CollegeHumor: Fix video URLs, and the title is in the XML now!

[videosite.git] / videosite / YouTubeGrabber.pm

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index dc830c6..be1add1 100644 (file)
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -20,12 +20,15 @@ use videosite::JSArrayParser;
 use strict;
 
 my %preflist = (
-    'insane' => [38, 37, 22, 35, 18, 34, 6, 5],
-    'hd' => [37, 22, 35, 18, 34, 6, 5, 38],
-    'h264' => [18, 34, 37, 22, 35, 6, 5, 38],
-    'high' => [34, 35, 18, 37, 22, 6, 5, 38],
-    'normal' => [6, 5, 34, 35, 18, 22, 37, 38]);
+    'insane' => [38, 37, 22, 35, 18, 34, 6, 5, 43],
+    'hd' => [37, 22, 35, 18, 34, 6, 5, 38, 43],
+    'h264' => [18, 34, 37, 22, 35, 6, 5, 38, 43],
+    'high' => [34, 35, 18, 37, 22, 6, 5, 38, 43],
+    'normal' => [6, 5, 34, 35, 18, 22, 37, 38, 43]);
 my %videoformats = (
+    43 => 'webm,360p',
+    44 => 'webm,480p',
+    45 => 'webm,720p',
     38 => 'mp4,h264,4k',
     37 => 'mp4,h264,1080p',
     35 => 'flv,h264,large',
@@ -70,9 +73,9 @@ sub _parse {
     my $id;
     my $res;
 
-    my $url =~ m|$pattern|;
-    my $url = $1;
-    my $id = $2;
+    $url =~ m|$pattern|;
+    $url = $1;
+    $id = $2;
 
     $self->debug("Matched id %s from pattern %s", $id, $pattern);
 
@@ -93,7 +96,7 @@ sub _parse_by_video_info {
     my $quality = $self->_getval('QUALITY');
     my $metadata;
     my $videourl;
-    my $ua = $self->{_ua};
+    my $ua = $self->ua();
     my $preflist;
     my $r;
     my $content;
@@ -141,27 +144,12 @@ sub _parse_by_video_info {
     $urls =~ s/%(..)/chr(hex($1))/ge;
     $urls = { split /[\|,]/, $urls };
 
-    foreach (keys(%{$urls})) {
-        if (exists($videoformats{$_})) {
-            $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls->{$_});
-        } else {
-            $self->error('Unknown format %s: %s', $_, $urls->{$_});
-        }
-    }
-
-    foreach (@{$preflist}) {
-        if (exists($urls->{$_})) {
-            $self->debug("Selected URL with quality level %s", $_);
-            $metadata->{'DLURL'} = $urls->{$_};
-            last;
-        }
-    }
-
-    $self->debug('URL found: %s', $metadata->{'DLURL'});
+    $self->__pick_url($urls, $preflist, $metadata);
 
     $metadata->{'TITLE'} = $content->{'title'};
     $metadata->{'TITLE'} =~ s/\+/ /g;
     $metadata->{'TITLE'} =~ s/%(..)/chr(hex($1))/ge;
+    $metadata->{'TITLE'} = decode("utf8", $metadata->{'TITLE'});
 
     $self->debug('Title found: %s', $metadata->{'TITLE'});
 
@@ -182,7 +170,7 @@ sub _parse_by_scrape {
     my $p;
     my $e;
     my $tag;
-    my $ua = $self->{_ua};
+    my $ua = $self->ua();
     my $r;
     my $videourl;
     my $quality = $self->_getval('QUALITY');
@@ -227,9 +215,8 @@ sub _parse_by_scrape {
             if (exists($tag->[1]->{'name'}) and ('title' eq $tag->[1]->{'name'})) {
                 $metadata->{'TITLE'} = $tag->[1]->{'content'};
                 # Convert HTML entities in the title. This is a bit convoluted.
-                $metadata->{'TITLE'} = encode("utf8",
-                                         decode_entities(
-                                           decode("utf8", $metadata->{'TITLE'})));
+                $metadata->{'TITLE'} = decode_entities(
+                                           decode("utf8", $metadata->{'TITLE'}));
                     
                 $self->debug('Title found: %s', $metadata->{'TITLE'});
             }
@@ -334,23 +321,7 @@ sub _parse_by_scrape {
             }
 
             if (%urls) {
-                foreach (keys(%urls)) {
-                    if (exists($videoformats{$_})) {
-                        $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls{$_});
-                    } else {
-                        $self->error('Unknown format %s: %s', $_, $urls{$_});
-                    }
-                }
-
-                foreach (@{$preflist}) {
-                    if (exists($urls{$_})) {
-                        $self->debug("Selected URL with quality level %s", $_);
-                        $metadata->{'DLURL'} = $urls{$_};
-                        last;
-                    }
-                }
-
-                $self->debug('URL found: %s', $metadata->{'DLURL'});
+                $self->__pick_url(\%urls, $preflist, $metadata);
                 last SWF_ARGS;
             }
         } elsif ('div' eq $tag->[0]) {
@@ -480,4 +451,29 @@ sub __login {
     return ($ua->get($videourl), $cookie);
 }
 
+sub __pick_url {
+    my $self = shift;
+    my $urls = shift;
+    my $preflist = shift;
+    my $metadata = shift;
+
+    foreach (keys(%{$urls})) {
+        if (exists($videoformats{$_})) {
+            $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls->{$_});
+        } else {
+            $self->error('Unknown format %s: %s', $_, $urls->{$_});
+        }
+    }
+
+    foreach (@{$preflist}) {
+        if (exists($urls->{$_})) {
+            $self->debug("Selected URL with quality level %s", $_);
+            $metadata->{'DLURL'} = $urls->{$_};
+            last;
+        }
+    }
+
+    $self->debug('URL found: %s', $metadata->{'DLURL'});
+}
+
 1;