icuSources/tools/gensprep/filterRFC3454.pl

   1 #/usr/bin/perl
   2 # Copyright (c) 2001-2004 International Business Machines
   3 # Corporation and others. All Rights Reserved.
   4
   5 ####################################################################################
   6 # filterRFC3454.pl:
   7 # This tool filters the RFC-3454 txt file for StringPrep tables and creates a table
   8 # to be used in NamePrepProfile
   9 #
  10 # Author: Ram Viswanadha
  11 #
  12 ####################################################################################
  13
  14 use File::Find;
  15 use File::Basename;
  16 use IO::File;
  17 use Cwd;
  18 use File::Copy;
  19 use Getopt::Long;
  20 use File::Path;
  21 use File::Copy;
  22
  23 $copyright = "#################\n# This file was generated from RFC 3454 (http://www.ietf.org/rfc/rfc3454.txt)\n# Copyright (C) The Internet Society (2002).  All Rights Reserved. \n###################\n\n";
  24 $warning = "###################\n# WARNING: This table is generated by filterRFC3454.pl tool. DO NOT EDIT \n###################\n\n";
  25 #run the program)
  26 main();
  27
  28 #---------------------------------------------------------------------
  29 # The main program
  30
  31 sub main(){
  32   GetOptions(
  33            "--sourcedir=s" => \$sourceDir,
  34            "--destdir=s" => \$destDir,
  35            "--src-filename=s" => \$srcFileName,
  36            "--dest-filename=s" => \$destFileName,
  37            "--A1"  => \$a1,
  38            "--B1"  => \$b1,
  39            "--B2"  => \$b2,
  40                    "--B3"  => \$b3,
  41            "--C11" => \$c11,
  42            "--C12" => \$c12,
  43            "--C21" => \$c21,
  44            "--C22" => \$c22,
  45            "--C3"  => \$c3,
  46            "--C4"  => \$c4,
  47            "--C5"  => \$c5,
  48            "--C6"  => \$c6,
  49            "--C7"  => \$c7,
  50            "--C8"  => \$c8,
  51            "--C9"  => \$c9,
  52            "--ldh-chars" => \$writeLDHChars,
  53            "--iscsi" => \$writeISCSIChars,
  54            );
  55   usage() unless defined $sourceDir;
  56   usage() unless defined $destDir;
  57   usage() unless defined $srcFileName;
  58   usage() unless defined $destFileName;
  59
  60   $infile = $sourceDir."/".$srcFileName;
  61   $inFH = IO::File->new($infile,"r")
  62             or die  "could not open the file $infile for reading: $! \n";
  63   $outfile = $destDir."/".$destFileName;
  64
  65   unlink($outfile);
  66   $outFH = IO::File->new($outfile,"a")
  67             or die  "could not open the file $outfile for writing: $! \n";
  68   print $outFH  $copyright;
  69   print $outFH  $warning;
  70   close($outFH);
  71
  72   if(defined $b2 && defined $b3){
  73       die "ERROR: --B2 and --B3 are both specified\!\n";
  74   }
  75
  76   while(defined ($line=<$inFH>)){
  77       next unless $line=~ /Start\sTable/;
  78       if($line =~ /A.1/){
  79             createUnassignedTable($inFH,$outfile);
  80       }
  81       if($line =~ /B.1/ && defined $b1){
  82             createMapToNothing($inFH,$outfile);
  83       }
  84       if($line =~ /B.2/ && defined $b2){
  85             createCaseMapNorm($inFH,$outfile);
  86       }
  87       if($line =~ /B.3/ && defined $b3){
  88             createCaseMapNoNorm($inFH,$outfile);
  89       }
  90       if($line =~ /C.1.1/ && defined $c11 ){
  91             createProhibitedTable($inFH,$outfile,$line);
  92       }
  93       if($line =~ /C.1.2/ && defined $c12 ){
  94             createProhibitedTable($inFH,$outfile,$line);
  95       }
  96       if($line =~ /C.2.1/ && defined $c21 ){
  97             createProhibitedTable($inFH,$outfile,$line);
  98       }
  99       if($line =~ /C.2.2/ && defined $c22 ){
 100             createProhibitedTable($inFH,$outfile,$line);
 101       }
 102       if($line =~ /C.3/ && defined $c3 ){
 103             createProhibitedTable($inFH,$outfile,$line);
 104       }
 105       if($line =~ /C.4/ && defined $c4 ){
 106             createProhibitedTable($inFH,$outfile,$line);
 107       }
 108       if($line =~ /C.5/ && defined $c5 ){
 109             createProhibitedTable($inFH,$outfile,$line);
 110       }
 111       if($line =~ /C.6/ && defined $c6 ){
 112             createProhibitedTable($inFH,$outfile,$line);
 113       }
 114       if($line =~ /C.7/ && defined $c7 ){
 115             createProhibitedTable($inFH,$outfile,$line);
 116       }
 117       if($line =~ /C.8/ && defined $c8 ){
 118             createProhibitedTable($inFH,$outfile,$line);
 119       }
 120       if($line =~ /C.9/ && defined $c9 ){
 121             createProhibitedTable($inFH,$outfile,$line);
 122       }
 123   }
 124   if( defined $writeISCSIChars){
 125       create_iSCSIExtraProhibitedTable($inFH, $outfile);
 126   }
 127   close($inFH);
 128 }
 129
 130 #-----------------------------------------------------------------------
 131 sub readPrint{
 132     local ($inFH, $outFH,$comment, $table) = @_;
 133     $count = 0;
 134     print $outFH $comment."\n";
 135     while(defined ($line = <$inFH>)){
 136         next if $line =~ /Hoffman\s\&\sBlanchet/;  # ignore heading
 137         next if $line =~ /RFC\s3454/; # ignore heading
 138         next if $line =~ /\f/;  # ignore form feed
 139         next if $line eq "\n";  # ignore blank lines
 140         # break if "End Table" is found
 141         if( $line =~ /End\sTable/){
 142             print $outFH "\n# Total code points $count\n\n";
 143             return;
 144         }
 145         if($print==1){
 146             print $line;
 147         }
 148         $line =~ s/-/../;
 149         $line =~ s/^\s+//;
 150         if($line =~ /\;/){
 151         }else{
 152             $line =~ s/$/;/;
 153         }
 154         if($table =~ /A/ ){
 155             ($code, $noise) = split /;/ , $line;
 156             $line = $code."; ; UNASSIGNED\n";
 157         }elsif ( $table =~ /B\.1/ ){
 158             $line =~ s/Map to nothing/MAP/;
 159         }elsif ( $table =~ /B\.[23]/ ){
 160             $line =~ s/Case map/MAP/;
 161             $line =~ s/Additional folding/MAP/;
 162         }elsif ( $table =~ /C/ ) {
 163             ($code, $noise) = split /;/ , $line;
 164             $line = $code."; ; PROHIBITED\n";
 165         }
 166         if($line =~ /\.\./){
 167             ($code, $noise) = split /;/ , $line;
 168             ($startStr, $endStr ) = split /\.\./, $code;
 169             $start = atoi($startStr);
 170             $end   = atoi($endStr);
 171             #print $start."     ".$end."\n";
 172             while($start <= $end){
 173                 $count++;
 174                 $start++;
 175             }
 176         }else{
 177               $count++;
 178         }
 179         print $outFH $line;
 180     }
 181 }
 182 #-----------------------------------------------------------------------
 183 sub atoi {
 184     my $t;
 185     foreach my $d (split(//, shift())) {
 186         $t = $t * 16 + $d;
 187     }
 188     return $t;
 189 }
 190 #-----------------------------------------------------------------------
 191 sub createUnassignedTable{
 192     ($inFH,$outfile) = @_;
 193     $outFH = IO::File->new($outfile,"a")
 194             or die  "could not open the file $outfile for writing: $! \n";
 195     $comment = "# This table contains code points from Table A.1 from RFC 3454\n";
 196     readPrint($inFH,$outFH, $comment, "A");
 197     close($outFH);
 198 }
 199 #-----------------------------------------------------------------------
 200 sub createMapToNothing{
 201     ($inFH,$outfile) = @_;
 202     $outFH = IO::File->new($outfile,"a")
 203             or die  "could not open the file $outfile for writing: $! \n";
 204     $comment = "# This table contains code points from Table B.1 from RFC 3454\n";
 205     readPrint($inFH,$outFH,$comment, "B.1");
 206     close($outFH);
 207 }
 208 #-----------------------------------------------------------------------
 209 sub createCaseMapNorm{
 210     ($inFH,$outfile) = @_;
 211     $outFH = IO::File->new($outfile,"a")
 212             or die  "could not open the file $outfile for writing: $! \n";
 213     $comment = $warning."# This table contains code points from Table B.2 from RFC 3454\n";
 214     readPrint($inFH,$outFH,$comment, "B.2");
 215     close($outFH);
 216 }
 217 #-----------------------------------------------------------------------
 218 sub createCaseMapNoNorm{
 219     ($inFH,$outfile) = @_;
 220     $outFH = IO::File->new($outfile,"a")
 221             or die  "could not open the file $outfile for writing: $! \n";
 222     $comment = $warning."# This table contains code points from Table B.3 from RFC 3454\n";
 223     readPrint($inFH,$outFH,$comment, "B.3");
 224     close($outFH);
 225 }
 226 #-----------------------------------------------------------------------
 227 sub createProhibitedTable{
 228     ($inFH,$outfile,$line) = @_;
 229     $line =~ s/Start//;
 230     $line =~ s/-//g;
 231     $comment = "# code points from $line";
 232
 233     $outFH = IO::File->new($outfile, "a")
 234             or die  "could not open the file $outfile for writing: $! \n";
 235     readPrint($inFH,$outFH,$comment, "C");
 236     close($outFH);
 237 }
 238
 239 #-----------------------------------------------------------------------
 240 sub create_iSCSIExtraProhibitedTable{
 241     ($inFH,$outfile,$line) = @_;
 242     $comment ="# Additional prohibitions from draft-ietf-ips-iscsi-string-prep-06.txt\n";
 243
 244     $outFH = IO::File->new($outfile, "a")
 245             or die  "could not open the file $outfile for writing: $! \n";
 246     print $outFH $comment;
 247     print $outFH "0021..002C; ; PROHIBITED\n";
 248     print $outFH "002F; ; PROHIBITED\n";
 249     print $outFH "003B..0040; ; PROHIBITED\n";
 250     print $outFH "005B..0060; ; PROHIBITED\n";
 251     print $outFH "007B..007E; ; PROHIBITED\n";
 252     print $outFH "3002; ; PROHIBITED\n";
 253     print $outFH "\n# Total code points 30\n";
 254     close($outFH);
 255 }
 256 #-----------------------------------------------------------------------
 257 sub usage {
 258     print << "END";
 259 Usage:
 260 filterRFC3454.pl
 261 Options:
 262         --sourcedir=<directory>
 263         --destdir=<directory>
 264         --src-filename=<name of RFC file>
 265         --dest-filename=<name of destination file>
 266         --A1             Generate data for table A.1
 267         --B1             Generate data for table B.1
 268         --B2             Generate data for table B.2
 269         --B3             Generate data for table B.3
 270         --C11            Generate data for table C.1.1
 271         --C12            Generate data for table C.1.2
 272         --C21            Generate data for table C.2.1
 273         --C22            Generate data for table C.2.2
 274         --C3             Generate data for table C.3
 275         --C4             Generate data for table C.4
 276         --C5             Generate data for table C.5
 277         --C6             Generate data for table C.6
 278         --C7             Generate data for table C.7
 279         --C8             Generate data for table C.8
 280         --C9             Generate data for table C.9
 281         --iscsi          Generate data for extra prohibited iSCSI chars
 282
 283 Note, --B2 and --B3 are mutually exclusive.
 284
 285 e.g.: filterRFC3454.pl --sourcedir=. --destdir=./output --src-filename=rfc3454.txt  --dest-filename=NamePrepProfile.txt --A1 --B1 --B2 --C12 --C22 --C3 --C4 --C5 --C6 --C7 --C8 --C9
 286
 287 filterRFC3454.pl filters the RFC file and creates String prep table files.
 288 The RFC text can be downloaded from ftp://ftp.rfc-editor.org/in-notes/rfc3454.txt
 289
 290 END
 291   exit(0);
 292 }
 293
 294