Changeset 5231 for lang/perl/Encode-JP-Mobile/trunk/tools/softbank-scrape-autosjis.pl

Timestamp:

01/22/08 07:53:40 (10 months ago)

Author:

tokuhirom

Message:

r5458@skinny (orig r5198): chiba | 2008-01-22 01:00:22 +0900
softbank-scrape-autosjisの役割を
dat/softbank-unicode2sjis_auto.yaml
を作るだけとした。
で、そのsoftbank-unicode2sjis_auto.yamlを同梱

Files:

• lang/perl/Encode-JP-Mobile/trunk/tools/softbank-scrape-autosjis.pl (modified) (3 diffs)

lang/perl/Encode-JP-Mobile/trunk/tools/softbank-scrape-autosjis.pl

@@ -6 +6 @@
 use FindBin;
 
+use Data::Dumper;
+$Data::Dumper::Terse++;
+
 # how to make 103-111-HTML_2.0.0.txt
 # 1. get PDF from http://www2.developers.softbankmobile.co.jp/dp/tool_dl/download.php?docid=120&companyid=
@@ -12 +15 @@
 
 my $pdf_text_file = shift or die "Usage: softbank-scrape-autosjis.pl 103-111-HTML_2.0.0.txt";
-my $fh =file($pdf_text_file)->openr;
+my $pdf_fh =file($pdf_text_file)->openr;
 
 my %map;
-while (my $line = <$fh>) {
+while (my $line = <$pdf_fh>) {
     chomp $line;
+    next if $line !~ /^&#\d\d\d\d\d;\s*&#x/;
+
     my @codes = split /\s+/, $line;
-
-    if ( @codes != 4 || $codes[0] =~ /^&#x/ ) {
-        next;
-    }
+    next if @codes != 4;
 
     my $unicode  = strip_entity_ref_mark($codes[1]);
@@ -28 +30 @@
     $map{ $unicode } = $shiftjis;
 }
+close $pdf_fh;
 
+print Dump(\%map);
 
-
-my $table_file = "$FindBin::Bin/../dat/softbank-table.yaml";
-my $table = YAML::LoadFile($table_file);
-
-for my $emoji (@$table) {
-    $emoji->{sjis_auto} = $map{ $emoji->{unicode} };
-}
-
-print Dump $table;
 
 sub strip_entity_ref_mark {