| 1 | package Anime::ANN; |
| 2 | |
| 3 | use POSIX; |
| 4 | use Digest::MD5; |
| 5 | use LWP::UserAgent; |
| 6 | use HTML::Entities; |
| 7 | |
| 8 | $ver = "0.1"; |
| 9 | $ua = LWP::UserAgent->new; |
| 10 | $ua->agent("ANNData/$ver "); |
| 11 | $usecache = 1; |
| 12 | |
| 13 | sub _get |
| 14 | { |
| 15 | my($uri, $cname, $res); |
| 16 | ($uri) = @_; |
| 17 | |
| 18 | $cname = $ENV{"HOME"} . "/.ann/"; |
| 19 | mkdir $cname unless -e $cname; |
| 20 | $cname .= "cache/"; |
| 21 | mkdir $cname unless -e $cname; |
| 22 | $cname .= Digest::MD5::md5_hex $uri; |
| 23 | if($usecache && -e $cname) { |
| 24 | my(@s); |
| 25 | @s = stat $cname; |
| 26 | if((time - $s[9]) < 86400) { |
| 27 | my($buf); |
| 28 | open CACHE, "<:utf8", $cname; |
| 29 | $res .= $buf while read CACHE, $buf, 1024; |
| 30 | close CACHE; |
| 31 | return $res; |
| 32 | } |
| 33 | } |
| 34 | |
| 35 | $res = $ua->request(HTTP::Request->new("GET", "$uri")); |
| 36 | |
| 37 | if(open CACHE, ">:utf8", $cname) { |
| 38 | print CACHE $res->decoded_content; |
| 39 | close CACHE; |
| 40 | } |
| 41 | |
| 42 | return undef unless $res->is_success; |
| 43 | return $res->decoded_content; |
| 44 | } |
| 45 | |
| 46 | sub getlist |
| 47 | { |
| 48 | my($name, $il, $html, @ret); |
| 49 | ($name) = @_; |
| 50 | |
| 51 | $name = ($name =~ /^(the\s+)?(.*)$/i)[1]; |
| 52 | $il = uc(($name =~ /^\W*(.)/)[0]); |
| 53 | $il = "9" if (!($il =~ /[A-Z]/)); |
| 54 | if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) { |
| 55 | return undef; |
| 56 | } |
| 57 | |
| 58 | # The only way to recognize entries that seems sure is to look |
| 59 | # after the "HOVERLINE" class. |
| 60 | |
| 61 | while($html =~ /<A\s+CLASS="HOVERLINE"\s+HREF="([^"]+)"[^>]*>(<FONT[^>]*>)?(<small>[^<]*<\/small>)?\s*([^<]+)<\//ig) { |
| 62 | if((substr "" . lc $4 , 0, length $name) eq lc $name) { |
| 63 | push @ret, $4; |
| 64 | } |
| 65 | } |
| 66 | # push @ret, $1 while $html =~ /<A\s.*CLASS=HOVERLINE\s.*>.*<FONT.*>([^<>]*$name[^<>]*)<\/FONT/ig; |
| 67 | |
| 68 | return @ret; |
| 69 | } |
| 70 | |
| 71 | sub getid |
| 72 | { |
| 73 | my($name, $il, $html, $url); |
| 74 | ($name) = @_; |
| 75 | |
| 76 | $name = ($name =~ /^(the\s+)?(.*)$/i)[1]; |
| 77 | $il = uc(($name =~ /^\W*(.)/)[0]); |
| 78 | $il = "9" if (!($il =~ /[A-Z]/)); |
| 79 | if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) { |
| 80 | return undef; |
| 81 | } |
| 82 | |
| 83 | # The only way to recognize entries that seems sure is to look |
| 84 | # after the "HOVERLINE" class. |
| 85 | |
| 86 | while($html =~ /<A\s+CLASS="HOVERLINE"\s+HREF="([^"]+)"[^>]*>(<FONT[^>]*>)?(<small>[^<]*<\/small>)?\s*([^<]+)<\//ig) { |
| 87 | if((substr "" . lc $4 , 0, length $name) eq lc $name) { |
| 88 | return ($1 =~ /id=(\d+)$/)[0]; |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | return undef; |
| 93 | } |
| 94 | |
| 95 | sub geturl |
| 96 | { |
| 97 | my($id); |
| 98 | ($id) = @_; |
| 99 | |
| 100 | return "http://www.animenewsnetwork.com/encyclopedia/anime.php?id=$id"; |
| 101 | } |
| 102 | |
| 103 | sub getthemes |
| 104 | { |
| 105 | my($html, $kind, @ret); |
| 106 | ($html, $kind) = @_; |
| 107 | |
| 108 | if($html =~ /$kind theme:<\/strong>\s*\n/igc) { |
| 109 | my(@parts, $ct, $buf); |
| 110 | while($html =~ /\G\s*\<div class=\"tab\"\>(([^<>]|\<i\>|<\/i>)+)(<span[^<>]*>[^<>]*<img[^<>]*>[^<>]*<\/span>)?<\/div>/igc) { |
| 111 | $buf = $1; |
| 112 | # 0 1 2 3 4 5 6 7 8 9 10 1112 |
| 113 | if(@parts = ($buf =~ /(\#(\d+):)?\s*\"([^\"\(]+\S)(\s*\((\<i\>(.*)<\/i>( - \s*)?)?([^<>]+)?\))?\"\s+by\s+([^\(]*[^\(\s])(\s*\(eps? (\d+)(-(\d+))?\))?/i)) { |
| 114 | $ct = {}; |
| 115 | $ct->{"num"} = $parts[1] if defined $parts[1]; |
| 116 | if(defined $parts[5]) { |
| 117 | $ct->{"tit"} = decode_entities($parts[5]); |
| 118 | $ct->{"jat"} = decode_entities($parts[2]) if defined $parts[2]; |
| 119 | } else { |
| 120 | $ct->{"tit"} = decode_entities($parts[2]) if defined $parts[2]; |
| 121 | } |
| 122 | $ct->{"ent"} = decode_entities($parts[7]) if defined $parts[7]; |
| 123 | $ct->{"prf"} = decode_entities($parts[8]) if defined $parts[8]; |
| 124 | $ct->{"fep"} = $parts[10] if defined $parts[10]; |
| 125 | $ct->{"lep"} = $parts[12] if defined $parts[12]; |
| 126 | push @ret, $ct; |
| 127 | } |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | return \@ret; |
| 132 | } |
| 133 | |
| 134 | sub getseries |
| 135 | { |
| 136 | my($id, $buf, $html, %ret); |
| 137 | ($id) = @_; |
| 138 | |
| 139 | if(!($html = _get geturl $id)) { |
| 140 | return undef; |
| 141 | } |
| 142 | |
| 143 | $ret{"url"} = geturl $id; |
| 144 | ($buf) = ($html =~ /\<title\>([^<]*) - Anime News Network<\/title>/); |
| 145 | if($buf =~ /\([^\)]+\)$/) { |
| 146 | ($ret{"name"}, $ret{"type"}) = ($buf =~ /^(.*[^\s])\s*\(([^\)]+)\)$/); |
| 147 | } else { |
| 148 | $ret{"name"} = $buf; |
| 149 | } |
| 150 | if(($buf) = ($html =~ /vintage:<\/strong>\s*\n\s*<span>([^<]+)</is)) { |
| 151 | $ret{"vtg"} = $buf; |
| 152 | } |
| 153 | if(($buf) = ($html =~ /number of episodes:<\/strong>\s*\n\s*<span>([^<]+)</is)) { |
| 154 | $ret{"eps"} = $buf; |
| 155 | } |
| 156 | if(($buf) = ($html =~ /genres:<\/strong>\s*\n\s*([^<]+)</is)) { |
| 157 | $ret{"gnr"} = [split /, /, $buf]; |
| 158 | } |
| 159 | $buf = getthemes $html, "opening"; |
| 160 | $ret{"op"} = $buf if(@{$buf}); |
| 161 | $buf = getthemes $html, "ending"; |
| 162 | $ret{"ed"} = $buf if(@{$buf}); |
| 163 | |
| 164 | return \%ret; |
| 165 | } |