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/[\s\n]+/ /g;
  16         $s =~ s/\\n/\n/g;
  17         $s =~ s/\\t/\t/g;
  18         $s =~ s/\\(.)/$1/g;
  19         $s =~ s/^ //;
  20         $s =~ s/ $//;
  21         return $s;
  22 }
  23
  24 sub dec_tpl_str
  25 {
  26         my $s = shift;
  27         $s =~ s/-$//;
  28         $s =~ s/[\s\n]+/ /g;
  29         $s =~ s/^ //;
  30         $s =~ s/ $//;
  31         return $s;
  32 }
  33
  34
  35 if( open F, "find @ARGV -type f '(' -name '*.htm' -o -name '*.lua' ')' |" )
  36 {
  37         while( defined( my $file = readline F ) )
  38         {
  39                 chomp $file;
  40
  41                 if( open S, "< $file" )
  42                 {
  43                         local $/ = undef;
  44                         my $raw = <S>;
  45                         close S;
  46
  47
  48                         my $text = $raw;
  49
  50                         while( $text =~ s/ ^ .*? (?:translate|translatef|i18n|_) [\n\s]* \( /(/sgx )
  51                         {
  52                                 ( my $code, $text ) = extract_bracketed($text, q{('")});
  53
  54                                 $code =~ s/\\\n/ /g;
  55                                 $code =~ s/^\([\n\s]*//;
  56                                 $code =~ s/[\n\s]*\)$//;
  57
  58                                 my $res = "";
  59                                 my $sub = "";
  60
  61                                 if( $code =~ /^['"]/ )
  62                                 {
  63                                         while( defined $sub )
  64                                         {
  65                                                 ( $sub, $code ) = extract_delimited($code, q{'"}, q{\s*(?:\.\.\s*)?});
  66
  67                                                 if( length($sub) < 1 ) {
  68                                                         undef $sub;
  69                                                 }
  70
  71                                                 if( defined $sub )
  72                                                 {
  73                                                         $res .= substr $sub, 1, length($sub) - 2;
  74                                                 }
  75                                         }
  76                                 }
  77                                 elsif( $code =~ /^(\[=*\[)/ )
  78                                 {
  79                                         my $stag = quotemeta $1;
  80                                         my $etag = $stag;
  81                                            $etag =~ s/\[/]/g;
  82
  83                                         ( $res ) = extract_tagged($code, $stag, $etag);
  84
  85                                         $res =~ s/^$stag//;
  86                                         $res =~ s/$etag$//;
  87                                 }
  88
  89                                 $res = dec_lua_str($res);
  90                                 $stringtable{$res}++ if $res;
  91                         }
  92
  93
  94                         $text = $raw;
  95
  96                         while( $text =~ s/ ^ .*? <% -? [:_] /<%/sgx )
  97                         {
  98                                 ( my $code, $text ) = extract_tagged($text, '<%', '%>');
  99
 100                                 if( defined $code )
 101                                 {
 102                                         $code = dec_tpl_str(substr $code, 2, length($code) - 4);
 103                                         $stringtable{$code}++;
 104                                 }
 105                         }
 106                 }
 107         }
 108
 109         close F;
 110 }
 111
 112
 113 if( open C, "| msgcat -" )
 114 {
 115         printf C "msgid \"\"\nmsgstr \"Content-Type: text/plain; charset=UTF-8\"\n\n";
 116
 117         foreach my $key ( sort keys %stringtable )
 118         {
 119                 if( length $key )
 120                 {
 121                         $key =~ s/"/\\"/g;
 122                         printf C "msgid \"%s\"\nmsgstr \"\"\n\n", $key;
 123                 }
 124         }
 125
 126         close C;
 127 }