Changeset 684

Timestamp:

05/24/07 07:22:38 (6 years ago)

Author:

paul

Message:

Allow \n into the xml file for use in imdb_augment_data etc

Fix print because some urls give errors:
Invalid conversion in printf: "%C" at Shepherd/Common.pm line 171.
Invalid conversion in printf: "%T" at Shepherd/Common.pm line 171.
Invalid conversion in printf: "%-" at Shepherd/Common.pm line 171.
Invalid conversion in printf: "%M" at Shepherd/Common.pm line 171.

Add sub urlify and sub translate_category.

Files:

• references/Shepherd/Common.pm (modified) (3 diffs)

references/Shepherd/Common.pm

@@ -169 +169 @@
         if ($cnf{debug})
         {
-            printf "Fetching $urlname%s...\n",
+            print "Fetching $urlname";
+            printf "%s...\n",
                    ($cnf{debug} > 1 ? " (attempt ".($failures+1)." of ".($cnf{retries}+1).")" : '');
         }
@@ -300 +301 @@
     elsif (defined $$x) {
         $$x =~ s/&(#(\d+)|(.*?));/ $2 ? chr($2) : $amp{$3}||' ' /eg;
-        $$x =~ s/[^\x20-\x7f]/ /g;
+        $$x =~ s/[^\x20-\x7f\x0a]/ /g;
         $$x =~ s/(^\s+|\s+$)//g;
     }
@@ -418 +419 @@
 ##########################################################################
 
+sub urlify
+{
+    my $str = shift;
+    $str =~ s/([^A-Za-z0-9])/sprintf("%%%02X", ord($1))/seg;
+    $str =~ s/%20/+/g;
+    $str =~ s/%2D/-/g;
+    return $str;
+}
+
+##############################################################################
+
+sub translate_category
+{
+    my $genre = shift;
+    my %translation = (
+        'Sport' => 'sports',
+        'Soap Opera' => 'Soap',
+        'Science and Technology' => 'Science/Nature',
+        'Real Life' => 'Reality',
+        'Cartoon' => 'Animation',
+        'Family' => 'Children',
+        'Murder' => 'Crime' );
+    return $translation{$genre} if defined $translation{$genre};
+    return $genre;
+}
+
+##########################################################################
+
 1;