YouTube: Correctly parse url_encoded_fmt_stream_map from the web scape, too

[videosite.git] / videosite / YouTubeGrabber.pm

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index eca896b..fd3e0c0 100644 (file)
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -26,7 +26,9 @@ my %preflist = (
     'high' => [34, 35, 18, 37, 22, 6, 5, 38, 43],
     'normal' => [6, 5, 34, 35, 18, 22, 37, 38, 43]);
 my %videoformats = (
-    43 => 'webm',
+    43 => 'webm,360p',
+    44 => 'webm,480p',
+    45 => 'webm,720p',
     38 => 'mp4,h264,4k',
     37 => 'mp4,h264,1080p',
     35 => 'flv,h264,large',
@@ -71,9 +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);
 
@@ -132,21 +134,27 @@ sub _parse_by_video_info {
         return undef;
     }
 
-    unless(exists($content->{'fmt_url_map'}) and exists($content->{'title'})) {
-        $self->debug("No fmt_url_map or no title found");
+    if (exists($content->{'fmt_url_map'})) {
+        # Decode fmt_url_map
+        $urls = $self->decode_hexurl($content->{'fmt_url_map'});
+        $urls = { split /[\|,]/, $urls };
+    } elsif (exists($content->{'url_encoded_fmt_stream_map'})) {
+        $urls = $self->_decode_url_encoded_fmt_stream_map($content->{'url_encoded_fmt_stream_map'});
+    } else {
+        $self->debug("No URL data found");
         return undef;
     }
 
-    # Decode fmt_url_map
-    $urls = $content->{'fmt_url_map'};
-    $urls =~ s/%(..)/chr(hex($1))/ge;
-    $urls = { split /[\|,]/, $urls };
+    unless(exists($content->{'title'})) {
+        $self->debug("No title found");
+        return undef;
+    }
 
     $self->__pick_url($urls, $preflist, $metadata);
 
     $metadata->{'TITLE'} = $content->{'title'};
     $metadata->{'TITLE'} =~ s/\+/ /g;
-    $metadata->{'TITLE'} =~ s/%(..)/chr(hex($1))/ge;
+    $metadata->{'TITLE'} = $self->decode_hexurl($metadata->{'TITLE'});
     $metadata->{'TITLE'} = decode("utf8", $metadata->{'TITLE'});
 
     $self->debug('Title found: %s', $metadata->{'TITLE'});
@@ -242,7 +250,7 @@ sub _parse_by_scrape {
 
                     $self->debug("Video has fmt_url_map: %s", $urls);
 
-                    $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+                    $urls = $self->decode_hexurl($urls);
                     %urls = split(/[\|,]/, $urls);
                     $self->debug("Pagetype: old (SWF_ARGS), fmt_url_map");
 
@@ -254,7 +262,7 @@ sub _parse_by_scrape {
                         my @fmt;
 
                         $self->debug('Video has fmt_map');
-                        $fmt =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+                        $fmt = $self->decode_hexurl($fmt);
                         @fmt = split(/,/, $fmt);
                         foreach (@fmt) {
                             @_=split(/\//);
@@ -279,10 +287,10 @@ sub _parse_by_scrape {
                 my $urls = $1;
                 $self->debug("Video has fmt_url_map: %s", $urls);
 
-                $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+                $urls = $self->decode_hexurl($urls);
                 %urls = split(/[\|,]/, $urls);
                 $self->debug("Pagetype: 2010 (swfHTML), fmt_url_map");
-            } elsif ($e =~ m|\x27PLAYER_CONFIG\x27:\s+(.+)\}\);|) {
+            } elsif ($e =~ m|\x27PLAYER_CONFIG\x27:\s+(.+)(?:\}\);)?|) {
                 my $args = $1;
                 $self->debug("Found PLAYER_CONFIG: %s", $args);
 
@@ -307,11 +315,13 @@ sub _parse_by_scrape {
 
                     %urls = split(/[\|,]/, $urls);
                     foreach (keys(%urls)) {
-                        my $u = $urls{$_};
-                        $u =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
-                        $urls{$_} = $u;
+                        $urls{$_} = $self->decode_hexurl($urls{$_});
                     }
                     $self->debug("Pagetype: 2011 (PLAYER_CONFIG), fmt_url_map");
+                } elsif (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
+                    %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
+
+                    $self->debug("Pagetype: 2011 (PLAYER_CONFIG), url_encoded_fmt_stream_map");
                 } else {
                     $self->error('fmt_url_map not found in PLAYER_CONFIG');
                     return undef;
@@ -449,6 +459,33 @@ sub __login {
     return ($ua->get($videourl), $cookie);
 }
 
+# Take an encoded url_encoded_fmt_stream_map and return a hash
+# matching video IDs to download URLs
+sub _decode_url_encoded_fmt_stream_map {
+    my $self = shift;
+    my $data = shift;
+    my $dataencoded = shift;
+    my @data;
+
+    $data = $self->decode_hexurl($data) if (defined($dataencoded) && $dataencoded);
+    # This will
+    # - Split the decoded string into segments (along ,)
+    # - Interpret each segment as a concatenated key-value list (key and value separated by =, pairs separated by &
+    # - URL-decode each key and value _again_
+    #
+    # @data will be an array of hash references
+    
+    @data = map { { map { $self->decode_hexurl($_) } split /[=&]/ } } split /,/, $data;
+    $self->debug("_decode_url_encoded_fmt_stream_map() decoded %s", Dumper(\@data));
+
+    # From each array entry, pick the itag and the url values and return that
+    # as a hash reference
+    
+    return { map { $_->{'itag'}, $_->{'url'} } @data };
+}
+
+
+
 sub __pick_url {
     my $self = shift;
     my $urls = shift;
@@ -475,3 +512,4 @@ sub __pick_url {
 }
 
 1;
+