build/i18n-scan.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5 use Text::Balanced qw(extract_tagged gen_delimited_pat);
   6 use POSIX;
   7
   8 POSIX::setlocale(POSIX::LC_ALL, "C");
   9
  10 @ARGV >= 1 || die "Usage: $0 <source directory>\n";
  11
  12
  13 my %stringtable;
  14
  15 sub dec_lua_str
  16 {
  17         my $s = shift;
  18         $s =~ s/\\n/\n/g;
  19         $s =~ s/\\t/\t/g;
  20         $s =~ s/\\(.)/$1/sg;
  21         $s =~ s/[\s\n]+/ /g;
  22         $s =~ s/^ //;
  23         $s =~ s/ $//;
  24         return $s;
  25 }
  26
  27 sub dec_tpl_str
  28 {
  29         my $s = shift;
  30         $s =~ s/-$//;
  31         $s =~ s/[\s\n]+/ /g;
  32         $s =~ s/^ //;
  33         $s =~ s/ $//;
  34         $s =~ s/\\/\\\\/g;
  35         return $s;
  36 }
  37
  38 if( open F, "find @ARGV -type f '(' -name '*.htm' -o -name '*.lua' -o -name '*.js' ')' | sort |" )
  39 {
  40         while( defined( my $file = readline F ) )
  41         {
  42                 chomp $file;
  43
  44                 if( open S, "< $file" )
  45                 {
  46                         local $/ = undef;
  47                         my $raw = <S>;
  48                         close S;
  49
  50                         my $text = $raw;
  51                         my $line = 1;
  52
  53                         while ($text =~ s/ ^ (.*?) (?:translate|translatef|i18n|_) ([\n\s]*) \( //sgx)
  54                         {
  55                                 my ($prefix, $suffix) = ($1, $2);
  56                                 my $code;
  57                                 my $res = "";
  58                                 my $sub = "";
  59
  60                                 $line += () = $prefix =~ /\n/g;
  61
  62                                 my $position = "$file:$line";
  63
  64                                 $line += () = $suffix =~ /\n/g;
  65
  66                                 while (defined $sub)
  67                                 {
  68                                         undef $sub;
  69
  70                                         if ($text =~ /^ ([\n\s]*(?:\.\.[\n\s]*)?) (\[=*\[) /sx)
  71                                         {
  72                                                 my $ws = $1;
  73                                                 my $stag = quotemeta $2;
  74                                                 (my $etag = $stag) =~ y/[/]/;
  75
  76                                                 ($sub, $text) = extract_tagged($text, $stag, $etag, q{\s*(?:\.\.\s*)?});
  77
  78                                                 $line += () = $ws =~ /\n/g;
  79
  80                                                 if (defined($sub) && length($sub)) {
  81                                                         $line += () = $sub =~ /\n/g;
  82
  83                                                         $sub =~ s/^$stag//;
  84                                                         $sub =~ s/$etag$//;
  85                                                         $res .= $sub;
  86                                                 }
  87                                         }
  88                                         elsif ($text =~ /^ ([\n\s]*(?:\.\.[\n\s]*)?) (['"]) /sx)
  89                                         {
  90                                                 my $ws = $1;
  91                                                 my $quote = $2;
  92                                                 my $re = gen_delimited_pat($quote, '\\');
  93
  94                                                 if ($text =~ m/\G\s*(?:\.\.\s*)?($re)/gcs)
  95                                                 {
  96                                                         $sub = $1;
  97                                                         $text = substr $text, pos $text;
  98                                                 }
  99
 100                                                 $line += () = $ws =~ /\n/g;
 101
 102                                                 if (defined($sub) && length($sub)) {
 103                                                         $line += () = $sub =~ /\n/g;
 104
 105                                                         $sub =~ s/^$quote//;
 106                                                         $sub =~ s/$quote$//;
 107                                                         $res .= $sub;
 108                                                 }
 109                                         }
 110                                 }
 111
 112                                 if (defined($res))
 113                                 {
 114                                         $res = dec_lua_str($res);
 115
 116                                         if ($res) {
 117                                                 $stringtable{$res} ||= [ ];
 118                                                 push @{$stringtable{$res}}, $position;
 119                                         }
 120                                 }
 121                         }
 122
 123
 124                         $text = $raw;
 125                         $line = 1;
 126
 127                         while( $text =~ s/ ^ (.*?) <% -? [:_] /<%/sgx )
 128                         {
 129                                 $line += () = $1 =~ /\n/g;
 130
 131                                 ( my $code, $text ) = extract_tagged($text, '<%', '%>');
 132
 133                                 if( defined $code )
 134                                 {
 135                                         my $position = "$file:$line";
 136
 137                                         $line += () = $code =~ /\n/g;
 138
 139                                         $code = dec_tpl_str(substr $code, 2, length($code) - 4);
 140
 141                                         $stringtable{$code} ||= [];
 142                                         push @{$stringtable{$code}}, $position;
 143                                 }
 144                         }
 145                 }
 146         }
 147
 148         close F;
 149 }
 150
 151
 152 if( open C, "| msgcat -" )
 153 {
 154         printf C "msgid \"\"\nmsgstr \"Content-Type: text/plain; charset=UTF-8\"\n\n";
 155
 156         foreach my $key ( sort keys %stringtable )
 157         {
 158                 if( length $key )
 159                 {
 160                         my @positions = @{$stringtable{$key}};
 161
 162                         $key =~ s/\\/\\\\/g;
 163                         $key =~ s/\n/\\n/g;
 164                         $key =~ s/\t/\\t/g;
 165                         $key =~ s/"/\\"/g;
 166
 167                         printf C "#: %s\nmsgid \"%s\"\nmsgstr \"\"\n\n",
 168                                 join(' ', @positions), $key;
 169                 }
 170         }
 171
 172         close C;
 173 }