--- scripts/reblog2couchdb.pl	2008/08/08 19:19:12	10
+++ scripts/reblog2couchdb.pl	2008/08/28 21:53:49	15
@@ -7,6 +7,8 @@
 
 $|++;
 
+my $debug = @ARGV ? 1 : 0;
+
 my $database = 'reblog';
 
 my $dbi = "DBI:mysql:database=$database";
@@ -20,16 +22,29 @@
 		feeds.title as feed_title,
 		feeds.link as feed_link,
 		feeds.description as feed_description
---		t.value_long as tags
 	from items
 	join items_userdata on items.id = item_id
 	join feeds on items.feed_id = feeds.id
---	left outer join items_userdata as t on items.id = t.item_id and t.label='tags'
 	where items.id > ? and items_userdata.label = 'published' and items_userdata.value_numeric = 1
 	order by items.id asc
 	limit 1000
 };
 
+my $sql_tags = qq{
+select
+	items_userdata.item_id,
+	t.value_long as tags,
+	items_userdata.timestamp
+from items_userdata
+join items_userdata as t
+	on items_userdata.item_id = t.item_id and t.label='tags'
+where
+	items_userdata.item_id > ? and
+	items_userdata.label = 'published' and
+	items_userdata.value_numeric = 1
+order by items_userdata.item_id asc
+};
+
 my $c = CouchDB::Client->new(uri => 'http://localhost:5984/');
 
 $c->testConnection or die "The server cannot be reached";
@@ -44,6 +59,7 @@
 print "status ",dump( $status->{data} ), "\n";
 
 my $last_row = $status->{data}->{last_row_id} || 0;
+$last_row = 0 if $debug;
 
 sub commit_last_row {
 	warn "commit_last_row $last_row\n";
@@ -51,32 +67,46 @@
 	$status->update;
 }
 
-my $dbh = DBI->connect($dbi,"","") || die $DBI::errstr;
+my $dbh = DBI->connect($dbi,"","",{ RaiseError => 1 });
 
 print "Fetching items from $dbi id > $last_row\n";
 
-my $sth = $dbh->prepare($sql) || die $dbh->errstr();
-$sth->execute( $last_row ) || die $sth->errstr();
+my $sth = $dbh->prepare($sql);
+$sth->execute( $last_row );
 
 warn dump( $sth->{NAME} );
 
 print "found ",$sth->rows," items to process...\n";
 
-my $pk = 'id';
+my $sth_tags = $dbh->prepare($sql_tags);
+$sth_tags->execute( $last_row );
+print "found ",$sth_tags->rows, " tags found...\n";
 
 my $count = 0;
 
+my $row_tags = $sth_tags->fetchrow_hashref();
+
 while (my $row = $sth->fetchrow_hashref() ) {
 	my $_id = $row->{_id} || die "row needs _id";
 	my $doc = $db->newDoc( $_id );
 
+	while ( $row_tags && $row_tags->{item_id} < $row->{_id} ) {
+		$row_tags = $sth_tags->fetchrow_hashref();
+		warn "## got tags: ",dump( $row_tags ) if $debug;
+	}
+
 	sub row2doc {
 		my ( $row, $doc ) = @_;
 		my $a = delete( $row->{xml} );
 		$doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a;
-		my $a = delete( $row->{content} );
+		$a = delete( $row->{content} );
 		$doc->addAttachment( 'content.html', 'text/html', $a ) if $a;
+		if ( $row_tags && $row_tags->{item_id} == $row->{_id} ) {
+			$row->{tags} = [ split(/\s+/, $row_tags->{tags} ) ];
+			warn "++ ",$row->{item_id}, dump( $row->{tags} );
+		}
 		$doc->{data} = $row;
+		warn "## ",dump( $row ) if $debug;
 		return $doc;
 	}