Guess RSS type by using konsume-xml instead of reading content by hand

api/src/main/java/com/readrops/api/localfeed/LocalRSSDataSource.kt

@@ -2,6 +2,7 @@ package com.readrops.api.localfeed
 
 import android.accounts.NetworkErrorException
 import androidx.annotation.WorkerThread
+import com.gitlab.mvysny.konsumexml.konsumeXml
 import com.readrops.api.localfeed.json.JSONFeedAdapter
 import com.readrops.api.localfeed.json.JSONItemsAdapter
 import com.readrops.api.utils.ApiUtils
@@ -49,10 +50,17 @@ class LocalRSSDataSource(private val httpClient: OkHttpClient): KoinComponent {
                 var type = LocalRSSHelper.getRSSType(contentType)
 
                 val bodyArray = response.peekBody(Long.MAX_VALUE).bytes()
+                val konsumer = ByteArrayInputStream(bodyArray).konsumeXml()
 
                 // if we can't guess type based on content-type header, we use the content
-                if (type == LocalRSSHelper.RSSType.UNKNOWN)
+                if (type == LocalRSSHelper.RSSType.UNKNOWN) {
-                    type = LocalRSSHelper.getRSSContentType(ByteArrayInputStream(bodyArray))
+                    val rootKonsumer = konsumer.nextElement(LocalRSSHelper.RSS_ROOT_NAMES)
+
+                    if (rootKonsumer != null) {
+                        type = LocalRSSHelper.guessRSSType(rootKonsumer)
+                        konsumer.close()
+                    }
+                }
                 // if we can't guess type even with the content, we are unable to go further
                 if (type == LocalRSSHelper.RSSType.UNKNOWN) throw UnknownFormatException("Unable to guess $url RSS type")
 
@@ -85,8 +93,15 @@ class LocalRSSDataSource(private val httpClient: OkHttpClient): KoinComponent {
 
             var type = LocalRSSHelper.getRSSType(contentType)
 
-            if (type == LocalRSSHelper.RSSType.UNKNOWN)
+            if (type == LocalRSSHelper.RSSType.UNKNOWN) {
-                type = LocalRSSHelper.getRSSContentType(response.body?.byteStream()!!) // stream is closed in helper method
+                val konsumer = response.body!!.byteStream().konsumeXml().apply {
+                    val rootKonsumer = nextElement(LocalRSSHelper.RSS_ROOT_NAMES)
+
+                    rootKonsumer?.let { type = LocalRSSHelper.guessRSSType(rootKonsumer) }
+                }
+
+                konsumer.close()
+            }
 
             type != LocalRSSHelper.RSSType.UNKNOWN
         } else false

api/src/main/java/com/readrops/api/localfeed/LocalRSSHelper.kt

@@ -1,5 +1,8 @@
 package com.readrops.api.localfeed
 
+import com.gitlab.mvysny.konsumexml.Konsumer
+import com.gitlab.mvysny.konsumexml.Names
+import com.readrops.api.utils.extensions.checkRoot
 import java.io.InputStream
 
 object LocalRSSHelper {
@@ -14,6 +17,11 @@ object LocalRSSHelper {
     private const val RSS_2_REGEX = "rss.*version=\"2.0\""
     private const val ATOM_REGEX = "<feed.* xmlns=\"http://www.w3.org/2005/Atom\""
 
+    const val RSS_1_ROOT_NAME = "rdf"
+    const val RSS_2_ROOT_NAME = "rss"
+    const val ATOM_ROOT_NAME = "feed"
+    val RSS_ROOT_NAMES = Names.of(RSS_1_ROOT_NAME, RSS_2_ROOT_NAME, ATOM_ROOT_NAME)
+
     /**
      * Guess RSS type based on content-type header
      */
@@ -56,6 +64,13 @@ object LocalRSSHelper {
         return if (type != null) getRSSType(type) != RSSType.UNKNOWN else false
     }
 
+    fun guessRSSType(konsumer: Konsumer): RSSType = when {
+        konsumer.checkRoot(RSS_1_ROOT_NAME) -> RSSType.RSS_1
+        konsumer.checkRoot(RSS_2_ROOT_NAME) -> RSSType.RSS_2
+        konsumer.checkRoot(ATOM_ROOT_NAME) -> RSSType.ATOM
+        else -> RSSType.UNKNOWN
+    }
+
     enum class RSSType {
         RSS_1,
         RSS_2,

api/src/main/java/com/readrops/api/utils/extensions/KonsumerExtensions.kt

@@ -19,3 +19,10 @@ fun Konsumer.nullableTextRecursively(): String? {
     val text = textRecursively()
     return if (text.isNotEmpty()) text.trim() else null
 }
+
+fun Konsumer.checkRoot(name: String): Boolean = try {
+    checkCurrent(name)
+    true
+} catch (e: Exception) {
+    false
+}