Add content validation for disclaimer

NDLANO · Dec 16, 2024 · d754dc5 · d754dc5
1 parent d3c8a06
commit d754dc5
Show file tree

Hide file tree

Showing 4 changed files with 86 additions and 0 deletions.
diff --git a/article-api/src/main/scala/no/ndla/articleapi/validation/ContentValidator.scala b/article-api/src/main/scala/no/ndla/articleapi/validation/ContentValidator.scala
@@ -50,6 +50,7 @@ trait ContentValidator {
     def validateArticle(article: Article, isImported: Boolean): Try[Article] = {
       val validationErrors = validateArticleContent(article.content) ++
         article.introduction.flatMap(i => validateIntroduction(i)) ++
+        validateArticleDisclaimer(article.disclaimer.getOrElse(Seq.empty)) ++
         validateMetaDescription(article.metaDescription, isImported) ++
         validateTitle(article.title) ++
         validateCopyright(article.copyright) ++
@@ -89,6 +90,14 @@ trait ContentValidator {
       }) ++ validateNonEmpty("content", contents)
     }
 
+    private def validateArticleDisclaimer(disclaimers: Seq[Disclaimer]): Seq[ValidationMessage] = {
+      disclaimers.flatMap(disclaimer => {
+        val field = s"disclaimer.${disclaimer.language}"
+        TextValidator.validate(field, disclaimer.disclaimer, allLegalTags).toList ++
+          validateLanguage("content.language", disclaimer.language)
+      })
+    }
+
     private def rootElementContainsOnlySectionBlocks(field: String, html: String): Option[ValidationMessage] = {
       val legalTopLevelTag = "section"
       val topLevelTags     = stringToJsoupDocument(html).children().asScala.map(_.tagName())

diff --git a/article-api/src/test/scala/no/ndla/articleapi/validation/ContentValidatorTest.scala b/article-api/src/test/scala/no/ndla/articleapi/validation/ContentValidatorTest.scala
@@ -15,6 +15,7 @@ import no.ndla.common.model.domain.{
   ArticleMetaImage,
   Author,
   Description,
+  Disclaimer,
   Introduction,
   RequiredLibrary,
   Tag,
@@ -30,6 +31,8 @@ class ContentValidatorTest extends UnitSuite with TestEnvironment {
   val validDocument             = """<section><h1>heisann</h1><h2>heia</h2></section>"""
   val validIntroduction         = """<p>heisann <span lang="en">heia</span></p><p>hopp</p>"""
   val invalidDocument           = """<section><invalid></invalid></section>"""
+  val validDisclaimer =
+    """<p><strong>hallo!</strong><ndlaembed data-content-id="123" data-open-in="current-context" data-resource="content-link" data-content-type="article">test</ndlaembed></p>"""
 
   test("validateArticle does not throw an exception on a valid document") {
     val article = TestData.sampleArticleWithByNcSa.copy(content = Seq(ArticleContent(validDocument, "nb")))
@@ -77,6 +80,41 @@ class ContentValidatorTest extends UnitSuite with TestEnvironment {
     contentValidator.validateArticle(article, false).isSuccess should be(true)
   }
 
+  test("validateArticle should throw an error if disclaimer contains illegal HTML tags") {
+    val article = TestData.sampleArticleWithByNcSa.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      disclaimer = Some(Seq(Disclaimer("<p><hallo>hei</hallo></p>", "nb")))
+    )
+    val result = contentValidator.validateArticle(article, false)
+    result.failed.get.asInstanceOf[ValidationException].errors.length should be(1)
+    result.failed.get.asInstanceOf[ValidationException].errors.head.message should be(
+      "The content contains illegal tags and/or attributes. Allowed HTML tags are: h3, msgroup, a, article, sub, sup, mtext, msrow, tbody, mtd, pre, thead, figcaption, mover, msup, semantics, ol, span, mroot, munder, h4, mscarries, dt, nav, mtr, ndlaembed, li, br, mrow, merror, mphantom, u, audio, ul, maligngroup, mfenced, annotation, div, strong, section, i, mspace, malignmark, mfrac, code, h2, td, aside, em, mstack, button, dl, th, tfoot, math, tr, b, blockquote, msline, col, annotation-xml, mstyle, caption, mpadded, mo, mlongdiv, msubsup, p, munderover, maction, menclose, h1, details, mmultiscripts, msqrt, mscarry, mstac, mi, mglyph, mlabeledtr, mtable, mprescripts, summary, mn, msub, ms, table, colgroup, dd"
+    )
+  }
+
+  test("validateArticle should not throw an error if disclaimer contains legal HTML tags") {
+    val article = TestData.sampleArticleWithByNcSa.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      disclaimer = Some(
+        Seq(
+          Disclaimer(
+            validDisclaimer,
+            "nb"
+          )
+        )
+      )
+    )
+    contentValidator.validateArticle(article, false).isSuccess should be(true)
+  }
+
+  test("validateArticle should not throw an error if disclaimer contains plain text") {
+    val article = TestData.sampleArticleWithByNcSa.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      disclaimer = Some(Seq(Disclaimer("disclaimer", "nb")))
+    )
+    contentValidator.validateArticle(article, false).isSuccess should be(true)
+  }
+
   test("validateArticle should throw an error if metaDescription contains HTML tags") {
     val article = TestData.sampleArticleWithByNcSa.copy(
       content = Seq(ArticleContent(validDocument, "nb")),

diff --git a/draft-api/src/main/scala/no/ndla/draftapi/validation/ContentValidator.scala b/draft-api/src/main/scala/no/ndla/draftapi/validation/ContentValidator.scala
@@ -84,6 +84,7 @@ trait ContentValidator {
         if (shouldValidateEntireArticle)
           article.content.flatMap(c => validateArticleContent(c)) ++
             article.introduction.flatMap(i => validateIntroduction(i)) ++
+            validateArticleDisclaimer(article.disclaimer.getOrElse(Seq.empty)) ++
             article.metaDescription.flatMap(m => validateMetaDescription(m)) ++
             validateTitles(article.title) ++
             article.copyright.map(x => validateCopyright(x)).toSeq.flatten ++
@@ -164,6 +165,13 @@ trait ContentValidator {
         validateLanguage("content.language", content.language)
     }
 
+    private def validateArticleDisclaimer(disclaimers: Seq[Disclaimer]): Seq[ValidationMessage] = {
+      disclaimers.flatMap(disclaimer => {
+        TextValidator.validate("disclaimer", disclaimer.disclaimer, allLegalTags).toList ++
+          validateLanguage("disclaimer.language", disclaimer.language)
+      })
+    }
+
     private def rootElementContainsOnlySectionBlocks(field: String, html: String): Option[ValidationMessage] = {
       val legalTopLevelTag = "section"
       val topLevelTags     = stringToJsoupDocument(html).children().asScala.map(_.tagName())

diff --git a/draft-api/src/test/scala/no/ndla/draftapi/validation/ContentValidatorTest.scala b/draft-api/src/test/scala/no/ndla/draftapi/validation/ContentValidatorTest.scala
@@ -21,6 +21,8 @@ class ContentValidatorTest extends UnitSuite with TestEnvironment {
   override val converterService: ConverterService = new ConverterService
   val validDocument                               = """<section><h1>heisann</h1><h2>heia</h2></section>"""
   val invalidDocument                             = """<section><invalid></invalid></section>"""
+  val validDisclaimer =
+    """<p><strong>hallo!</strong><ndlaembed data-content-id="123" data-open-in="current-context" data-resource="content-link" data-content-type="article">test</ndlaembed></p>"""
 
   val articleToValidate: Draft =
     TestData.sampleArticleWithByNcSa.copy(responsible = Some(Responsible("hei", TestData.today)))
@@ -58,7 +60,36 @@ class ContentValidatorTest extends UnitSuite with TestEnvironment {
     contentValidator.validateArticle(article).isSuccess should be(true)
   }
 
+  test("validateArticle should throw an error if disclaimer contains illegal HTML tags") {
+    val article = articleToValidate.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      disclaimer = Some(Seq(Disclaimer("<p><hallo>hei</hallo></p>", "nb")))
+    )
+    val result = contentValidator.validateArticle(article)
+    result.failed.get.asInstanceOf[ValidationException].errors.length should be(1)
+    result.failed.get.asInstanceOf[ValidationException].errors.head.message should be(
+      "The content contains illegal tags and/or attributes. Allowed HTML tags are: h3, msgroup, a, article, sub, sup, mtext, msrow, tbody, mtd, pre, thead, figcaption, mover, msup, semantics, ol, span, mroot, munder, h4, mscarries, dt, nav, mtr, ndlaembed, li, br, mrow, merror, mphantom, u, audio, ul, maligngroup, mfenced, annotation, div, strong, section, i, mspace, malignmark, mfrac, code, h2, td, aside, em, mstack, button, dl, th, tfoot, math, tr, b, blockquote, msline, col, annotation-xml, mstyle, caption, mpadded, mo, mlongdiv, msubsup, p, munderover, maction, menclose, h1, details, mmultiscripts, msqrt, mscarry, mstac, mi, mglyph, mlabeledtr, mtable, mprescripts, summary, mn, msub, ms, table, colgroup, dd"
+    )
+  }
+
+  test("validateArticle should not throw an error if disclaimer contains legal HTML tags") {
+    val article = articleToValidate.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      disclaimer = Some(Seq(Disclaimer(validDisclaimer, "nb")))
+    )
+    contentValidator.validateArticle(article).isSuccess should be(true)
+
+  }
+
   test("validateArticle should throw an error if metaDescription contains HTML tags") {
+    val article = articleToValidate.copy(
+      content = Seq(ArticleContent(validDocument, "nb")),
+      metaDescription = Seq(Description(validDisclaimer, "nb"))
+    )
+    contentValidator.validateArticle(article).isFailure should be(true)
+  }
+
+  test("validateArticle should throw an error if metaDescription contains plain text") {
     val article = articleToValidate.copy(
       content = Seq(ArticleContent(validDocument, "nb")),
       metaDescription = Seq(Description(validDocument, "nb"))