wikidata/migration/grep2swml.pl

#!/usr/bin/perl
use strict;
use warnings;

use Encode::EUCJPSW;

my @entry;
while (<>) {
  if (s/^([^:]+)://) {
    my $path = $1;
    if ($path =~ s[^page/][]) {
      $path =~ s/\.txt$//;
      $path =~ s[\.ns/][//]g;
      $path =~ s/([0-9A-Fa-f]{2})/pack 'C', hex $1/ge;
      $path = Encode::decode ('euc-jp-sw', $path);
      $path = "[[$path]]";
    }
    push @entry, [$path, Encode::decode ('euc-jp-sw', $_)];
  } else {
    warn $_;
  }
}

sub e ($) {
  my $s = shift;
  $s =~ s/([\[\]']|^[,-=*])/+{qw[' 1 , 1 - 1 = 1 * 1]}->{$1} ? "[SPAN[$1]]" : "'''$1'''"/ge;
  return $s;
} # e

binmode STDOUT, ':encoding(utf8)';
for (1..$#entry) {
  printf "[%d] %s\n[PRE(code)[\n%s]PRE]\n\n", $_, $entry[$_]->[0], e ($entry[$_]->[1]);
}
1	#!/usr/bin/perl
2	use strict;
3	use warnings;
4
5	use Encode::EUCJPSW;
6
7	my @entry;
8	while (<>) {
9	if (s/^([^:]+)://) {
10	my $path = $1;
11	if ($path =~ s[^page/][]) {
12	$path =~ s/\.txt$//;
13	$path =~ s[\.ns/][//]g;
14	$path =~ s/([0-9A-Fa-f]{2})/pack 'C', hex $1/ge;
15	$path = Encode::decode ('euc-jp-sw', $path);
16	$path = "[[$path]]";
17	}
18	push @entry, [$path, Encode::decode ('euc-jp-sw', $_)];
19	} else {
20	warn $_;
21	}
22	}
23
24	sub e ($) {
25	my $s = shift;
26	$s =~ s/([\[\]']\|^[,-=])/+{qw[' 1 , 1 - 1 = 1 1]}->{$1} ? "[SPAN[$1]]" : "'''$1'''"/ge;
27	return $s;
28	} # e
29
30	binmode STDOUT, ':encoding(utf8)';
31	for (1..$#entry) {
32	printf "[%d] %s\n[PRE(code)[\n%s]PRE]\n\n", $_, $entry[$_]->[0], e ($entry[$_]->[1]);
33	}