hexadecimalify
1.308 + {
1.309 + my $string=shift;
1.310 + my $result='';
1.311 + my $lengthOfString=length($string);
1.312 + my $i;
1.313 + for ($i=0; $i<$lengthOfString; ++$i)
1.314 + {
1.315 + $result.=sprintf("\\x%02x", (unpack('C', substr($string, $i, 1)))[0]);
1.316 + }
1.317 + return $result;
1.318 + }
1.319 +
1.320 +sub readSourceFile
1.321 + {
1.322 + my $fileHandle=shift;
1.323 + my $fileName=shift;
1.324 + my $foreignCharacterCodes=shift;
1.325 + my $unicodeCharacterCodes=shift;
1.326 + my $columns=shift;
1.327 + my $cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed=shift;
1.328 + my $privateUseUnicodeCharacterSlotsUsed=shift;
1.329 + my $subtract=shift;
1.330 + my $foreignCharacterCodeProcessingCode='';
1.331 + if (!(($columns->[0]>0) && ($columns->[1]>0) && ($columns->[2]>0) && ($columns->[1]<=$columns->[0]) && ($columns->[2]<=$columns->[0]) && ($columns->[1]!=$columns->[2])))
1.332 + {
1.333 + close($fileHandle);
1.334 + die("Error: bad \"-columns\" data\n");
1.335 + }
1.336 + my $patternOfLineContainingCharacterCodes=join('\s+', ('0x([0-9a-f]+)') x $columns->[0]);
1.337 + my $line;
1.338 + my $strippedDownLine;
1.339 + for (;;)
1.340 + {
1.341 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.342 + if (($strippedDownLine eq '')||(substr($strippedDownLine,0,1) eq chr(26))) # if there are no more lines in the file or if we encountered EOF character
1.343 + {
1.344 + last;
1.345 + }
1.346 + if ($strippedDownLine=~/^SET_FOREIGN_CHARACTER_CODE_PROCESSING_CODE\s+(.*)$/i)
1.347 + {
1.348 + $foreignCharacterCodeProcessingCode=$1;
1.349 + }
1.350 + elsif ($strippedDownLine=~/^$patternOfLineContainingCharacterCodes$/i)
1.351 + {
1.352 + no strict 'refs'; # so that we can use symbolic references for $1, $2, etc
1.353 + my $foreignCharacterCode=hex(${$columns->[1]});
1.354 + my $unicodeCharacterCode=hex(${$columns->[2]});
1.355 + use strict 'refs';
1.356 + if ($foreignCharacterCodeProcessingCode ne '')
1.357 + {
1.358 + $foreignCharacterCode=eval($foreignCharacterCodeProcessingCode);
1.359 + }
1.360 + my $handleConversionPair=1;
1.361 + if ((($unicodeCharacterCode>=0xe000) && ($unicodeCharacterCode<=0xf8ff)) || (($unicodeCharacterCode>=0xf0000) && ($unicodeCharacterCode<=0x10ffff)))
1.362 + {
1.363 + if ($cutOutAnyPrivateUseUnicodeCharacterSlotsBeingUsed)
1.364 + {
1.365 + $handleConversionPair=0;
1.366 + }
1.367 + else
1.368 + {
1.369 + if ($subtract)
1.370 + {
1.371 + delete $privateUseUnicodeCharacterSlotsUsed->{$unicodeCharacterCode};
1.372 + }
1.373 + else
1.374 + {
1.375 + $privateUseUnicodeCharacterSlotsUsed->{$unicodeCharacterCode}=1;
1.376 + }
1.377 + }
1.378 + }
1.379 + if ($handleConversionPair)
1.380 + {
1.381 + if ($subtract)
1.382 + {
1.383 + if (!defined($foreignCharacterCodes->{$unicodeCharacterCode}->{$foreignCharacterCode}))
1.384 + {
1.385 + close($fileHandle);
1.386 + die('Error: cannot subtract conversion pair ['.sprintf('foreign 0x%x, Unicode 0x%04x', $foreignCharacterCode, $unicodeCharacterCode)."] as it does not occur in \"$fileName\"\n");
1.387 + }
1.388 + if (!defined($unicodeCharacterCodes->{$foreignCharacterCode}->{$unicodeCharacterCode}))
1.389 + {
1.390 + close($fileHandle);
1.391 + die('Error: cannot subtract conversion pair ['.sprintf('Unicode 0x%04x, foreign 0x%x', $unicodeCharacterCode, $foreignCharacterCode)."] as it does not occur in \"$fileName\"\n");
1.392 + }
1.393 + delete $foreignCharacterCodes->{$unicodeCharacterCode}->{$foreignCharacterCode};
1.394 + if (keys(%{$foreignCharacterCodes->{$unicodeCharacterCode}})==0)
1.395 + {
1.396 + delete $foreignCharacterCodes->{$unicodeCharacterCode};
1.397 + }
1.398 + delete $unicodeCharacterCodes->{$foreignCharacterCode}->{$unicodeCharacterCode};
1.399 + if (keys(%{$unicodeCharacterCodes->{$foreignCharacterCode}})==0)
1.400 + {
1.401 + delete $unicodeCharacterCodes->{$foreignCharacterCode};
1.402 + }
1.403 + }
1.404 + else
1.405 + {
1.406 + if (defined($foreignCharacterCodes->{$unicodeCharacterCode}->{$foreignCharacterCode}))
1.407 + {
1.408 + close($fileHandle);
1.409 + die('Error: same conversion pair ['.sprintf('foreign 0x%x, Unicode 0x%04x', $foreignCharacterCode, $unicodeCharacterCode)."] occurs more than once in \"$fileName\"\n");
1.410 + }
1.411 + if (defined($unicodeCharacterCodes->{$foreignCharacterCode}->{$unicodeCharacterCode}))
1.412 + {
1.413 + close($fileHandle);
1.414 + die('Error: same conversion pair ['.sprintf('Unicode 0x%04x, foreign 0x%x', $unicodeCharacterCode, $foreignCharacterCode)."] occurs more than once in \"$fileName\"\n");
1.415 + }
1.416 + $foreignCharacterCodes->{$unicodeCharacterCode}->{$foreignCharacterCode}=1;
1.417 + $unicodeCharacterCodes->{$foreignCharacterCode}->{$unicodeCharacterCode}=1;
1.418 + }
1.419 + }
1.420 + }
1.421 + elsif ($line!~/^\s*0x([0-9a-f]+)\s*#\s*undefined.*$/i)
1.422 + {
1.423 + close($fileHandle);
1.424 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.425 + }
1.426 + }
1.427 + }
1.428 +
1.429 +sub readHeaderFromControlFile
1.430 + {
1.431 + my $fileHandle=shift;
1.432 + my $fileName=shift;
1.433 + my $generateSourceCode=shift;
1.434 + my $uid=shift;
1.435 + my $endiannessAsText=shift;
1.436 + my $endianness=shift;
1.437 + my $replacementForUnconvertibleUnicodeCharacters=shift;
1.438 + my $flattenHashAndSave=shift;
1.439 + my $line;
1.440 + my $strippedDownLine;
1.441 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.442 + if ($strippedDownLine=~/^UID\s+0x([0-9a-f]+)$/i)
1.443 + {
1.444 + if ($generateSourceCode)
1.445 + {
1.446 + print(STDERR "Warning: \"UID\" keyword should not be used with \"-generateSourceCode\" flag - specify the UID in the MMP file\n");
1.447 + }
1.448 + $$uid=hex($1);
1.449 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.450 + }
1.451 + else
1.452 + {
1.453 + if (!$generateSourceCode && !$flattenHashAndSave)
1.454 + {
1.455 + close($fileHandle);
1.456 + die("Error: unexpected line in \"$fileName\" (\"UID\" keyword expected):\n $line\n");
1.457 + }
1.458 + }
1.459 + if ($strippedDownLine=~/^Name\s+"(.+?)"$/i)
1.460 + {
1.461 + print(STDERR "Warning: obsolete keyword \"Name\" used\n");
1.462 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.463 + }
1.464 + if ($strippedDownLine!~/^Endianness\s+(\w+)$/i)
1.465 + {
1.466 + close($fileHandle);
1.467 + die("Error: unexpected line in \"$fileName\" (\"Endianness\" keyword expected):\n $line\n");
1.468 + }
1.469 + $$endiannessAsText=$1;
1.470 + if ($$endiannessAsText=~/Unspecified/i)
1.471 + {
1.472 + $$endianness=0; # SCnvConversionData::EUnspecified
1.473 + }
1.474 + elsif ($$endiannessAsText=~/FixedLittleEndian/i)
1.475 + {
1.476 + $$endianness=1; # SCnvConversionData::EFixedLittleEndian
1.477 + }
1.478 + elsif ($$endiannessAsText=~/FixedBigEndian/i)
1.479 + {
1.480 + $$endianness=2; # SCnvConversionData::EFixedBigEndian
1.481 + }
1.482 + else
1.483 + {
1.484 + close($fileHandle);
1.485 + die("Error: \"$$endiannessAsText\" is not a legal value for \"Endianness\"\n");
1.486 + }
1.487 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.488 + if ($strippedDownLine!~/^ReplacementForUnconvertibleUnicodeCharacters\s+(.*?)$/i)
1.489 + {
1.490 + close($fileHandle);
1.491 + die("Error: unexpected line in \"$fileName\" (\"ReplacementForUnconvertibleUnicodeCharacters\" keyword expected):\n $line\n");
1.492 + }
1.493 + $$replacementForUnconvertibleUnicodeCharacters='';
1.494 + my $remainderOfXxx=$1;
1.495 + while ($remainderOfXxx ne '')
1.496 + {
1.497 + if ($remainderOfXxx!~/^0x([0-9a-f]{1,2})\s*(.*)$/i)
1.498 + {
1.499 + close($fileHandle);
1.500 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.501 + }
1.502 + $$replacementForUnconvertibleUnicodeCharacters.=pack("C", hex($1));
1.503 + $remainderOfXxx=$2;
1.504 + }
1.505 + my @temp=&nextNonEmptyStrippedDownLine($fileHandle);
1.506 + if ($temp[1]=~/^ForeignCharacterCodeProcessingCode/i)
1.507 + {
1.508 + print(STDERR "Warning: obsolete keyword \"ForeignCharacterCodeProcessingCode\" used\n");
1.509 + }
1.510 + else
1.511 + {
1.512 + ungetNonEmptyStrippedDownLine(@temp)
1.513 + }
1.514 + }
1.515 +
1.516 +sub readForeignVariableByteDataFromControlFile
1.517 + {
1.518 + my $fileHandle=shift;
1.519 + my $fileName=shift;
1.520 + my $foreignVariableByteData=shift;
1.521 + my $line;
1.522 + my $strippedDownLine;
1.523 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.524 + if ($strippedDownLine!~/^StartForeignVariableByteData$/i)
1.525 + {
1.526 + close($fileHandle);
1.527 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.528 + }
1.529 +
1.530 + for (;;)
1.531 + {
1.532 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.533 + if ($strippedDownLine=~/^EndForeignVariableByteData$/i)
1.534 + {
1.535 + last;
1.536 + }
1.537 + if ($strippedDownLine!~/^0x([0-9a-f]+)\s+0x([0-9a-f]+)\s+(\d+)$/i)
1.538 + {
1.539 + close($fileHandle);
1.540 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.541 + }
1.542 + my $firstInitialByteValueInRange=hex($1);
1.543 + my $lastInitialByteValueInRange=hex($2);
1.544 + my $numberOfSubsequentBytes=$3;
1.545 + if ($firstInitialByteValueInRange>0xff)
1.546 + {
1.547 + close($fileHandle);
1.548 + die("Error: firstInitialByteValueInRange ".sprintf("0x%02x", $firstInitialByteValueInRange)." does not fit in a single byte\n");
1.549 + }
1.550 + if ($lastInitialByteValueInRange>0xff)
1.551 + {
1.552 + close($fileHandle);
1.553 + die("Error: lastInitialByteValueInRange ".sprintf("0x%02x", $lastInitialByteValueInRange)." does not fit in a single byte\n");
1.554 + }
1.555 + if ($lastInitialByteValueInRange<$firstInitialByteValueInRange)
1.556 + {
1.557 + close($fileHandle);
1.558 + die("Error: lastInitialByteValueInRange ".sprintf("0x%02x", $lastInitialByteValueInRange)." is less than firstInitialByteValueInRange ".sprintf("0x%02x", $firstInitialByteValueInRange)."\n");
1.559 + }
1.560 + push(@$foreignVariableByteData, [$firstInitialByteValueInRange, $lastInitialByteValueInRange, $numberOfSubsequentBytes]);
1.561 + }
1.562 + }
1.563 +
1.564 +sub readOneDirectionDataFromControlFile
1.565 + {
1.566 + my $fileHandle=shift;
1.567 + my $fileName=shift;
1.568 + my $oneDirectionData=shift;
1.569 + my $preferredCharacterCodesForConflictResolution=shift;
1.570 + my $additionalSubsetTables=shift;
1.571 + my $outputIsUnicode=shift;
1.572 + my $source=$outputIsUnicode? 'foreign': 'Unicode';
1.573 + my $target=$outputIsUnicode? 'Unicode': 'foreign';
1.574 + my $middlePortionOfKeyWords=$outputIsUnicode? "ForeignToUnicode": "UnicodeToForeign";
1.575 + my $extraPatternToMatch=$outputIsUnicode? '()': '\s+(\d+)';
1.576 + my $line;
1.577 + my $strippedDownLine;
1.578 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.579 + if ($strippedDownLine!~/^Start${middlePortionOfKeyWords}Data$/i)
1.580 + {
1.581 + close($fileHandle);
1.582 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.583 + }
1.584 + my $doingConflictResolution=0;
1.585 + for (;;)
1.586 + {
1.587 + ($line, $strippedDownLine)=&nextNonEmptyStrippedDownLine($fileHandle);
1.588 + if ($strippedDownLine=~/^End${middlePortionOfKeyWords}Data$/i)
1.589 + {
1.590 + last;
1.591 + }
1.592 + if ($strippedDownLine=~/^ConflictResolution$/i)
1.593 + {
1.594 + $doingConflictResolution=1;
1.595 + }
1.596 + elsif ($doingConflictResolution)
1.597 + {
1.598 + if ($strippedDownLine!~/^0x([0-9a-f]+)\s+0x([0-9a-f]+)$/i)
1.599 + {
1.600 + close($fileHandle);
1.601 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.602 + }
1.603 + my $sourceCharacterCodeToResolve=hex($1);
1.604 + my $targetCharacterCodePreferred=hex($2);
1.605 + $preferredCharacterCodesForConflictResolution->{$sourceCharacterCodeToResolve}=$targetCharacterCodePreferred;
1.606 + }
1.607 + elsif ($strippedDownLine=~/^(Start|End)AdditionalSubsetTable\s+(.*)$/i)
1.608 + {
1.609 + my $prefix=$1;
1.610 + my $nameOfAdditionalSubsetTable=$2;
1.611 + my $index=$prefix=~(/^Start$/i)? 0: 1;
1.612 + if (!$outputIsUnicode)
1.613 + {
1.614 + $index+=2;
1.615 + }
1.616 + if (defined($additionalSubsetTables{$nameOfAdditionalSubsetTable}->[$index]))
1.617 + {
1.618 + close($fileHandle);
1.619 + die("Error: multiple redefinition of \"${prefix}AdditionalSubsetTable $nameOfAdditionalSubsetTable\"\n");
1.620 + }
1.621 + $additionalSubsetTables{$nameOfAdditionalSubsetTable}->[$index]=@$oneDirectionData;
1.622 + }
1.623 + else
1.624 + {
1.625 + if ($strippedDownLine!~/^(\d+)\s+(\d+)\s+0x([0-9a-f]+)\s+0x([0-9a-f]+)\s+(\w+)$extraPatternToMatch\s+\{(.*?)\}$/i)
1.626 + {
1.627 + close($fileHandle);
1.628 + die("Error: unexpected line in \"$fileName\":\n $line\n");
1.629 + }
1.630 + my $includePriority=$1;
1.631 + my $searchPriority=$2;
1.632 + my $firstInputCharacterCodeInRange=hex($3);
1.633 + my $lastInputCharacterCodeInRange=hex($4);
1.634 + my $algorithmAsText=$5;
1.635 + my $sizeOfOutputCharacterCodeInBytes=$6;
1.636 + my $parameters=$7;
1.637 + if ($lastInputCharacterCodeInRange<$firstInputCharacterCodeInRange)
1.638 + {
1.639 + close($fileHandle);
1.640 + die("Error: lastInputCharacterCodeInRange ".sprintf("0x%02x", $lastInputCharacterCodeInRange)." is less than firstInputCharacterCodeInRange ".sprintf("0x%02x", $firstInputCharacterCodeInRange)."\n");
1.641 + }
1.642 + my $algorithm=&algorithm($algorithmAsText);
1.643 + if ($algorithm<0)
1.644 + {
1.645 + close($fileHandle);
1.646 + die("Error: unexpected algorithm \"$algorithmAsText\"\n");
1.647 + }
1.648 + my $rangeData=[$includePriority, $searchPriority, $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange, $algorithm];
1.649 + if (!$outputIsUnicode)
1.650 + {
1.651 + push(@$rangeData, $sizeOfOutputCharacterCodeInBytes);
1.652 + }
1.653 + push(@$rangeData, $parameters);
1.654 + push(@$oneDirectionData, $rangeData);
1.655 + }
1.656 + }
1.657 + }
1.658 +
1.659 +sub warnIfAnyPrivateUseUnicodeCharacterSlotsBeingUsed
1.660 + {
1.661 + my $privateUseUnicodeCharacterSlotsUsed=shift;
1.662 + my @sortedPrivateUseUnicodeCharacterSlotsUsed=sort({$a<=>$b} keys(%$privateUseUnicodeCharacterSlotsUsed));
1.663 + if (@sortedPrivateUseUnicodeCharacterSlotsUsed>0)
1.664 + {
1.665 + my $lastPrivateUseUnicodeCharacterSlotUsed=$sortedPrivateUseUnicodeCharacterSlotsUsed[0];
1.666 + my $asText=sprintf('0x%04x', $lastPrivateUseUnicodeCharacterSlotUsed);
1.667 + my @asText=($asText);
1.668 + my $i;
1.669 + for ($i=1; $i<@sortedPrivateUseUnicodeCharacterSlotsUsed; ++$i) # this loop starts from 1 not 0 as we have already dealt with $sortedPrivateUseUnicodeCharacterSlotsUsed[0]
1.670 + {
1.671 + ($sortedPrivateUseUnicodeCharacterSlotsUsed[$i]>$lastPrivateUseUnicodeCharacterSlotUsed) or die("Error: internal error 1\n");
1.672 + if ($sortedPrivateUseUnicodeCharacterSlotsUsed[$i]>$lastPrivateUseUnicodeCharacterSlotUsed+1)
1.673 + {
1.674 + $asText=sprintf('0x%04x', $lastPrivateUseUnicodeCharacterSlotUsed);
1.675 + if ($asText[-1] ne $asText)
1.676 + {
1.677 + $asText[-1].='-'.$asText;
1.678 + }
1.679 + push(@asText, sprintf('0x%04x', $sortedPrivateUseUnicodeCharacterSlotsUsed[$i]));
1.680 + }
1.681 + $lastPrivateUseUnicodeCharacterSlotUsed=$sortedPrivateUseUnicodeCharacterSlotsUsed[$i];
1.682 + }
1.683 + $asText=sprintf('0x%04x', $lastPrivateUseUnicodeCharacterSlotUsed);
1.684 + if ($asText[-1] ne $asText)
1.685 + {
1.686 + $asText[-1].='-'.$asText;
1.687 + }
1.688 + print(STDERR 'Warning: the following private-use Unicode character slots were used: ['.join(', ', @asText)."]\n");
1.689 + }
1.690 + }
1.691 +
1.692 +sub resolveConflictsAndFlattenArraysToScalars
1.693 + {
1.694 + my $characterCodes=shift;
1.695 + my $preferredCharacterCodesForConflictResolution=shift;
1.696 + my $source=shift;
1.697 + my $target=shift;
1.698 + my $sourceCharacterCode;
1.699 + my $candidateTargetCharacterCodes;
1.700 + while (($sourceCharacterCode, $candidateTargetCharacterCodes)=each(%$characterCodes))
1.701 + {
1.702 + my @candidateTargetCharacterCodes=keys(%$candidateTargetCharacterCodes);
1.703 + if (@candidateTargetCharacterCodes<1)
1.704 + {
1.705 + die("Error: internal error 2\n");
1.706 + }
1.707 + if (@candidateTargetCharacterCodes==1)
1.708 + {
1.709 + $characterCodes->{$sourceCharacterCode}=$candidateTargetCharacterCodes[0];
1.710 + }
1.711 + else
1.712 + {
1.713 + if (!defined($preferredCharacterCodesForConflictResolution->{$sourceCharacterCode}))
1.714 + {
1.715 + die("Error: no preferred $target character code is specified for conflict resolution for the $source character code ".sprintf("0x%08x", $sourceCharacterCode)."\n");
1.716 + }
1.717 + my $preferredCharacterCodeIsNotACandidateForConflictResolution=1;
1.718 + my $candidateTargetCharacterCode;
1.719 + foreach $candidateTargetCharacterCode (@candidateTargetCharacterCodes)
1.720 + {
1.721 + if ($preferredCharacterCodesForConflictResolution->{$sourceCharacterCode}==$candidateTargetCharacterCode)
1.722 + {
1.723 + $preferredCharacterCodeIsNotACandidateForConflictResolution=0;
1.724 + last;
1.725 + }
1.726 + }
1.727 + if ($preferredCharacterCodeIsNotACandidateForConflictResolution)
1.728 + {
1.729 + die("Error: the preferred $target character code ".sprintf("0x%08x", $preferredCharacterCodesForConflictResolution->{$sourceCharacterCode})." is not a candidate for conflict resolution for the $source character code ".sprintf("0x%08x", $sourceCharacterCode)."\n");
1.730 + }
1.731 + $characterCodes->{$sourceCharacterCode}=$preferredCharacterCodesForConflictResolution->{$sourceCharacterCode};
1.732 + delete $preferredCharacterCodesForConflictResolution->{$sourceCharacterCode};
1.733 + }
1.734 + }
1.735 + my $numberOfPreferredCharacterCodesForConflictResolution=keys(%$preferredCharacterCodesForConflictResolution);
1.736 + if ($numberOfPreferredCharacterCodesForConflictResolution!=0)
1.737 + {
1.738 + print(STDERR "Warning: there are $numberOfPreferredCharacterCodesForConflictResolution $target preferred character codes specified for which there are no conflicts to resolve\n");
1.739 + }
1.740 + }
1.741 +
1.742 +sub checkForeignVariableByteData
1.743 + {
1.744 + my $endianness=shift;
1.745 + my $foreignVariableByteData=shift;
1.746 + my $foreignToUnicodeData=shift;
1.747 + my $rangeData;
1.748 + my %initialForeignBytes=();
1.749 + foreach $rangeData (@$foreignVariableByteData)
1.750 + {
1.751 + my $initialByte;
1.752 + for ($initialByte=$rangeData->[0]; $initialByte<=$rangeData->[1]; ++$initialByte)
1.753 + {
1.754 + if (defined($initialForeignBytes{$initialByte}))
1.755 + {
1.756 + die("Error: the number of bytes subsequent to the initial foreign-byte $initialForeignBytes{$initialByte} is defined more than once\n");
1.757 + }
1.758 + $initialForeignBytes{$initialByte}=1;
1.759 + }
1.760 + }
1.761 +# if ($endianness!=0) # unfortunately, nothing can be checked if the $endianness is 0 (SCnvConversionData::EUnspecified)
1.762 +# {
1.763 +# foreach $rangeData (@$foreignToUnicodeData)
1.764 +# {
1.765 +# my $inputCharacterCode;
1.766 +# for ($inputCharacterCode=$rangeData->[2]; $inputCharacterCode<=$rangeData->[3]; ++$inputCharacterCode)
1.767 +# {
1.768 +# my $initialByte;
1.769 +# if ($endianness==1) # SCnvConversionData::EFixedLittleEndian
1.770 +# {
1.771 +# $initialByte=($inputCharacterCode&0xff);
1.772 +# }
1.773 +# elsif ($endianness==2) # SCnvConversionData::EFixedBigEndian
1.774 +# {
1.775 +# $initialByte=($inputCharacterCode&0xff00)>>8; ## this is hard-coded and needs to be done properly!
1.776 +# }
1.777 +# else
1.778 +# {
1.779 +# die("Error: internal error ??\n");
1.780 +# }
1.781 +# if (!defined($initialForeignBytes{$initialByte}))
1.782 +# {
1.783 +# die("Error: no number-of-subsequent-bytes is specified for the initial byte $initialByte\n");
1.784 +# }
1.785 +# }
1.786 +# }
1.787 +# }
1.788 + }
1.789 +
1.790 +sub writeSourceCodeHeader
1.791 + {
1.792 + my $fileHandle=shift;
1.793 + my $fileName=shift;
1.794 + my $replacementForUnconvertibleUnicodeCharacters=shift;
1.795 + while ($fileName=~/^.*\\(.*)$/i)
1.796 + {
1.797 + $fileName=$1;
1.798 + }
1.799 + print($fileHandle "// $fileName\n//\n// Copyright (c) Nokia Corporation and/or its subsidiary(-ies) ".(1900+(gmtime(time))[5]).". All rights reserved.\n//\n\n");
1.800 + print($fileHandle "#include <e32std.h>\n#include <convdata.h>\n#include <convgeneratedcpp.h>\n\n#define ARRAY_LENGTH(aArray) (sizeof(aArray)/sizeof((aArray)\[0\]))\n\n#pragma warning (disable: 4049) // compiler limit : terminating line number emission\n\n");
1.801 + if (!$omitReplacementForUnconvertibleUnicodeCharacters)
1.802 + {
1.803 + print($fileHandle "_LIT8(KLit8ReplacementForUnconvertibleUnicodeCharacters, \"".&hexadecimalify($replacementForUnconvertibleUnicodeCharacters)."\");\n\n");
1.804 + print($fileHandle "GLDEF_C const TDesC8& ReplacementForUnconvertibleUnicodeCharacters_internal()\n\t{\n\treturn KLit8ReplacementForUnconvertibleUnicodeCharacters;\n\t}\n\n");
1.805 + }
1.806 + }
1.807 +
1.808 +sub writeSourceCodeForeignVariableByteData
1.809 + {
1.810 + my $sourceCodeOfTopLevelStructures=shift;
1.811 + my $foreignVariableByteData=shift;
1.812 + push(@$sourceCodeOfTopLevelStructures, "LOCAL_D const SCnvConversionData::SVariableByteData::SRange foreignVariableByteDataRanges[]=\n\t\{\n");
1.813 + my $indexOfLastRange=$#$foreignVariableByteData;
1.814 + my $i;
1.815 + for ($i=0; $i<=$indexOfLastRange; ++$i)
1.816 + {
1.817 + my $rangeData=$foreignVariableByteData->[$i];
1.818 + if (@$rangeData!=3)
1.819 + {
1.820 + die("Error: internal error 3\n");
1.821 + }
1.822 + my $firstInitialByteValueInRange=$rangeData->[0];
1.823 + my $lastInitialByteValueInRange=$rangeData->[1];
1.824 + if ($lastInitialByteValueInRange<$firstInitialByteValueInRange)
1.825 + {
1.826 + die("Error: internal error 4\n");
1.827 + }
1.828 + my $numberOfSubsequentBytes=$rangeData->[2];
1.829 + push(@$sourceCodeOfTopLevelStructures, "\t\t\{\n\t\t".sprintf("0x%02x", $firstInitialByteValueInRange).",\n\t\t".sprintf("0x%02x", $lastInitialByteValueInRange).",\n\t\t$numberOfSubsequentBytes,\n\t\t0\n\t\t\}");
1.830 + if ($i<$indexOfLastRange)
1.831 + {
1.832 + push(@$sourceCodeOfTopLevelStructures, ',');
1.833 + }
1.834 + push(@$sourceCodeOfTopLevelStructures, "\n");
1.835 + }
1.836 + push(@$sourceCodeOfTopLevelStructures, "\t\};\n\n");
1.837 + }
1.838 +
1.839 +sub writeSourceCodeOneDirectionData
1.840 + {
1.841 + my $sourceCodeOfTopLevelStructures=shift;
1.842 + my $sourceCodeOfOneDirectionIndexedTables16=shift;
1.843 + my $sourceCodeOfOneDirectionKeyedTables1616=shift;
1.844 + my $sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_indexedEntries=shift;
1.845 + my $sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_keyedEntries=shift;
1.846 + # new for 32 bit encoding begin
1.847 + my $sourceCodeOfOneDirectionIndexedTables32=shift;
1.848 + my $sourceCodeOfOneDirectionKeyedTables3232=shift;
1.849 + my $sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_indexedEntries=shift;
1.850 + my $sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_keyedEntries=shift;
1.851 + # new for 32 bit encoding end
1.852 +
1.853 + my $oneDirectionData=shift;
1.854 + my $characterCodes=shift;
1.855 + my $outputIsUnicode=shift;
1.856 + push(@$sourceCodeOfTopLevelStructures, 'LOCAL_D const SCnvConversionData::SOneDirectionData::SRange '.($outputIsUnicode? 'foreignToUnicodeDataRanges': 'unicodeToForeignDataRanges')."[]=\n\t\{\n");
1.857 + my $formatForInputCharacters=$outputIsUnicode? '0x%02x': '0x%04x';
1.858 + my $formatForOutputCharacters=$outputIsUnicode? '0x%04x': '0x%02x';
1.859 + my $indexOfLastRange=$#$oneDirectionData;
1.860 + my $i;
1.861 + for ($i=0; $i<=$indexOfLastRange; ++$i)
1.862 + {
1.863 + my $rangeData=$oneDirectionData->[$i];
1.864 + # $rangeData is $includePriority, $searchPriority, $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange, $algorithm[, $sizeOfOutputCharacterCodeInBytes], $parameters
1.865 + if (scalar(@$rangeData)!=($outputIsUnicode? 6: 7))
1.866 + {
1.867 + die("Error: internal error 5\n");
1.868 + }
1.869 + my $firstInputCharacterCodeInRange=$rangeData->[2];
1.870 + my $lastInputCharacterCodeInRange=$rangeData->[3];
1.871 + if ($lastInputCharacterCodeInRange<$firstInputCharacterCodeInRange)
1.872 + {
1.873 + die("Error: internal error 6\n");
1.874 + }
1.875 + my $algorithmAsText=''; # set by the if-elsif stuff below
1.876 + my $sizeOfOutputCharacterCodeInBytesIfForeign=$outputIsUnicode? 0: $rangeData->[5];
1.877 + my $parameters=$rangeData->[$outputIsUnicode? 5: 6];
1.878 + my $word1=0; # set by the if-elsif stuff below
1.879 + my $algorithm=$rangeData->[4];
1.880 + if ($algorithm==0) # Direct
1.881 + {
1.882 + $algorithmAsText='Direct';
1.883 + my $characterCode;
1.884 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.885 + {
1.886 + if (!defined($characterCodes->{$characterCode}))
1.887 + {
1.888 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)."\n");
1.889 + }
1.890 + if ($characterCodes->{$characterCode}!=$characterCode)
1.891 + {
1.892 + die("Error: the conversion from ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)." to ".($outputIsUnicode? 'Unicode': 'foreign')." character code ".sprintf("0x%08x", $characterCodes->{$characterCode})." is not a direct conversion\n");
1.893 + }
1.894 + delete $characterCodes->{$characterCode};
1.895 + }
1.896 + }
1.897 + elsif ($algorithm==1) # Offset
1.898 + {
1.899 + $algorithmAsText='Offset';
1.900 + my $offset=$characterCodes->{$firstInputCharacterCodeInRange}-$firstInputCharacterCodeInRange;
1.901 + delete $characterCodes->{$firstInputCharacterCodeInRange};
1.902 + my $characterCode;
1.903 + for ($characterCode=$firstInputCharacterCodeInRange+1; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.904 + {
1.905 + if (!defined($characterCodes->{$characterCode}))
1.906 + {
1.907 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x (0x%08x-0x%08x)", $characterCode, $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange)."\n");
1.908 + }
1.909 + if ($characterCodes->{$characterCode}-$characterCode!=$offset)
1.910 + {
1.911 + die("Error: the conversion from ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)." to ".($outputIsUnicode? 'Unicode': 'foreign')." character code ".sprintf("0x%08x", $characterCodes->{$characterCode})." has a different offset from the previous one in the range\n");
1.912 + }
1.913 + delete $characterCodes->{$characterCode};
1.914 + }
1.915 + $word1="STATIC_CAST(TUint, $offset)";
1.916 + }
1.917 + elsif ($algorithm==2) # IndexedTable16
1.918 + {
1.919 + $algorithmAsText='IndexedTable16';
1.920 + my $nameOfNextOneDirectionIndexedTable16='indexedTable16_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionIndexedTables16+2);
1.921 + my $sourceCodeOfNextOneDirectionIndexedTable16=[];
1.922 + push(@$sourceCodeOfNextOneDirectionIndexedTable16, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SIndexedTable16::SEntry $nameOfNextOneDirectionIndexedTable16\[\]=\n\t\{\n");
1.923 + my $characterCode;
1.924 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.925 + {
1.926 + if (!defined($characterCodes->{$characterCode}))
1.927 + {
1.928 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)."\n");
1.929 + }
1.930 + push(@$sourceCodeOfNextOneDirectionIndexedTable16, "\t\t\{\n\t\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode})."\n\t\t\}");
1.931 + if ($characterCode<$lastInputCharacterCodeInRange)
1.932 + {
1.933 + push(@$sourceCodeOfNextOneDirectionIndexedTable16, ',');
1.934 + }
1.935 + push(@$sourceCodeOfNextOneDirectionIndexedTable16, "\n");
1.936 + delete $characterCodes->{$characterCode};
1.937 + }
1.938 + push(@$sourceCodeOfNextOneDirectionIndexedTable16, "\t\};\n\n");
1.939 + push(@$sourceCodeOfOneDirectionIndexedTables16, $sourceCodeOfNextOneDirectionIndexedTable16);
1.940 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionIndexedTable16)";
1.941 + }
1.942 + elsif ($algorithm==3) # KeyedTable1616
1.943 + {
1.944 + $algorithmAsText='KeyedTable1616';
1.945 + my $nameOfNextOneDirectionKeyedTable1616='keyedTable1616_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables1616+2);
1.946 + my $sourceCodeOfNextOneDirectionKeyedTable1616=[];
1.947 + push(@$sourceCodeOfNextOneDirectionKeyedTable1616, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable1616::SEntry $nameOfNextOneDirectionKeyedTable1616\[\]=\n\t\{\n");
1.948 + my @characterCodes=grep(($_>=$firstInputCharacterCodeInRange) && ($_<=$lastInputCharacterCodeInRange), sort({$a<=>$b} keys(%$characterCodes)));
1.949 + if (@characterCodes==0)
1.950 + {
1.951 + die("Error: There are no ".($outputIsUnicode? 'foreign': 'Unicode').'-to-'.($outputIsUnicode? 'Unicode': 'foreign')." characters to convert using KeyedTable1616 (range ".sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).").\n");
1.952 + }
1.953 + if ($characterCodes[0]!=$firstInputCharacterCodeInRange)
1.954 + {
1.955 + print(STDERR 'Warning: the specified start of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[0])."\n");
1.956 + }
1.957 + if ($characterCodes[-1]!=$lastInputCharacterCodeInRange)
1.958 + {
1.959 + print(STDERR 'Warning: the specified end of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[-1])."\n");
1.960 + }
1.961 + my $characterCode;
1.962 + foreach $characterCode (@characterCodes)
1.963 + {
1.964 + if (defined($characterCodes->{$characterCode}))
1.965 + {
1.966 + push(@$sourceCodeOfNextOneDirectionKeyedTable1616, "\t\t\{\n\t\t".sprintf($formatForInputCharacters, $characterCode).",\n\t\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode})."\n\t\t\}");
1.967 + if ($characterCode<$characterCodes[-1])
1.968 + {
1.969 + push(@$sourceCodeOfNextOneDirectionKeyedTable1616, ',');
1.970 + }
1.971 + push(@$sourceCodeOfNextOneDirectionKeyedTable1616, "\n");
1.972 + delete $characterCodes->{$characterCode};
1.973 + }
1.974 + }
1.975 + push(@$sourceCodeOfNextOneDirectionKeyedTable1616, "\t\};\n\n");
1.976 + push(@$sourceCodeOfOneDirectionKeyedTables1616, $sourceCodeOfNextOneDirectionKeyedTable1616);
1.977 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionKeyedTable1616)";
1.978 + }
1.979 + elsif ($algorithm==4) # KeyedTable16OfIndexedTables16
1.980 + {
1.981 + $algorithmAsText='KeyedTable16OfIndexedTables16';
1.982 + my $nameOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries='keyedTables16OfIndexedTables16_keyedEntries_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_keyedEntries+2);
1.983 + my $sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries=[];
1.984 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable16OfIndexedTables16::SKeyedEntry $nameOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries\[\]=\n\t\{\n");
1.985 + my @characterCodes=grep(($_>=$firstInputCharacterCodeInRange) && ($_<=$lastInputCharacterCodeInRange), sort({$a<=>$b} keys(%$characterCodes)));
1.986 + if (@characterCodes==0)
1.987 + {
1.988 + die("Error: There are no ".($outputIsUnicode? 'foreign': 'Unicode').'-to-'.($outputIsUnicode? 'Unicode': 'foreign')." characters to convert using KeyedTable16OfIndexedTables16 (range ".sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).").\n");
1.989 + }
1.990 + if ($characterCodes[0]!=$firstInputCharacterCodeInRange)
1.991 + {
1.992 + print(STDERR 'Warning: the specified start of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[0])."\n");
1.993 + }
1.994 + if ($characterCodes[-1]!=$lastInputCharacterCodeInRange)
1.995 + {
1.996 + print(STDERR 'Warning: the specified end of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[-1])."\n");
1.997 + }
1.998 + my @characterCodeRanges=();
1.999 + my $minimumNumberOfEntriesPerIndexedTable=($parameters ne '')? $parameters: 0;
1.1000 + my $firstInputCharacterCodeInIndexedTable=$characterCodes[0];
1.1001 + my $previousCharacterCode=$firstInputCharacterCodeInIndexedTable;
1.1002 + my $characterCode;
1.1003 + foreach $characterCode (@characterCodes)
1.1004 + {
1.1005 + ($characterCode>=$previousCharacterCode) or die("Error: internal error 7\n");
1.1006 + if ($characterCode>$previousCharacterCode+1)
1.1007 + {
1.1008 + if (($previousCharacterCode-$firstInputCharacterCodeInIndexedTable)+1>=$minimumNumberOfEntriesPerIndexedTable)
1.1009 + {
1.1010 + push(@characterCodeRanges, [$firstInputCharacterCodeInIndexedTable, $previousCharacterCode]);
1.1011 + }
1.1012 + $firstInputCharacterCodeInIndexedTable=$characterCode;
1.1013 + }
1.1014 + $previousCharacterCode=$characterCode;
1.1015 + }
1.1016 + push(@characterCodeRanges, [$firstInputCharacterCodeInIndexedTable, $previousCharacterCode]);
1.1017 + @characterCodes=();
1.1018 + my $characterCodeRange;
1.1019 + foreach $characterCodeRange (@characterCodeRanges)
1.1020 + {
1.1021 + my $nameOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries='keyedTables16OfIndexedTables16_indexedEntries_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_indexedEntries+2);
1.1022 + my $sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries=[];
1.1023 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, "LOCAL_D const TUint16 $nameOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries\[\]=\n\t\{\n");
1.1024 + my $characterCode;
1.1025 + my $lastInputCharacterCodeInIndexedTable=$characterCodeRange->[1];
1.1026 + for ($characterCode=$characterCodeRange->[0]; $characterCode<=$lastInputCharacterCodeInIndexedTable; ++$characterCode)
1.1027 + {
1.1028 + if (!defined($characterCodes->{$characterCode}))
1.1029 + {
1.1030 + die("Error: internal error 8\n");
1.1031 + }
1.1032 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, "\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode}));
1.1033 + if ($characterCode<$lastInputCharacterCodeInIndexedTable)
1.1034 + {
1.1035 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, ',');
1.1036 + }
1.1037 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, "\n");
1.1038 + delete $characterCodes->{$characterCode};
1.1039 + }
1.1040 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, "\t\};\n\n");
1.1041 + push(@$sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_indexedEntries, $sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries);
1.1042 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, "\t\t\{\n\t\t".sprintf($formatForInputCharacters, $characterCodeRange->[0]).",\n\t\t".sprintf($formatForInputCharacters, $characterCodeRange->[1]).",\n\t\t$nameOfNextOneDirectionKeyedTables16OfIndexedTables16_indexedEntries\n\t\t\}");
1.1043 + if ($characterCodeRange->[1]<$characterCodeRanges[-1]->[1])
1.1044 + {
1.1045 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, ',');
1.1046 + }
1.1047 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, "\n");
1.1048 + }
1.1049 + push(@$sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, "\t\};\n\n");
1.1050 + push(@$sourceCodeOfOneDirectionKeyedTables16OfIndexedTables16_keyedEntries, $sourceCodeOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries);
1.1051 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionKeyedTables16OfIndexedTables16_keyedEntries)";
1.1052 + }
1.1053 + elsif ($algorithm==5) # IndexedTable32
1.1054 + {
1.1055 + $algorithmAsText='IndexedTable32';
1.1056 + my $nameOfNextOneDirectionIndexedTable32='indexedTable32_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionIndexedTables32+2);
1.1057 + my $sourceCodeOfNextOneDirectionIndexedTable32=[];
1.1058 + push(@$sourceCodeOfNextOneDirectionIndexedTable32, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SIndexedTable32::SEntry $nameOfNextOneDirectionIndexedTable32\[\]=\n\t\{\n");
1.1059 + my $characterCode;
1.1060 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.1061 + {
1.1062 + if (!defined($characterCodes->{$characterCode}))
1.1063 + {
1.1064 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)."\n");
1.1065 + }
1.1066 + push(@$sourceCodeOfNextOneDirectionIndexedTable32, "\t\t\{\n\t\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode})."\n\t\t\}");
1.1067 + if ($characterCode<$lastInputCharacterCodeInRange)
1.1068 + {
1.1069 + push(@$sourceCodeOfNextOneDirectionIndexedTable32, ',');
1.1070 + }
1.1071 + push(@$sourceCodeOfNextOneDirectionIndexedTable32, "\n");
1.1072 + delete $characterCodes->{$characterCode};
1.1073 + }
1.1074 + push(@$sourceCodeOfNextOneDirectionIndexedTable32, "\t\};\n\n");
1.1075 + push(@$sourceCodeOfOneDirectionIndexedTables32, $sourceCodeOfNextOneDirectionIndexedTable32);
1.1076 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionIndexedTable32)";
1.1077 + }
1.1078 + elsif ($algorithm==6) # KeyedTable3232
1.1079 + {
1.1080 + $algorithmAsText='KeyedTable3232';
1.1081 + my $nameOfNextOneDirectionKeyedTable3232='keyedTable3232_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables3232+2);
1.1082 + my $sourceCodeOfNextOneDirectionKeyedTable3232=[];
1.1083 + push(@$sourceCodeOfNextOneDirectionKeyedTable3232, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable3232::SEntry $nameOfNextOneDirectionKeyedTable3232\[\]=\n\t\{\n");
1.1084 + my @characterCodes=grep(($_>=$firstInputCharacterCodeInRange) && ($_<=$lastInputCharacterCodeInRange), sort({$a<=>$b} keys(%$characterCodes)));
1.1085 + if (@characterCodes==0)
1.1086 + {
1.1087 + die("Error: There are no ".($outputIsUnicode? 'foreign': 'Unicode').'-to-'.($outputIsUnicode? 'Unicode': 'foreign')." characters to convert using KeyedTable3232 (range ".sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).").\n");
1.1088 + }
1.1089 + if ($characterCodes[0]!=$firstInputCharacterCodeInRange)
1.1090 + {
1.1091 + print(STDERR 'Warning: the specified start of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf($formatForInputCharacters, $characterCodes[0])."\n");
1.1092 + }
1.1093 + if ($characterCodes[-1]!=$lastInputCharacterCodeInRange)
1.1094 + {
1.1095 + print(STDERR 'Warning: the specified end of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf($formatForInputCharacters, $characterCodes[-1])."\n");
1.1096 + }
1.1097 + my $characterCode;
1.1098 + foreach $characterCode (@characterCodes)
1.1099 + {
1.1100 + if (defined($characterCodes->{$characterCode}))
1.1101 + {
1.1102 + push(@$sourceCodeOfNextOneDirectionKeyedTable3232, "\t\t\{\n\t\t".sprintf($formatForInputCharacters, $characterCode).",\n\t\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode})."\n\t\t\}");
1.1103 + if ($characterCode<$characterCodes[-1])
1.1104 + {
1.1105 + push(@$sourceCodeOfNextOneDirectionKeyedTable3232, ',');
1.1106 + }
1.1107 + push(@$sourceCodeOfNextOneDirectionKeyedTable3232, "\n");
1.1108 + delete $characterCodes->{$characterCode};
1.1109 + }
1.1110 + }
1.1111 + push(@$sourceCodeOfNextOneDirectionKeyedTable3232, "\t\};\n\n");
1.1112 + push(@$sourceCodeOfOneDirectionKeyedTables3232, $sourceCodeOfNextOneDirectionKeyedTable3232);
1.1113 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionKeyedTable3232)";
1.1114 + }
1.1115 + elsif ($algorithm==7) # KeyedTable32OfIndexedTables32
1.1116 + {
1.1117 + $algorithmAsText='KeyedTable32OfIndexedTables32';
1.1118 + my $nameOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries='keyedTables32OfIndexedTables32_keyedEntries_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_keyedEntries+2);
1.1119 + my $sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries=[];
1.1120 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, "LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable32OfIndexedTables32::SKeyedEntry $nameOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries\[\]=\n\t\{\n");
1.1121 + my @characterCodes=grep(($_>=$firstInputCharacterCodeInRange) && ($_<=$lastInputCharacterCodeInRange), sort({$a<=>$b} keys(%$characterCodes)));
1.1122 + if (@characterCodes==0)
1.1123 + {
1.1124 + die("Error: There are no ".($outputIsUnicode? 'foreign': 'Unicode').'-to-'.($outputIsUnicode? 'Unicode': 'foreign')." characters to convert using KeyedTable32OfIndexedTables32 (range ".sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).").\n");
1.1125 + }
1.1126 + if ($characterCodes[0]!=$firstInputCharacterCodeInRange)
1.1127 + {
1.1128 + print(STDERR 'Warning: the specified start of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[0])."\n");
1.1129 + }
1.1130 + if ($characterCodes[-1]!=$lastInputCharacterCodeInRange)
1.1131 + {
1.1132 + print(STDERR 'Warning: the specified end of the '.($outputIsUnicode? 'foreign': 'Unicode').' range '.sprintf("$formatForInputCharacters-$formatForInputCharacters", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange).' could actually be '.sprintf("$formatForInputCharacters", $characterCodes[-1])."\n");
1.1133 + }
1.1134 + my @characterCodeRanges=();
1.1135 + my $minimumNumberOfEntriesPerIndexedTable=($parameters ne '')? $parameters: 0;
1.1136 + my $firstInputCharacterCodeInIndexedTable=$characterCodes[0];
1.1137 + my $previousCharacterCode=$firstInputCharacterCodeInIndexedTable;
1.1138 + my $characterCode;
1.1139 + foreach $characterCode (@characterCodes)
1.1140 + {
1.1141 + ($characterCode>=$previousCharacterCode) or die("Error: internal error 7\n");
1.1142 + if ($characterCode>$previousCharacterCode+1)
1.1143 + {
1.1144 + if (($previousCharacterCode-$firstInputCharacterCodeInIndexedTable)+1>=$minimumNumberOfEntriesPerIndexedTable)
1.1145 + {
1.1146 + push(@characterCodeRanges, [$firstInputCharacterCodeInIndexedTable, $previousCharacterCode]);
1.1147 + }
1.1148 + $firstInputCharacterCodeInIndexedTable=$characterCode;
1.1149 + }
1.1150 + $previousCharacterCode=$characterCode;
1.1151 + }
1.1152 + push(@characterCodeRanges, [$firstInputCharacterCodeInIndexedTable, $previousCharacterCode]);
1.1153 + @characterCodes=();
1.1154 + my $characterCodeRange;
1.1155 + foreach $characterCodeRange (@characterCodeRanges)
1.1156 + {
1.1157 + my $nameOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries='keyedTables32OfIndexedTables32_indexedEntries_'.($outputIsUnicode? 'foreignToUnicode': 'unicodeToForeign').'_'.($#$sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_indexedEntries+2);
1.1158 + my $sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries=[];
1.1159 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, "LOCAL_D const TUint32 $nameOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries\[\]=\n\t\{\n");
1.1160 + my $characterCode;
1.1161 + my $lastInputCharacterCodeInIndexedTable=$characterCodeRange->[1];
1.1162 + for ($characterCode=$characterCodeRange->[0]; $characterCode<=$lastInputCharacterCodeInIndexedTable; ++$characterCode)
1.1163 + {
1.1164 + if (!defined($characterCodes->{$characterCode}))
1.1165 + {
1.1166 + die("Error: internal error 8\n");
1.1167 + }
1.1168 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, "\t".sprintf($formatForOutputCharacters, $characterCodes->{$characterCode}));
1.1169 + if ($characterCode<$lastInputCharacterCodeInIndexedTable)
1.1170 + {
1.1171 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, ',');
1.1172 + }
1.1173 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, "\n");
1.1174 + delete $characterCodes->{$characterCode};
1.1175 + }
1.1176 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, "\t\};\n\n");
1.1177 + push(@$sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_indexedEntries, $sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries);
1.1178 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, "\t\t\{\n\t\t".sprintf($formatForInputCharacters, $characterCodeRange->[0]).",\n\t\t".sprintf($formatForInputCharacters, $characterCodeRange->[1]).",\n\t\t$nameOfNextOneDirectionKeyedTables32OfIndexedTables32_indexedEntries\n\t\t\}");
1.1179 + if ($characterCodeRange->[1]<$characterCodeRanges[-1]->[1])
1.1180 + {
1.1181 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, ',');
1.1182 + }
1.1183 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, "\n");
1.1184 + }
1.1185 + push(@$sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, "\t\};\n\n");
1.1186 + push(@$sourceCodeOfOneDirectionKeyedTables32OfIndexedTables32_keyedEntries, $sourceCodeOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries);
1.1187 + $word1="UData_S$algorithmAsText($nameOfNextOneDirectionKeyedTables32OfIndexedTables32_keyedEntries)";
1.1188 + }
1.1189 + else
1.1190 + {
1.1191 + die("Error: internal error 9\n");
1.1192 + }
1.1193 + push(@$sourceCodeOfTopLevelStructures, "\t\t\{\n\t\t".sprintf($formatForInputCharacters, $firstInputCharacterCodeInRange).",\n\t\t".sprintf($formatForInputCharacters, $lastInputCharacterCodeInRange).",\n\t\tSCnvConversionData::SOneDirectionData::SRange::E$algorithmAsText,\n\t\t".$sizeOfOutputCharacterCodeInBytesIfForeign.",\n\t\t0,\n\t\t\t\{\n\t\t\t".$word1."\n\t\t\t\}\n\t\t\}");
1.1194 + if ($i<$indexOfLastRange)
1.1195 + {
1.1196 + push(@$sourceCodeOfTopLevelStructures, ',');
1.1197 + }
1.1198 + push(@$sourceCodeOfTopLevelStructures, "\n");
1.1199 + }
1.1200 + my @characterCodes=sort({$a<=>$b} keys(%$characterCodes));
1.1201 + if (@characterCodes>0)
1.1202 + {
1.1203 + die('The following '.($outputIsUnicode? 'foreign': 'Unicode').' characters have no conversion algorithm specified: ['.join(', ', map(sprintf($formatForInputCharacters, $_), @characterCodes))."\]\n");
1.1204 + }
1.1205 + push(@$sourceCodeOfTopLevelStructures, "\t\};\n\n");
1.1206 + }
1.1207 +
1.1208 +sub writeSourceCodeFinalStuff
1.1209 + {
1.1210 + my $fileHandle=shift;
1.1211 + my $sourceCodeOfForeignToUnicodeIndexedTables16=shift;
1.1212 + my $sourceCodeOfForeignToUnicodeKeyedTables1616=shift;
1.1213 + my $sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_indexedEntries=shift;
1.1214 + my $sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_keyedEntries=shift;
1.1215 + my $sourceCodeOfUnicodeToForeignIndexedTables16=shift;
1.1216 + my $sourceCodeOfUnicodeToForeignKeyedTables1616=shift;
1.1217 + my $sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_indexedEntries=shift;
1.1218 + my $sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_keyedEntries=shift;
1.1219 +
1.1220 + my $sourceCodeOfForeignToUnicodeIndexedTables32=shift;
1.1221 + my $sourceCodeOfForeignToUnicodeKeyedTables3232=shift;
1.1222 + my $sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_indexedEntries=shift;
1.1223 + my $sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_keyedEntries=shift;
1.1224 + my $sourceCodeOfUnicodeToForeignIndexedTables32=shift;
1.1225 + my $sourceCodeOfUnicodeToForeignKeyedTables3232=shift;
1.1226 + my $sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_indexedEntries=shift;
1.1227 + my $sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_keyedEntries=shift;
1.1228 +
1.1229 + my $sourceCodeOfTopLevelStructures=shift;
1.1230 + my $endiannessAsText=shift;
1.1231 + my $additionalSubsetTables=shift;
1.1232 + my $sourceCodeChunk;
1.1233 + my $arrayOfSourceCodeChunks;
1.1234 +
1.1235 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeIndexedTables16)
1.1236 + {
1.1237 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1238 + {
1.1239 + print($fileHandle $sourceCodeChunk);
1.1240 + }
1.1241 + }
1.1242 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables1616)
1.1243 + {
1.1244 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1245 + {
1.1246 + print($fileHandle $sourceCodeChunk);
1.1247 + }
1.1248 + }
1.1249 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_indexedEntries)
1.1250 + {
1.1251 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1252 + {
1.1253 + print($fileHandle $sourceCodeChunk);
1.1254 + }
1.1255 + }
1.1256 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables16OfIndexedTables16_keyedEntries)
1.1257 + {
1.1258 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1259 + {
1.1260 + print($fileHandle $sourceCodeChunk);
1.1261 + }
1.1262 + }
1.1263 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignIndexedTables16)
1.1264 + {
1.1265 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1266 + {
1.1267 + print($fileHandle $sourceCodeChunk);
1.1268 + }
1.1269 + }
1.1270 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables1616)
1.1271 + {
1.1272 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1273 + {
1.1274 + print($fileHandle $sourceCodeChunk);
1.1275 + }
1.1276 + }
1.1277 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_indexedEntries)
1.1278 + {
1.1279 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1280 + {
1.1281 + print($fileHandle $sourceCodeChunk);
1.1282 + }
1.1283 + }
1.1284 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables16OfIndexedTables16_keyedEntries)
1.1285 + {
1.1286 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1287 + {
1.1288 + print($fileHandle $sourceCodeChunk);
1.1289 + }
1.1290 + }
1.1291 + # for 32 bit encoding begin
1.1292 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeIndexedTables32)
1.1293 + {
1.1294 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1295 + {
1.1296 + print($fileHandle $sourceCodeChunk);
1.1297 + }
1.1298 + }
1.1299 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables3232)
1.1300 + {
1.1301 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1302 + {
1.1303 + print($fileHandle $sourceCodeChunk);
1.1304 + }
1.1305 + }
1.1306 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_indexedEntries)
1.1307 + {
1.1308 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1309 + {
1.1310 + print($fileHandle $sourceCodeChunk);
1.1311 + }
1.1312 + }
1.1313 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfForeignToUnicodeKeyedTables32OfIndexedTables32_keyedEntries)
1.1314 + {
1.1315 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1316 + {
1.1317 + print($fileHandle $sourceCodeChunk);
1.1318 + }
1.1319 + }
1.1320 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignIndexedTables32)
1.1321 + {
1.1322 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1323 + {
1.1324 + print($fileHandle $sourceCodeChunk);
1.1325 + }
1.1326 + }
1.1327 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables3232)
1.1328 + {
1.1329 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1330 + {
1.1331 + print($fileHandle $sourceCodeChunk);
1.1332 + }
1.1333 + }
1.1334 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_indexedEntries)
1.1335 + {
1.1336 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1337 + {
1.1338 + print($fileHandle $sourceCodeChunk);
1.1339 + }
1.1340 + }
1.1341 + foreach $arrayOfSourceCodeChunks (@$sourceCodeOfUnicodeToForeignKeyedTables32OfIndexedTables32_keyedEntries)
1.1342 + {
1.1343 + foreach $sourceCodeChunk (@$arrayOfSourceCodeChunks)
1.1344 + {
1.1345 + print($fileHandle $sourceCodeChunk);
1.1346 + }
1.1347 + }
1.1348 + # for 32 bit encoding end
1.1349 + foreach $sourceCodeChunk (@$sourceCodeOfTopLevelStructures)
1.1350 + {
1.1351 + print($fileHandle $sourceCodeChunk);
1.1352 + }
1.1353 +
1.1354 + print($fileHandle "GLDEF_D const SCnvConversionData conversionData=\n\t\{\n\tSCnvConversionData::E$endiannessAsText,\n\t\t\{\n\t\tARRAY_LENGTH(foreignVariableByteDataRanges),\n\t\tforeignVariableByteDataRanges\n\t\t\},\n\t\t\{\n\t\tARRAY_LENGTH(foreignToUnicodeDataRanges),\n\t\tforeignToUnicodeDataRanges\n\t\t\},\n\t\t\{\n\t\tARRAY_LENGTH(unicodeToForeignDataRanges),\n\t\tunicodeToForeignDataRanges\n\t\t\},\n\tNULL,\n\tNULL\n\t\};\n\n");
1.1355 +
1.1356 + my $additionalSubsetTableName;
1.1357 + my $additionalSubsetTableData;
1.1358 + while (($additionalSubsetTableName, $additionalSubsetTableData)=each(%$additionalSubsetTables))
1.1359 + {
1.1360 + (defined($additionalSubsetTableData->[0]) && defined($additionalSubsetTableData->[1]) && defined($additionalSubsetTableData->[2]) && defined($additionalSubsetTableData->[3])) or die("Error: incomplete definition of \"$additionalSubsetTableName\"\n");
1.1361 + print($fileHandle "GLREF_D const SCnvConversionData $additionalSubsetTableName;\n");
1.1362 + print($fileHandle "GLDEF_D const SCnvConversionData $additionalSubsetTableName=\n\t\{\n\tSCnvConversionData::E$endiannessAsText,\n\t\t\{\n\t\tARRAY_LENGTH(foreignVariableByteDataRanges),\n\t\tforeignVariableByteDataRanges\n\t\t\},\n\t\t\{\n\t\t$additionalSubsetTableData->[1]-$additionalSubsetTableData->[0],\n\t\tforeignToUnicodeDataRanges+$additionalSubsetTableData->[0]\n\t\t\},\n\t\t\{\n\t\t$additionalSubsetTableData->[3]-$additionalSubsetTableData->[2],\n\t\tunicodeToForeignDataRanges+$additionalSubsetTableData->[2]\n\t\t\}\n\t\};\n\n");
1.1363 + }
1.1364 + }
1.1365 +
1.1366 +sub writeBinaryHeader
1.1367 + {
1.1368 + my $fileHandle=shift;
1.1369 + my $uid=shift;
1.1370 + my $endianness=shift;
1.1371 + my $replacementForUnconvertibleUnicodeCharacters=shift;
1.1372 + &writeUids($fileHandle, 0x100011bd, $uid, 0);
1.1373 + &write32($fileHandle, 1); # version number of the file format
1.1374 + &write32($fileHandle, 0); # not currently used
1.1375 + &write8($fileHandle, 0); # number of Unicode characters in the name (which is now derived from the file-name, hence why this is zero)
1.1376 + &write8($fileHandle, $endianness);
1.1377 + &write8($fileHandle, length($replacementForUnconvertibleUnicodeCharacters));
1.1378 + &writeString($fileHandle, $replacementForUnconvertibleUnicodeCharacters);
1.1379 + }
1.1380 +
1.1381 +sub writeBinaryForeignVariableByteData
1.1382 + {
1.1383 + my $fileHandle=shift;
1.1384 + my $foreignVariableByteData=shift;
1.1385 + &writePositiveIntegerCompacted30($fileHandle, scalar(@$foreignVariableByteData));
1.1386 + my $rangeData;
1.1387 + foreach $rangeData (@$foreignVariableByteData)
1.1388 + {
1.1389 + if (@$rangeData!=3)
1.1390 + {
1.1391 + die("Error: internal error 10\n");
1.1392 + }
1.1393 + my $firstInitialByteValueInRange=$rangeData->[0];
1.1394 + my $lastInitialByteValueInRange=$rangeData->[1];
1.1395 + if ($lastInitialByteValueInRange<$firstInitialByteValueInRange)
1.1396 + {
1.1397 + die("Error: internal error 11\n");
1.1398 + }
1.1399 + &write8($fileHandle, $firstInitialByteValueInRange);
1.1400 + &write8($fileHandle, $lastInitialByteValueInRange);
1.1401 + &write8($fileHandle, $rangeData->[2]); # numberOfSubsequentBytes
1.1402 + }
1.1403 + }
1.1404 +
1.1405 +sub writeBinaryOneDirectionData
1.1406 + {
1.1407 + my $fileHandle=shift;
1.1408 + my $oneDirectionData=shift;
1.1409 + my $characterCodes=shift;
1.1410 + my $outputIsUnicode=shift;
1.1411 + &writePositiveIntegerCompacted30($fileHandle, scalar(@$oneDirectionData));
1.1412 + my $rangeData;
1.1413 + foreach $rangeData (@$oneDirectionData)
1.1414 + {
1.1415 +## $rangeData is $includePriority, $searchPriority, $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange, $algorithm[, $sizeOfOutputCharacterCodeInBytes], $parameters
1.1416 + if (scalar(@$rangeData)!=($outputIsUnicode? 6: 7))
1.1417 + {
1.1418 + die("Error: internal error 12\n");
1.1419 + }
1.1420 + my $firstInputCharacterCodeInRange=$rangeData->[2];
1.1421 + my $lastInputCharacterCodeInRange=$rangeData->[3];
1.1422 + if ($lastInputCharacterCodeInRange<$firstInputCharacterCodeInRange)
1.1423 + {
1.1424 + die("Error: internal error 13\n");
1.1425 + }
1.1426 + &writePositiveIntegerCompacted30($fileHandle, $firstInputCharacterCodeInRange);
1.1427 + &writePositiveIntegerCompacted30($fileHandle, $lastInputCharacterCodeInRange);
1.1428 + my $algorithm=$rangeData->[4];
1.1429 + &write8($fileHandle, $algorithm);
1.1430 + if (!$outputIsUnicode)
1.1431 + {
1.1432 + &write8($fileHandle, $rangeData->[5]); # sizeOfOutputCharacterCodeInBytesIfForeign
1.1433 + }
1.1434 + if ($algorithm==0) # Direct
1.1435 + {
1.1436 + my $characterCode;
1.1437 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.1438 + {
1.1439 + if (!defined($characterCodes->{$characterCode}))
1.1440 + {
1.1441 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)."\n");
1.1442 + }
1.1443 + if ($characterCodes->{$characterCode}!=$characterCode)
1.1444 + {
1.1445 + die("Error: the conversion from ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)." to ".($outputIsUnicode? 'Unicode': 'foreign')." character code ".sprintf("0x%08x", $characterCodes->{$characterCode})." is not a direct conversion\n");
1.1446 + }
1.1447 + delete $characterCodes->{$characterCode};
1.1448 + }
1.1449 + }
1.1450 + elsif ($algorithm==1) # Offset
1.1451 + {
1.1452 + my $offset=$characterCodes->{$firstInputCharacterCodeInRange}-$firstInputCharacterCodeInRange;
1.1453 + delete $characterCodes->{$firstInputCharacterCodeInRange};
1.1454 + my $characterCode;
1.1455 + for ($characterCode=$firstInputCharacterCodeInRange+1; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.1456 + {
1.1457 + if (!defined($characterCodes->{$characterCode}))
1.1458 + {
1.1459 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x (0x%08x-0x%08x)", $characterCode, $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange)."\n");
1.1460 + }
1.1461 + if ($characterCodes->{$characterCode}-$characterCode!=$offset)
1.1462 + {
1.1463 + die("Error: the conversion from ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)." to ".($outputIsUnicode? 'Unicode': 'foreign')." character code ".sprintf("0x%08x", $characterCodes->{$characterCode})." has a different offset from the previous one in the range\n");
1.1464 + }
1.1465 + delete $characterCodes->{$characterCode};
1.1466 + }
1.1467 + &writeSignedIntegerCompacted29($fileHandle, $offset);
1.1468 + }
1.1469 + elsif ($algorithm==2) # IndexedTable16
1.1470 + {
1.1471 + my $characterCode;
1.1472 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.1473 + {
1.1474 + if (!defined($characterCodes->{$characterCode}))
1.1475 + {
1.1476 + die("Error: There is no conversion defined for ".($outputIsUnicode? 'foreign': 'Unicode')." character code ".sprintf("0x%08x", $characterCode)."\n");
1.1477 + }
1.1478 + &write16($fileHandle, $characterCodes->{$characterCode});
1.1479 + delete $characterCodes->{$characterCode};
1.1480 + }
1.1481 + }
1.1482 + elsif ($algorithm==3) # KeyedTable1616
1.1483 + {
1.1484 + my $characterCode;
1.1485 + my @table=();
1.1486 + for ($characterCode=$firstInputCharacterCodeInRange; $characterCode<=$lastInputCharacterCodeInRange; ++$characterCode)
1.1487 + {
1.1488 + if (defined($characterCodes->{$characterCode}))
1.1489 + {
1.1490 + push(@table, [$characterCode, $characterCodes->{$characterCode}]);
1.1491 + delete $characterCodes->{$characterCode};
1.1492 + }
1.1493 + }
1.1494 + my $firstIteration=1;
1.1495 + my $lastKey;
1.1496 + &writePositiveIntegerCompacted30($fileHandle, scalar(@table));
1.1497 + if ($table[0][0]!=$firstInputCharacterCodeInRange)
1.1498 + {
1.1499 + die("Error: no conversion is specified for the first ".($outputIsUnicode? 'foreign': 'Unicode')." character code in the KeyedTable1616 range ".sprintf("0x%08x to 0x%08x", $firstInputCharacterCodeInRange, $lastInputCharacterCodeInRange)."\n");
1.1500 + }
1.1501 + my $pair;
1.1502 + foreach $pair (@table)
1.1503 + {
1.1504 + my $key=$pair->[0];
1.1505 + if ($firstIteration)
1.1506 + {
1.1507 + $firstIteration=0;
1.1508 + }
1.1509 + else
1.1510 + {
1.1511 + if ($key<=$lastKey)
1.1512 + {
1.1513 + die("Error: internal error 14\n");
1.1514 + }
1.1515 + &writePositiveIntegerCompacted15($fileHandle, $key-$lastKey);
1.1516 + }
1.1517 + &write16($fileHandle, $pair->[1]);
1.1518 + $lastKey=$key;
1.1519 + }
1.1520 + }
1.1521 + elsif ($algorithm==4) # KeyedTable16OfIndexedTables16
1.1522 + {
1.1523 + die("Error: \"KeyedTable16OfIndexedTables16\" is only supported if generating source code\n");
1.1524 + }
1.1525 + else
1.1526 + {
1.1527 + die("Error: internal error 15\n");
1.1528 + }
1.1529 + }
1.1530 + my @characterCodes=sort({$a<=>$b} keys(%$characterCodes));
1.1531 + if (@characterCodes>0)
1.1532 + {
1.1533 + die('The following '.($outputIsUnicode? 'foreign': 'Unicode').'characters have no conversion algorithm specified: ['.join(', ', map({sprintf('0x%x', $_)} @characterCodes))."\]\n");
1.1534 + }
1.1535 + }
1.1536 +