Merge pull request #148 from dop251/filters-pr

Implemented negative filters for title and description.

README.md

@@ -77,7 +77,7 @@ vimeo = [ # Multiple keys will be rotated.
   # custom = { cover_art = "{IMAGE_URL}}", category = "TV", explicit = true, lang = "en" } # Optional feed customizations
   # max_height = "720" # Optional maximal height of video, example: 720, 1080, 1440, 2160, ...
   # cron_schedule = "@every 12h" # Optional cron expression format. If set then overwrite 'update_period'. See details below
-  # filters = { title = "regex for title here" } # Optional Golang regexp format. If set, then only download episodes with matching titles.
+  # filters = { title = "regex for title here", not_title = "regex for negative title match", description = "...", not_description = "..." } # Optional Golang regexp format. If set, then only download matching episodes.
   # opml = true|false # Optional inclusion of the feed in the OPML file (default value: false)
   # clean = { keep_last = 10 } # Keep last 10 episodes (order desc by PubDate)
 

cmd/podsync/updater.go

@@ -124,6 +124,40 @@ func (u *Updater) updateFeed(ctx context.Context, feedConfig *config.Feed) error
 	return nil
 }
 
+func (u *Updater) matchRegexpFilter(pattern, str string, negative bool, logger log.FieldLogger) bool {
+	if pattern != "" {
+		matched, err := regexp.MatchString(pattern, str)
+		if err != nil {
+			logger.Warnf("pattern %q is not a valid")
+		} else {
+			if matched == negative {
+				logger.Infof("skipping due to mismatch")
+				return false
+			}
+		}
+	}
+	return true
+}
+
+func (u *Updater) matchFilters(episode *model.Episode, filters *config.Filters) bool {
+	logger := log.WithFields(log.Fields{"episode_id": episode.ID})
+	if !u.matchRegexpFilter(filters.Title, episode.Title, false, logger.WithField("filter", "title")) {
+		return false
+	}
+	if !u.matchRegexpFilter(filters.NotTitle, episode.Title, true, logger.WithField("filter", "not_title")) {
+		return false
+	}
+
+	if !u.matchRegexpFilter(filters.Description, episode.Description, false, logger.WithField("filter", "description")) {
+		return false
+	}
+	if !u.matchRegexpFilter(filters.NotDescription, episode.Description, true, logger.WithField("filter", "not_description")) {
+		return false
+	}
+
+	return true
+}
+
 func (u *Updater) downloadEpisodes(ctx context.Context, feedConfig *config.Feed) error {
 	var (
 		feedID       = feedConfig.ID
@@ -140,16 +174,8 @@ func (u *Updater) downloadEpisodes(ctx context.Context, feedConfig *config.Feed)
 			return nil
 		}
 
-		if feedConfig.Filters.Title != "" {
-			matched, err := regexp.MatchString(feedConfig.Filters.Title, episode.Title)
-			if err != nil {
-				log.Warnf("pattern %q is not a valid filter for %q Title", feedConfig.Filters.Title, feedConfig.ID)
-			} else {
-				if !matched {
-					log.Infof("skipping %q due to lack of match with %q", episode.Title, feedConfig.Filters.Title)
-					return nil
-				}
-			}
+		if !u.matchFilters(episode, &feedConfig.Filters) {
+			return nil
 		}
 
 		// Limit the number of episodes downloaded at once

pkg/config/config.go

@@ -45,7 +45,10 @@ type Feed struct {
 }
 
 type Filters struct {
-	Title string `toml:"title"`
+	Title          string `toml:"title"`
+	NotTitle       string `toml:"not_title"`
+	Description    string `toml:"description"`
+	NotDescription string `toml:"not_description"`
 	// More filters to be added here
 }