From 08640b27d55f6382af6f08d0789b8109279ae825 Mon Sep 17 00:00:00 2001
From: Frédéric Guillot <f@miniflux.net>
Date: Tue, 19 Mar 2024 21:24:30 -0700
Subject: Ensure enclosure URLs are always absolute

---
 internal/reader/rss/parser_test.go | 170 +++++++++++++++++++++++++++++--------
 1 file changed, 135 insertions(+), 35 deletions(-)

(limited to 'internal/reader/rss/parser_test.go')

diff --git a/internal/reader/rss/parser_test.go b/internal/reader/rss/parser_test.go
index e3f8450f..8d84e582 100644
--- a/internal/reader/rss/parser_test.go
+++ b/internal/reader/rss/parser_test.go
@@ -1016,15 +1016,11 @@ func TestParseEntryWithEnclosures(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
-	}
-
-	if feed.Entries[0].URL != "http://www.example.org/entries/1" {
-		t.Errorf("Incorrect entry URL, got: %s", feed.Entries[0].URL)
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
 
 	if len(feed.Entries[0].Enclosures) != 1 {
-		t.Errorf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
 
 	if feed.Entries[0].Enclosures[0].URL != "http://www.example.org/myaudiofile.mp3" {
@@ -1065,15 +1061,11 @@ func TestParseEntryWithIncorrectEnclosureLength(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
-	}
-
-	if feed.Entries[0].URL != "http://www.example.org/entries/1" {
-		t.Errorf("Incorrect entry URL, got: %s", feed.Entries[0].URL)
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
 
 	if len(feed.Entries[0].Enclosures) != 2 {
-		t.Errorf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
 
 	if feed.Entries[0].Enclosures[0].URL != "http://www.example.org/myaudiofile.mp3" {
@@ -1093,6 +1085,39 @@ func TestParseEntryWithIncorrectEnclosureLength(t *testing.T) {
 	}
 }
 
+func TestParseEntryWithDuplicatedEnclosureURL(t *testing.T) {
+	data := `<?xml version="1.0" encoding="utf-8"?>
+		<rss version="2.0">
+		<channel>
+		<title>My Podcast Feed</title>
+		<link>http://example.org</link>
+		<item>
+			<title>Podcasting with RSS</title>
+			<link>http://www.example.org/entries/1</link>
+			<enclosure url="http://www.example.org/myaudiofile.mp3" type="audio/mpeg" />
+			<enclosure url="   http://www.example.org/myaudiofile.mp3   " type="audio/mpeg" />
+		</item>
+		</channel>
+		</rss>`
+
+	feed, err := Parse("https://example.org/", bytes.NewReader([]byte(data)))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if len(feed.Entries) != 1 {
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
+	}
+
+	if len(feed.Entries[0].Enclosures) != 1 {
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+	}
+
+	if feed.Entries[0].Enclosures[0].URL != "http://www.example.org/myaudiofile.mp3" {
+		t.Errorf("Incorrect enclosure URL, got: %s", feed.Entries[0].Enclosures[0].URL)
+	}
+}
+
 func TestParseEntryWithEmptyEnclosureURL(t *testing.T) {
 	data := `<?xml version="1.0" encoding="utf-8"?>
 		<rss version="2.0">
@@ -1106,7 +1131,7 @@ func TestParseEntryWithEmptyEnclosureURL(t *testing.T) {
 			<description>An overview of RSS podcasting</description>
 			<pubDate>Fri, 15 Jul 2005 00:00:00 -0500</pubDate>
 			<guid isPermaLink="true">http://www.example.org/entries/1</guid>
-			<enclosure url="" length="0"/>
+			<enclosure url=" " length="0"/>
 		</item>
 		</channel>
 		</rss>`
@@ -1117,15 +1142,47 @@ func TestParseEntryWithEmptyEnclosureURL(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
 
-	if feed.Entries[0].URL != "http://www.example.org/entries/1" {
-		t.Errorf("Incorrect entry URL, got: %s", feed.Entries[0].URL)
+	if len(feed.Entries[0].Enclosures) != 0 {
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
+}
 
-	if len(feed.Entries[0].Enclosures) != 0 {
-		t.Errorf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+func TestParseEntryWithRelativeEnclosureURL(t *testing.T) {
+	data := `<?xml version="1.0" encoding="utf-8"?>
+		<rss version="2.0">
+		<channel>
+		<title>My Podcast Feed</title>
+		<link>http://example.org</link>
+		<author>some.email@example.org</author>
+		<item>
+			<title>Podcasting with RSS</title>
+			<link>http://www.example.org/entries/1</link>
+			<description>An overview of RSS podcasting</description>
+			<pubDate>Fri, 15 Jul 2005 00:00:00 -0500</pubDate>
+			<guid isPermaLink="true">http://www.example.org/entries/1</guid>
+			<enclosure url=" /files/file.mp3  "/>
+		</item>
+		</channel>
+		</rss>`
+
+	feed, err := Parse("https://example.org/", bytes.NewReader([]byte(data)))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if len(feed.Entries) != 1 {
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
+	}
+
+	if len(feed.Entries[0].Enclosures) != 1 {
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+	}
+
+	if feed.Entries[0].Enclosures[0].URL != "http://example.org/files/file.mp3" {
+		t.Errorf("Incorrect enclosure URL, got: %q", feed.Entries[0].Enclosures[0].URL)
 	}
 }
 
@@ -1154,15 +1211,11 @@ func TestParseEntryWithFeedBurnerEnclosures(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
-	}
-
-	if feed.Entries[0].URL != "http://www.example.org/entries/1" {
-		t.Errorf("Incorrect entry URL, got: %s", feed.Entries[0].URL)
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
 
 	if len(feed.Entries[0].Enclosures) != 1 {
-		t.Errorf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
 
 	if feed.Entries[0].Enclosures[0].URL != "http://example.org/67ca416c-f22a-4228-a681-68fc9998ec10/File.mp3" {
@@ -1178,6 +1231,42 @@ func TestParseEntryWithFeedBurnerEnclosures(t *testing.T) {
 	}
 }
 
+func TestParseEntryWithFeedBurnerEnclosuresAndRelativeURL(t *testing.T) {
+	data := `<?xml version="1.0" encoding="utf-8"?>
+		<rss version="2.0" xmlns:feedburner="http://rssnamespace.org/feedburner/ext/1.0">
+		<channel>
+		<title>My Example Feed</title>
+		<link>http://example.org</link>
+		<item>
+			<title>Example Item</title>
+			<link>http://www.example.org/entries/1</link>
+			<enclosure
+				url="http://feedproxy.google.com/~r/example/~5/lpMyFSCvubs/File.mp3"
+				length="76192460"
+				type="audio/mpeg" />
+			<feedburner:origEnclosureLink>/67ca416c-f22a-4228-a681-68fc9998ec10/File.mp3</feedburner:origEnclosureLink>
+		</item>
+		</channel>
+		</rss>`
+
+	feed, err := Parse("https://example.org/", bytes.NewReader([]byte(data)))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if len(feed.Entries) != 1 {
+		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
+	}
+
+	if len(feed.Entries[0].Enclosures) != 1 {
+		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
+	}
+
+	if feed.Entries[0].Enclosures[0].URL != "http://example.org/67ca416c-f22a-4228-a681-68fc9998ec10/File.mp3" {
+		t.Errorf("Incorrect enclosure URL, got: %s", feed.Entries[0].Enclosures[0].URL)
+	}
+}
+
 func TestParseEntryWithRelativeURL(t *testing.T) {
 	data := `<?xml version="1.0" encoding="utf-8"?>
 		<rss version="2.0">
@@ -1389,7 +1478,7 @@ func TestParseEntryWithMediaGroup(t *testing.T) {
 		<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/">
 		<channel>
 		<title>My Example Feed</title>
-		<link>http://example.org</link>
+		<link>https://example.org</link>
 		<item>
 			<title>Example Item</title>
 			<link>http://www.example.org/entries/1</link>
@@ -1400,7 +1489,9 @@ func TestParseEntryWithMediaGroup(t *testing.T) {
 				<media:content type="application/x-bittorrent" url="https://example.org/file2.torrent" isDefault="true"></media:content>
 				<media:content type="application/x-bittorrent" url="https://example.org/file3.torrent"></media:content>
 				<media:content type="application/x-bittorrent" url="https://example.org/file4.torrent"></media:content>
-				<media:content type="application/x-bittorrent" url="https://example.org/file5.torrent" fileSize="42"></media:content>
+				<media:content type="application/x-bittorrent" url="https://example.org/file4.torrent"></media:content>
+				<media:content type="application/x-bittorrent" url=" file5.torrent  " fileSize="42"></media:content>
+				<media:content type="application/x-bittorrent" url="  " fileSize="42"></media:content>
 				<media:rating>nonadult</media:rating>
 			</media:group>
 			<media:thumbnail url="https://example.org/image.jpg" height="122" width="223"></media:thumbnail>
@@ -1453,15 +1544,19 @@ func TestParseEntryWithMediaContent(t *testing.T) {
 		<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/">
 		<channel>
 		<title>My Example Feed</title>
-		<link>http://example.org</link>
+		<link>https://example.org</link>
 		<item>
 			<title>Example Item</title>
 			<link>http://www.example.org/entries/1</link>
 			<media:thumbnail url="https://example.org/thumbnail.jpg" />
+			<media:thumbnail url="https://example.org/thumbnail.jpg" />
+			<media:thumbnail url=" thumbnail.jpg  " />
+			<media:thumbnail url="   " />
 			<media:content url="https://example.org/media1.jpg" medium="image">
 				<media:title type="html">Some Title for Media 1</media:title>
 			</media:content>
-			<media:content url="https://example.org/media2.jpg" medium="image" />
+			<media:content url="   /media2.jpg   " medium="image" />
+			<media:content url="    " medium="image" />
 		</item>
 		</channel>
 		</rss>`
@@ -1472,9 +1567,9 @@ func TestParseEntryWithMediaContent(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
-	if len(feed.Entries[0].Enclosures) != 3 {
+	if len(feed.Entries[0].Enclosures) != 4 {
 		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
 
@@ -1483,6 +1578,7 @@ func TestParseEntryWithMediaContent(t *testing.T) {
 		mimeType string
 		size     int64
 	}{
+		{"https://example.org/thumbnail.jpg", "image/*", 0},
 		{"https://example.org/thumbnail.jpg", "image/*", 0},
 		{"https://example.org/media1.jpg", "image/*", 0},
 		{"https://example.org/media2.jpg", "image/*", 0},
@@ -1508,11 +1604,14 @@ func TestParseEntryWithMediaPeerLink(t *testing.T) {
 		<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/">
 		<channel>
 		<title>My Example Feed</title>
-		<link>http://example.org</link>
+		<link>https://website.example.org</link>
 		<item>
 			<title>Example Item</title>
 			<link>http://www.example.org/entries/1</link>
-			<media:peerLink type="application/x-bittorrent" href="http://www.example.org/file.torrent" />
+			<media:peerLink type="application/x-bittorrent" href="https://www.example.org/file.torrent" />
+			<media:peerLink type="application/x-bittorrent" href="https://www.example.org/file.torrent" />
+			<media:peerLink type="application/x-bittorrent" href="  file2.torrent   " />
+			<media:peerLink type="application/x-bittorrent" href="    " />
 		</item>
 		</channel>
 		</rss>`
@@ -1523,10 +1622,10 @@ func TestParseEntryWithMediaPeerLink(t *testing.T) {
 	}
 
 	if len(feed.Entries) != 1 {
-		t.Errorf("Incorrect number of entries, got: %d", len(feed.Entries))
+		t.Fatalf("Incorrect number of entries, got: %d", len(feed.Entries))
 	}
 
-	if len(feed.Entries[0].Enclosures) != 1 {
+	if len(feed.Entries[0].Enclosures) != 2 {
 		t.Fatalf("Incorrect number of enclosures, got: %d", len(feed.Entries[0].Enclosures))
 	}
 
@@ -1535,7 +1634,8 @@ func TestParseEntryWithMediaPeerLink(t *testing.T) {
 		mimeType string
 		size     int64
 	}{
-		{"http://www.example.org/file.torrent", "application/x-bittorrent", 0},
+		{"https://www.example.org/file.torrent", "application/x-bittorrent", 0},
+		{"https://website.example.org/file2.torrent", "application/x-bittorrent", 0},
 	}
 
 	for index, enclosure := range feed.Entries[0].Enclosures {
-- 
cgit v1.2.3