rewrite icon crawling

src/storage/feed.go

@@ -76,10 +76,10 @@ func (s *Storage) UpdateFeedIcon(feedId int64, icon *[]byte) bool {
 }
 
 func (s *Storage) ListFeeds() []Feed {
-	result := make([]Feed, 0, 0)
+	result := make([]Feed, 0)
 	rows, err := s.db.Query(`
 		select id, folder_id, title, description, link, feed_link,
-		       ifnull(icon, '') != '' as has_icon
+		       ifnull(length(icon), 0) > 0 as has_icon
 		from feeds
 		order by title collate nocase
 	`)
@@ -107,6 +107,36 @@ func (s *Storage) ListFeeds() []Feed {
 	return result
 }
 
+func (s *Storage) ListFeedsMissingIcons() []Feed {
+	result := make([]Feed, 0)
+	rows, err := s.db.Query(`
+		select id, folder_id, title, description, link, feed_link
+		from feeds
+		where icon is null
+	`)
+	if err != nil {
+		log.Print(err)
+		return result
+	}
+	for rows.Next() {
+		var f Feed
+		err = rows.Scan(
+			&f.Id,
+			&f.FolderId,
+			&f.Title,
+			&f.Description,
+			&f.Link,
+			&f.FeedLink,
+		)
+		if err != nil {
+			log.Print(err)
+			return result
+		}
+		result = append(result, f)
+	}
+	return result
+}
+
 func (s *Storage) GetFeed(id int64) *Feed {
 	var f Feed
 	err := s.db.QueryRow(`

src/worker/crawler.go

@@ -76,8 +76,16 @@ func DiscoverFeed(candidateUrl string) (*DiscoverResult, error) {
 	return result, nil
 }
 
-func findFavicon(websiteUrl, feedUrl string) (*[]byte, error) {
-	candidateUrls := make([]string, 0)
+var emptyIcon = make([]byte, 0)
+var imageTypes = map[string]bool{
+	"image/x-icon": true,
+	"image/png":    true,
+	"image/jpeg":   true,
+	"image/gif":    true,
+}
+
+func findFavicon(siteUrl, feedUrl string) (*[]byte, error) {
+	urls := make([]string, 0)
 
 	favicon := func(link string) string {
 		u, err := url.Parse(link)
@@ -87,49 +95,43 @@ func findFavicon(websiteUrl, feedUrl string) (*[]byte, error) {
 		return fmt.Sprintf("%s://%s/favicon.ico", u.Scheme, u.Host)
 	}
 
-	if len(websiteUrl) != 0 {
-		res, err := client.get(websiteUrl)
-		if err != nil {
-			return nil, err
-		}
-		body, err := ioutil.ReadAll(res.Body)
+	if siteUrl != "" {
+		if res, err := client.get(siteUrl); err == nil {
 			defer res.Body.Close()
-		if err != nil {
-			return nil, err
-		}
-		candidateUrls = append(candidateUrls, scraper.FindIcons(string(body), websiteUrl)...)
-		if c := favicon(websiteUrl); len(c) != 0 {
-			candidateUrls = append(candidateUrls, c)
+			if body, err := ioutil.ReadAll(res.Body); err == nil {
+				urls = append(urls, scraper.FindIcons(string(body), siteUrl)...)
+				if c := favicon(siteUrl); c != "" {
+					urls = append(urls, c)
+				}
 			}
 		}
-	if c := favicon(feedUrl); len(c) != 0 {
-		candidateUrls = append(candidateUrls, c)
 	}
 
-	imageTypes := [4]string{
-		"image/x-icon",
-		"image/png",
-		"image/jpeg",
-		"image/gif",
+	if c := favicon(feedUrl); c != "" {
+		urls = append(urls, c)
 	}
-	for _, url := range candidateUrls {
-		res, err := client.get(url)
+
+	for _, u := range urls {
+		res, err := client.get(u)
 		if err != nil {
 			continue
 		}
 		defer res.Body.Close()
-		if res.StatusCode == 200 {
-			if content, err := ioutil.ReadAll(res.Body); err == nil {
+		if res.StatusCode != 200 {
+			continue
+		}
+
+		content, err := ioutil.ReadAll(res.Body)
+		if err != nil {
+			continue
+		}
+
 		ctype := http.DetectContentType(content)
-				for _, itype := range imageTypes {
-					if ctype == itype {
+		if imageTypes[ctype] {
 			return &content, nil
 		}
 	}
-			}
-		}
-	}
-	return nil, nil
+	return &emptyIcon, nil
 }
 
 func ConvertItems(items []parser.Item, feed storage.Feed) []storage.Item {

src/worker/worker.go

@@ -41,11 +41,9 @@ func (w *Worker) StartFeedCleaner() {
 
 func (w *Worker) FindFavicons() {
 	go func() {
-		for _, feed := range w.db.ListFeeds() {
-			if !feed.HasIcon {
+		for _, feed := range w.db.ListFeedsMissingIcons() {
 			w.FindFeedFavicon(feed)
 		}
-		}
 	}()
 }