remove feed hacks from storage

src/parser/atom.go

@@ -81,10 +81,11 @@ func ParseAtom(r io.Reader) (*Feed, error) {
 		SiteURL: firstNonEmpty(srcfeed.Links.First("alternate"), srcfeed.Links.First("")),
 	}
 	for _, srcitem := range srcfeed.Entries {
+		link := firstNonEmpty(srcitem.OrigLink, srcitem.Links.First("alternate"), srcitem.Links.First(""))
 		dstfeed.Items = append(dstfeed.Items, Item{
-			GUID:     firstNonEmpty(srcitem.ID),
+			GUID:     firstNonEmpty(srcitem.ID, link),
 			Date:     dateParse(firstNonEmpty(srcitem.Published, srcitem.Updated)),
-			URL:      firstNonEmpty(srcitem.OrigLink, srcitem.Links.First("alternate"), srcitem.Links.First("")),
+			URL:      link,
 			Title:    srcitem.Title.Text(),
 			Content:  firstNonEmpty(srcitem.Content.String(), srcitem.Summary.String(), srcitem.firstMediaDescription()),
 			ImageURL: srcitem.firstMediaThumbnail(),

src/parser/feed.go

@@ -70,6 +70,7 @@ func Parse(r io.Reader) (*Feed, error) {
 func (feed *Feed) cleanup() {
 	feed.Title = strings.TrimSpace(feed.Title)
 	feed.SiteURL = strings.TrimSpace(feed.SiteURL)
+
 	for i, item := range feed.Items {
 		feed.Items[i].GUID = strings.TrimSpace(item.GUID)
 		feed.Items[i].URL = strings.TrimSpace(item.URL)

src/parser/json.go

@@ -46,7 +46,7 @@ func ParseJSON(data io.Reader) (*Feed, error) {
 	}
 	for _, srcitem := range srcfeed.Items {
 		dstfeed.Items = append(dstfeed.Items, Item{
-			GUID:    srcitem.ID,
+			GUID:    firstNonEmpty(srcitem.ID, srcitem.URL),
 			Date:    dateParse(firstNonEmpty(srcitem.DatePublished, srcitem.DateModified)),
 			URL:     srcitem.URL,
 			Title:   srcitem.Title,

src/storage/feed.go

@@ -2,9 +2,7 @@ package storage
 
 import (
 	"database/sql"
-	"html"
 	"log"
-	"net/url"
 )
 
 type Feed struct {
@@ -19,22 +17,9 @@ type Feed struct {
 }
 
 func (s *Storage) CreateFeed(title, description, link, feedLink string, folderId *int64) *Feed {
-	title = html.UnescapeString(title)
-	// WILD: fallback to `feed.link` -> `feed.feed_link` -> "<???>" if title is missing
-	if title == "" {
-		title = link
-		// use domain if possible
-		linkUrl, err := url.Parse(link)
-		if err == nil && linkUrl.Host != "" && len(linkUrl.Path) <= 1 {
-			title = linkUrl.Host
-		}
-	}
 	if title == "" {
 		title = feedLink
 	}
-	if title == "" {
-		title = "<???>"
-	}
 	result, err := s.db.Exec(`
 		insert into feeds (title, description, link, feed_link, folder_id) 
 		values (?, ?, ?, ?, ?)

src/storage/item.go

@@ -3,7 +3,6 @@ package storage
 import (
 	"encoding/json"
 	"fmt"
-	"html"
 	"log"
 	"strings"
 	"time"
@@ -81,14 +80,6 @@ func (s *Storage) CreateItems(items []Item) bool {
 	now := time.Now()
 
 	for _, item := range items {
-		// WILD: some feeds provide only `item.date_updated` (without `item.date_created`)
-		if item.Date == nil {
-			item.Date = item.DateUpdated
-		}
-		// WILD: `item.guid` is not always present
-		if item.GUID == "" {
-			item.GUID = item.Link
-		}
 		_, err = tx.Exec(`
 			insert into items (
 				guid, feed_id, title, link, description,
@@ -99,7 +90,7 @@ func (s *Storage) CreateItems(items []Item) bool {
 			values (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
 			on conflict (feed_id, guid) do update set
 			date_updated = ?, date_arrived = ?`,
-			item.GUID, item.FeedId, html.UnescapeString(item.Title), item.Link, item.Description,
+			item.GUID, item.FeedId, item.Title, item.Link, item.Description,
 			item.Content, item.Author,
 			item.Date, item.DateUpdated, now,
 			UNREAD, item.Image, item.PodcastURL,