maciej/yarr
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

remove html tags from titles

Browse source
This commit is contained in:
nkanaev 2021-11-10 10:54:12 +00:00
parent 77c7f938f1
commit 26b87dee98
3 changed files with 26 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
src/parser/feed.go
View file

@ -9,6 +9,8 @@ import (
"net/url"
"strings"
"time"
"github.com/nkanaev/yarr/src/content/htmlutil"
)
var UnknownFormat = errors.New("unknown feed format")
@ -80,7 +82,7 @@ func (feed *Feed) cleanup() {
for i, item := range feed.Items {
feed.Items[i].GUID = strings.TrimSpace(item.GUID)
feed.Items[i].URL = strings.TrimSpace(item.URL)
feed.Items[i].Title = strings.TrimSpace(item.Title)
feed.Items[i].Title = strings.TrimSpace(htmlutil.ExtractText(item.Title))
feed.Items[i].Content = strings.TrimSpace(item.Content)
if item.ImageURL != "" && strings.Contains(item.Content, item.ImageURL) {