Add method of looking up talk urls on ACT conferences and adding to the talk data

diff --git a/lib/App/IdiotBox/Importer.pm b/lib/App/IdiotBox/Importer.pm
index b7ba153..0bd54f7 100644 (file)
--- a/lib/App/IdiotBox/Importer.pm
+++ b/lib/App/IdiotBox/Importer.pm
@@ -8,6 +8,9 @@ use ExtUtils::MakeMaker qw(prompt);
 use File::Spec::Functions qw(catfile catdir);
 use POSIX qw(strftime);
 
+use Data::ICal;
+use Text::LevenshteinXS 'distance';
+
 sub log_info (&) { print $_[0]->(), "\n"; }
 
 sub run {
@@ -77,6 +80,15 @@ sub run {
     $info->{name} = prompt('What is the name of this talk?', $info->{name});
     (my $slug = lc $info->{name}) =~ s/ /-/g;
     $info->{slug} = prompt('What is the slug for this talk?', $slug);
+    my $act_data = {};
+    $act_data = get_act_talk_data($info->{name}, $bucket->{act_url})
+      if $bucket->{act_url};
+    if(exists $act_data->{talk_id}) {
+        $info->{act_talk_id} = $act_data->{talk_id};
+        $info->{details} = $act_data->{description};
+    }
+        
+    $info->{details} = prompt('Enter a talk description', $info->{details});
     $info->{bucket} = $bucket;
     $info->{announcement} = $ann;
     $videos{$video_file} = bless($info, 'App::IdiotBox::Video');
@@ -96,6 +108,7 @@ sub run {
   $bucket->videos->add($_) for values %videos;
 }
 
+## Bug, this doesn't cope with files named foo.bar.5.23.mkv etc (multiple periods)
 sub video_files_from_dir {
   my ($class, $dir) = @_;
   my %videos;
@@ -106,4 +119,56 @@ sub video_files_from_dir {
   \%videos;
 }
 
+## Parse the act timetable.ics to find talk data and description
+## (best guess based on talk name)
+sub get_act_talk_data {
+  my ($talk_name, $base_url) = @_;
+
+  my $act_url = URI->new($base_url . '/timetable.ics');
+  my $timetable < io($act_url);
+
+  my $act_schedule = Data::ICal->new(data => $timetable)
+    or die "Can't parse timetable.ics: $!";
+
+  my $target = "Things I learned from users";
+
+  my $mangled_target = uc $target;
+  $mangled_target =~ s/\W//g;
+
+  my @possibles;
+
+  for my $entry (@{$act_schedule->entries}) {
+    my $title = $entry->property('SUMMARY')->[0]->decoded_value;
+
+    my $description = $entry->property('DESCRIPTION')->[0]->decoded_value;
+
+    my $eid = $entry->property('URL')->[0]->decoded_value;
+    $eid =~ s!/$!!;
+    $eid =~ m!.*/(.*?)$! or die "Couldn't find last component in $eid";
+    $eid = $1;
+
+    my $mangled_title = uc $title;
+    $mangled_title =~ s/\W//g;
+
+    my $score = (length($mangled_title) - distance($mangled_title, $mangled_target))/length($mangled_title);
+    # print "Score for $mangled_title vs $mangled_target = $score\n";
+
+    push @possibles, {
+                      score => $score,
+                      title => $title,
+                      description => $description,
+                      event_id => $eid
+                     };
+  }
+
+  @possibles = sort {$b->{score} <=> $a->{score}} @possibles;
+
+
+  print STDERR "ye2010/$target/ http://conferences.yapceurope.org/ye2010/talk/$possibles[0]{event_id}\n";
+
+  return $possibles[0];
+
+}
+
+
 1;

diff --git a/share/sql/idiotbox-2.0-sqlite.sql b/share/sql/idiotbox-2.0-sqlite.sql
new file mode 100644 (file)
index 0000000..ed4535d
--- /dev/null
+++ b/share/sql/idiotbox-2.0-sqlite.sql
@@ -0,0 +1,24 @@
+CREATE TABLE buckets (
+  slug TEXT NOT NULL PRIMARY KEY,
+  name TEXT NOT NULL,
+  act_url TEXT 
+);
+
+CREATE TABLE announcements (
+  id INTEGER NOT NULL PRIMARY KEY,
+  made_at DATETIME NOT NULL,
+  bucket_slug TEXT REFERENCES buckets(slug)
+);
+
+CREATE TABLE videos (
+  slug TEXT NOT NULL,
+  bucket_slug TEXT NOT NULL REFERENCES buckets(slug),
+  name TEXT NOT NULL,
+  author TEXT NOT NULL,
+  act_talk_id INTEGER,
+  details TEXT NOT NULL DEFAULT '',
+  announcement_id INTEGER NOT NULL,
+  PRIMARY KEY (slug, bucket_slug),
+  FOREIGN KEY (announcement_id, bucket_slug)
+    REFERENCES announcements(id, bucket_slug)
+);