X-Git-Url: http://dolda2000.com/gitweb/?a=blobdiff_plain;f=ANN.pm;h=1406c4584045c8f4d636875d9ac481aa7166f38e;hb=3e60094e9bdf9494bcddb4ebc036e39ae5d9b5ce;hp=dfa4b41f7a6d2aaac05e8a82b50ca3b27abd3423;hpb=66cb73cd2c475094d76911764addaaf3cc97ea3b;p=utils.git

diff --git a/ANN.pm b/ANN.pm
index dfa4b41..1406c45 100644
--- a/ANN.pm
+++ b/ANN.pm
@@ -57,7 +57,12 @@ sub getlist
     # 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;
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>([^<]+)<\//ig) {
+	if((substr "" . lc $2 , 0, length $name) eq lc $name) {
+	    push @ret, $2;
+	}
+    }
+    # push @ret, $1 while $html =~ /<A\s.*CLASS=HOVERLINE\s.*>.*<FONT.*>([^<>]*$name[^<>]*)<\/FONT/ig;
     
     return @ret;
 }
@@ -76,9 +81,13 @@ sub getid
     # The only way to recognize entries that seems sure is to look
     # after the "HOVERLINE" class.
     
-    (($url) = ($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*$name/i)) || return;
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>([^<]+)<\//ig) {
+	if((substr "" . lc $2 , 0, length $name) eq lc $name) {
+	    return ($1 =~ /id=(\d+)$/)[0];
+	}
+    }
     
-    return ($url =~ /\?id=(\d+)$/)[0];
+    return undef;
 }
 
 sub geturl
@@ -98,8 +107,8 @@ sub getthemes
 	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+\((\<i\>(.*)<\/i>(;\s*)?)?([^<>]+)?\))?\"\s+by\s+([^\(]*[^\(\s])(\s*\(eps (\d+)-(\d+)?\))?/i)) {
+	    #                     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 $parts[5]) {
@@ -142,6 +151,9 @@ sub getseries
     if(($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";