二重規制やっと解けた記念
sub scrapeimgurl(@) {
# make each image's url from a thumbnail page.
my ( $ofn ) = @_;
my $size = ( -s $ofn );
my $content = "";
open( DM, "<", $ofn ) || die $!;
read( DM, $content, $size );
close ( DM );
$content =~ s/<img/¥n<img/g;
my @html = split /¥n/, $content;
my ( $retLink, $retName ) = ( "", "" );
my %tmp = ();
for ( @html ) {
$tmp{$1}++ if ( m|(<img src=".*?[a-f0-9]{40}-.*?" class=)| );
}
my $url = "";
while ( my ( $key, $val ) = ( each %tmp ) ) {
$url = $key, last if ( $val == 1 );
}
if ( $url =~ /src="(.*?)"/ ) {
( $retLink = $1 ) =~ s/&amp;/&/g;
if ( $url =~ m{alt="(.+?)"} ) {
( $retName = $1 ) =~ tr/¥/ /_/d;
} else {
$retName = "unknown.jpg";
}
}
return ( $retLink, $retName );
}