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         $s =~ s/-$//;
  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' -or -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                                 $code =~ s/^\(//; $code =~ s/\)$//;
  54
  55                                 my $res = "";
  56                                 my $sub = "";
  57
  58                                 while( defined $sub )
  59                                 {
  60                                         ( $sub, $code ) = extract_delimited($code, q{'"}, q{\s*(?:\.\.\s*)?});
  61
  62                                         if( defined $sub )
  63                                         {
  64                                                 $res .= substr $sub, 1, length($sub) - 2;
  65                                         }
  66                                 }
  67
  68                                 $res = dec_lua_str($res);
  69                                 $stringtable{$res}++;
  70                         }
  71
  72
  73                         $text = $raw;
  74
  75                         while( $text =~ s/ ^ .*? <% [:_] -? /<%/sgx )
  76                         {
  77                                 ( my $code, $text ) = extract_tagged($text, '<%', '%>');
  78
  79                                 if( defined $code )
  80                                 {
  81                                         $code = dec_tpl_str(substr $code, 2, length($code) - 4);
  82                                         $stringtable{$code}++;
  83                                 }
  84                         }
  85                 }
  86         }
  87
  88         close F;
  89 }
  90
  91
  92 if( open C, "| msgcat -" )
  93 {
  94         printf C "msgid \"\"\nmsgstr \"Content-Type: text/plain; charset=UTF-8\"\n\n";
  95
  96         foreach my $key ( sort keys %stringtable )
  97         {
  98                 if( length $key )
  99                 {
 100                         $key =~ s/"/\\"/g;
 101                         printf C "msgid \"%s\"\nmsgstr \"\"\n\n", $key;
 102                 }
 103         }
 104
 105         close C;
 106 }