From 26b87dee98cec2e1bb46f7656edfb7483f476171 Mon Sep 17 00:00:00 2001
From: nkanaev <nkanaev@live.com>
Date: Wed, 10 Nov 2021 10:54:12 +0000
Subject: [PATCH] remove html tags from titles

---
 doc/todo.txt           |  2 --
 src/parser/feed.go     |  4 +++-
 src/parser/rss_test.go | 23 +++++++++++++++++++++++
 3 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/doc/todo.txt b/doc/todo.txt
index 05cdc7b..fb20fb8 100644
--- a/doc/todo.txt
+++ b/doc/todo.txt
@@ -1,5 +1,3 @@
-- strip out html tags from titles
-  https://www.aldaily.com/feed/
 - feedlist keyboard navigation is flaky in "unread" section
 - windows cli mode not working
   incorporate changes from:
diff --git a/src/parser/feed.go b/src/parser/feed.go
index bf77764..3846764 100644
--- a/src/parser/feed.go
+++ b/src/parser/feed.go
@@ -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) {
diff --git a/src/parser/rss_test.go b/src/parser/rss_test.go
index 65d0b41..e6eb8a1 100644
--- a/src/parser/rss_test.go
+++ b/src/parser/rss_test.go
@@ -180,3 +180,26 @@ func TestRSSPodcastDuplicated(t *testing.T) {
 		t.Fatal("item.audio_url must be unset if present in the content")
 	}
 }
+
+func TestRSSTitleHTMLTags(t *testing.T) {
+	feed, _ := Parse(strings.NewReader(`
+		<?xml version="1.0" encoding="UTF-8"?>
+		<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/">
+			<channel>
+				<item>
+					<title>&lt;p&gt;title in p&lt;/p&gt;</title>
+				</item>
+				<item>
+					<title>very &lt;strong&gt;strong&lt;/strong&gt; title</title>
+				</item>
+			</channel>
+		</rss>
+	`))
+	have := []string{feed.Items[0].Title, feed.Items[1].Title}
+	want := []string{"title in p", "very strong title"}
+	for i := 0; i < len(want); i++ {
+		if want[i] != have[i] {
+			t.Errorf("title doesn't match\nwant: %#v\nhave: %#v\n", want[i], have[i])
+		}
+	}
+}