uj

   1 #!/usr/bin/env perl
   2
   3 # uj
   4 # File ID: f8d8d376-5d46-11df-8ae0-90e6ba3022ac
   5 # Gjentar editering av filer som inneholder ulovlig UTF-8
   6
   7 use strict;
   8 use warnings;
   9
  10 my @err_list = @ARGV;
  11 my @new_list = ();
  12 my $error_found = 0;
  13 my $Editor = "";
  14
  15 if (defined($ENV{EDITOR})) {
  16         $Editor = $ENV{EDITOR};
  17 } else {
  18         warn("\$EDITOR er ikke definert, setter den til \"vim\"\n");
  19         $Editor = "vim";
  20 }
  21
  22 while (scalar(@err_list)) {
  23         if (scalar(@new_list)) {
  24                 print("Ulovlige sekvenser ute og går. Gjelder ", join(" ", @new_list), "\nTrykk ENTER eller CTRL-C...");
  25                 getc(STDIN);
  26         }
  27         my $Cmd = join(" ", $Editor, @err_list);
  28         # print("Kjører \"$Cmd\"\n");
  29         system("$Cmd");
  30         @new_list = ();
  31         # print("Starter med ", join(" ", @err_list), "\n");
  32         for my $File (@err_list) {
  33                 if (file_legal($File)) {
  34                         # print("OK: $File\n");
  35                 } else {
  36                         # print("FEIL: $File\n");
  37                         push(@new_list, $File);
  38                 }
  39         }
  40         @err_list = ();
  41         @err_list = @new_list;
  42 }
  43
  44 sub file_legal {
  45         # {{{
  46         my $File = shift;
  47
  48         # print("file_legal(\"$File\")\n");
  49         $error_found = 0;
  50         if (open(FP, $File)) {
  51                 while (!$error_found && ($_ = <FP>)) {
  52                         s/([\xFC-\xFD][\x80-\xBF][\x80-\xBF][\x80-\xBF][\x80-\xBF][\x80-\xBF])/decode_char($1)/ge;
  53                         s/([\xF8-\xFB][\x80-\xBF][\x80-\xBF][\x80-\xBF][\x80-\xBF])/decode_char($1)/ge;
  54                         s/([\xF0-\xF7][\x80-\xBF][\x80-\xBF][\x80-\xBF])/decode_char($1)/ge;
  55                         s/([\xE0-\xEF][\x80-\xBF][\x80-\xBF])/decode_char($1)/ge;
  56                         s/([\xC0-\xDF][\x80-\xBF])/decode_char($1)/ge;
  57                         /[\x80-\xFD]/ && ($error_found = 1);
  58                 }
  59                 close(FP);
  60         } else {
  61                 warn("$File: $!\n");
  62         }
  63         my $Retval = !$error_found;
  64         $error_found = 0;
  65         return($Retval);
  66         # }}}
  67 } # file_legal()
  68
  69 sub decode_char {
  70         # {{{
  71         my $Msg = shift;
  72         my $Val = "";
  73         if ($Msg =~ /^([\xC0-\xDF])([\x80-\xBF])/) {
  74                 if ($Msg =~ /^[\xC0-\xC1]/) {
  75                         $error_found = 1;
  76                 } else {
  77                         $Val = ((ord($1) & 0x1F) << 6) | (ord($2) & 0x3F);
  78                 }
  79         } elsif ($Msg =~ /^([\xE0-\xEF])([\x80-\xBF])([\x80-\xBF])/) {
  80                 if ($Msg =~ /^\xE0[\x80-\x9F]/) {
  81                         $error_found = 1;
  82                 } else {
  83                         $Val = ((ord($1) & 0x0F) << 12) |
  84                                ((ord($2) & 0x3F) <<  6) |
  85                                ( ord($3) & 0x3F);
  86                 }
  87         } elsif ($Msg =~ /^([\xF0-\xF7])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])/) {
  88                 if ($Msg =~ /^\xF0[\x80-\x8F]/) {
  89                         $error_found = 1;
  90                 } else {
  91                         $Val = ((ord($1) & 0x07) << 18) |
  92                                ((ord($2) & 0x3F) << 12) |
  93                                ((ord($3) & 0x3F) <<  6) |
  94                                ( ord($4) & 0x3F);
  95                 }
  96         } elsif ($Msg =~ /^([\xF8-\xFB])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])/) {
  97                 if ($Msg =~ /^\xF8[\x80-\x87]/) {
  98                         $error_found = 1;
  99                 } else {
 100                         $Val = ((ord($1) & 0x03) << 24) |
 101                                ((ord($2) & 0x3F) << 18) |
 102                                ((ord($3) & 0x3F) << 12) |
 103                                ((ord($4) & 0x3F) <<  6) |
 104                                ( ord($5) & 0x3F);
 105                 }
 106         } elsif ($Msg =~ /^([\xFC-\xFD])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])/) {
 107                 if ($Msg =~ /^\xFC[\x80-\x83]/) {
 108                         $error_found = 1;
 109                 } else {
 110                         $Val = ((ord($1) & 0x01) << 30) |
 111                                ((ord($2) & 0x3F) << 24) |
 112                                ((ord($3) & 0x3F) << 18) |
 113                                ((ord($4) & 0x3F) << 12) |
 114                                ((ord($5) & 0x3F) <<  6) |
 115                                ( ord($6) & 0x3F);
 116                 }
 117         }
 118         # printf("Val = 0x%X\n", $Val);
 119         if (($Val >= 0xD800 && $Val <= 0xDFFF) || ($Val eq 0xFFFE) || ($Val eq 0xFFFF)) {
 120                 $error_found = 1;
 121         }
 122         # print("error_found satt til \"$error_found\" i decode_char()\n");
 123         # return ($error_found);
 124         # }}}
 125 } # decode_char()
 126
 127 __END__
 128
 129 # End of file uj