Fix theme regex again.

[utils.git] / ANN.pm

package Anime::ANN;

use POSIX;
use Digest::MD5;
use LWP::UserAgent;
use HTML::Entities;

$ver = "0.1";
$ua = LWP::UserAgent->new;
$ua->agent("ANNData/$ver ");
$usecache = 1;

sub _get
{
    my($uri, $cname, $res);
    ($uri) = @_;

    $cname = $ENV{"HOME"} . "/.ann/";
    mkdir $cname unless -e $cname;
    $cname .= "cache/";
    mkdir $cname unless -e $cname;
    $cname .= Digest::MD5::md5_hex $uri;
    if($usecache && -e $cname) {
	my(@s);
	@s = stat $cname;
	if((time - $s[9]) < 86400) {
	    my($buf);
	    open CACHE, "<:utf8", $cname;
	    $res .= $buf while read CACHE, $buf, 1024;
	    close CACHE;
	    return $res;
	}
    }
    
    $res = $ua->request(HTTP::Request->new("GET", "$uri"));
    
    if(open CACHE, ">:utf8", $cname) {
	print CACHE $res->content;
	close CACHE;
    }
    
    return undef unless $res->is_success;
    return $res->content;
}

sub getlist
{
    my($name, $il, $html, @ret);
    ($name) = @_;
    
    $il = uc(($name =~ /^(.)/)[0]);
    $il = "9" if (!($il =~ /[A-Z]/));
    if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) {
	return undef;
    }
    
    # The only way to recognize entries that seems sure is to look
    # after the "HOVERLINE" class.
    
    push @ret, $1 while $html =~ /<A\s.*CLASS=HOVERLINE\s.*>.*<FONT.*>([^<>]*$name[^<>]*)<\/FONT/ig;
    
    return @ret;
}

sub getid
{
    my($name, $il, $html, $url);
    ($name) = @_;
    
    $il = uc(($name =~ /^(.)/)[0]);
    $il = "9" if (!($il =~ /[A-Z]/));
    if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) {
	return undef;
    }
    
    # The only way to recognize entries that seems sure is to look
    # after the "HOVERLINE" class.
    
    (($url) = ($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*$name/i)) || return;
    
    return ($url =~ /\?id=(\d+)$/)[0];
}

sub geturl
{
    my($id);
    ($id) = @_;
    
    return "http://www.animenewsnetwork.com/encyclopedia/anime.php?id=$id";
}

sub getthemes
{
    my($html, $kind, @ret);
    ($html, $kind) = @_;
    
    if($html =~ /$kind theme:<\/b>\n/igc) {
	my(@parts, $ct, $buf);
	while($html =~ /\G\<br\>&nbsp; &nbsp; (([^<>]|\<i\>|<\/i>)+)/igc) {
	    $buf = $1;
	    #                     0  1            2           3     4     5        6        7                      8              9         10    11
	    if(@parts = ($buf =~ /(\#(\d+):)?\s*\"([^\"\(]+\S)(\s*\((\<i\>(.*)<\/i>(;\s*)?)?([^<>]+)?\))?\"\s+by\s+([^\(]*[^\(\s])(\s*\(eps (\d+)-(\d+)?\))?/i)) {
		$ct = {};
		$ct->{"num"} = $parts[1] if defined $parts[1];
		if(defined $parts[5]) {
		    $ct->{"tit"} = decode_entities($parts[5]);
		    $ct->{"jat"} = decode_entities($parts[2]) if defined $parts[2];
		} else {
		    $ct->{"tit"} = decode_entities($parts[2]) if defined $parts[2];
		}
		$ct->{"ent"} = decode_entities($parts[7]) if defined $parts[7];
		$ct->{"prf"} = decode_entities($parts[8]) if defined $parts[8];
		$ct->{"fep"} = $parts[10] if defined $parts[10];
		$ct->{"lep"} = $parts[11] if defined $parts[11];
		push @ret, $ct;
	    }
	}
    }
    
    return \@ret;
}

sub getseries
{
    my($id, $buf, $html, %ret);
    ($id) = @_;
    
    if(!($html = _get geturl $id)) {
	return undef;
    }
    
    $ret{"url"} = geturl $id;
    ($buf) = ($html =~ /\<TITLE\>Anime News Network - ([^<]*)<\/TITLE>/);
    if($buf =~ /\([^\)]+\)$/) {
	($ret{"name"}, $ret{"type"}) = ($buf =~ /^(.*[^\s])\s*\(([^\)]+)\)$/);
    } else {
	$ret{"name"} = $buf;
    }
    if(($buf) = ($html =~ /vintage:<\/b>\n([^<]+)</is)) {
	$ret{"vtg"} = $buf;
    }
    if(($buf) = ($html =~ /number of episodes:<\/b>\n([^<]+)</is)) {
	$ret{"eps"} = $buf;
    }
    if(($buf) = ($html =~ /genres:<\/b>\n([^<]+)</is)) {
	$ret{"gnr"} = [split /, /, $buf];
    }
    $buf = getthemes $html, "opening";
    $ret{"op"} = $buf if(@{$buf});
    $buf = getthemes $html, "ending";
    $ret{"ed"} = $buf if(@{$buf});
    
    return \%ret;
}