Changeset Navigation

Changeset 487

Timestamp:

04/02/06 11:41:30

Author:

miyagawa

Message:

support Guess in decode_content

Files:

r481	r487
5	5
6	6	use Encode ();
	7	use Encode::Guess;
7	8	use List::Util qw(min);
8	9	use HTML::Entities;
…	…
36	37	my $content = $res->content;
37	38
	39	# 1) get charset from HTTP Content-Type header
38	40	my $charset = ($res->http_response->content_type =~ /charset=([\w\-]+)/)[0];
39		unless ($charset) {
40		$charset = ( $content =~ m!<meta http-equiv="Content-Type" content=".*charset=([\w\-]+)"! )[0] \|\| "utf-8";
41		}
	41
	42	# 2) if there's not, try META tag
	43	$charset \|\|= ( $content =~ m!<meta http-equiv="Content-Type" content=".*charset=([\w\-]+)"!i )[0];
	44
	45	# 3) if there's not still, try Guess
	46	# xxx it supports Japanese only
	47	my @guess = qw(utf-8 euc-jp shift_jis);
	48	$charset = guess_encoding($content, @guess);
42	49
43	50	return Encode::decode($charset, $content);