src/components/dom_distiller/content/distiller_page_web_contents_browsertest.cc

   1 // Copyright 2013 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/memory/weak_ptr.h"
   6 #include "base/path_service.h"
   7 #include "base/run_loop.h"
   8 #include "base/values.h"
   9 #include "components/dom_distiller/content/distiller_page_web_contents.h"
  10 #include "components/dom_distiller/content/web_contents_main_frame_observer.h"
  11 #include "components/dom_distiller/core/distiller_page.h"
  12 #include "content/public/browser/browser_context.h"
  13 #include "content/public/browser/navigation_controller.h"
  14 #include "content/public/browser/render_frame_host.h"
  15 #include "content/public/browser/web_contents_observer.h"
  16 #include "content/public/test/content_browser_test.h"
  17 #include "content/shell/browser/shell.h"
  18 #include "grit/component_resources.h"
  19 #include "net/test/embedded_test_server/embedded_test_server.h"
  20 #include "testing/gmock/include/gmock/gmock.h"
  21 #include "ui/base/resource/resource_bundle.h"
  22
  23 using content::ContentBrowserTest;
  24 using testing::ContainsRegex;
  25 using testing::HasSubstr;
  26 using testing::Not;
  27
  28 namespace dom_distiller {
  29
  30 const char* kSimpleArticlePath = "/simple_article.html";
  31 const char* kVideoArticlePath = "/video_article.html";
  32
  33 class DistillerPageWebContentsTest : public ContentBrowserTest {
  34  public:
  35   // ContentBrowserTest:
  36   virtual void SetUpOnMainThread() OVERRIDE {
  37     AddComponentsResources();
  38     SetUpTestServer();
  39     ContentBrowserTest::SetUpOnMainThread();
  40   }
  41
  42   void DistillPage(const base::Closure& quit_closure, const std::string& url) {
  43     quit_closure_ = quit_closure;
  44     distiller_page_->DistillPage(
  45         embedded_test_server()->GetURL(url),
  46         dom_distiller::proto::DomDistillerOptions(),
  47         base::Bind(&DistillerPageWebContentsTest::OnPageDistillationFinished,
  48                    this));
  49   }
  50
  51   void OnPageDistillationFinished(scoped_ptr<DistilledPageInfo> distilled_page,
  52                                   bool distillation_successful) {
  53     page_info_ = distilled_page.Pass();
  54     quit_closure_.Run();
  55   }
  56
  57  private:
  58   void AddComponentsResources() {
  59     base::FilePath pak_file;
  60     base::FilePath pak_dir;
  61     PathService::Get(base::DIR_MODULE, &pak_dir);
  62     pak_file = pak_dir.Append(FILE_PATH_LITERAL("components_resources.pak"));
  63     ui::ResourceBundle::GetSharedInstance().AddDataPackFromPath(
  64         pak_file, ui::SCALE_FACTOR_NONE);
  65   }
  66
  67   void SetUpTestServer() {
  68     base::FilePath path;
  69     PathService::Get(base::DIR_SOURCE_ROOT, &path);
  70     path = path.AppendASCII("components/test/data/dom_distiller");
  71     embedded_test_server()->ServeFilesFromDirectory(path);
  72     ASSERT_TRUE(embedded_test_server()->InitializeAndWaitUntilReady());
  73   }
  74
  75  protected:
  76   void RunUseCurrentWebContentsTest(const std::string& url,
  77                                     bool expect_new_web_contents,
  78                                     bool setup_main_frame_observer,
  79                                     bool wait_for_document_loaded);
  80
  81   DistillerPageWebContents* distiller_page_;
  82   base::Closure quit_closure_;
  83   scoped_ptr<DistilledPageInfo> page_info_;
  84 };
  85
  86 // Use this class to be able to leak the WebContents, which is needed for when
  87 // the current WebContents is used for distillation.
  88 class TestDistillerPageWebContents : public DistillerPageWebContents {
  89  public:
  90   TestDistillerPageWebContents(
  91       content::BrowserContext* browser_context,
  92       const gfx::Size& render_view_size,
  93       scoped_ptr<SourcePageHandleWebContents> optional_web_contents_handle,
  94       bool expect_new_web_contents)
  95       : DistillerPageWebContents(browser_context, render_view_size,
  96                                  optional_web_contents_handle.Pass()),
  97         expect_new_web_contents_(expect_new_web_contents),
  98         new_web_contents_created_(false) {}
  99
 100   virtual void CreateNewWebContents(const GURL& url) OVERRIDE {
 101     ASSERT_EQ(true, expect_new_web_contents_);
 102     new_web_contents_created_ = true;
 103     // DistillerPageWebContents::CreateNewWebContents resets the scoped_ptr to
 104     // the WebContents, so intentionally leak WebContents here, since it is
 105     // owned by the shell.
 106     content::WebContents* web_contents = web_contents_.release();
 107     web_contents->GetLastCommittedURL();
 108     DistillerPageWebContents::CreateNewWebContents(url);
 109   }
 110
 111   virtual ~TestDistillerPageWebContents() {
 112     if (!expect_new_web_contents_) {
 113       // Intentionally leaking WebContents, since it is owned by the shell.
 114       content::WebContents* web_contents = web_contents_.release();
 115       web_contents->GetLastCommittedURL();
 116     }
 117   }
 118
 119   bool new_web_contents_created() { return new_web_contents_created_; }
 120
 121  private:
 122   bool expect_new_web_contents_;
 123   bool new_web_contents_created_;
 124 };
 125
 126 // Helper class to know how far in the loading process the current WebContents
 127 // has come. It will call the callback either after
 128 // DidCommitProvisionalLoadForFrame or DocumentLoadedInFrame is called for the
 129 // main frame, based on the value of |wait_for_document_loaded|.
 130 class WebContentsMainFrameHelper : public content::WebContentsObserver {
 131  public:
 132   WebContentsMainFrameHelper(content::WebContents* web_contents,
 133                              const base::Closure& callback,
 134                              bool wait_for_document_loaded)
 135       : WebContentsObserver(web_contents),
 136         callback_(callback),
 137         wait_for_document_loaded_(wait_for_document_loaded) {}
 138
 139   virtual void DidCommitProvisionalLoadForFrame(
 140       content::RenderFrameHost* render_frame_host,
 141       const GURL& url,
 142       content::PageTransition transition_type) OVERRIDE {
 143     if (wait_for_document_loaded_)
 144       return;
 145     if (!render_frame_host->GetParent())
 146       callback_.Run();
 147   }
 148
 149   virtual void DocumentLoadedInFrame(
 150       content::RenderFrameHost* render_frame_host) OVERRIDE {
 151     if (wait_for_document_loaded_) {
 152       if (!render_frame_host->GetParent())
 153         callback_.Run();
 154     }
 155   }
 156
 157  private:
 158   base::Closure callback_;
 159   bool wait_for_document_loaded_;
 160 };
 161
 162 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, BasicDistillationWorks) {
 163   DistillerPageWebContents distiller_page(
 164       shell()->web_contents()->GetBrowserContext(),
 165       shell()->web_contents()->GetContainerBounds().size(),
 166       scoped_ptr<SourcePageHandleWebContents>());
 167   distiller_page_ = &distiller_page;
 168
 169   base::RunLoop run_loop;
 170   DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
 171   run_loop.Run();
 172
 173   EXPECT_EQ("Test Page Title", page_info_.get()->title);
 174   EXPECT_THAT(page_info_.get()->html, HasSubstr("Lorem ipsum"));
 175   EXPECT_THAT(page_info_.get()->html, Not(HasSubstr("questionable content")));
 176   EXPECT_EQ("", page_info_.get()->next_page_url);
 177   EXPECT_EQ("", page_info_.get()->prev_page_url);
 178 }
 179
 180 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeLinks) {
 181   DistillerPageWebContents distiller_page(
 182       shell()->web_contents()->GetBrowserContext(),
 183       shell()->web_contents()->GetContainerBounds().size(),
 184       scoped_ptr<SourcePageHandleWebContents>());
 185   distiller_page_ = &distiller_page;
 186
 187   base::RunLoop run_loop;
 188   DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
 189   run_loop.Run();
 190
 191   // A relative link should've been updated.
 192   EXPECT_THAT(page_info_.get()->html,
 193               ContainsRegex("href=\"http://127.0.0.1:.*/relativelink.html\""));
 194   EXPECT_THAT(page_info_.get()->html,
 195               HasSubstr("href=\"http://www.google.com/absolutelink.html\""));
 196 }
 197
 198 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeImages) {
 199   DistillerPageWebContents distiller_page(
 200       shell()->web_contents()->GetBrowserContext(),
 201       shell()->web_contents()->GetContainerBounds().size(),
 202       scoped_ptr<SourcePageHandleWebContents>());
 203   distiller_page_ = &distiller_page;
 204
 205   base::RunLoop run_loop;
 206   DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
 207   run_loop.Run();
 208
 209   // A relative link should've been updated.
 210   EXPECT_THAT(page_info_.get()->html,
 211               ContainsRegex("src=\"http://127.0.0.1:.*/relativeimage.png\""));
 212   EXPECT_THAT(page_info_.get()->html,
 213               HasSubstr("src=\"http://www.google.com/absoluteimage.png\""));
 214 }
 215
 216
 217 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, HandlesRelativeVideos) {
 218   DistillerPageWebContents distiller_page(
 219       shell()->web_contents()->GetBrowserContext(),
 220       shell()->web_contents()->GetContainerBounds().size(),
 221       scoped_ptr<SourcePageHandleWebContents>());
 222   distiller_page_ = &distiller_page;
 223
 224   base::RunLoop run_loop;
 225   DistillPage(run_loop.QuitClosure(), kVideoArticlePath);
 226   run_loop.Run();
 227
 228   // A relative source/track should've been updated.
 229   EXPECT_THAT(
 230       page_info_.get()->html,
 231       ContainsRegex("src=\"http://127.0.0.1:.*/relative_video.mp4\""));
 232   EXPECT_THAT(
 233       page_info_.get()->html,
 234       ContainsRegex("src=\"http://127.0.0.1:.*/relative_track_en.vtt\""));
 235   EXPECT_THAT(
 236       page_info_.get()->html,
 237       HasSubstr("src=\"http://www.google.com/absolute_video.ogg\""));
 238   EXPECT_THAT(
 239       page_info_.get()->html,
 240       HasSubstr("src=\"http://www.google.com/absolute_track_fr.vtt\""));
 241 }
 242
 243 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, VisibilityDetection) {
 244   DistillerPageWebContents distiller_page(
 245       shell()->web_contents()->GetBrowserContext(),
 246       shell()->web_contents()->GetContainerBounds().size(),
 247       scoped_ptr<SourcePageHandleWebContents>());
 248   distiller_page_ = &distiller_page;
 249
 250   // visble_style.html and invisible_style.html only differ by the visibility
 251   // internal stylesheet.
 252
 253   {
 254     base::RunLoop run_loop;
 255     DistillPage(run_loop.QuitClosure(), "/visible_style.html");
 256     run_loop.Run();
 257     EXPECT_THAT(page_info_.get()->html, HasSubstr("Lorem ipsum"));
 258   }
 259
 260   {
 261     base::RunLoop run_loop;
 262     DistillPage(run_loop.QuitClosure(), "/invisible_style.html");
 263     run_loop.Run();
 264     EXPECT_THAT(page_info_.get()->html, Not(HasSubstr("Lorem ipsum")));
 265   }
 266 }
 267
 268 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
 269                        UsingCurrentWebContentsWrongUrl) {
 270   std::string url("/bogus");
 271   bool expect_new_web_contents = true;
 272   bool setup_main_frame_observer = true;
 273   bool wait_for_document_loaded = true;
 274   RunUseCurrentWebContentsTest(url,
 275                                expect_new_web_contents,
 276                                setup_main_frame_observer,
 277                                wait_for_document_loaded);
 278 }
 279
 280 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
 281                        UsingCurrentWebContentsNoMainFrameObserver) {
 282   std::string url(kSimpleArticlePath);
 283   bool expect_new_web_contents = true;
 284   bool setup_main_frame_observer = false;
 285   bool wait_for_document_loaded = true;
 286   RunUseCurrentWebContentsTest(url,
 287                                expect_new_web_contents,
 288                                setup_main_frame_observer,
 289                                wait_for_document_loaded);
 290 }
 291
 292 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
 293                        UsingCurrentWebContentsNotFinishedLoadingYet) {
 294   std::string url(kSimpleArticlePath);
 295   bool expect_new_web_contents = false;
 296   bool setup_main_frame_observer = true;
 297   bool wait_for_document_loaded = false;
 298   RunUseCurrentWebContentsTest(url,
 299                                expect_new_web_contents,
 300                                setup_main_frame_observer,
 301                                wait_for_document_loaded);
 302 }
 303
 304 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest,
 305                        UsingCurrentWebContentsReadyForDistillation) {
 306   std::string url(kSimpleArticlePath);
 307   bool expect_new_web_contents = false;
 308   bool setup_main_frame_observer = true;
 309   bool wait_for_document_loaded = true;
 310   RunUseCurrentWebContentsTest(url,
 311                                expect_new_web_contents,
 312                                setup_main_frame_observer,
 313                                wait_for_document_loaded);
 314 }
 315
 316 void DistillerPageWebContentsTest::RunUseCurrentWebContentsTest(
 317     const std::string& url,
 318     bool expect_new_web_contents,
 319     bool setup_main_frame_observer,
 320     bool wait_for_document_loaded) {
 321   content::WebContents* current_web_contents = shell()->web_contents();
 322   if (setup_main_frame_observer) {
 323     dom_distiller::WebContentsMainFrameObserver::CreateForWebContents(
 324         current_web_contents);
 325   }
 326   base::RunLoop url_loaded_runner;
 327   WebContentsMainFrameHelper main_frame_loaded(current_web_contents,
 328                                                url_loaded_runner.QuitClosure(),
 329                                                wait_for_document_loaded);
 330   current_web_contents->GetController().LoadURL(
 331       embedded_test_server()->GetURL(url),
 332       content::Referrer(),
 333       content::PAGE_TRANSITION_TYPED,
 334       std::string());
 335   url_loaded_runner.Run();
 336
 337   scoped_ptr<content::WebContents> old_web_contents_sptr(current_web_contents);
 338   scoped_ptr<SourcePageHandleWebContents> source_page_handle(
 339       new SourcePageHandleWebContents(old_web_contents_sptr.Pass()));
 340
 341   TestDistillerPageWebContents distiller_page(
 342       shell()->web_contents()->GetBrowserContext(),
 343       shell()->web_contents()->GetContainerBounds().size(),
 344       source_page_handle.Pass(),
 345       expect_new_web_contents);
 346   distiller_page_ = &distiller_page;
 347
 348   base::RunLoop run_loop;
 349   DistillPage(run_loop.QuitClosure(), kSimpleArticlePath);
 350   run_loop.Run();
 351
 352   // Sanity check of distillation process.
 353   EXPECT_EQ(expect_new_web_contents, distiller_page.new_web_contents_created());
 354   EXPECT_EQ("Test Page Title", page_info_.get()->title);
 355 }
 356
 357 IN_PROC_BROWSER_TEST_F(DistillerPageWebContentsTest, MarkupInfo) {
 358   DistillerPageWebContents distiller_page(
 359       shell()->web_contents()->GetBrowserContext(),
 360       shell()->web_contents()->GetContainerBounds().size(),
 361       scoped_ptr<SourcePageHandleWebContents>());
 362   distiller_page_ = &distiller_page;
 363
 364   base::RunLoop run_loop;
 365   DistillPage(run_loop.QuitClosure(), "/markup_article.html");
 366   run_loop.Run();
 367
 368   EXPECT_THAT(page_info_.get()->html, HasSubstr("Lorem ipsum"));
 369   EXPECT_EQ("Marked-up Markup Test Page Title", page_info_.get()->title);
 370
 371   const DistilledPageInfo::MarkupInfo& markup_info = page_info_->markup_info;
 372   EXPECT_EQ("Marked-up Markup Test Page Title", markup_info.title);
 373   EXPECT_EQ("Article", markup_info.type);
 374   EXPECT_EQ("http://test/markup.html", markup_info.url);
 375   EXPECT_EQ("This page tests Markup Info.", markup_info.description);
 376   EXPECT_EQ("Whoever Published", markup_info.publisher);
 377   EXPECT_EQ("Copyright 2000-2014 Whoever Copyrighted", markup_info.copyright);
 378   EXPECT_EQ("Whoever Authored", markup_info.author);
 379
 380   const DistilledPageInfo::MarkupArticle& markup_article = markup_info.article;
 381   EXPECT_EQ("Whatever Section", markup_article.section);
 382   EXPECT_EQ("July 23, 2014", markup_article.published_time);
 383   EXPECT_EQ("2014-07-23T23:59", markup_article.modified_time);
 384   EXPECT_EQ("", markup_article.expiration_time);
 385   ASSERT_EQ(1U, markup_article.authors.size());
 386   EXPECT_EQ("Whoever Authored", markup_article.authors[0]);
 387
 388   ASSERT_EQ(2U, markup_info.images.size());
 389
 390   const DistilledPageInfo::MarkupImage& markup_image1 = markup_info.images[0];
 391   EXPECT_EQ("http://test/markup1.jpeg", markup_image1.url);
 392   EXPECT_EQ("https://test/markup1.jpeg", markup_image1.secure_url);
 393   EXPECT_EQ("jpeg", markup_image1.type);
 394   EXPECT_EQ("", markup_image1.caption);
 395   EXPECT_EQ(600, markup_image1.width);
 396   EXPECT_EQ(400, markup_image1.height);
 397
 398   const DistilledPageInfo::MarkupImage& markup_image2 = markup_info.images[1];
 399   EXPECT_EQ("http://test/markup2.gif", markup_image2.url);
 400   EXPECT_EQ("https://test/markup2.gif", markup_image2.secure_url);
 401   EXPECT_EQ("gif", markup_image2.type);
 402   EXPECT_EQ("", markup_image2.caption);
 403   EXPECT_EQ(1000, markup_image2.width);
 404   EXPECT_EQ(600, markup_image2.height);
 405 }
 406
 407 }  // namespace dom_distiller