MigrationTools-utf8.patch

   1 diff -ruN migrationtools-47.orig/migrate_common.ph migrationtools-47/migrate_common.ph
   2 --- migrationtools-47.orig/migrate_common.ph    2006-01-25 05:18:16.000000000 +0100
   3 +++ migrationtools-47/migrate_common.ph 2007-04-07 00:05:52.000000000 +0200
   4 @@ -89,6 +89,9 @@
   5  # such as person.
   6  $EXTENDED_SCHEMA = 0;
   7
   8 +# Comment this out if your ldap server does not support UTF8 encoding
   9 +$USE_UTF8 = 1;
  10 +
  11  #
  12  # allow environment variables to override predefines
  13  #
  14 diff -ruN migrationtools-47.orig/migrate_passwd.pl migrationtools-47/migrate_passwd.pl
  15 --- migrationtools-47.orig/migrate_passwd.pl    2006-01-25 05:18:16.000000000 +0100
  16 +++ migrationtools-47/migrate_passwd.pl 2007-04-07 00:06:13.000000000 +0200
  17 @@ -36,6 +36,7 @@
  18  #
  19  # Thanks to Peter Jacob Slot <peter@vision.auk.dk>.
  20  #
  21 +# UTF8 support by Jonas Smedegaard <dr@jones.dk>.
  22
  23  require 'migrate_common.ph';
  24
  25 @@ -53,28 +54,6 @@
  26         next if /^#/;
  27         next if /^\+/;
  28
  29 -       s/Ä/Ae/g;
  30 -       s/Ë/Ee/g;
  31 -       s/Ï/Ie/g;
  32 -       s/Ö/Oe/g;
  33 -       s/Ü/Ue/g;
  34 -
  35 -       s/ä/ae/g;
  36 -       s/ë/ee/g;
  37 -       s/ï/ie/g;
  38 -       s/ö/oe/g;
  39 -       s/ü/ue/g;
  40 -       s/ÿ/ye/g;
  41 -       s/ß/ss/g;
  42 -       s/é/e/g;
  43 -
  44 -       s/Æ/Ae/g;
  45 -       s/æ/ae/g;
  46 -       s/Ø/Oe/g;
  47 -       s/ø/oe/g;
  48 -       s/Å/Ae/g;
  49 -       s/å/ae/g;
  50 -
  51         local($user, $pwd, $uid, $gid, $gecos, $homedir, $shell) = split(/:/);
  52
  53         if ($use_stdout) {
  54 @@ -100,25 +79,25 @@
  55         $sn = $tmp[$#tmp];
  56         pop(@tmp);
  57         $givenname=join(' ',@tmp);
  58 -
  59 +
  60         print $HANDLE "dn: uid=$user,$NAMINGCONTEXT\n";
  61         print $HANDLE "uid: $user\n";
  62 -       print $HANDLE "cn: $cn\n";
  63 +       &print_utf8($HANDLE, "cn", $cn);
  64
  65         if ($EXTENDED_SCHEMA) {
  66                 if ($wphone) {
  67 -                       print $HANDLE "telephoneNumber: $wphone\n";
  68 +                       &print_utf8($HANDLE, "telephoneNumber", $wphone);
  69                 }
  70                 if ($office) {
  71 -                       print $HANDLE "roomNumber: $office\n";
  72 +                       &print_utf8($HANDLE, "roomNumber", $office);
  73                 }
  74                 if ($hphone) {
  75 -                       print $HANDLE "homePhone: $hphone\n";
  76 +                       &print_utf8($HANDLE, "homePhone", $hphone);
  77                 }
  78                 if ($givenname) {
  79 -                       print $HANDLE "givenName: $givenname\n";
  80 +                       &print_utf8($HANDLE, "givenName", $givenname);
  81                 }
  82 -               print $HANDLE "sn: $sn\n";
  83 +               &print_utf8($HANDLE, "sn", $sn);
  84                 if ($DEFAULT_MAIL_DOMAIN) {
  85                         print $HANDLE "mail: $user\@$DEFAULT_MAIL_DOMAIN\n";
  86                 }
  87 @@ -174,7 +153,7 @@
  88         }
  89
  90         if ($gecos) {
  91 -               print $HANDLE "gecos: $gecos\n";
  92 +               &print_ascii($HANDLE, "gecos", $gecos);
  93         }
  94
  95         print $HANDLE "\n";
  96 @@ -225,3 +204,177 @@
  97         }
  98  }
  99
 100 +sub print_utf8
 101 +{
 102 +       my($HANDLE, $attribute, $content) = @_;
 103 +
 104 +       if (&validate_ascii($content)) {
 105 +               print $HANDLE "$attribute: $content\n";
 106 +       } elsif ($USE_UTF8) {
 107 +#              $content = &recode_custom_to_utf8($content);
 108 +               $content = &recode_latin1_to_utf8($content);
 109 +               if (&validate_utf8($content)) {
 110 +                       $content = &encode_base64($content, "");
 111 +                       print $HANDLE "$attribute\:: $content\n";
 112 +               } else {
 113 +                       die "ERROR: Illegal character(s) in UTF-8 string: \"$content\"";
 114 +               }
 115 +       } else {
 116 +               &print_ascii($HANDLE, "$attribute", "$content");
 117 +       }
 118 +}
 119 +
 120 +sub print_ascii
 121 +{
 122 +       my($HANDLE, $attribute, $content) = @_;
 123 +
 124 +       if (&validate_utf8($content)) {
 125 +               $content = &recode_utf8_to_latin1($content);
 126 +       } else {
 127 +               $content = &recode_latin1_to_utf8($content);
 128 +               $content = &recode_utf8_to_latin1($content);
 129 +       }
 130 +       $content = &recode_custom_to_ascii($content);
 131 +       if (&validate_ascii($content)) {
 132 +               print $HANDLE "$attribute: $content\n";
 133 +       } else {
 134 +               my $badchars = $content;
 135 +               for ($badchars) {
 136 +                       s/[\x20-\x7E]//g;
 137 +               }
 138 +               die "ERROR: Illegal character(s) \"$badchars\" in ASCII string: \"$content\"";
 139 +       }
 140 +}
 141 +
 142 +sub recode_latin1_to_utf8
 143 +{
 144 +       my ($content) = @_;
 145 +       for ($content) {
 146 +               s/([\x80-\xFF])/chr(0xC0|ord($1)>>6).chr(0x80|ord($1)&0x3F)/eg;
 147 +       }
 148 +       return ($content)
 149 +}
 150 +
 151 +sub recode_utf8_to_latin1
 152 +{
 153 +       my ($content) = @_;
 154 +       for ($content) {
 155 +               s/([\xC2\xC3])([\x80-\xBF])/chr(ord($1)<<6&0xC0|ord($2)&0x3F)/eg;
 156 +       }
 157 +       return ($content)
 158 +}
 159 +
 160 +sub recode_custom_to_ascii
 161 +{
 162 +       my ($content) = @_;
 163 +       for ($content) {
 164 +               s/\xc0/A/g; # latin capital letter a with grave
 165 +               s/\xc1/A/g; # latin capital letter a with acute
 166 +               s/\xc2/A/g; # latin capital letter a with circumflex
 167 +               s/\xc3/A/g; # latin capital letter a with tilde
 168 +               s/\xc4/Ae/g; # latin capital letter a with diaeresis
 169 +               s/\xc5/Aa/g; # latin capital letter a with ring above
 170 +               s/\xc6/Ae/g; # latin capital letter ae
 171 +               s/\xc7/C/g; # latin capital letter c with cedilla
 172 +               s/\xc8/E/g; # latin capital letter e with grave
 173 +               s/\xc9/E/g; # latin capital letter e with acute
 174 +               s/\xca/E/g; # latin capital letter e with circumflex
 175 +               s/\xcb/Ee/g; # latin capital letter e with diaeresis
 176 +               s/\xcc/I/g; # latin capital letter i with grave
 177 +               s/\xcd/I/g; # latin capital letter i with acute
 178 +               s/\xce/I/g; # latin capital letter i with circumflex
 179 +               s/\xcf/Ie/g; # latin capital letter i with diaeresis
 180 +               s/\xd0/Dh/g; # latin capital letter eth (icelandic)
 181 +               s/\xd1/N/g; # latin capital letter n with tilde
 182 +               s/\xd2/O/g; # latin capital letter o with grave
 183 +               s/\xd3/O/g; # latin capital letter o with acute
 184 +               s/\xd4/O/g; # latin capital letter o with circumflex
 185 +               s/\xd5/O/g; # latin capital letter o with tilde
 186 +               s/\xd6/Oe/g; # latin capital letter o with diaeresis
 187 +               s/\xd8/Oe/g; # latin capital letter o with stroke
 188 +               s/\xd9/U/g; # latin capital letter u with grave
 189 +               s/\xda/U/g; # latin capital letter u with acute
 190 +               s/\xdb/U/g; # latin capital letter u with circumflex
 191 +               s/\xdc/Ue/g; # latin capital letter u with diaeresis
 192 +               s/\xdd/Y/g; # latin capital letter y with acute
 193 +               s/\xde/TH/g; # latin capital letter thorn (icelandic)
 194 +               s/\xdf/ss/g; # latin small letter sharp s (german)
 195 +               s/\xe0/a/g; # latin small letter a with grave
 196 +               s/\xe1/a/g; # latin small letter a with acute
 197 +               s/\xe2/a/g; # latin small letter a with circumflex
 198 +               s/\xe3/a/g; # latin small letter a with tilde
 199 +               s/\xe4/ae/g; # latin small letter a with diaeresis
 200 +               s/\xe5/aa/g; # latin small letter a with ring above
 201 +               s/\xe6/ae/g; # latin small letter ae
 202 +               s/\xe7/c/g; # latin small letter c with cedilla
 203 +               s/\xe8/e/g; # latin small letter e with grave
 204 +               s/\xe9/e/g; # latin small letter e with acute
 205 +               s/\xea/e/g; # latin small letter e with circumflex
 206 +               s/\xeb/ee/g; # latin small letter e with diaeresis
 207 +               s/\xec/i/g; # latin small letter i with grave
 208 +               s/\xed/i/g; # latin small letter i with acute
 209 +               s/\xee/i/g; # latin small letter i with circumflex
 210 +               s/\xef/ii/g; # latin small letter i with diaeresis
 211 +               s/\xf0/dh/g; # latin small letter eth (icelandic)
 212 +               s/\xf1/n/g; # latin small letter n with tilde
 213 +               s/\xf2/o/g; # latin small letter o with grave
 214 +               s/\xf3/o/g; # latin small letter o with acute
 215 +               s/\xf4/o/g; # latin small letter o with circumflex
 216 +               s/\xf5/o/g; # latin small letter o with tilde
 217 +               s/\xf6/oe/g; # latin small letter o with diaeresis
 218 +               s/\xf8/oe/g; # latin small letter o with stroke
 219 +               s/\xf9/u/g; # latin small letter u with grave
 220 +               s/\xfa/u/g; # latin small letter u with acute
 221 +               s/\xfb/u/g; # latin small letter u with circumflex
 222 +               s/\xfc/ue/g; # latin small letter u with diaeresis
 223 +               s/\xfd/y/g; # latin small letter y with acute
 224 +               s/\xfe/th/g; # latin small letter thorn (icelandic)
 225 +               s/\xff/ye/g; # latin small letter y with diaeresis
 226 +       }
 227 +       return ($content);
 228 +}
 229 +
 230 +sub encode_base64
 231 +# Found in email by Baruzzi Giovanni <giovanni.baruzzi@allianz-leben.de> on openldap mailinglist
 232 +
 233 +# Historically this module has been implemented as pure perl code.
 234 +# The XS implementation runs about 20 times faster, but the Perl
 235 +# code might be more portable, so it is still here.
 236 +{
 237 +       my $res = "";
 238 +       my $eol = $_[1];
 239 +       $eol = "\n" unless defined $eol;
 240 +       pos($_[0]) = 0; # ensure start at the beginning
 241 +       while ($_[0] =~ /(.{1,45})/gs) {
 242 +               $res .= substr(pack('u', $1), 1);
 243 +               chop($res);
 244 +       }
 245 +       $res =~ tr|` -_|AA-Za-z0-9+/|;               # `# help emacs
 246 +       # fix padding at the end
 247 +       my $padding = (3 - length($_[0]) % 3) % 3;
 248 +       $res =~ s/.{$padding}$/'=' x $padding/e if $padding;
 249 +       # break encoded string into lines of no more than 76 characters each
 250 +       if (length $eol) {
 251 +               $res =~ s/(.{1,76})/$1$eol/g;
 252 +       }
 253 +       $res;
 254 +}
 255 +
 256 +sub validate_ascii
 257 +{
 258 +       my ($content) = @_;
 259 +       $content =~ /^[\x20-\x7E]*$/;
 260 +}
 261 +
 262 +sub validate_utf8
 263 +{
 264 +       my ($content) = @_;
 265 +       if (&validate_ascii($content)) {
 266 +               return 1;
 267 +       }
 268 +       if ($] >= 5.8) {
 269 +               ## No Perl support for UTF-8! ;-/
 270 +               return undef;
 271 +       }
 272 +       $content =~ /^[\x20-\x7E\x{0080}-\x{FFFF}]*$/;
 273 +}