Ensure XML parser uses response encoding (#672)

jocmp · Jan 4, 2025 · 7e266fc · 7e266fc
1 parent 88eff2f
commit 7e266fc
Show file tree

Hide file tree

Showing 5 changed files with 136 additions and 7 deletions.
diff --git a/rssparser/src/main/kotlin/com/jocmp/rssparser/internal/DefaultParser.kt b/rssparser/src/main/kotlin/com/jocmp/rssparser/internal/DefaultParser.kt
@@ -67,7 +67,7 @@ internal class DefaultParser(
 
 private fun tryXmlParse(input: ParserInput): Document? {
     return try {
-        Jsoup.parse(input.inputStream(), null, "", JsoupParser.xmlParser())
+        Jsoup.parse(input.inputStream(), input.charset?.toString(), "", JsoupParser.xmlParser())
     } catch (e: IOException) {
         null
     }

diff --git a/rssparser/src/test/kotlin/com/jocmp/rssparser/BaseParserTest.kt b/rssparser/src/test/kotlin/com/jocmp/rssparser/BaseParserTest.kt
@@ -1,11 +1,12 @@
 package com.jocmp.rssparser
 
-import com.jocmp.rssparser.model.RssItem
+import com.jocmp.rssparser.model.ItunesChannelData
+import com.jocmp.rssparser.model.ItunesItemData
 import com.jocmp.rssparser.model.RssChannel
 import com.jocmp.rssparser.model.RssImage
-import com.jocmp.rssparser.model.ItunesItemData
-import com.jocmp.rssparser.model.ItunesChannelData
+import com.jocmp.rssparser.model.RssItem
 import kotlinx.coroutines.test.runTest
+import java.nio.charset.Charset
 import kotlin.test.BeforeTest
 import kotlin.test.Test
 import kotlin.test.assertEquals
@@ -38,14 +39,15 @@ abstract class BaseParserTest(
     val articleCategories: List<String> = emptyList(),
     val articleCommentsUrl: String? = null,
     val articleItunesData: ItunesItemData? = null,
+    val charset: Charset? = null
 ) {
 
     private lateinit var channel: RssChannel
     private lateinit var article: RssItem
 
     @BeforeTest
     fun setUp() = runTest {
-        val input = readFileFromResources(feedPath)
+        val input = readFileFromResources(feedPath, charset = charset)
         channel = ParserFactory.build().parse(input)
         article = channel.items[0]
     }

diff --git a/rssparser/src/test/kotlin/com/jocmp/rssparser/TestUtils.kt b/rssparser/src/test/kotlin/com/jocmp/rssparser/TestUtils.kt
@@ -2,11 +2,13 @@ package com.jocmp.rssparser
 
 import com.jocmp.rssparser.internal.ParserInput
 import java.io.File
+import java.nio.charset.Charset
 
 internal fun readFileFromResources(
-    resourceName: String
+    resourceName: String,
+    charset: Charset? = null,
 ): ParserInput {
     val file = File("src/test/resources/$resourceName")
 
-    return ParserInput(file.readBytes(), charset = null)
+    return ParserInput(file.readBytes(), charset = charset)
 }
diff --git a/rssparser/src/test/kotlin/com/jocmp/rssparser/rss/XmlParserAccentsTest.kt b/rssparser/src/test/kotlin/com/jocmp/rssparser/rss/XmlParserAccentsTest.kt
@@ -0,0 +1,20 @@
+package com.jocmp.rssparser.rss
+
+import com.jocmp.rssparser.BaseParserTest
+import com.jocmp.rssparser.model.RssImage
+import java.nio.charset.Charset
+
+class XmlParserAccentsTest  : BaseParserTest(
+    charset = Charset.forName("ISO-8859-1"),
+    feedPath = "feed-test-accents.xml",
+    channelTitle = "UOL Noticias",
+    channelLink = "http://noticias.uol.com.br/",
+    channelImage = RssImage(
+        url = "http://rss.i.uol.com.br/uol_rss.gif"
+    ),
+    channelDescription = "Últimas Notícias",
+    articleTitle = "Artur Jorge não é mais técnico do Botafogo",
+    articleLink = "https://noticias.uol.com.br/ultimas-noticias/afp/2025/01/03/artur-jorge-nao-e-mais-tecnico-do-botafogo.htm",
+    articlePubDate = "Sex, 03 Jan 2025 23:47:02 -0300",
+    articleDescription = "O português Artur Jorge não continuará como treinador do Botafogo, anunciou nesta sexta-feira (3) o atual campeão brasileiro e da Copa Libertadores.",
+)
diff --git a/rssparser/src/test/resources/feed-test-accents.xml b/rssparser/src/test/resources/feed-test-accents.xml