dedupe: save last see timestamp

also, allow to ignore date and/or body

1 files changed, 30 insertions, 15 deletions

diff --git a/lib/Feed/Role/DeDupe.pm b/lib/Feed/Role/DeDupe.pm
index 641c898..164e3ec 100644
--- a/lib/Feed/Role/DeDupe.pm
+++ b/lib/Feed/Role/DeDupe.pm
@@ -20,11 +20,23 @@ has dbh => (
     lazy_build => 1,
 );
 
-has ['_find_sth','_insert_sth'] => (
+has ['_insert_sth','_update_sth'] => (
     is => 'ro',
     lazy_build => 1,
 );
 
+has dedupe_use_body => (
+    is => 'ro',
+    isa => 'Bool',
+    default => 1,
+);
+
+has dedupe_use_date => (
+    is => 'ro',
+    isa => 'Bool',
+    default => 1,
+);
+
 sub _build_dbh {
     my ($self) = @_;
 
@@ -57,22 +69,22 @@ $dbh->commit;
     return;
 }
 
-sub _build__find_sth {
+sub _build__insert_sth {
     my ($self) = @_;
 
     return $self->dbh->prepare(<<'SQL');
-SELECT COUNT(*)
-FROM seen_items
-WHERE id=?
+INSERT INTO seen_items(id,last_seen)
+VALUES (?,datetime('now'))
 SQL
 }
 
-sub _build__insert_sth {
+sub _build__update_sth {
     my ($self) = @_;
 
     return $self->dbh->prepare(<<'SQL');
-INSERT INTO seen_items(id)
-VALUES (?)
+UPDATE seen_items
+SET last_seen = datetime('now')
+WHERE id = ?
 SQL
 }
 
@@ -109,8 +121,7 @@ sub seen_already {
 
     my $id = $self->_entry_id($e);
 
-    $self->_find_sth->execute($id);
-    my ($count) = $self->_find_sth->fetchrow_array;
+    my $count = $self->_update_sth->execute($id);
 
     $self->log->trace("seen_already - end ($count)");
 
@@ -134,12 +145,16 @@ sub _entry_id {
 
     my $id = $e->id;
 
-    my $date = $e->modified // $e->issued;
-    if ($date) { $id .= '-' . $date->iso8601 }
+    if ($self->dedupe_use_date) {
+        my $date = $e->modified // $e->issued;
+        if ($date) { $id .= '-' . $date->iso8601 }
+    }
 
-    my $body = $e->content->body;
-    my $content_digest = sha1_base64(encode('utf-8',$body));
-    $id .= '-' . $content_digest;
+    if ($self->dedupe_use_body) {
+        my $body = $e->content->body;
+        my $content_digest = sha1_base64(encode('utf-8',$body));
+        $id .= '-' . $content_digest;
+    }
 
     return encode('utf-8',$id);
 }