Add support for Schema.org/Recipe

javatests/org/chromium/distiller/SchemaOrgParserAccessorTest.java

 // 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 org.chromium.distiller;
 
 import com.google.gwt.dom.client.AnchorElement;
 import com.google.gwt.dom.client.Element;
 
 public class SchemaOrgParserAccessorTest extends DomDistillerJsTestCase {
@@ skipping 433 matching lines @@
         assertEquals(1, images.length);
         assertEquals(expectedImage, images[0].url);
         MarkupParser.Article article = parser.getArticle();
         assertEquals(expectedDatePublished, article.publishedTime);
         assertEquals("", article.expirationTime);
         assertEquals(1, article.authors.length);
         assertEquals(expectedAuthor, article.authors[0]);
     }
 
     public void testSameItempropDifferentValues() {
-        String expectedAuthor = "Author 1";
+        String expectedAuthor = "Author 1, Author 2";
         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 itemprop=\"name\">Author 1" +
                     "</span>" +
                     "<span itemprop=\"name\">Author 2" +
                     "</span>" +
                 "</div>" +
             "</div>";
 
         Element rootDiv = TestUtil.createDiv(0);
         rootDiv.setInnerHTML(htmlStr);
         mBody.appendChild(rootDiv);
 
@@ skipping 64 matching lines @@
 
         parser = new SchemaOrgParserAccessor(mRoot);
         assertEquals("", parser.getType());
         assertEquals(expectedAuthor, parser.getAuthor());
         assertEquals(null, parser.getArticle());
 
         // Remove anchor from parent, so that other testcases won't be affected.
         link.removeFromParent();
     }
 
+    public void testSingleRecipe() {
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Recipe\">" +

[wychen, 2016/03/14 22:58:42]

Lots of duplicated text here. Maybe saving the common things in strings and
concatenate when needed?

+                "<h1 itemprop=\"name\">Bat Rice Recipe</h1>" +
+                "<h2 itemprop=\"description\">" +
+                    "Testing RECIPE with required recipe schema.org attributes" +
+                    "</h2>" +
+                "<a itemprop=\"url\" " +
+                    "href=\"http://dummy/test_article_with_embedded_items.html\">" +
+                    "test results" + "</a>" +
+                "<img itemprop=\"image\" " +
+                    "src=\"http://dummy/test_article_with_embedded_items.jpeg\" />" +
+                "<div itemprop=\"author\">Whoever authored</div>" +
+                "<div itemprop=\"recipeYield\" content=\"4\">" +
+                    "4 servings</div>" +
+                "<time itemprop=\"prepTime\" datetime=\"PT15M\">" +
+                    "15 Minutes" + "</time>" +
+                "<time itemprop=\"cookTime\" datetime=\"PT45M\">" +
+                    "45 Minutes" + "</time>" +
+                "<time itemprop=\"totalTime\" datetime=\"PT9H\">" +
+                    "9 Hours</time>" +
+                "<span itemprop=\"ingredients\">4 bats</span>" +
+                "<span itemprop=\"ingredients\">1/4 milk</span>" +
+                "<span itemprop=\"ingredients\">1 rice</span>" +
+                "<span itemprop=\"ingredients\">1 salt</span>" +
+                "<span itemprop=\"recipeInstructions\">" +
+                    "Mix everything and wait for cook</span>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+        SchemaOrgParserAccessor parser = new SchemaOrgParserAccessor(mRoot);
+        String result = parser.getStructuredData();
+
+        String expected =
+            "<img src='http://dummy/test_article_with_embedded_items.jpeg' />" +
+            "<p>Author: Whoever authored</p><p>Description: " +
+                "Testing RECIPE with required recipe schema.org attributes</p>" +
+            "<p>Serves: 4 servings</p>" +
+            "<p>Prep time: 15 minute(s)</p>" +
+            "<p>Cook time: 45 minute(s)</p>" +
+            "<p>Total time: 9 hour(s)</p>" +
+            "Ingredients: <ul><li>4 bats</li>" +
+            "<li>1/4 milk</li>" +
+            "<li>1 rice</li>" +
+            "<li>1 salt</li>" +
+            "</ul><p>Instructions: <br />" +
+                "Mix everything and wait for cook</p>";
+        assertEquals(expected, result);
+    }
+
+    public void testSingleRecipeWithEmbeddedPerson() {
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Recipe\">" +
+                "<h1 itemprop=\"name\">Bat Rice Recipe</h1>" +
+                "<h2 itemprop=\"description\">" +
+                    "Testing RECIPE with required recipe schema.org attributes</h2>" +
+                "<a itemprop=\"url\" href=\"" +
+                    "http://dummy/test_article_with_embedded_items.html\">" +
+                    "test results" + "</a>" +
+                "<img itemprop=\"image\" " +
+                    "src=\"http://dummy/test_article_with_embedded_items.jpeg\" />" +
+                "<span itemprop=\"author\" " +
+                    "itemscope itemtype=\"http://schema.org/Person\">" +
+                    "<span itemprop=\"givenName\">Guest</span>" +
+                    "<span itemprop=\"familyName\">Smith</span>" +
+                "</span>" +
+                "<div itemprop=\"recipeYield\" content=\"4\">4 servings</div>" +
+                "<time itemprop=\"prepTime\" datetime=\"PT15M\">1" +
+                    "5 Minutes" + "</time>" +
+                "<time itemprop=\"cookTime\" datetime=\"PT45M\">" +
+                    "45 Minutes" + "</time>" +
+                "<time itemprop=\"totalTime\" datetime=\"PT9H\">" +
+                    "9 Hours</time>" +
+                "<span itemprop=\"ingredients\">4 bats</span>" +
+                "<span itemprop=\"ingredients\">1/4 milk</span>" +
+                "<span itemprop=\"ingredients\">1 rice</span>" +
+                "<span itemprop=\"ingredients\">1 salt</span>" +
+                "<span itemprop=\"recipeInstructions\">" +
+                    "Mix everything and wait for cook</span>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+        SchemaOrgParserAccessor parser = new SchemaOrgParserAccessor(mRoot);
+        String result = parser.getStructuredData();
+
+        String expected =
+            "<img src='http://dummy/test_article_with_embedded_items.jpeg' />" +
+            "<p>Author: <span>Guest Smith</span></p>" +
+            "<p>Description: " +
+                "Testing RECIPE with required recipe schema.org attributes</p>" +
+            "<p>Serves: 4 servings</p>" +
+            "<p>Prep time: 15 minute(s)</p>" +
+            "<p>Cook time: 45 minute(s)</p>" +
+            "<p>Total time: 9 hour(s)</p>" +
+            "Ingredients: <ul><li>4 bats</li>" +
+            "<li>1/4 milk</li>" +
+            "<li>1 rice</li>" +
+            "<li>1 salt</li>" +
+            "</ul><p>Instructions: <br />" +
+                    "Mix everything and wait for cook</p>";
+        assertEquals(expected, result);
+    }
+
+    public void testMoreThanOneSchemaOrg() {
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Recipe\">" +
+                "<h1 itemprop=\"name\">Bat Rice Recipe</h1>" +
+                "<h2 itemprop=\"description\">" +
+                    "Testing RECIPE with required recipe schema.org attributes" +
+                    "</h2>" +
+                "<a itemprop=\"url\" href=\"" +
+                    "http://dummy/test_article_with_embedded_items.html\">" +
+                    "test results" + "</a>" +
+                "<img itemprop=\"image\" src=\"" +
+                    "http://dummy/test_article_with_embedded_items.jpeg\" />" +
+                "<div itemprop=\"author\">Whoever authored</div>" +
+                "<div itemprop=\"recipeYield\" content=\"4\">" +
+                    "4 servings</div>" +
+                "<time itemprop=\"prepTime\" datetime=\"PT15M\">" +
+                    "15 Minutes" + "</time>" +
+                "<time itemprop=\"cookTime\" datetime=\"PT45M\">" +
+                    "45 Minutes" + "</time>" +
+                "<time itemprop=\"totalTime\" datetime=\"PT9H\">" +
+                    "9 Hours</time>" +
+                "<span itemprop=\"ingredients\">4 bats</span>" +
+                "<span itemprop=\"ingredients\">1/4 milk</span>" +
+                "<span itemprop=\"ingredients\">1 rice</span>" +
+                "<span itemprop=\"ingredients\">1 salt</span>" +
+                "<span itemprop=\"recipeInstructions\">" +
+                    "Mix everything and wait for cook</span>" +
+            "</div>" +
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\" " +
+                    "itemprop=\"author\">Creator: " +
+                    "<span itemprop=\"name\"> User </span>" +
+                "</div>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+        SchemaOrgParserAccessor parser = new SchemaOrgParserAccessor(mRoot);
+        String result = parser.getStructuredData();
+
+        assertEquals("", result);
+    }
+
+    public void testMultipleSchemaOrgWithMainEntity() {
+        String htmlStr =
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Recipe\">" +
+                "<meta itemprop=\"mainEntityOfPage\" content=\"true\">" +
+                "<h1 itemprop=\"name\">Bat Rice Recipe</h1>" +
+                "<h2 itemprop=\"description\">" +
+                    "Testing RECIPE with required recipe schema.org attributes" +
+                    "</h2>" +
+                "<a itemprop=\"url\" href=\"" +
+                    "http://dummy/test_article_with_embedded_items.html\">" +
+                    "test results" + "</a>" +
+                "<img itemprop=\"image\" src=\"" +
+                    "http://dummy/test_article_with_embedded_items.jpeg\" />" +
+                "<div itemprop=\"author\">Whoever authored</div>" +
+                "<div itemprop=\"recipeYield\" content=\"4\">" +
+                    "4 servings</div>" +
+                "<time itemprop=\"prepTime\" datetime=\"PT15M\">" +
+                    "15 Minutes" + "</time>" +
+                "<time itemprop=\"cookTime\" datetime=\"PT45M\">" +
+                    "45 Minutes" + "</time>" +
+                "<time itemprop=\"totalTime\" datetime=\"PT9H\">" +
+                    "9 Hours</time>" +
+                "<span itemprop=\"ingredients\">4 bats</span>" +
+                "<span itemprop=\"ingredients\">1/4 milk</span>" +
+                "<span itemprop=\"ingredients\">1 rice</span>" +
+                "<span itemprop=\"ingredients\">1 salt</span>" +
+                "<span itemprop=\"recipeInstructions\">" +
+                    "Mix everything and wait for cook</span>" +
+            "</div>" +
+            "<div id=\"1\" itemscope itemtype=\"http://schema.org/Article\">" +
+                "<div id=\"2\" itemscope itemtype=\"http://schema.org/Person\" " +
+                    "itemprop=\"author\">Creator: " +
+                    "<span itemprop=\"name\"> User </span>" +
+                "</div>" +
+            "</div>";
+
+        Element rootDiv = TestUtil.createDiv(0);
+        rootDiv.setInnerHTML(htmlStr);
+        mBody.appendChild(rootDiv);
+        SchemaOrgParserAccessor parser = new SchemaOrgParserAccessor(mRoot);
+        String result = parser.getStructuredData();
+
+        String expected =
+            "<img src='http://dummy/test_article_with_embedded_items.jpeg' />" +
+                "<p>Author: Whoever authored</p><p>Description: " +
+                    "Testing RECIPE with required recipe schema.org attributes" +
+                    "</p>" +
+                "<p>Serves: 4 servings</p>" +
+                "<p>Prep time: 15 minute(s)</p>" +
+                "<p>Cook time: 45 minute(s)</p>" +
+                "<p>Total time: 9 hour(s)</p>" +
+                "Ingredients: <ul><li>4 bats</li>" +
+                "<li>1/4 milk</li>" +
+                "<li>1 rice</li>" +
+                "<li>1 salt</li>" +
+                "</ul><p>Instructions: <br />" +
+                    "Mix everything and wait for cook</p>";
+        assertEquals(expected, result);
+    }
+
     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);
     }
 }