Fix slow migration of published_at field (#240)

Author: mgdigital

internal/dhtcrawler/persist.go

@@ -218,6 +218,8 @@ func (c *crawler) runPersistSources(ctx context.Context) {
 					DoUpdates: clause.AssignmentColumns([]string{
 						string(c.dao.TorrentsTorrentSource.Seeders.ColumnName()),
 						string(c.dao.TorrentsTorrentSource.Leechers.ColumnName()),
+						// sets to null, fixes torrents indexed before 0.8.0 with published_at 0001-01-01 00:00:00+00:
+						string(c.dao.TorrentsTorrentSource.PublishedAt.ColumnName()),
 						string(c.dao.TorrentsTorrentSource.UpdatedAt.ColumnName()),
 					}),
 				},

internal/model/torrents.go

@@ -54,11 +54,13 @@ func (t Torrent) Leechers() NullUint {
 	return leechers
 }
 
+var cutoff = time.Date(2000, 1, 1, 0, 0, 0, 0, time.UTC)
+
 func (t Torrent) PublishedAt() time.Time {
 	publishedAt := t.CreatedAt
 	for _, source := range t.Sources {
 		dt := source.CreatedAt
-		if source.PublishedAt.Valid {
+		if source.PublishedAt.Valid && source.PublishedAt.Time.After(cutoff) {
 			dt = source.PublishedAt.Time
 		}
 		if dt.Before(publishedAt) {

migrations/00017_ordering_fields.sql

@@ -4,9 +4,6 @@
 -- Fix an egregious error that has been there from day 1:
 update torrents_torrent_sources set seeders = leechers, leechers = seeders where source = 'dht';
 
--- Fix dates that were not set to null:
-update torrents_torrent_sources set published_at = null where published_at = '0001-01-01 00:00:00+00';
-
 -- Drop some fairly useless columns that take up a lot of space:
 alter table torrents_torrent_sources drop column bfsd;
 alter table torrents_torrent_sources drop column bfpe;