Description: fix ISBNdb book plugin
Author: Marc Deslauriers <marc.deslauriers@canonical.com>
Forwarded: http://forums.gcstar.org/viewtopic.php?id=18072

Index: gcstar-1.7.0/lib/gcstar/GCPlugins/GCbooks/GCISBNdb.pm
===================================================================
--- gcstar-1.7.0.orig/lib/gcstar/GCPlugins/GCbooks/GCISBNdb.pm	2011-07-17 08:23:00.000000000 -0400
+++ gcstar-1.7.0/lib/gcstar/GCPlugins/GCbooks/GCISBNdb.pm	2013-08-07 16:47:41.752463960 -0400
@@ -42,49 +42,51 @@
         if ($self->{parsingList})
         {
 
-            if (($tagname eq 'div') && ($attr->{class} eq 'bookInfo') && ($self->{searchField} ne 'isbn'))
+            if (($tagname eq 'div') && ($attr->{class} eq 'bookSnippetBasic'))
             {
                 $self->{isBook} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/d/book/") >= 0) && ($self->{isBook}))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/book/") >= 0) && ($self->{isBook}))
             {
                 $self->{itemIdx}++;
-                $self->{itemsList}[$self->{itemIdx}]->{url} = "http://isbndb.com" . $attr->{href};
+                $self->{itemsList}[$self->{itemIdx}]->{url} = $attr->{href};
+            }
+            elsif (($tagname eq 'h1') && ($self->{isBook}))
+            {
                 $self->{isTitle} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/d/person/") >= 0) && ($self->{isBook}))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/author/") >= 0) && ($self->{isBook}))
             {
                 $self->{isAuthor} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/d/publisher/") >= 0) && ($self->{isBook}))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/publisher/") >= 0) && ($self->{isBook}))
             {
                 $self->{isPublisher} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{onclick},"isbndbTrackBuy") >= 0) && ($self->{itemIdx} eq '-1'))
-            {
-                $self->{itemIdx}++;
-                $self->{itemsList}[$self->{itemIdx}]->{url} = $self->{loadedUrl} ;
-            }
-            elsif (($tagname eq 'span') && ($attr->{class} eq 'inactive'))
+            elsif ($tagname eq 'span')
             {
                 $self->{isBook} = 0 ;
             }
         }
         else
         {
-            if ($tagname eq 'title')
+            if (($tagname eq 'div') && ($attr->{class} eq 'bookSnippetBasic'))
+            {
+                $self->{isBook} = 1 ;
+            }
+            if (($tagname eq 'h1') && ($self->{isBook}))
             {
                 $self->{isTitle} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/d/person/") >= 0))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/author/") >= 0))
             {
                 $self->{isAuthor} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/d/publisher/") >= 0))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/publisher/") >= 0))
             {
                 $self->{isPublisher} = 1 ;
             }
-            elsif (($tagname eq 'a') && ( index($attr->{href},"/c/Library_Shelves/Dewey") >= 0))
+            elsif (($tagname eq 'a') && ( index($attr->{href},"/subject/") >= 0))
             {
                 $self->{isGenre} = 1 ;
             }
@@ -104,6 +106,10 @@
                    $self->{curInfo}->{cover} = $array[0];
                 }
             }
+            elsif (($tagname eq 'div') && ($attr->{id} eq 'rightColumn'))
+            {
+                $self->{isBook} = 0 ;
+            }
         }
     }
 
@@ -127,14 +133,34 @@
             }
             elsif ($self->{isAuthor})
             {
+                my @nom_prenom = split(/,/,$origtext);
+                # Enleve les blancs en debut de chaine
+                $nom_prenom[0] =~ s/^\s//;
+                $nom_prenom[1] =~ s/^\s//;
+                # Enleve les blancs en fin de chaine
+                $nom_prenom[0] =~ s/\s+$//;
+                $nom_prenom[1] =~ s/\s+$//;
                 if ($self->{itemsList}[$self->{itemIdx}]->{authors} eq '')
                 {
-                   $self->{itemsList}[$self->{itemIdx}]->{authors} = $origtext;
+                   if ($nom_prenom[1] ne '')
+                   {
+                      $self->{itemsList}[$self->{itemIdx}]->{authors} = $nom_prenom[1] ." " . $nom_prenom[0];
+                   }
+                   else
+                   {
+                      $self->{itemsList}[$self->{itemIdx}]->{authors} = $nom_prenom[0];
+                   }
                 }
                 else
                 {
-                   $self->{itemsList}[$self->{itemIdx}]->{authors} .= ', ';
-                   $self->{itemsList}[$self->{itemIdx}]->{authors} .= $origtext;
+                   if ($nom_prenom[1] ne '')
+                   {
+                      $self->{itemsList}[$self->{itemIdx}]->{authors} .= ", " . $nom_prenom[1] ." " . $nom_prenom[0];
+                   }
+                   else
+                   {
+                      $self->{itemsList}[$self->{itemIdx}]->{authors} .= ", " . $nom_prenom[0];
+                   }
                 }
                 $self->{isAuthor} = 0 ;
             }
@@ -157,8 +183,35 @@
             }
             elsif ($self->{isAuthor})
             {
-                $self->{curInfo}->{authors} .= $origtext;
-                $self->{curInfo}->{authors} .= ",";
+                my @nom_prenom = split(/,/,$origtext);
+                # Enleve les blancs en debut de chaine
+                $nom_prenom[0] =~ s/^\s//;
+                $nom_prenom[1] =~ s/^\s//;
+                # Enleve les blancs en fin de chaine
+                $nom_prenom[0] =~ s/\s+$//;
+                $nom_prenom[1] =~ s/\s+$//;
+                if ($self->{curInfo}->{authors} eq '')
+                {
+                   if ($nom_prenom[1] ne '')
+                   {
+                      $self->{curInfo}->{authors} = $nom_prenom[1] ." " . $nom_prenom[0];
+                   }
+                   else
+                   {
+                      $self->{curInfo}->{authors} = $nom_prenom[0];
+                   }
+                }
+                else
+                {
+                   if ($nom_prenom[1] ne '')
+                   {
+                      $self->{curInfo}->{authors} .= ", " . $nom_prenom[1] ." " . $nom_prenom[0];
+                   }
+                   else
+                   {
+                      $self->{curInfo}->{authors} .= ", " . $nom_prenom[0];
+                   }
+                }
                 $self->{isAuthor} = 0 ;
             }
             elsif ($self->{isPublisher})
@@ -168,9 +221,9 @@
             }
             elsif ($self->{isAnalyse})
             {
-                $self->{isFormat} = 1 if ($origtext =~ m/Book Details:/i);
-                $self->{isDescription} = 1 if ($origtext =~ m/Notes:/i);
-                $self->{isDescription} = 1 if ($origtext =~ m/Summary:/i);
+                $self->{isFormat} = 1 if ($origtext =~ m/Book Details/i);
+                $self->{isDescription} = 1 if ($origtext =~ m/Notes/i);
+                $self->{isDescription} = 1 if ($origtext =~ m/Summary/i);
 
                 $self->{isAnalyse} = 0 ;
             }
@@ -233,14 +286,25 @@
             }
             elsif ($self->{isGenre})
             {
-                 my @array = split(/--/,$origtext);
+                 $origtext =~ s/ -- /\//gi;
 
-                 $self->{curInfo}->{genre} = $array[1];
                  # Enleve les blancs en debut de chaine
-                 $self->{curInfo}->{genre} =~ s/^\s+//;
+                 $origtext =~ s/^\s+//;
+                 # Enleve les blancs en fin de chaine
+                 $origtext =~ s/\s+$//g;
+
+                 if ($self->{curInfo}->{genre} eq '')
+                 {
+                     $self->{curInfo}->{genre} = $origtext;
+                 }
+                 else
+                 {
+                     $self->{curInfo}->{genre} .= ", " . $origtext;
+                 }
                  $self->{isGenre} = 0 ;
             }
-            elsif (($origtext =~ m/ISBN:/i) && ($self->{curInfo}->{isbn} eq ''))
+            elsif ((($origtext =~ m/ISBN10:/i) && ($self->{curInfo}->{isbn} eq '')) ||
+                    ($origtext =~ m/ISBN13:/i))
             {
                 my @array = split(/:/,$origtext);
 
@@ -297,7 +361,10 @@
         else
         {
             $html =~ s|<u>||gi;
-            $html =~ s|<li>|\n* |gi;
+            $html =~ s|<li>||gi;
+            $html =~ s|</li>||gi;
+            $html =~ s|<ul>||gi;
+            $html =~ s|</ul>||gi;
             $html =~ s|<br>|\n|gi;
             $html =~ s|<br />|\n|gi;
             $html =~ s|<b>||gi;
@@ -312,7 +379,7 @@
             $html =~ s|&#133;|...|gi;
             $html =~ s|\x{8C}|OE|gi;
             $html =~ s|\x{9C}|oe|gi;
-
+            $html =~ s|\x{94}|\n|gi;
         }
         
         return $html;
@@ -321,14 +388,13 @@
     sub getSearchUrl
     {
 		my ($self, $word) = @_;
-	
         if ($self->{searchField} eq 'isbn')
         {
-           return "http://isbndb.com/search-all.html?kw=" .$word;
+           return "http://isbndb.com/search/all?query=" .$word;
         }
         else
         {
-           return "http://isbndb.com/search-title.html?kw=" .$word ."&isn=";
+           return "http://isbndb.com/search/books/1?query=" .$word;
         }
 
     }
@@ -348,7 +414,7 @@
     sub getCharset
     {
         my $self = shift;
-        return "ISO-8859-15";
+        return "UTF-8";
     }
 
     sub getAuthor