- Extract URLs in description from `descriptionHtml` field. (https://github.com/trizen/fair-viewer/issues/6) - Added the `--api-host=<...>` command-line option for changing the API host. - Fixed the `--info=<URL>` option.

Signed-off-by: Jesús <heckyel@hyperbola.info>

3 files changed, 43 insertions, 11 deletions

diff --git a/bin/fair-viewer b/bin/fair-viewer
index bbd18fc..f6159bc 100755
--- a/bin/fair-viewer
+++ b/bin/fair-viewer
@@ -796,7 +796,7 @@ usage: $execname [options] ([url] | [keywords])
        --use-colors!    : enable or disable the ANSI colors for text
 
  * Other
-   --invidious!         : use the API of invidio.us to get the streaming URLs
+   --api-host=s         : set an API host from https://instances.invidio.us/
    --proxy=s            : set HTTP(S)/SOCKS proxy: 'proto://domain.tld:port/'
                           If authentication required,
                           use 'proto://user:pass\@domain.tld:port/'
@@ -1593,7 +1593,8 @@ sub parse_arguments {
         'merge-into-mkv|mkv-merge!'           => \$opt{merge_into_mkv},
         'merge-with-captions|merge-captions!' => \$opt{merge_with_captions},
 
-        'invidious!'                         => \$opt{use_invidious_api},
+        'api-host=s' => \$opt{api_host},
+
         'convert-command|convert-cmd=s'      => \$opt{convert_cmd},
         'dash-m4a|dash-mp4-audio|dash-mp4a!' => \$opt{dash_mp4_audio},
         'dash-segmented!'                    => \$opt{dash_segmented},
@@ -2127,7 +2128,7 @@ sub get_and_print_video_info {
 
         if ($yv_utils->has_entries($info)) {
             local $opt{show_video_info} = 1;
-            print_video_info($info->{results}{items}[0]);
+            print_video_info($info->{results});
         }
         else {
             warn_cant_do('get info for', $videoID);
@@ -2279,7 +2280,7 @@ sub general_options {
         $callback->($request);
     }
     elsif ($option =~ /^(?:R|refresh)\z/ and defined $url) {
-        @{$results} = @{$yv_obj->_get_results($url)->{results}{items}};
+        @{$results} = @{$yv_obj->_get_results($url)->{results}};
     }
     elsif ($option eq 'login') {
         authenticate();
@@ -3374,7 +3375,7 @@ sub autoplay {
     while (1) {
         get_and_play_video_ids($video_id) || return;
         my $related = $yv_obj->related_to_videoID($video_id);
-        (my @video_ids = grep { !$seen{$_}++ } map { $yv_utils->get_video_id($_) } @{$related->{results}{items}}) || return;
+        (my @video_ids = grep { !$seen{$_}++ } map { $yv_utils->get_video_id($_) } @{$related->{results}}) || return;
         $video_id = $opt{shuffle} ? $video_ids[rand @video_ids] : $video_ids[0];
     }
 
@@ -3878,7 +3879,7 @@ sub print_videos {
                 #}
             }
             elsif ($opt =~ /^(?:R|refresh)\z/) {
-                @{$videos} = @{$yv_obj->_get_results($url)->{results}{items}};
+                @{$videos} = @{$yv_obj->_get_results($url)->{results}};
                 $results->{has_extra_info} = 0;
             }
             elsif ($opt =~ /^(?:r|return)\z/) {
diff --git a/lib/WWW/FairViewer.pm b/lib/WWW/FairViewer.pm
index d117b95..4b57f68 100644
--- a/lib/WWW/FairViewer.pm
+++ b/lib/WWW/FairViewer.pm
@@ -92,8 +92,8 @@ my %valid_options = (
     access_token  => {valid => [qr/^.{15}/], default => undef},
     refresh_token => {valid => [qr/^.{15}/], default => undef},
 
-    authentication_file => {valid => [qr/^./],         default => undef},
-    api_host            => {valid => [qr{^https?://}], default => "https://invidio.us"},
+    authentication_file => {valid => [qr/^./],             default => undef},
+    api_host            => {valid => [qr{[-\w]+\.[-\w]+}], default => "https://invidio.us"},
 
     # No input value allowed
     api_path         => {valid => q[], default => '/api/v1/'},
@@ -135,6 +135,7 @@ sub basic_video_info_fields {
           title
           videoId
           description
+          descriptionHtml
           published
           publishedText
           viewCount
@@ -493,7 +494,16 @@ sub _append_url_args {
 
 sub get_api_url {
     my ($self) = @_;
-    join('', $self->get_api_host, $self->get_api_path);
+
+    my $host = $self->get_api_host;
+
+    $host =~ s{/+\z}{};    # remove trailing '/'
+
+    if ($host =~ m{^[-\w]+(?>\.[-\w]+)+\z}) {    # no protocol specified
+        $host = 'https://' . $host;              # default to HTTPS
+    }
+
+    join('', $host, $self->get_api_path);
 }
 
 sub _simple_feeds_url {
diff --git a/lib/WWW/FairViewer/Utils.pm b/lib/WWW/FairViewer/Utils.pm
index 5ca1106..09c0174 100644
--- a/lib/WWW/FairViewer/Utils.pm
+++ b/lib/WWW/FairViewer/Utils.pm
@@ -452,8 +452,29 @@ Get description.
 
 sub get_description {
     my ($self, $info) = @_;
-    my $desc = $info->{description};
-    (defined($desc) and $desc =~ /\S/) ? $desc : 'No description available...';
+
+    my $desc = $info->{descriptionHtml} // '';
+
+    require URI::Escape;
+    require HTML::Entities;
+
+    $desc =~ s{<a href="/redirect\?(.*?)".*?>.*?</a>}{
+        my $url = $1;
+        if ($url =~ /(?:^|;)q=([^&]+)/) {
+            URI::Escape::uri_unescape($1);
+        }
+        else {
+            $url;
+        }
+    }segi;
+
+    $desc =~ s/<br>/\n/gi;
+    $desc =~ s{<a href=".*?".*?>(.*?)</a>}{$1}sgi;
+    $desc =~ s/<.*?>//gs;
+
+    $desc = HTML::Entities::decode_entities($desc);
+
+    ($desc =~ /\S/) ? $desc : 'No description available...';
 }
 
 =head2 get_title($info)