ANN: Fixed name detection bug in presence of "The".

diff --git a/ANN.pm b/ANN.pm
index 8b311f4..f10e869 100644 (file)
--- a/ANN.pm
+++ b/ANN.pm
@@ -58,9 +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=\"([^\"]+)\".*>(\<small\>.*\<\/small\>)?([^<]+)<\//ig) {
-       if((substr "" . lc $3 , 0, length $name) eq lc $name) {
-           push @ret, $3;
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\"[^>]*>(<FONT[^>]*>)?(\<small\>.*\<\/small\>)?\s*([^<]+)<\//ig) {
+       if((substr "" . lc $4 , 0, length $name) eq lc $name) {
+           push @ret, $4;
        }
     }
     # push @ret, $1 while $html =~ /<A\s.*CLASS=HOVERLINE\s.*>.*<FONT.*>([^<>]*$name[^<>]*)<\/FONT/ig;
@@ -83,8 +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=\"([^\"]+)\".*>(\<small\>.*\<\/small\>)?([^<]+)<\//ig) {
-       if((substr "" . lc $3 , 0, length $name) eq lc $name) {
+    while($html =~ /<A\s.*CLASS=HOVERLINE\s.*HREF=\"([^\"]+)\"[^>]*>(<FONT[^>]*>)?(\<small\>.*\<\/small\>)?\s*([^<]+)<\//ig) {
+       if((substr "" . lc $4 , 0, length $name) eq lc $name) {
            return ($1 =~ /id=(\d+)$/)[0];
        }
     }