From: fredrik <fredrik@959494ce-11ee-0310-bf91-de5d638817bd>
Date: Thu, 18 May 2006 01:39:21 +0000 (+0000)
Subject: Update for new ann html schema.
X-Git-Url: http://git.dolda2000.com/gitweb/?a=commitdiff_plain;h=3b5018db546d352a46cae32a64f17a72196a1275;p=utils.git

Update for new ann html schema.


git-svn-id: svn+ssh://svn.dolda2000.com/srv/svn/repos/src/utils@608 959494ce-11ee-0310-bf91-de5d638817bd
---

diff --git a/ANN.pm b/ANN.pm
index 1406c45..746cf15 100644
--- a/ANN.pm
+++ b/ANN.pm
@@ -48,6 +48,7 @@ sub getlist
     my($name, $il, $html, @ret);
     ($name) = @_;
     
+    $name = ($name =~ /^(the\s+)?(.*)$/i)[1];
     $il = uc(($name =~ /^(.)/)[0]);
     $il = "9" if (!($il =~ /[A-Z]/));
     if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) {
@@ -57,9 +58,9 @@ sub getlist
     # The only way to recognize entries that seems sure is to look
     # after the "HOVERLINE" class.
     
-    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>([^<]+)<\//ig) {
-	if((substr "" . lc $2 , 0, length $name) eq lc $name) {
-	    push @ret, $2;
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>(\<small\>.*\<\/small\>)?([^<]+)<\//ig) {
+	if((substr "" . lc $3 , 0, length $name) eq lc $name) {
+	    push @ret, $3;
 	}
     }
     # push @ret, $1 while $html =~ /<A\s.*CLASS=HOVERLINE\s.*>.*<FONT.*>([^<>]*$name[^<>]*)<\/FONT/ig;
@@ -72,6 +73,7 @@ sub getid
     my($name, $il, $html, $url);
     ($name) = @_;
     
+    $name = ($name =~ /^(the\s+)?(.*)$/i)[1];
     $il = uc(($name =~ /^(.)/)[0]);
     $il = "9" if (!($il =~ /[A-Z]/));
     if(!($html = _get "http://www.animenewsnetwork.com/encyclopedia/anime.php?list=$il")) {
@@ -81,8 +83,8 @@ sub getid
     # The only way to recognize entries that seems sure is to look
     # after the "HOVERLINE" class.
     
-    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>([^<]+)<\//ig) {
-	if((substr "" . lc $2 , 0, length $name) eq lc $name) {
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\".*>(\<small\>.*\<\/small\>)?([^<]+)<\//ig) {
+	if((substr "" . lc $3 , 0, length $name) eq lc $name) {
 	    return ($1 =~ /id=(\d+)$/)[0];
 	}
     }
@@ -103,12 +105,12 @@ sub getthemes
     my($html, $kind, @ret);
     ($html, $kind) = @_;
     
-    if($html =~ /$kind theme:<\/b>\n/igc) {
+    if($html =~ /$kind theme:<\/strong>\s*\n/igc) {
 	my(@parts, $ct, $buf);
-	while($html =~ /\G\<br\>&nbsp; &nbsp; (([^<>]|\<i\>|<\/i>)+)/igc) {
+	while($html =~ /\G\s*\<div\>(([^<>]|\<i\>|<\/i>)+)<\/div>/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)) {
+	    #                     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]) {
@@ -145,13 +147,13 @@ sub getseries
     } else {
 	$ret{"name"} = $buf;
     }
-    if(($buf) = ($html =~ /vintage:<\/b>\n([^<]+)</is)) {
+    if(($buf) = ($html =~ /vintage:<\/strong>\s*\n\s*<span>([^<]+)</is)) {
 	$ret{"vtg"} = $buf;
     }
-    if(($buf) = ($html =~ /number of episodes:<\/b>\n([^<]+)</is)) {
+    if(($buf) = ($html =~ /number of episodes:<\/strong>\s*\n\s*<span>([^<]+)</is)) {
 	$ret{"eps"} = $buf;
     }
-    if(($buf) = ($html =~ /genres:<\/b>\n([^<]+)</is)) {
+    if(($buf) = ($html =~ /genres:<\/strong>\s*\n\s*([^<]+)</is)) {
 	$ret{"gnr"} = [split /, /, $buf];
     }
     $buf = getthemes $html, "opening";