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         $s =~ s/\\/\\\\/g;
  32         return $s;
  33 }
  34
  35
  36 if( open F, "find @ARGV -type f '(' -name '*.htm' -o -name '*.lua' ')' |" )
  37 {
  38         while( defined( my $file = readline F ) )
  39         {
  40                 chomp $file;
  41
  42                 if( open S, "< $file" )
  43                 {
  44                         local $/ = undef;
  45                         my $raw = <S>;
  46                         close S;
  47
  48
  49                         my $text = $raw;
  50
  51                         while( $text =~ s/ ^ .*? (?:translate|translatef|i18n|_) [\n\s]* \( /(/sgx )
  52                         {
  53                                 ( my $code, $text ) = extract_bracketed($text, q{('")});
  54
  55                                 $code =~ s/\\\n/ /g;
  56                                 $code =~ s/^\([\n\s]*//;
  57                                 $code =~ s/[\n\s]*\)$//;
  58
  59                                 my $res = "";
  60                                 my $sub = "";
  61
  62                                 if( $code =~ /^['"]/ )
  63                                 {
  64                                         while( defined $sub )
  65                                         {
  66                                                 ( $sub, $code ) = extract_delimited($code, q{'"}, q{\s*(?:\.\.\s*)?});
  67
  68                                                 if( defined $sub && length($sub) > 2 )
  69                                                 {
  70                                                         $res .= substr $sub, 1, length($sub) - 2;
  71                                                 }
  72                                                 else
  73                                                 {
  74                                                         undef $sub;
  75                                                 }
  76                                         }
  77                                 }
  78                                 elsif( $code =~ /^(\[=*\[)/ )
  79                                 {
  80                                         my $stag = quotemeta $1;
  81                                         my $etag = $stag;
  82                                            $etag =~ s/\[/]/g;
  83
  84                                         ( $res ) = extract_tagged($code, $stag, $etag);
  85
  86                                         $res =~ s/^$stag//;
  87                                         $res =~ s/$etag$//;
  88                                 }
  89
  90                                 $res = dec_lua_str($res);
  91                                 $stringtable{$res}++ if $res;
  92                         }
  93
  94
  95                         $text = $raw;
  96
  97                         while( $text =~ s/ ^ .*? <% -? [:_] /<%/sgx )
  98                         {
  99                                 ( my $code, $text ) = extract_tagged($text, '<%', '%>');
 100
 101                                 if( defined $code )
 102                                 {
 103                                         $code = dec_tpl_str(substr $code, 2, length($code) - 4);
 104                                         $stringtable{$code}++;
 105                                 }
 106                         }
 107                 }
 108         }
 109
 110         close F;
 111 }
 112
 113
 114 if( open C, "| msgcat -" )
 115 {
 116         printf C "msgid \"\"\nmsgstr \"Content-Type: text/plain; charset=UTF-8\"\n\n";
 117
 118         foreach my $key ( sort keys %stringtable )
 119         {
 120                 if( length $key )
 121                 {
 122                         $key =~ s/"/\\"/g;
 123                         printf C "msgid \"%s\"\nmsgstr \"\"\n\n", $key;
 124                 }
 125         }
 126
 127         close C;
 128 }