Add support for alternate relative link on Atom feeds

prof18 · Jan 2, 2025 · 165c226 · 165c226
1 parent 1910992
commit 165c226
Show file tree

Hide file tree

Showing 18 changed files with 119 additions and 30 deletions.
diff --git a/gradle/libs.versions.toml b/gradle/libs.versions.toml
@@ -1,5 +1,5 @@
 [versions]
-android-gradle-plugin = "8.7.3"
+android-gradle-plugin = "8.2.2"
 android-compile-sdk = "34"
 android-min-sdk = "21"
 android-target-sdk = "34"

diff --git a/rssparser/src/androidMain/kotlin/com/prof18/rssparser/internal/AndroidXmlParser.kt b/rssparser/src/androidMain/kotlin/com/prof18/rssparser/internal/AndroidXmlParser.kt
@@ -53,7 +53,7 @@ internal class AndroidXmlParser(
                     if (xmlPullParser.contains(RssKeyword.Rss)) {
                         rssChannel = extractRSSContent(xmlPullParser)
                     } else if (xmlPullParser.contains(AtomKeyword.Atom)) {
-                        rssChannel = extractAtomContent(xmlPullParser)
+                        rssChannel = extractAtomContent(xmlPullParser, input)
                     }
                 }
                 eventType = xmlPullParser.next()

diff --git a/rssparser/src/androidMain/kotlin/com/prof18/rssparser/internal/atom/AtomParser.kt b/rssparser/src/androidMain/kotlin/com/prof18/rssparser/internal/atom/AtomParser.kt
@@ -19,6 +19,7 @@ package com.prof18.rssparser.internal.atom
 
 import com.prof18.rssparser.internal.AtomKeyword
 import com.prof18.rssparser.internal.ChannelFactory
+import com.prof18.rssparser.internal.ParserInput
 import com.prof18.rssparser.internal.attributeValue
 import com.prof18.rssparser.internal.contains
 import com.prof18.rssparser.internal.nextTrimmedText
@@ -30,6 +31,7 @@ import org.xmlpull.v1.XmlPullParserException
 
 internal fun CoroutineScope.extractAtomContent(
     xmlPullParser: XmlPullParser,
+    input: ParserInput,
 ): RssChannel {
     val channelFactory = ChannelFactory()
 
@@ -41,7 +43,6 @@ internal fun CoroutineScope.extractAtomContent(
 
     // Start parsing the xml
     loop@ while (eventType != XmlPullParser.END_DOCUMENT && isActive) {
-
         // Start parsing the item
         when {
             eventType == XmlPullParser.START_TAG -> when {
@@ -154,10 +155,19 @@ internal fun CoroutineScope.extractAtomContent(
                         val rel = xmlPullParser.attributeValue(
                             AtomKeyword.Link.Rel
                         )
+                        val link = if (input.baseUrl != null &&
+                            rel == AtomKeyword.Link.Rel.Alternate.value &&
+                            // Some feeds have full links
+                            href?.startsWith("/") == true
+                        ) {
+                            input.baseUrl + href
+                        } else {
+                            href
+                        }
                         if (rel != AtomKeyword.Link.Edit.value && rel != AtomKeyword.Link.Self.value) {
                             when {
-                                insideItem -> channelFactory.articleBuilder.link(href)
-                                else -> channelFactory.channelBuilder.link(href)
+                                insideItem -> channelFactory.articleBuilder.link(link)
+                                else -> channelFactory.channelBuilder.link(link)
                             }
                         }
                     }

diff --git a/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/IosXmlFetcher.kt b/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/IosXmlFetcher.kt
@@ -19,10 +19,11 @@ import kotlin.coroutines.resumeWithException
 internal class IosXmlFetcher(
     private val nsUrlSession: NSURLSession,
 ) : XmlFetcher {
-
     override suspend fun fetchXml(url: String): ParserInput = suspendCancellableCoroutine { continuation ->
+        val nsUrl = NSURL(string = url)
+        val baseUrl = nsUrl.scheme + "://" + nsUrl.host
         val task = nsUrlSession.dataTaskWithURL(
-            url = NSURL(string = url)
+            url = nsUrl
         ) { data: NSData?, response: NSURLResponse?, error: NSError? ->
             if (error != null) {
                 val throwable = Throwable(
@@ -36,7 +37,7 @@ internal class IosXmlFetcher(
                 )
                 continuation.resumeWithException(exception)
             } else if (data != null) {
-                continuation.resume(ParserInput(data))
+                continuation.resume(ParserInput(data, baseUrl))
             }
         }
 

diff --git a/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/IosXmlParser.kt b/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/IosXmlParser.kt
@@ -29,6 +29,7 @@ internal class IosXmlParser(
         val parser = NSXMLParser(stream)
         suspendCancellableCoroutine { continuation ->
             parser.delegate = NSXMLParserDelegate(
+                baseFeedUrl = input.baseUrl,
                 onEnd = { rssChannel ->
                     stream.close()
                     continuation.resume(rssChannel)
@@ -60,6 +61,7 @@ internal class IosXmlParser(
 }
 
 private class NSXMLParserDelegate(
+    private val baseFeedUrl: String?,
     private val onEnd: (RssChannel) -> Unit,
     private val onError: (NSXMLParsingException) -> Unit,
 ) : NSObject(), NSXMLParserDelegateProtocol {
@@ -80,7 +82,7 @@ private class NSXMLParserDelegate(
             }
 
             AtomKeyword.Atom.value -> {
-                feedHandler = AtomFeedHandler()
+                feedHandler = AtomFeedHandler(baseFeedUrl)
             }
 
             else -> {

diff --git a/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/ParserInput.kt b/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/ParserInput.kt
@@ -3,5 +3,6 @@ package com.prof18.rssparser.internal
 import platform.Foundation.NSData
 
 internal actual data class ParserInput(
-    val data: NSData
+    val data: NSData,
+    val baseUrl: String? = null,
 )
diff --git a/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/atom/AtomFeedHandler.kt b/rssparser/src/appleMain/kotlin/com/prof18/rssparser/internal/atom/AtomFeedHandler.kt
@@ -6,7 +6,9 @@ import com.prof18.rssparser.internal.FeedHandler
 import com.prof18.rssparser.internal.getValueOrNull
 import com.prof18.rssparser.model.RssChannel
 
-internal class AtomFeedHandler : FeedHandler {
+internal class AtomFeedHandler(
+    private val baseFeedUrl: String?,
+) : FeedHandler {
 
     private var currentElement: String? = null
 
@@ -28,18 +30,28 @@ internal class AtomFeedHandler : FeedHandler {
                 if (isInsideItem) {
                     val category = attributes.getValueOrNull(
                         AtomKeyword.Entry.Term.value,
-                        ) as? String
+                    ) as? String
                     channelFactory.articleBuilder.addCategory(category)
                 }
             }
 
             AtomKeyword.Link.value -> {
                 val href = attributes.getValueOrNull(AtomKeyword.Link.Href.value) as? String
                 val rel = attributes.getValueOrNull(AtomKeyword.Link.Rel.value) as? String
+                val link = if (
+                    baseFeedUrl != null &&
+                    rel == AtomKeyword.Link.Rel.Alternate.value &&
+                    // Some feeds have full links
+                    href?.startsWith("/") == true
+                ) {
+                    baseFeedUrl + href
+                } else {
+                    href
+                }
                 if (rel != AtomKeyword.Link.Edit.value && rel != AtomKeyword.Link.Self.value) {
                     when {
-                        isInsideItem -> channelFactory.articleBuilder.link(href)
-                        else -> channelFactory.channelBuilder.link(href)
+                        isInsideItem -> channelFactory.articleBuilder.link(link)
+                        else -> channelFactory.channelBuilder.link(link)
                     }
                 }
             }

diff --git a/rssparser/src/appleTest/kotlin/com/prof18/rssparser/AppleTestUtils.kt b/rssparser/src/appleTest/kotlin/com/prof18/rssparser/AppleTestUtils.kt
@@ -13,17 +13,17 @@ import platform.posix.getenv
 
 @OptIn(ExperimentalForeignApi::class)
 internal actual fun readFileFromResources(
-    resourceName: String
+    resourceName: String,
 ): ParserInput {
     val s = getenv("TEST_RESOURCES_ROOT")?.toKString()
     val path = "$s/${resourceName}"
     val data = NSData.dataWithContentsOfFile(path)
-    return ParserInput(requireNotNull(data))
+    return ParserInput(requireNotNull(value = data), baseUrl = BASE_FEED_URL)
 }
 
 @OptIn(ExperimentalForeignApi::class, BetaInteropApi::class)
 internal actual fun readFileFromResourcesAsString(
-    resourceName: String
+    resourceName: String,
 ): String {
     val s = getenv("TEST_RESOURCES_ROOT")?.toKString()
     val path = "$s/${resourceName}"

diff --git a/rssparser/src/commonJvmAndroidMain/kotlin/com/prof18/rssparser/internal/JvmXmlFetcher.kt b/rssparser/src/commonJvmAndroidMain/kotlin/com/prof18/rssparser/internal/JvmXmlFetcher.kt
@@ -17,8 +17,10 @@ internal class JvmXmlFetcher(
 
     override suspend fun fetchXml(url: String): ParserInput {
         val request = createRequest(url)
+        val baseUrl = request.url.scheme + "://" + request.url.host
         return ParserInput(
-            inputStream = callFactory.newCall(request).awaitForInputStream()
+            inputStream = callFactory.newCall(request).awaitForInputStream(),
+            baseUrl = baseUrl,
         )
     }
 

diff --git a/rssparser/src/commonJvmAndroidMain/kotlin/com/prof18/rssparser/internal/ParserInput.kt b/rssparser/src/commonJvmAndroidMain/kotlin/com/prof18/rssparser/internal/ParserInput.kt
@@ -3,5 +3,6 @@ package com.prof18.rssparser.internal
 import java.io.InputStream
 
 internal actual data class ParserInput(
-    val inputStream: InputStream
+    val inputStream: InputStream,
+    val baseUrl: String? = null,
 )
diff --git a/rssparser/src/commonJvmAndroidTest/kotlin/com/prof18/rssparser/JvmAndroidTestUtils.kt b/rssparser/src/commonJvmAndroidTest/kotlin/com/prof18/rssparser/JvmAndroidTestUtils.kt
@@ -10,7 +10,8 @@ internal actual fun readFileFromResources(
     val path = System.getenv("TEST_RESOURCES_ROOT")
     val file = File("$path/$resourceName")
     return ParserInput(
-        inputStream = FileInputStream(file)
+        inputStream = FileInputStream(file),
+        baseUrl = BASE_FEED_URL,
     )
 }
 

diff --git a/rssparser/src/commonMain/kotlin/com/prof18/rssparser/internal/AtomKeyword.kt b/rssparser/src/commonMain/kotlin/com/prof18/rssparser/internal/AtomKeyword.kt
@@ -3,15 +3,18 @@ package com.prof18.rssparser.internal
 internal sealed class AtomKeyword(val value: String) {
     data object Atom : AtomKeyword("feed")
     data object Title : AtomKeyword("title")
-    data object Icon: AtomKeyword("icon")
+    data object Icon : AtomKeyword("icon")
     data object Link : AtomKeyword("link") {
         data object Href : AtomKeyword("href")
-        data object Rel : AtomKeyword("rel")
+        data object Rel : AtomKeyword("rel") {
+            data object Alternate : AtomKeyword("alternate")
+        }
         data object Edit : AtomKeyword("edit")
         data object Self : AtomKeyword("self")
     }
+
     data object Subtitle : AtomKeyword("subtitle")
-    data object Updated: AtomKeyword("updated")
+    data object Updated : AtomKeyword("updated")
     data object Entry {
         data object Item : AtomKeyword("entry")
         data object Guid : AtomKeyword("id")

diff --git a/rssparser/src/commonTest/kotlin/com/prof18/rssparser/TestUtils.kt b/rssparser/src/commonTest/kotlin/com/prof18/rssparser/TestUtils.kt
@@ -11,3 +11,5 @@ internal expect fun readFileFromResourcesAsString(
 ): String
 
 internal expect val currentTarget: CurrentTarget
+
+internal const val BASE_FEED_URL = "https://www.base-feed-url.com"
diff --git a/rssparser/src/commonTest/kotlin/com/prof18/rssparser/atom/XmlParserAtomRelativeLink.kt b/rssparser/src/commonTest/kotlin/com/prof18/rssparser/atom/XmlParserAtomRelativeLink.kt
@@ -0,0 +1,18 @@
+package com.prof18.rssparser.atom
+
+import com.prof18.rssparser.BASE_FEED_URL
+import com.prof18.rssparser.BaseXmlParserTest
+
+class XmlParserAtomRelativeLink : BaseXmlParserTest(
+    feedPath = "atom-relative-url.xml",
+    channelTitle = "Aurimas Liutikas",
+    channelLink = "$BASE_FEED_URL/",
+    channelDescription = "A personal page of Aurimas Liutikas.",
+    channelLastBuildDate = "2024-12-19T06:13:47+00:00",
+    articleTitle = "What The Distribution - What Is Inside Gradle Distribution Zips?",
+    articlePubDate = "2024-12-18T00:00:00+00:00",
+    articleLink = "$BASE_FEED_URL/2024/12/18/What-The-Distribution.html",
+    articleGuid = "/2024/12/18/What-The-Distribution",
+    articleDescription = "When I saw this post",
+    articleAuthor = "",
+)
diff --git a/rssparser/src/commonTest/resources/atom-relative-url.xml b/rssparser/src/commonTest/resources/atom-relative-url.xml
@@ -0,0 +1,22 @@
+<feed>
+    <generator uri="https://jekyllrb.com/" version="3.10.0">Jekyll</generator>
+    <link href="/feed.xml" rel="self" type="application/atom+xml"/>
+    <link href="/" rel="alternate" type="text/html"/>
+    <updated>2024-12-19T06:13:47+00:00</updated>
+    <id>/feed.xml</id>
+    <title type="html">Aurimas Liutikas</title>
+    <subtitle>A personal page of Aurimas Liutikas. </subtitle>
+    <entry>
+        <title type="html">
+            What The Distribution - What Is Inside Gradle Distribution Zips?
+        </title>
+        <link href="/2024/12/18/What-The-Distribution.html" rel="alternate" type="text/html" title="What The Distribution - What Is Inside Gradle Distribution Zips?"/>
+        <published>2024-12-18T00:00:00+00:00</published>
+        <updated>2024-12-18T00:00:00+00:00</updated>
+        <id>/2024/12/18/What-The-Distribution</id>
+        <author>
+            <name/>
+        </author>
+        <summary type="html">When I saw this post</summary>
+    </entry>
+</feed>
diff --git a/rssparser/src/jvmMain/kotlin/com/prof18/rssparser/internal/JvmXmlParser.kt b/rssparser/src/jvmMain/kotlin/com/prof18/rssparser/internal/JvmXmlParser.kt
@@ -21,7 +21,7 @@ internal class JvmXmlParser(
     override suspend fun parseXML(input: ParserInput): RssChannel = withContext(dispatcher) {
         try {
             val parser = SAXParserFactory.newInstance().newSAXParser()
-            val handler = SaxFeedHandler()
+            val handler = SaxFeedHandler(input.baseUrl)
 
             if (charset != null) {
                 val inputSource = InputSource(input.inputStream).apply {
@@ -46,11 +46,14 @@ internal class JvmXmlParser(
     override fun generateParserInputFromString(rawRssFeed: String): ParserInput {
         val cleanedXml = rawRssFeed.trim()
         val inputStream: InputStream = cleanedXml.byteInputStream(charset ?: Charsets.UTF_8)
-        return ParserInput(inputStream)
+        // TODO: changeme
+        return ParserInput(inputStream, "TODO")
     }
 }
 
-private class SaxFeedHandler : DefaultHandler() {
+private class SaxFeedHandler (
+    private val feedUrl: String?,
+) : DefaultHandler() {
     private var feedHandler: FeedHandler? = null
     private val textBuilder: StringBuilder = StringBuilder()
 
@@ -70,7 +73,7 @@ private class SaxFeedHandler : DefaultHandler() {
                 feedHandler = RssFeedHandler()
             }
             AtomKeyword.Atom.value -> {
-                feedHandler = AtomFeedHandler()
+                feedHandler = AtomFeedHandler(feedUrl)
             }
             else -> feedHandler?.onStartRssElement(qName, attributes)
         }

diff --git a/rssparser/src/jvmMain/kotlin/com/prof18/rssparser/internal/atom/AtomFeedHandler.kt b/rssparser/src/jvmMain/kotlin/com/prof18/rssparser/internal/atom/AtomFeedHandler.kt
@@ -6,7 +6,9 @@ import com.prof18.rssparser.internal.FeedHandler
 import com.prof18.rssparser.model.RssChannel
 import org.xml.sax.Attributes
 
-internal class AtomFeedHandler : FeedHandler {
+internal class AtomFeedHandler(
+    private val baseFeedUrl: String?,
+) : FeedHandler {
 
     private var isInsideItem = false
     private var isInsideChannel = true
@@ -29,9 +31,18 @@ internal class AtomFeedHandler : FeedHandler {
                 val href = attributes?.getValue(AtomKeyword.Link.Href.value)
                 val rel = attributes?.getValue(AtomKeyword.Link.Rel.value)
                 if (rel != AtomKeyword.Link.Edit.value && rel != AtomKeyword.Link.Self.value) {
+                    val link = if (baseFeedUrl != null &&
+                        rel == AtomKeyword.Link.Rel.Alternate.value &&
+                        // Some feeds have full links
+                        href?.startsWith("/") == true
+                    ) {
+                        baseFeedUrl + href
+                    } else {
+                        href
+                    }
                     when {
-                        isInsideItem -> channelFactory.articleBuilder.link(href)
-                        else -> channelFactory.channelBuilder.link(href)
+                        isInsideItem -> channelFactory.articleBuilder.link(link)
+                        else -> channelFactory.channelBuilder.link(link)
                     }
                 }
             }

diff --git a/...pp.xcodeproj/project.xcworkspace/xcuserdata/mg.xcuserdatad/UserInterfaceState.xcuserstate b/...pp.xcodeproj/project.xcworkspace/xcuserdata/mg.xcuserdatad/UserInterfaceState.xcuserstate
Original file line number	Diff line number	Diff line change
Expand Up		@@ -11,3 +11,5 @@ internal expect fun readFileFromResourcesAsString(
		): String

		internal expect val currentTarget: CurrentTarget

		internal const val BASE_FEED_URL = "https://www.base-feed-url.com"