recognize and parse Schema.org Markup

test/com/dom_distiller/client/SchemaOrgParserTest.java

Index: test/com/dom_distiller/client/SchemaOrgParserTest.java
diff --git a/test/com/dom_distiller/client/SchemaOrgParserTest.java b/test/com/dom_distiller/client/SchemaOrgParserTest.java
new file mode 100644
index 0000000000000000000000000000000000000000..169d90b253e6e2c13c0c464eca67e787d6d087f8
--- /dev/null
+++ b/test/com/dom_distiller/client/SchemaOrgParserTest.java
@@ -0,0 +1,574 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+package com.dom_distiller.client;
+
+import com.google.gwt.dom.client.Document;
+import com.google.gwt.dom.client.AnchorElement;
+import com.google.gwt.dom.client.Element;
+import com.google.gwt.dom.client.NodeList;
+
+import com.google.gwt.junit.client.GWTTestCase;
+
+public class SchemaOrgParserTest extends GWTTestCase {
+    @Override
+    public String getModuleName() {
+        return "com.dom_distiller.DomDistillerJUnit";
+    }
+
+    public void testImageWithEmbeddedPublisher() {
+        String expectedTitle = "Testcase for IMAGE";
+        String expectedDescription = "Testing IMAGE with embedded publisher";
+        String expectedUrl = "http://test_image_with_embedded_item.html";
+        String expectedPublisher = "Whatever Image Incorporated";
+        String expectedCopyrightYear = "1999-2022";
+        String expectedCopyrightHolder = "Whoever Image Copyrighted";
+        String expectedFormat = "jpeg";
+        String expectedCaption = "A test for IMAGE with embedded publisher";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/ImageObject\">" +
+                "<h1 itemprop=\"headline\">" + expectedTitle +
+                "</h1>" +
+                "<h2 itemprop=\"description\">" + expectedDescription +
+                "</h2>" +
+                "<a itemprop=\"contentUrl\" href=\"" + expectedUrl + "\">test results" +
+                "</a>" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Organization\"" +
+                    " itemprop=\"publisher\">Publisher: " +
+                    "<span itemprop=\"name\">" + expectedPublisher +
+                    "</span>" +
+                "</div>" +
+                "<div id=\"3\">" +
+                    "<span itemprop=\"copyrightYear\">" + expectedCopyrightYear +
+                    "</span>" +
+                    "<span itemprop=\"copyrightHolder\">" + expectedCopyrightHolder +
+                    "</span>" +
+                "</div>" +
+                "<span itemprop=\"encodingFormat\">" + expectedFormat +
+                "</span>" +
+                "<span itemprop=\"caption\">" + expectedCaption +
+                "</span>" +
+                "<meta itemprop=\"representativeOfPage\" content=\"true\">" +
+                "<meta itemprop=\"width\" content=\"600\">" +
+                "<meta itemprop=\"height\" content=\"400\">" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("IMAGE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertEquals(expectedDescription, parser.getDescription());
+        assertEquals("", parser.getUrl());
+        assertEquals(expectedPublisher, parser.getPublisher());
+        assertEquals(null, parser.getArticle());
+        assertEquals("", parser.getAuthor());
+        assertEquals("Copyright " + expectedCopyrightYear + " " + expectedCopyrightHolder,
+                parser.getCopyright());
+        MarkupParser.Image[] images = parser.getImages();
+        assertEquals(1, images.length);
+        MarkupParser.Image image = images[0];
+        assertEquals(expectedUrl, image.image);
+        assertEquals(expectedUrl, image.url);
+        assertEquals(null, image.secureUrl);
+        assertEquals(expectedFormat, image.type);
+        assertEquals(expectedCaption, image.caption);
+        assertEquals(600, image.width);
+        assertEquals(400, image.height);
+    }
+
+    public void test2Images() {
+        String expectedTitle1 = "Testcase for 1st IMAGE";
+        String expectedDescription1 = "Testing 1st IMAGE";
+        String expectedUrl1 = "http://test_1st image.html";
+        String expectedPublisher1 = "Whatever 1st Image Incorporated";
+        String expectedCopyrightYear1 = "1000-1999";
+        String expectedCopyrightHolder1 = "Whoever 1st Image Copyrighted";
+        String expectedFormat1 = "jpeg";
+        String expectedCaption1 = "A test for 1st IMAGE";
+        String expectedTitle2 = "Testcase for 2nd IMAGE";
+        String expectedDescription2 = "Testing 2nd IMAGE";
+        String expectedUrl2 = "http://test_2nd image.html";
+        String expectedPublisher2 = "Whatever 2nd Image Incorporated";
+        String expectedCopyrightYear2 = "2000-2999";
+        String expectedCopyrightHolder2 = "Whoever 2nd Image Copyrighted";
+        String expectedFormat2 = "gif";
+        String expectedCaption2 = "A test for 2nd IMAGE";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/ImageObject\">" +
+                "<h1 itemprop=\"headline\">" + expectedTitle1 +
+                "</h1>" +
+                "<h2 itemprop=\"description\">" + expectedDescription1 +
+                "</h2>" +
+                "<a itemprop=\"contentUrl\" href=\"" + expectedUrl1 + "\">1st test results" +
+                "</a>" +
+                "<div id=\"2\" itemprop=\"publisher\">" + expectedPublisher1 +
+                "</div>" +
+                "<div id=\"3\">" +
+                    "<span itemprop=\"copyrightYear\">" + expectedCopyrightYear1 +
+                    "</span>" +
+                    "<span itemprop=\"copyrightHolder\">" + expectedCopyrightHolder1 +
+                    "</span>" +
+                "</div>" +
+                "<span itemprop=\"encodingFormat\">" + expectedFormat1 +
+                "</span>" +
+                "<span itemprop=\"caption\">" + expectedCaption1 +
+                "</span>" +
+                "<meta itemprop=\"representativeOfPage\" content=\"false\">" +
+                "<meta itemprop=\"width\" content=\"400\">" +
+                "<meta itemprop=\"height\" content=\"300\">" +
+            "</div>" +
+            "<div id=\"4\" itemscope itemtype=\"http://schema.org/ImageObject\">" +
+                "<h3 itemprop=\"headline\">" + expectedTitle2 +
+                "</h3>" +
+                "<h4 itemprop=\"description\">" + expectedDescription2 +
+                "</h4>" +
+                "<a itemprop=\"contentUrl\" href=\"" + expectedUrl2 + "\">2nd test results" +
+                "</a>" +
+                "<div id=\"5\" itemprop=\"publisher\">" + expectedPublisher2 +
+                "</div>" +
+                "<div id=\"6\">" +
+                    "<span itemprop=\"copyrightYear\">" + expectedCopyrightYear2 +
+                    "</span>" +
+                    "<span itemprop=\"copyrightHolder\">" + expectedCopyrightHolder2 +
+                    "</span>" +
+                "</div>" +
+                "<span itemprop=\"encodingFormat\">" + expectedFormat2 +
+                "</span>" +
+                "<span itemprop=\"caption\">" + expectedCaption2 +
+                "</span>" +
+                "<meta itemprop=\"representativeOfPage\" content=\"true\">" +
+                "<meta itemprop=\"width\" content=\"1000\">" +
+                "<meta itemprop=\"height\" content=\"600\">" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        // The basic properties of Thing should be from the first image that was
+        // inserted.
+        assertEquals("IMAGE", parser.getType());
+        assertEquals(expectedTitle1, parser.getTitle());
+        assertEquals(expectedDescription1, parser.getDescription());
+        assertEquals("", parser.getUrl());
+        assertEquals(expectedPublisher1, parser.getPublisher());
+        assertEquals(null, parser.getArticle());
+        assertEquals("", parser.getAuthor());
+        assertEquals("Copyright " + expectedCopyrightYear1 + " " + expectedCopyrightHolder1,
+                     parser.getCopyright());
+
+        MarkupParser.Image[] images = parser.getImages();
+        assertEquals(2, images.length);
+        // The 2nd image that was inserted is representative of page, so the
+        // images should be swapped in |images|.
+        MarkupParser.Image image = images[0];
+        assertEquals(expectedUrl2, image.image);
+        assertEquals(expectedUrl2, image.url);
+        assertEquals(null, image.secureUrl);
+        assertEquals(expectedFormat2, image.type);
+        assertEquals(expectedCaption2, image.caption);
+        assertEquals(1000, image.width);
+        assertEquals(600, image.height);
+        image = images[1];
+        assertEquals(expectedUrl1, image.image);
+        assertEquals(expectedUrl1, image.url);
+        assertEquals(null, image.secureUrl);
+        assertEquals(expectedFormat1, image.type);
+        assertEquals(expectedCaption1, image.caption);
+        assertEquals(400, image.width);
+        assertEquals(300, image.height);
+    }
+
+    public void testArticleWithEmbeddedAuthorAndPublisher() {
+        String expectedTitle = "Testcase for ARTICLE";
+        String expectedDescription = "Testing ARTICLE with embedded author and publisher";
+        String expectedUrl = "http://test_article_with_embedded_items.html";
+        String expectedImage = "http://test_article_with_embedded_items.jpeg";
+        String expectedAuthor = "Whoever authored";
+        String expectedPublisher = "Whatever Article Incorporated";
+        String expectedDatePublished = "April 15, 2014";
+        String expectedTimeModified = "2014-04-16T23:59";
+        String expectedCopyrightYear = "2000-2014";
+        String expectedCopyrightHolder = "Whoever Article Copyrighted";
+        String expectedSection = "Romance thriller";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<h1 itemprop=\"headline\">" + expectedTitle +
+                "</h1>" +
+                "<h2 itemprop=\"description\">" + expectedDescription +
+                "</h2>" +
+                "<a itemprop=\"url\" href=\"" + expectedUrl + "\">test results" +
+                "</a>" +
+                "<img itemprop=\"image\" src=\"" + expectedImage + "\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\"" +
+                    " itemprop=\"author\">Author: " +
+                    "<span itemprop=\"name\">" + expectedAuthor +
+                    "</span>" +
+                "</div>" +
+                "<div id=\"3\" itemscope itemtype=\"http://schema.org/Organization\"" +
+                    " itemprop=\"publisher\">Publisher: " +
+                    "<span itemprop=\"name\">" + expectedPublisher +
+                    "</span>" +
+                "</div>" +
+                "<span itemprop=\"datePublished\">" + expectedDatePublished +
+                "</span>" +
+                "<time itemprop=\"dateModified\" datetime=\"" + expectedTimeModified +
+                    "\">April 16, 2014 11:59pm" +
+                "</time>" +
+                "<span itemprop=\"copyrightYear\">" + expectedCopyrightYear +
+                "</span>" +
+                "<span itemprop=\"copyrightHolder\">" + expectedCopyrightHolder +
+                "</span>" +
+                "<span itemprop=\"articleSection\">" + expectedSection +
+                "</span>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertEquals(expectedDescription, parser.getDescription());
+        assertEquals(expectedUrl, parser.getUrl());
+        assertEquals(expectedAuthor, parser.getAuthor());
+        assertEquals(expectedPublisher, parser.getPublisher());
+        assertEquals( "Copyright " + expectedCopyrightYear + " " + expectedCopyrightHolder,
+                parser.getCopyright());
+        MarkupParser.Image[] images = parser.getImages();
+        assertEquals(1, images.length);
+        assertEquals(expectedImage, images[0].image);
+        assertEquals(expectedImage, images[0].url);
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals(expectedDatePublished, article.publishedTime);
+        assertEquals(expectedTimeModified, article.modifiedTime);
+        assertEquals(null, article.expirationTime);
+        assertEquals(expectedSection, article.section);
+        assertEquals(1, article.authors.length);
+        assertEquals(expectedAuthor, article.authors[0]);
+    }
+
+    // This verifies that properties are retrieved from the first schema.org type that defines the
+    // properties requested, i.e. if a property doesn't exist in the first type but does in the
+    // second, it should be available upon request.
+    public void testPropertiesFromFirstAvailableType() {
+        String expectedTitle = "Testcase for properties from first available type";
+        String expectedImage = "http://test_properties_from_1st_available_type.jpeg";
+        String expectedAuthor = "Whoever authored";
+        String expectedPublisher = "Whatever Article Incorporated";
+        String expectedDescription = "The organization that published this article";
+        String expectedUrl = "http://article_publisher.html";
+        String expectedDatePublished = "April 15, 2014";
+        String expectedTimeModified = "2014-04-16T23:59";
+        String expectedCopyrightYear = "2000-2014";
+        String expectedCopyrightHolder = "Whoever Article Copyrighted";
+        String expectedSection = "Romance thriller";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<h1 itemprop=\"headline\">" + expectedTitle +
+                "</h1>" +
+                "<img itemprop=\"image\" src=\"" + expectedImage + "\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\"" +
+                    " itemprop=\"author\">Author: " +
+                    "<span itemprop=\"name\">" + expectedAuthor +
+                    "</span>" +
+                "</div>" +
+                "<div id=\"3\" itemscope itemtype=\"http://schema.org/Organization\"" +
+                    " itemprop=\"publisher\">Publisher: " +
+                    "<span itemprop=\"name\">" + expectedPublisher +
+                    "</span>" +
+                    "<span itemprop=\"description\">" + expectedDescription +
+                    "</span>" +
+                    "<a itemprop=\"url\" href=\"" + expectedUrl + "\">More on Publisher" +
+                    "</a>" +
+                "</div>" +
+                "<span itemprop=\"datePublished\">" + expectedDatePublished +
+                "</span>" +
+                "<time itemprop=\"dateModified\" datetime=\"" + expectedTimeModified +
+                    "\">April 16, 2014 11:59pm" +
+                "</time>" +
+                "<span itemprop=\"copyrightYear\">" + expectedCopyrightYear +
+                "</span>" +
+                "<span itemprop=\"copyrightHolder\">" + expectedCopyrightHolder +
+                "</span>" +
+                "<span itemprop=\"articleSection\">" + expectedSection +
+                "</span>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertEquals(expectedDescription, parser.getDescription());
+        assertEquals(expectedUrl, parser.getUrl());
+        assertEquals(expectedPublisher, parser.getPublisher());
+        assertEquals(expectedAuthor, parser.getAuthor());
+        assertEquals("Copyright " + expectedCopyrightYear + " " + expectedCopyrightHolder,
+                     parser.getCopyright());
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals(expectedDatePublished, article.publishedTime);
+        assertEquals(expectedTimeModified, article.modifiedTime);
+        assertEquals(null, article.expirationTime);
+        assertEquals(expectedSection, article.section);
+        assertEquals(1, article.authors.length);
+        assertEquals(expectedAuthor, article.authors[0]);
+    }
+
+    public void testItemscopeInHTMLTag() {
+        setItemScopeAndType(mRoot, "Article");
+
+        String expectedTitle = "Testcase for ItemScope in HTML tag";
+        Element h = TestUtil.createHeading(1, expectedTitle);
+        setItemProp(h, "headline");
+        mBody.appendChild(h);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertTrue(parser.getArticle() != null);
+
+        // Remove "itemscope" and "itemtype" attributes in <html> tag, so that
+        // other testcases won't be affected.
+        mRoot.removeAttribute("ITEMSCOPE");
+        mRoot.removeAttribute("ITEMTYPE");
+    }
+
+    public void testSupportedWithUnsupportedItemprop() {
+        String expectedTitle = "Testcase for Supported With Unsupported Itemprop";
+        String expectedSection = "Testing";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<span itemprop=\"headline\">" + expectedTitle +
+                "</span>" +
+                "<span itemprop=\"articleSection\">" + expectedSection +
+                "</span>" +
+                // Add unsupported AggregateRating to supported Article as itemprop.
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/AggregateRating\"" +
+                    " itemprop=\"aggregateRating\">Ratings: " +
+                    "<span itemprop=\"ratingValue\">9.9" +
+                    "</span>" +
+                    // Add supported Person to unsupported AggregateRating as itemprop.
+                    "<div id=\"3\" itemscope itemtype=\"http://schema.org/Person\"" +
+                        " itemprop=\"author\">Author: " +
+                        "<span itemprop=\"name\">Whoever authored" +
+                        "</span>" +
+                    "</div>" +
+                "</div>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertEquals("", parser.getDescription());
+        assertEquals("", parser.getUrl());
+        assertEquals("", parser.getAuthor());
+        assertEquals("", parser.getPublisher());
+        assertEquals("", parser.getCopyright());
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals("", article.publishedTime);
+        assertEquals("", article.modifiedTime);
+        assertEquals(null, article.expirationTime);
+        assertEquals(expectedSection, article.section);
+        assertEquals(0, article.authors.length);
+    }
+
+    public void testUnsupportedWithSupportedItemprop() {
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Movie\">" +
+                "<span itemprop=\"headline\">Testcase for Unsupported With Supported Itemprop" +
+                "</span>" +
+                // Add supported Person to unsupported Movie as itemprop.
+                "<div id=\"3\" itemscope itemtype=\"http://schema.org/Person\"" +
+                    " itemprop=\"publisher\">Publisher: " +
+                    "<span itemprop=\"name\">Whoever published" +
+                    "</span>" +
+                "</div>" +
+            "</div>";
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("", parser.getType());
+        assertEquals("", parser.getTitle());
+        assertEquals("", parser.getDescription());
+        assertEquals("", parser.getUrl());
+        assertEquals("", parser.getAuthor());
+        assertEquals("", parser.getPublisher());
+        assertEquals("", parser.getCopyright());
+        assertEquals(null, parser.getArticle());
+        assertEquals(null, parser.getImages());
+    }
+
+    public void testUnsupportedWithNestedSupported() {
+        String expectedTitle = "Testcase for ARTICLE nested in Unsupported Type";
+        String expectedDescription = "Testing ARTICLE that is nested within unsupported type";
+        String expectedUrl = "http://test_article_with_embedded_items.html";
+        String expectedImage = "http://test_article_with_embedded_items.jpeg";
+        String expectedAuthor = "Whoever authored";
+        String expectedPublisher = "Whatever Article Incorporated";
+        String expectedDatePublished = "April 15, 2014";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Movie\">" +
+                "<span itemprop=\"headline\">Testcase for Unsupported With Supported Itemprop" +
+                "</span>" +
+                // Add supported Article to unsupported Movie as a non-itemprop.
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Article\">" +
+                    "<span itemprop=\"headline\">" + expectedTitle +
+                    "</span>" +
+                    "<span itemprop=\"description\">" + expectedDescription +
+                    "</span>" +
+                    "<a itemprop=\"url\" href=\"" + expectedUrl + "\">test results" +
+                    "</a>" +
+                    "<img itemprop=\"image\" src=\"" + expectedImage + "\">" +
+                    "<div id=\"3\" itemscope itemtype=\"http://schema.org/Person\"" +
+                        " itemprop=\"author\">Author: " +
+                        "<span itemprop=\"name\">" + expectedAuthor +
+                        "</span>" +
+                    "</div>" +
+                    "<div id=\"4\" itemscope itemtype=\"http://schema.org/Organization\"" +
+                        " itemprop=\"publisher\">Publisher: " +
+                        "<span itemprop=\"name\">" + expectedPublisher +
+                        "</span>" +
+                    "</div>" +
+                    "<span itemprop=\"datePublished\">" + expectedDatePublished +
+                    "</span>" +
+                "</div>" +
+            "</div>";
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedTitle, parser.getTitle());
+        assertEquals(expectedDescription, parser.getDescription());
+        assertEquals(expectedUrl, parser.getUrl());
+        assertEquals(expectedAuthor, parser.getAuthor());
+        assertEquals(expectedPublisher, parser.getPublisher());
+        assertEquals("", parser.getCopyright());
+        MarkupParser.Image[] images = parser.getImages();
+        assertEquals(1, images.length);
+        assertEquals(expectedImage, images[0].image);
+        assertEquals(expectedImage, images[0].url);
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals(expectedDatePublished, article.publishedTime);
+        assertEquals(null, article.expirationTime);
+        assertEquals(1, article.authors.length);
+        assertEquals(expectedAuthor, article.authors[0]);
+    }
+
+    public void testSameItempropDifferentValues() {
+        String expectedAuthor = "Author 1";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\"" +
+                    " itemprop=\"author\">Authors: " +
+                    "<span itemprop=\"name\">" + expectedAuthor +
+                    "</span>" +
+                    "<span itemprop=\"name\">Author 2" +
+                    "</span>" +
+                "</div>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedAuthor, parser.getAuthor());
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals(1, article.authors.length);
+        assertEquals(expectedAuthor, article.authors[0]);
+   }
+
+    public void testItempropWithMultiProperties() {
+        String expectedPerson = "Person foo";
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\"" +
+                    " itemprop=\"author publisher\">" +
+                    "<span itemprop=\"name\">" + expectedPerson +
+                    "</span>" +
+                "</div>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("ARTICLE", parser.getType());
+        assertEquals(expectedPerson, parser.getAuthor());
+        assertEquals(expectedPerson, parser.getPublisher());
+        MarkupParser.Article article = parser.getArticle();
+        assertEquals(1, article.authors.length);
+        assertEquals(expectedPerson, article.authors[0]);
+   }
+
+    public void testAuthorFromRelAttribute() {
+        String expectedAuthor = "Chromium Authors";
+        AnchorElement link = TestUtil.createAnchor("http://rel_author.html", expectedAuthor);
+        link.setRel("author");
+        mBody.appendChild(link);
+
+        SchemaOrgParser parser = new SchemaOrgParser(mRoot);
+        assertEquals("", parser.getType());
+        assertEquals(expectedAuthor, parser.getAuthor());
+
+        // Remove anchor from parent, so that other testcases won't be affected.
+        link.removeFromParent();
+   }
+
+    @Override
+    protected void gwtSetUp() throws Exception {
+        // Get root element.
+        mRoot = Document.get().getDocumentElement();
+
+        // Get <body> element.
+        NodeList<Element> bodies = mRoot.getElementsByTagName("BODY");
+        if (bodies.getLength() != 1)
+            throw new Exception("There shouldn't be more than 1 <body> tag");
+        mBody = bodies.getItem(0);
+
+        // Remove all meta tags, otherwise a testcase may run with the meta tags
+        // set up in a previous testcase, resulting in unexpected results.
+        NodeList<Element> allMeta = mRoot.getElementsByTagName("META");
+        for (int i = allMeta.getLength() - 1; i >= 0; i--) {
+            allMeta.getItem(i).removeFromParent();
+        }
+
+        // Remove all div tags, otherwise a testcase may run with the div tags
+        // set up in a previous testcase, resulting in unexpected results.
+        NodeList<Element> allDiv = mRoot.getElementsByTagName("DIV");
+        for (int i = allDiv.getLength() - 1; i >= 0; i--) {
+            allDiv.getItem(i).removeFromParent();
+        }
+    }
+
+    private void setItemScopeAndType(Element e, String type) {
+        e.setAttribute("ITEMSCOPE", "");
+        e.setAttribute("ITEMTYPE", "http://schema.org/" + type);
+    }
+
+    private void setItemProp(Element e, String name) {
+        e.setAttribute("itemprop", name);
+    }
+
+    private Element mRoot;
+    private Element mBody;
+}