Add support legacy rss format;

src/parser/rss.go

@@ -5,6 +5,7 @@
 package parser
 
 import (
+	"bytes"
 	"encoding/xml"
 	"io"
 	"path"
@@ -61,9 +62,15 @@ type rssEnclosure struct {
 }
 
 func ParseRSS(r io.Reader) (*Feed, error) {
+	body, err := io.ReadAll(r)
+	if err != nil {
+		return nil, err
+	}
+	body = stripRSSDefaultNamespace(body)
+
 	srcfeed := rssFeed{}
 
-	decoder := xmlDecoder(r)
+	decoder := xmlDecoder(bytes.NewReader(body))
 	decoder.DefaultSpace = "rss"
 	if err := decoder.Decode(&srcfeed); err != nil {
 		return nil, err

src/parser/rss_test.go

@@ -327,3 +327,39 @@ func TestRSSMultipleMedia(t *testing.T) {
 		t.Fatal("invalid rss")
 	}
 }
+
+// Feeds that declare a default namespace on the root <rss> element (e.g. the
+// legacy Userland namespace) must still parse — see sud.ua/rss/rss_news_uk.xml.
+func TestRSSDefaultNamespace(t *testing.T) {
+	have, _ := Parse(strings.NewReader(`
+		<?xml version="1.0" encoding="utf-8"?>
+		<rss xmlns="http://backend.userland.com/rss2" version="2.0" xmlns:yandex="https://sud.ua/">
+			<channel>
+				<title>Example</title>
+				<link>https://example.com/</link>
+				<item>
+					<title>Title 1</title>
+					<link>https://example.com/news/1</link>
+					<description>Description 1</description>
+				</item>
+			</channel>
+		</rss>
+	`))
+	want := &Feed{
+		Title:   "Example",
+		SiteURL: "https://example.com/",
+		Items: []Item{
+			{
+				GUID:    "https://example.com/news/1",
+				URL:     "https://example.com/news/1",
+				Title:   "Title 1",
+				Content: "Description 1",
+			},
+		},
+	}
+	if !reflect.DeepEqual(want, have) {
+		t.Logf("want: %#v", want)
+		t.Logf("have: %#v", have)
+		t.Fatal("default-namespaced rss not parsed")
+	}
+}

src/parser/util.go

@@ -23,6 +23,19 @@ func firstNonEmpty(vals ...string) string {
 
 var linkRe = regexp.MustCompile(`(https?:\/\/\S+)`)
 
+// RSS 2.0 elements live in no namespace, but some feeds (e.g. those declaring
+// the legacy Userland namespace `http://backend.userland.com/rss2`) put every
+// element in a default namespace. That prevents the namespaced struct tags in
+// ParseRSS from matching, so items come back blank. Strip the default namespace
+// declaration from the root <rss> element so its children fall back to the
+// decoder's DefaultSpace. Only a default `xmlns=` is matched (not
+// `xmlns:prefix=`), leaving extension namespaces intact.
+var rssDefaultNamespaceRe = regexp.MustCompile(`(<rss\b[^>]*?)\s+xmlns="[^"]*"`)
+
+func stripRSSDefaultNamespace(body []byte) []byte {
+	return rssDefaultNamespaceRe.ReplaceAll(body, []byte("$1"))
+}
+
 func plain2html(text string) string {
 	text = linkRe.ReplaceAllString(text, `<a href="$1">$1</a>`)
 	text = strings.ReplaceAll(text, "\n", "<br>")