Handle relative <link> elements in Atom parser — use xml:base when present to resolve. Add test case.

Modules/Parser/Sources/Parser/FeedParser/Feeds/XML/AtomParser.swift

@@ -25,6 +25,7 @@ final class AtomParser {
 		attributesStack.last
 	}
 
+	private var xmlBaseURL: URL?
 	private var parsingXHTML = false
 	private var xhtmlString: String?
 
@@ -90,6 +91,7 @@ private extension AtomParser {
 		static let type = "type"
 		static let length = "length"
 		static let xmlLang = "xml:lang"
+		static let xmlBase = "xml:base"
 	}
 
 	func currentString(_ saxParser: SAXParser) -> String? {
@@ -139,13 +141,23 @@ private extension AtomParser {
 		}
 	}
 
-	func addFeedLanguage() {
+	func addFeedAttributes() {
 
-		guard feed.language == nil, let currentAttributes else {
+		guard let currentAttributes else {
 			return
 		}
 
-		feed.language = currentAttributes[XMLString.xmlLang]
+		if feed.language == nil {
+			feed.language = currentAttributes[XMLString.xmlLang]
+		}
+
+		if xmlBaseURL == nil {
+			if let xmlBase = currentAttributes[XMLString.xmlBase] {
+				if let baseURL = URL(string: xmlBase) {
+					xmlBaseURL = baseURL
+				}
+			}
+		}
 	}
 
 	func addArticle() {
@@ -225,7 +237,8 @@ private extension AtomParser {
 		guard let urlString = attributes[XMLString.href], !urlString.isEmpty else {
 			return
 		}
-		
+		let resolvedURLString = linkResolvedAgainstXMLBase(urlString)
+
 		var rel = attributes[XMLString.rel]
 		if rel?.isEmpty ?? true {
 			rel = XMLString.alternate
@@ -233,21 +246,33 @@ private extension AtomParser {
 
 		if rel == XMLString.related {
 			if article.link == nil {
-				article.link = urlString
+				article.link = resolvedURLString
 			}
 		}
 		else if rel == XMLString.alternate {
 			if article.permalink == nil {
-				article.permalink = urlString
+				article.permalink = resolvedURLString
 			}
 		}
 		else if rel == XMLString.enclosure {
-			if let enclosure = enclosure(urlString, attributes) {
+			if let enclosure = enclosure(resolvedURLString, attributes) {
 				article.addEnclosure(enclosure)
 			}
 		}
 	}
 
+	func linkResolvedAgainstXMLBase(_ urlString: String) -> String {
+
+		guard let xmlBaseURL else {
+			return urlString
+		}
+
+		if let resolvedURL = URL(string: urlString, relativeTo: xmlBaseURL) {
+			return resolvedURL.absoluteString
+		}
+		return urlString
+	}
+
 	func enclosure(_ urlString: String, _ attributes: StringDictionary) -> RSSEnclosure? {
 
 		let enclosure = RSSEnclosure(url: urlString)
@@ -351,7 +376,7 @@ extension AtomParser: SAXParserDelegate {
 		}
 
 		if SAXEqualTags(localName, XMLName.feed) {
-			addFeedLanguage()
+			addFeedAttributes()
 		}
 
 		saxParser.beginStoringCharacters()

Modules/Parser/Tests/ParserTests/AtomParserTests.swift

@@ -109,4 +109,22 @@ final class AtomParserTests: XCTestCase {
 			XCTAssertNotEqual(article.title, "Default Title")
 		}
 	}
+
+	func testLinkElementsWithRelativeURLs() {
+
+		// This feed has <link> elements that look like this…
+		//		<link href="/en/publish/2022/07/01/great-moments-in-document-history-reimagining-the-declaration-of-independence-as-pdf"/>
+		// …and it also has, in the feed declaration…
+		// 		xml:base="https://blog.adobe.com"
+		// …and so the <link> values should be parsed as (for example):
+		// https://blog.adobe.com/en/publish/2022/07/01/great-moments-in-document-history-reimagining-the-declaration-of-independence-as-pdf
+		// Issue: https://github.com/Ranchero-Software/NetNewsWire/issues/3662
+
+		let d = parserData("adobe", "atom", "https://blog.adobe.com/feed.xml")
+		let parsedFeed = try! FeedParser.parse(d)!
+
+		for article in parsedFeed.items {
+			XCTAssertTrue(article.url!.hasPrefix("https://blog.adobe.com/en/publish/20"))
+		}
+	}
 }