build/i18n-scan.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5 use Text::Balanced qw(extract_bracketed extract_delimited extract_tagged);
   6
   7 @ARGV >= 1 || die "Usage: $0 <source direcory>\n";
   8
   9
  10 my %stringtable;
  11
  12 sub dec_lua_str
  13 {
  14         my $s = shift;
  15         $s =~ s/\\n/\n/g;
  16         $s =~ s/\\t/\n/g;
  17         $s =~ s/\\(.)/$1/g;
  18         $s =~ s/[\s\n]+/ /g;
  19         $s =~ s/^ //;
  20         $s =~ s/ $//;
  21         return $s;
  22 }
  23
  24 sub dec_tpl_str
  25 {
  26         my $s = shift;
  27         $s =~ s/[\s\n]+/ /g;
  28         $s =~ s/^ //;
  29         $s =~ s/ $//;
  30         return $s;
  31 }
  32
  33
  34 if( open F, "find @ARGV -type f '(' -name '*.htm' -or -name '*.lua' ')' |" )
  35 {
  36         while( defined( my $file = readline F ) )
  37         {
  38                 chomp $file;
  39
  40                 if( open S, "< $file" )
  41                 {
  42                         local $/ = undef;
  43                         my $raw = <S>;
  44                         close S;
  45
  46
  47                         my $text = $raw;
  48
  49                         while( $text =~ s/ ^ .*? (?:translate|translatef|i18n|_) [\n\s]* \( /(/sgx )
  50                         {
  51                                 ( my $code, $text ) = extract_bracketed($text, q{('")});
  52                                 $code =~ s/^\(//; $code =~ s/\)$//;
  53
  54                                 my $res = "";
  55                                 my $sub = "";
  56
  57                                 while( defined $sub )
  58                                 {
  59                                         ( $sub, $code ) = extract_delimited($code, q{'"}, q{\s*(?:\.\.\s*)?});
  60
  61                                         if( defined $sub )
  62                                         {
  63                                                 $res .= substr $sub, 1, length($sub) - 2;
  64                                         }
  65                                 }
  66
  67                                 $res = dec_lua_str($res);
  68                                 $stringtable{$res}++;
  69                         }
  70
  71
  72                         $text = $raw;
  73
  74                         while( $text =~ s/ ^ .*? <% [:_] -? /<%/sgx )
  75                         {
  76                                 ( my $code, $text ) = extract_tagged($text, '<%', '%>');
  77
  78                                 if( defined $code )
  79                                 {
  80                                         $code = dec_tpl_str(substr $code, 2, length($code) - 4);
  81                                         $stringtable{$code}++;
  82                                 }
  83                         }
  84                 }
  85         }
  86
  87         close F;
  88 }
  89
  90
  91 if( open C, "| msgcat -" )
  92 {
  93         printf C "msgid \"\"\nmsgstr \"Content-Type: text/plain; charset=UTF-8\"\n\n";
  94
  95         foreach my $key ( sort keys %stringtable )
  96         {
  97                 if( length $key )
  98                 {
  99                         $key =~ s/"/\\"/g;
 100                         printf C "msgid \"%s\"\nmsgstr \"\"\n\n", $key;
 101                 }
 102         }
 103
 104         close C;
 105 }