Parse JSON in Blink for CopylessPaste.

third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp

Index: third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp
diff --git a/third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp b/third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp
index e65edf0ee4b28723b4ce837e5a924173b82b2694..56bdbff6a1630a9516b67da74b284be259d94de3 100644
--- a/third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp
+++ b/third_party/WebKit/Source/modules/document_metadata/CopylessPasteExtractorTest.cpp
@@ -2,12 +2,17 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
-#include "modules/document_metadata/CopylessPasteExtractor.h"
 
 #include <memory>
+#include <string>
+#include <utility>
 #include "core/dom/Document.h"
 #include "core/dom/Element.h"
 #include "core/testing/DummyPageHolder.h"
+#include "modules/document_metadata/CopylessPasteExtractor.h"
+#include "platform/json/JSONValues.h"
+#include "platform/testing/URLTestHelpers.h"
+#include "public/platform/modules/document_metadata/copyless_paste.mojom-blink.h"
 #include "testing/gtest/include/gtest/gtest.h"
 #include "wtf/text/StringBuilder.h"
 
@@ -15,16 +20,18 @@ namespace blink {
 
 namespace {
 
+using mojom::blink::Entity;
+using mojom::blink::EntityPtr;
+using mojom::blink::Property;
+using mojom::blink::PropertyPtr;
+using mojom::blink::Values;
+using mojom::blink::ValuesPtr;
+using mojom::blink::WebPage;
+using mojom::blink::WebPagePtr;
+
 class CopylessPasteExtractorTest : public ::testing::Test {

[wychen, 2017/04/04 02:07:15]

Might make sense to add tests that trigger addProperty = false, like empty
array, mixed typed array, and nested array.

[dproctor, 2017/04/04 06:21:19]

Done.

  public:
-  CopylessPasteExtractorTest()
-      : m_content(
-            "\n"
-            "\n"
-            "{\"@type\": \"NewsArticle\","
-            "\"headline\": \"Special characters for ya >_<;\"\n"
-            "}\n"
-            "\n") {}
+  CopylessPasteExtractorTest() {}
 
  protected:
   void SetUp() override;
@@ -33,11 +40,25 @@ class CopylessPasteExtractorTest : public ::testing::Test {
 
   Document& document() const { return m_dummyPageHolder->document(); }
 
-  String extract() { return CopylessPasteExtractor::extract(document()); }
+  bool extract(WebPagePtr* page) {
+    return CopylessPasteExtractor::extract(document(), page);
+  }
 
   void setHtmlInnerHTML(const String&);
 
-  String m_content;
+  void setURL(const std::string);
+
+  void setTitle(const String&);
+
+  PropertyPtr createStringProperty(const String&, const String&);
+
+  PropertyPtr createBooleanProperty(const String&, const bool&);
+
+  PropertyPtr createLongProperty(const String&, const int64_t&);
+
+  PropertyPtr createEntityProperty(const String&, EntityPtr);
+
+  WebPagePtr createWebPage(const String&, const String&);
 
  private:
   std::unique_ptr<DummyPageHolder> m_dummyPageHolder;
@@ -51,59 +72,574 @@ void CopylessPasteExtractorTest::setHtmlInnerHTML(const String& htmlContent) {
   document().documentElement()->setInnerHTML((htmlContent));
 }
 
+void CopylessPasteExtractorTest::setURL(const std::string url) {
+  document().setURL(URLTestHelpers::toKURL(url));
+}
+
+void CopylessPasteExtractorTest::setTitle(const String& title) {
+  document().setTitle(title);
+}
+
+PropertyPtr CopylessPasteExtractorTest::createStringProperty(
+    const String& name,
+    const String& value) {
+  PropertyPtr p = Property::New();
+  p->name = name;
+  p->values = Values::New();
+  p->values->set_string_values(Vector<String>(1, value));
+  return p;
+}
+
+PropertyPtr CopylessPasteExtractorTest::createBooleanProperty(
+    const String& name,
+    const bool& value) {
+  PropertyPtr p = Property::New();
+  p->name = name;
+  p->values = Values::New();
+  p->values->set_bool_values(Vector<bool>(1, value));
+  return p;
+}
+
+PropertyPtr CopylessPasteExtractorTest::createLongProperty(
+    const String& name,
+    const int64_t& value) {
+  PropertyPtr p = Property::New();
+  p->name = name;
+  p->values = Values::New();
+  p->values->set_long_values(Vector<int64_t>(1, value));
+  return p;
+}
+
+PropertyPtr CopylessPasteExtractorTest::createEntityProperty(const String& name,
+                                                             EntityPtr value) {
+  PropertyPtr p = Property::New();
+  p->name = name;
+  p->values = Values::New();
+  p->values->set_entity_values(Vector<EntityPtr>());
+  p->values->get_entity_values().push_back(std::move(value));
+  return p;
+}
+
+WebPagePtr CopylessPasteExtractorTest::createWebPage(const String& url,
+                                                     const String& title) {
+  WebPagePtr wp = WebPage::New();
+  wp->url = url;
+  wp->title = title;
+  return wp;
+}
+
 TEST_F(CopylessPasteExtractorTest, empty) {
-  String extracted = extract();
-  String expected = "[]";
-  EXPECT_EQ(expected, extracted);
+  WebPagePtr page = WebPage::New();
+  ASSERT_FALSE(extract(&page));
+  EXPECT_EQ(WebPage::New(), page);
 }
 
 TEST_F(CopylessPasteExtractorTest, basic) {
   setHtmlInnerHTML(
       "<body>"
-      "<script type=\"application/ld+json\">" +
-      m_content +
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
       "</script>"
       "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
 
-  String extracted = extract();
-  String expected = "[" + m_content + "]";
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(
+      createStringProperty("name", "Special characters for ya >_<;"));
+
+  expected->entities.push_back(std::move(restaurant));
   EXPECT_EQ(expected, extracted);
 }
 
 TEST_F(CopylessPasteExtractorTest, header) {
   setHtmlInnerHTML(
       "<head>"
-      "<script type=\"application/ld+json\">" +
-      m_content +
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
       "</script>"
       "</head>");
 
-  String extracted = extract();
-  String expected = "[" + m_content + "]";
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(
+      createStringProperty("name", "Special characters for ya >_<;"));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, booleanValue) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"open\": true"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(createBooleanProperty("open", true));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, longValue) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"long\": 1"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(createLongProperty("long", 1ll));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, doubleValue) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"double\": 1.5"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(createStringProperty("double", "1.5"));
+
+  expected->entities.push_back(std::move(restaurant));
   EXPECT_EQ(expected, extracted);
 }
 
 TEST_F(CopylessPasteExtractorTest, multiple) {
   setHtmlInnerHTML(
       "<head>"
-      "<script type=\"application/ld+json\">" +
-      m_content +
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
       "</script>"
       "</head>"
       "<body>"
-      "<script type=\"application/ld+json\">" +
-      m_content +
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
       "</script>"
-      "<script type=\"application/ld+json\">" +
-      m_content +
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
       "</script>"
       "</body>");
 
-  String extracted = extract();
-  String expected = "[" + m_content + "," + m_content + "," + m_content + "]";
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  for (int i = 0; i < 3; ++i) {
+    EntityPtr restaurant = Entity::New();
+    restaurant->type = "Restaurant";
+    restaurant->properties.push_back(
+        createStringProperty("name", "Special characters for ya >_<;"));
+
+    expected->entities.push_back(std::move(restaurant));
+  }
   EXPECT_EQ(expected, extracted);
 }
 
-}  // namespace
+TEST_F(CopylessPasteExtractorTest, nested) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Ye ol greasy diner\","
+      "\"address\": {"
+      "\n"
+      "  \"streetAddress\": \"123 Big Oak Road\","
+      "  \"addressLocality\": \"San Francisco\""
+      "  }\n"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(
+      createStringProperty("name", "Ye ol greasy diner"));
+
+  EntityPtr address = Entity::New();
+  address->type = "Thing";
+  address->properties.push_back(
+      createStringProperty("streetAddress", "123 Big Oak Road"));
+  address->properties.push_back(
+      createStringProperty("addressLocality", "San Francisco"));
+
+  restaurant->properties.push_back(
+      createEntityProperty("address", std::move(address)));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
 
+TEST_F(CopylessPasteExtractorTest, repeated) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": [ \"First name\", \"Second name\" ]"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+
+  PropertyPtr name = Property::New();
+  name->name = "name";
+  name->values = Values::New();
+  Vector<String> nameValues;
+  nameValues.push_back("First name");
+  nameValues.push_back("Second name");
+  name->values->set_string_values(nameValues);
+
+  restaurant->properties.push_back(std::move(name));
+
+  expected->entities.push_back(std::move(restaurant));
+
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, repeatedObject) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"Ye ol greasy diner\","
+      "\"address\": ["
+      "\n"
+      "  {"
+      "  \"streetAddress\": \"123 Big Oak Road\","
+      "  \"addressLocality\": \"San Francisco\""
+      "  },\n"
+      "  {"
+      "  \"streetAddress\": \"123 Big Oak Road\","
+      "  \"addressLocality\": \"San Francisco\""
+      "  }\n"
+      "]\n"
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(
+      createStringProperty("name", "Ye ol greasy diner"));
+
+  PropertyPtr addressProperty = Property::New();
+  addressProperty->name = "address";
+  addressProperty->values = Values::New();
+  addressProperty->values->set_entity_values(Vector<EntityPtr>());
+  for (int i = 0; i < 2; ++i) {
+    EntityPtr address = Entity::New();
+    address->type = "Thing";
+    address->properties.push_back(
+        createStringProperty("streetAddress", "123 Big Oak Road"));
+    address->properties.push_back(
+        createStringProperty("addressLocality", "San Francisco"));
+    addressProperty->values->get_entity_values().push_back(std::move(address));
+  }
+  restaurant->properties.push_back(std::move(addressProperty));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, truncateLongString) {
+  String maxLengthString;
+  for (int i = 0; i < 200; ++i) {
+    maxLengthString.append("a");
+  }
+  String tooLongString(maxLengthString);
+  tooLongString.append("a");
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": \"" +
+      tooLongString +
+      "\""
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+  restaurant->properties.push_back(
+      createStringProperty("name", maxLengthString));
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, enforceTypeExists) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_FALSE(extract(&extracted));
+  WebPagePtr expected = WebPage::New();
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, enforceTypeWhitelist) {
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"UnsupportedType\","
+      "\"name\": \"Special characters for ya >_<;\""
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_FALSE(extract(&extracted));
+  WebPagePtr expected = WebPage::New();
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, truncateTooManyValuesInField) {
+  String largeRepeatedField = "[";
+  for (int i = 0; i < 101; ++i) {
+    largeRepeatedField.append("\"a\"");
+    if (i != 100) {
+      largeRepeatedField.append(", ");
+    }
+  }
+  largeRepeatedField.append("]");
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\","
+      "\"name\": " +
+      largeRepeatedField +
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+
+  PropertyPtr name = Property::New();
+  name->name = "name";
+  name->values = Values::New();
+  Vector<String> nameValues;
+  for (int i = 0; i < 100; ++i) {
+    nameValues.push_back("a");
+  }
+  name->values->set_string_values(nameValues);
+
+  restaurant->properties.push_back(std::move(name));
+
+  expected->entities.push_back(std::move(restaurant));
+
+  EXPECT_EQ(expected, extracted);
+}
+
+TEST_F(CopylessPasteExtractorTest, truncateTooManyFields) {
+  String tooManyFields;
+  for (int i = 0; i < 20; ++i) {
+    tooManyFields.append(String::format("\"%d\": \"a\"", i));
+    if (i != 19) {
+      tooManyFields.append(",\n");
+    }
+  }
+  setHtmlInnerHTML(
+      "<body>"
+      "<script type=\"application/ld+json\">"
+      "\n"
+      "\n"
+      "{\"@type\": \"Restaurant\"," +
+      tooManyFields +
+      "}\n"
+      "\n"
+      "</script>"
+      "</body>");
+  setURL("http://www.test.com/");
+  setTitle("My neat website about cool stuff");
+
+  WebPagePtr extracted = WebPage::New();
+  ASSERT_TRUE(extract(&extracted));
+
+  WebPagePtr expected =
+      createWebPage("http://www.test.com/", "My neat website about cool stuff");
+
+  EntityPtr restaurant = Entity::New();
+  restaurant->type = "Restaurant";
+
+  for (int i = 0; i < 19; ++i) {
+    restaurant->properties.push_back(
+        createStringProperty(String::number(i), "a"));
+  }
+
+  expected->entities.push_back(std::move(restaurant));
+  EXPECT_EQ(expected, extracted);
+}
+
+}  // namespace
 }  // namespace blink