content/renderer/dom_serializer_browsertest.cc

   1 // Copyright (c) 2010 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/bind.h"
   6 #include "base/command_line.h"
   7 #include "base/compiler_specific.h"
   8 #include "base/containers/hash_tables.h"
   9 #include "base/files/file_path.h"
  10 #include "base/files/file_util.h"
  11 #include "base/strings/string_util.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "content/public/common/content_switches.h"
  14 #include "content/public/renderer/render_view.h"
  15 #include "content/public/renderer/render_view_observer.h"
  16 #include "content/public/test/content_browser_test.h"
  17 #include "content/public/test/content_browser_test_utils.h"
  18 #include "content/public/test/routing_id_mangling_disabler.h"
  19 #include "content/public/test/test_utils.h"
  20 #include "content/renderer/savable_resources.h"
  21 #include "content/shell/browser/shell.h"
  22 #include "net/base/filename_util.h"
  23 #include "net/url_request/url_request_context.h"
  24 #include "third_party/WebKit/public/platform/WebCString.h"
  25 #include "third_party/WebKit/public/platform/WebData.h"
  26 #include "third_party/WebKit/public/platform/WebString.h"
  27 #include "third_party/WebKit/public/platform/WebURL.h"
  28 #include "third_party/WebKit/public/platform/WebVector.h"
  29 #include "third_party/WebKit/public/web/WebDocument.h"
  30 #include "third_party/WebKit/public/web/WebElement.h"
  31 #include "third_party/WebKit/public/web/WebElementCollection.h"
  32 #include "third_party/WebKit/public/web/WebLocalFrame.h"
  33 #include "third_party/WebKit/public/web/WebNode.h"
  34 #include "third_party/WebKit/public/web/WebNodeList.h"
  35 #include "third_party/WebKit/public/web/WebPageSerializer.h"
  36 #include "third_party/WebKit/public/web/WebPageSerializerClient.h"
  37 #include "third_party/WebKit/public/web/WebView.h"
  38
  39 using blink::WebCString;
  40 using blink::WebData;
  41 using blink::WebDocument;
  42 using blink::WebElement;
  43 using blink::WebElementCollection;
  44 using blink::WebFrame;
  45 using blink::WebLocalFrame;
  46 using blink::WebNode;
  47 using blink::WebNodeList;
  48 using blink::WebPageSerializer;
  49 using blink::WebPageSerializerClient;
  50 using blink::WebString;
  51 using blink::WebURL;
  52 using blink::WebView;
  53 using blink::WebVector;
  54
  55 namespace {
  56
  57 // The first RenderFrame is routing ID 1, and the first RenderView is 2.
  58 const int kRenderViewRoutingId = 2;
  59
  60 }
  61
  62 namespace content {
  63
  64 // Iterate recursively over sub-frames to find one with with a given url.
  65 WebFrame* FindSubFrameByURL(WebView* web_view, const GURL& url) {
  66   if (!web_view->mainFrame())
  67     return NULL;
  68
  69   std::vector<WebFrame*> stack;
  70   stack.push_back(web_view->mainFrame());
  71
  72   while (!stack.empty()) {
  73     WebFrame* current_frame = stack.back();
  74     stack.pop_back();
  75     if (GURL(current_frame->document().url()) == url)
  76       return current_frame;
  77     WebElementCollection all = current_frame->document().all();
  78     for (WebElement element = all.firstItem();
  79          !element.isNull(); element = all.nextItem()) {
  80       // Check frame tag and iframe tag
  81       if (!element.hasHTMLTagName("frame") && !element.hasHTMLTagName("iframe"))
  82         continue;
  83       WebFrame* sub_frame = WebLocalFrame::fromFrameOwnerElement(element);
  84       if (sub_frame)
  85         stack.push_back(sub_frame);
  86     }
  87   }
  88   return NULL;
  89 }
  90
  91 // Helper function that test whether the first node in the doc is a doc type
  92 // node.
  93 bool HasDocType(const WebDocument& doc) {
  94   WebNode node = doc.firstChild();
  95   if (node.isNull())
  96     return false;
  97   return node.nodeType() == WebNode::DocumentTypeNode;
  98 }
  99
 100   // Helper function for checking whether input node is META tag. Return true
 101 // means it is META element, otherwise return false. The parameter charset_info
 102 // return actual charset info if the META tag has charset declaration.
 103 bool IsMetaElement(const WebNode& node, std::string& charset_info) {
 104   if (!node.isElementNode())
 105     return false;
 106   const WebElement meta = node.toConst<WebElement>();
 107   if (!meta.hasHTMLTagName("meta"))
 108     return false;
 109   charset_info.erase(0, charset_info.length());
 110   // Check the META charset declaration.
 111   WebString httpEquiv = meta.getAttribute("http-equiv");
 112   if (LowerCaseEqualsASCII(httpEquiv, "content-type")) {
 113     std::string content = meta.getAttribute("content").utf8();
 114     int pos = content.find("charset", 0);
 115     if (pos > -1) {
 116       // Add a dummy charset declaration to charset_info, which indicates this
 117       // META tag has charset declaration although we do not get correct value
 118       // yet.
 119       charset_info.append("has-charset-declaration");
 120       int remaining_length = content.length() - pos - 7;
 121       if (!remaining_length)
 122         return true;
 123       int start_pos = pos + 7;
 124       // Find "=" symbol.
 125       while (remaining_length--)
 126         if (content[start_pos++] == L'=')
 127           break;
 128       // Skip beginning space.
 129       while (remaining_length) {
 130         if (content[start_pos] > 0x0020)
 131           break;
 132         ++start_pos;
 133         --remaining_length;
 134       }
 135       if (!remaining_length)
 136         return true;
 137       int end_pos = start_pos;
 138       // Now we find out the start point of charset info. Search the end point.
 139       while (remaining_length--) {
 140         if (content[end_pos] <= 0x0020 || content[end_pos] == L';')
 141           break;
 142         ++end_pos;
 143       }
 144       // Get actual charset info.
 145       charset_info = content.substr(start_pos, end_pos - start_pos);
 146       return true;
 147     }
 148   }
 149   return true;
 150 }
 151
 152 class LoadObserver : public RenderViewObserver {
 153  public:
 154   LoadObserver(RenderView* render_view, const base::Closure& quit_closure)
 155       : RenderViewObserver(render_view),
 156         quit_closure_(quit_closure) {}
 157
 158   void DidFinishLoad(blink::WebLocalFrame* frame) override {
 159     if (frame == render_view()->GetWebView()->mainFrame())
 160       quit_closure_.Run();
 161   }
 162
 163  private:
 164   base::Closure quit_closure_;
 165 };
 166
 167 class DomSerializerTests : public ContentBrowserTest,
 168                            public WebPageSerializerClient {
 169  public:
 170   DomSerializerTests()
 171     : serialized_(false),
 172       local_directory_name_(FILE_PATH_LITERAL("./dummy_files/")) {}
 173
 174   void SetUpCommandLine(base::CommandLine* command_line) override {
 175     command_line->AppendSwitch(switches::kSingleProcess);
 176 #if defined(OS_WIN)
 177     // Don't want to try to create a GPU process.
 178     command_line->AppendSwitch(switches::kDisableGpu);
 179 #endif
 180   }
 181
 182   // DomSerializerDelegate.
 183   virtual void didSerializeDataForFrame(const WebURL& frame_web_url,
 184                                         const WebCString& data,
 185                                         PageSerializationStatus status) {
 186
 187     GURL frame_url(frame_web_url);
 188     // If the all frames are finished saving, check all finish status
 189     if (status == WebPageSerializerClient::AllFramesAreFinished) {
 190       SerializationFinishStatusMap::iterator it =
 191           serialization_finish_status_.begin();
 192       for (; it != serialization_finish_status_.end(); ++it)
 193         ASSERT_TRUE(it->second);
 194       serialized_ = true;
 195       return;
 196     }
 197
 198     // Check finish status of current frame.
 199     SerializationFinishStatusMap::iterator it =
 200         serialization_finish_status_.find(frame_url.spec());
 201     // New frame, set initial status as false.
 202     if (it == serialization_finish_status_.end())
 203       serialization_finish_status_[frame_url.spec()] = false;
 204
 205     it = serialization_finish_status_.find(frame_url.spec());
 206     ASSERT_TRUE(it != serialization_finish_status_.end());
 207     // In process frame, finish status should be false.
 208     ASSERT_FALSE(it->second);
 209
 210     // Add data to corresponding frame's content.
 211     serialized_frame_map_[frame_url.spec()] += data.data();
 212
 213     // Current frame is completed saving, change the finish status.
 214     if (status == WebPageSerializerClient::CurrentFrameIsFinished)
 215       it->second = true;
 216   }
 217
 218   bool HasSerializedFrame(const GURL& frame_url) {
 219     return serialized_frame_map_.find(frame_url.spec()) !=
 220            serialized_frame_map_.end();
 221   }
 222
 223   const std::string& GetSerializedContentForFrame(
 224       const GURL& frame_url) {
 225     return serialized_frame_map_[frame_url.spec()];
 226   }
 227
 228   RenderView* GetRenderView() {
 229     // We could have the test on the UI thread get the WebContent's routing ID,
 230     // but we know this will be the first RV so skip that and just hardcode it.
 231     return RenderView::FromRoutingID(kRenderViewRoutingId);
 232   }
 233
 234   WebView* GetWebView() {
 235     return GetRenderView()->GetWebView();
 236   }
 237
 238   WebFrame* GetMainFrame() {
 239     return GetWebView()->mainFrame();
 240   }
 241
 242   // Load web page according to input content and relative URLs within
 243   // the document.
 244   void LoadContents(const std::string& contents,
 245                     const GURL& base_url,
 246                     const WebString encoding_info) {
 247     scoped_refptr<MessageLoopRunner> runner = new MessageLoopRunner;
 248     LoadObserver observer(GetRenderView(), runner->QuitClosure());
 249
 250     // If input encoding is empty, use UTF-8 as default encoding.
 251     if (encoding_info.isEmpty()) {
 252       GetMainFrame()->loadHTMLString(contents, base_url);
 253     } else {
 254       WebData data(contents.data(), contents.length());
 255
 256       // Do not use WebFrame.LoadHTMLString because it assumes that input
 257       // html contents use UTF-8 encoding.
 258       // TODO(darin): This should use WebFrame::loadData.
 259       WebFrame* web_frame = GetMainFrame();
 260
 261       ASSERT_TRUE(web_frame != NULL);
 262
 263       web_frame->loadData(data, "text/html", encoding_info, base_url);
 264     }
 265
 266     runner->Run();
 267   }
 268
 269   // Serialize page DOM according to specific page URL. The parameter
 270   // recursive_serialization indicates whether we will serialize all
 271   // sub-frames.
 272   void SerializeDomForURL(const GURL& page_url,
 273                           bool recursive_serialization) {
 274     // Find corresponding WebFrame according to page_url.
 275     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), page_url);
 276     ASSERT_TRUE(web_frame != NULL);
 277     WebVector<WebURL> links;
 278     links.assign(&page_url, 1);
 279     WebString file_path =
 280         base::FilePath(FILE_PATH_LITERAL("c:\\dummy.htm")).AsUTF16Unsafe();
 281     WebVector<WebString> local_paths;
 282     local_paths.assign(&file_path, 1);
 283     // Start serializing DOM.
 284     bool result = WebPageSerializer::serialize(web_frame->toWebLocalFrame(),
 285        recursive_serialization,
 286        static_cast<WebPageSerializerClient*>(this),
 287        links,
 288        local_paths,
 289        local_directory_name_.AsUTF16Unsafe());
 290     ASSERT_TRUE(result);
 291     ASSERT_TRUE(serialized_);
 292   }
 293
 294   void SerializeHTMLDOMWithDocTypeOnRenderer(const GURL& file_url) {
 295     // Make sure original contents have document type.
 296     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 297     ASSERT_TRUE(web_frame != NULL);
 298     WebDocument doc = web_frame->document();
 299     ASSERT_TRUE(HasDocType(doc));
 300     // Do serialization.
 301     SerializeDomForURL(file_url, false);
 302     // Load the serialized contents.
 303     ASSERT_TRUE(HasSerializedFrame(file_url));
 304     const std::string& serialized_contents =
 305         GetSerializedContentForFrame(file_url);
 306     LoadContents(serialized_contents, file_url,
 307                  web_frame->document().encoding());
 308     // Make sure serialized contents still have document type.
 309     web_frame = GetMainFrame();
 310     doc = web_frame->document();
 311     ASSERT_TRUE(HasDocType(doc));
 312   }
 313
 314   void SerializeHTMLDOMWithoutDocTypeOnRenderer(const GURL& file_url) {
 315     // Make sure original contents do not have document type.
 316     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 317     ASSERT_TRUE(web_frame != NULL);
 318     WebDocument doc = web_frame->document();
 319     ASSERT_TRUE(!HasDocType(doc));
 320     // Do serialization.
 321     SerializeDomForURL(file_url, false);
 322     // Load the serialized contents.
 323     ASSERT_TRUE(HasSerializedFrame(file_url));
 324     const std::string& serialized_contents =
 325         GetSerializedContentForFrame(file_url);
 326     LoadContents(serialized_contents, file_url,
 327                  web_frame->document().encoding());
 328     // Make sure serialized contents do not have document type.
 329     web_frame = GetMainFrame();
 330     doc = web_frame->document();
 331     ASSERT_TRUE(!HasDocType(doc));
 332   }
 333
 334   void SerializeXMLDocWithBuiltInEntitiesOnRenderer(
 335       const GURL& xml_file_url, const std::string& original_contents) {
 336     // Do serialization.
 337     SerializeDomForURL(xml_file_url, false);
 338     // Compare the serialized contents with original contents.
 339     ASSERT_TRUE(HasSerializedFrame(xml_file_url));
 340     const std::string& serialized_contents =
 341         GetSerializedContentForFrame(xml_file_url);
 342     ASSERT_EQ(original_contents, serialized_contents);
 343   }
 344
 345   void SerializeHTMLDOMWithAddingMOTWOnRenderer(
 346       const GURL& file_url, const std::string& original_contents) {
 347     // Make sure original contents does not have MOTW;
 348     std::string motw_declaration =
 349        WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 350     ASSERT_FALSE(motw_declaration.empty());
 351     // The encoding of original contents is ISO-8859-1, so we convert the MOTW
 352     // declaration to ASCII and search whether original contents has it or not.
 353     ASSERT_TRUE(std::string::npos == original_contents.find(motw_declaration));
 354
 355     // Do serialization.
 356     SerializeDomForURL(file_url, false);
 357     // Make sure the serialized contents have MOTW ;
 358     ASSERT_TRUE(HasSerializedFrame(file_url));
 359     const std::string& serialized_contents =
 360         GetSerializedContentForFrame(file_url);
 361     ASSERT_FALSE(std::string::npos ==
 362         serialized_contents.find(motw_declaration));
 363   }
 364
 365   void SerializeHTMLDOMWithNoMetaCharsetInOriginalDocOnRenderer(
 366       const GURL& file_url) {
 367     // Make sure there is no META charset declaration in original document.
 368     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 369     ASSERT_TRUE(web_frame != NULL);
 370     WebDocument doc = web_frame->document();
 371     ASSERT_TRUE(doc.isHTMLDocument());
 372     WebElement head_element = doc.head();
 373     ASSERT_TRUE(!head_element.isNull());
 374     // Go through all children of HEAD element.
 375     for (WebNode child = head_element.firstChild(); !child.isNull();
 376          child = child.nextSibling()) {
 377       std::string charset_info;
 378       if (IsMetaElement(child, charset_info))
 379         ASSERT_TRUE(charset_info.empty());
 380     }
 381     // Do serialization.
 382     SerializeDomForURL(file_url, false);
 383
 384     // Load the serialized contents.
 385     ASSERT_TRUE(HasSerializedFrame(file_url));
 386     const std::string& serialized_contents =
 387         GetSerializedContentForFrame(file_url);
 388     LoadContents(serialized_contents, file_url,
 389                  web_frame->document().encoding());
 390     // Make sure the first child of HEAD element is META which has charset
 391     // declaration in serialized contents.
 392     web_frame = GetMainFrame();
 393     ASSERT_TRUE(web_frame != NULL);
 394     doc = web_frame->document();
 395     ASSERT_TRUE(doc.isHTMLDocument());
 396     head_element = doc.head();
 397     ASSERT_TRUE(!head_element.isNull());
 398     WebNode meta_node = head_element.firstChild();
 399     ASSERT_TRUE(!meta_node.isNull());
 400     // Get meta charset info.
 401     std::string charset_info2;
 402     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 403     ASSERT_TRUE(!charset_info2.empty());
 404     ASSERT_EQ(charset_info2,
 405               std::string(web_frame->document().encoding().utf8()));
 406
 407     // Make sure no more additional META tags which have charset declaration.
 408     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 409          child = child.nextSibling()) {
 410       std::string charset_info;
 411       if (IsMetaElement(child, charset_info))
 412         ASSERT_TRUE(charset_info.empty());
 413     }
 414   }
 415
 416   void SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDocOnRenderer(
 417       const GURL& file_url) {
 418     // Make sure there are multiple META charset declarations in original
 419     // document.
 420     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 421     ASSERT_TRUE(web_frame != NULL);
 422     WebDocument doc = web_frame->document();
 423     ASSERT_TRUE(doc.isHTMLDocument());
 424     WebElement head_ele = doc.head();
 425     ASSERT_TRUE(!head_ele.isNull());
 426     // Go through all children of HEAD element.
 427     int charset_declaration_count = 0;
 428     for (WebNode child = head_ele.firstChild(); !child.isNull();
 429          child = child.nextSibling()) {
 430       std::string charset_info;
 431       if (IsMetaElement(child, charset_info) && !charset_info.empty())
 432         charset_declaration_count++;
 433     }
 434     // The original doc has more than META tags which have charset declaration.
 435     ASSERT_TRUE(charset_declaration_count > 1);
 436
 437     // Do serialization.
 438     SerializeDomForURL(file_url, false);
 439
 440     // Load the serialized contents.
 441     ASSERT_TRUE(HasSerializedFrame(file_url));
 442     const std::string& serialized_contents =
 443         GetSerializedContentForFrame(file_url);
 444     LoadContents(serialized_contents, file_url,
 445                  web_frame->document().encoding());
 446     // Make sure only first child of HEAD element is META which has charset
 447     // declaration in serialized contents.
 448     web_frame = GetMainFrame();
 449     ASSERT_TRUE(web_frame != NULL);
 450     doc = web_frame->document();
 451     ASSERT_TRUE(doc.isHTMLDocument());
 452     head_ele = doc.head();
 453     ASSERT_TRUE(!head_ele.isNull());
 454     WebNode meta_node = head_ele.firstChild();
 455     ASSERT_TRUE(!meta_node.isNull());
 456     // Get meta charset info.
 457     std::string charset_info2;
 458     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 459     ASSERT_TRUE(!charset_info2.empty());
 460     ASSERT_EQ(charset_info2,
 461               std::string(web_frame->document().encoding().utf8()));
 462
 463     // Make sure no more additional META tags which have charset declaration.
 464     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 465          child = child.nextSibling()) {
 466       std::string charset_info;
 467       if (IsMetaElement(child, charset_info))
 468         ASSERT_TRUE(charset_info.empty());
 469     }
 470   }
 471
 472   void SerializeHTMLDOMWithEntitiesInTextOnRenderer() {
 473     base::FilePath page_file_path = GetTestFilePath(
 474         "dom_serializer", "dom_serializer/htmlentities_in_text.htm");
 475     // Get file URL. The URL is dummy URL to identify the following loading
 476     // actions. The test content is in constant:original_contents.
 477     GURL file_url = net::FilePathToFileURL(page_file_path);
 478     ASSERT_TRUE(file_url.SchemeIsFile());
 479     // Test contents.
 480     static const char* const original_contents =
 481         "<html><body>&amp;&lt;&gt;\"\'</body></html>";
 482     // Load the test contents.
 483     LoadContents(original_contents, file_url, WebString());
 484
 485     // Get BODY's text content in DOM.
 486     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 487     ASSERT_TRUE(web_frame != NULL);
 488     WebDocument doc = web_frame->document();
 489     ASSERT_TRUE(doc.isHTMLDocument());
 490     WebElement body_ele = doc.body();
 491     ASSERT_TRUE(!body_ele.isNull());
 492     WebNode text_node = body_ele.firstChild();
 493     ASSERT_TRUE(text_node.isTextNode());
 494     ASSERT_TRUE(std::string(text_node.createMarkup().utf8()) ==
 495                 "&amp;&lt;&gt;\"\'");
 496     // Do serialization.
 497     SerializeDomForURL(file_url, false);
 498     // Compare the serialized contents with original contents.
 499     ASSERT_TRUE(HasSerializedFrame(file_url));
 500     const std::string& serialized_contents =
 501         GetSerializedContentForFrame(file_url);
 502     // Compare the serialized contents with original contents to make sure
 503     // they are same.
 504     // Because we add MOTW when serializing DOM, so before comparison, we also
 505     // need to add MOTW to original_contents.
 506     std::string original_str =
 507       WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 508     original_str += original_contents;
 509     // Since WebCore now inserts a new HEAD element if there is no HEAD element
 510     // when creating BODY element. (Please see
 511     // HTMLParser::bodyCreateErrorCheck.) We need to append the HEAD content and
 512     // corresponding META content if we find WebCore-generated HEAD element.
 513     if (!doc.head().isNull()) {
 514       WebString encoding = web_frame->document().encoding();
 515       std::string htmlTag("<html>");
 516       std::string::size_type pos = original_str.find(htmlTag);
 517       ASSERT_NE(std::string::npos, pos);
 518       pos += htmlTag.length();
 519       std::string head_part("<head>");
 520       head_part +=
 521           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 522       head_part += "</head>";
 523       original_str.insert(pos, head_part);
 524     }
 525     ASSERT_EQ(original_str, serialized_contents);
 526   }
 527
 528   void SerializeHTMLDOMWithEntitiesInAttributeValueOnRenderer() {
 529     base::FilePath page_file_path = GetTestFilePath(
 530         "dom_serializer", "dom_serializer/htmlentities_in_attribute_value.htm");
 531     // Get file URL. The URL is dummy URL to identify the following loading
 532     // actions. The test content is in constant:original_contents.
 533     GURL file_url = net::FilePathToFileURL(page_file_path);
 534     ASSERT_TRUE(file_url.SchemeIsFile());
 535     // Test contents.
 536     static const char* const original_contents =
 537         "<html><body title=\"&amp;&lt;&gt;&quot;&#39;\"></body></html>";
 538     // Load the test contents.
 539     LoadContents(original_contents, file_url, WebString());
 540     // Get value of BODY's title attribute in DOM.
 541     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 542     ASSERT_TRUE(web_frame != NULL);
 543     WebDocument doc = web_frame->document();
 544     ASSERT_TRUE(doc.isHTMLDocument());
 545     WebElement body_ele = doc.body();
 546     ASSERT_TRUE(!body_ele.isNull());
 547     WebString value = body_ele.getAttribute("title");
 548     ASSERT_TRUE(std::string(value.utf8()) == "&<>\"\'");
 549     // Do serialization.
 550     SerializeDomForURL(file_url, false);
 551     // Compare the serialized contents with original contents.
 552     ASSERT_TRUE(HasSerializedFrame(file_url));
 553     const std::string& serialized_contents =
 554         GetSerializedContentForFrame(file_url);
 555     // Compare the serialized contents with original contents to make sure
 556     // they are same.
 557     std::string original_str =
 558         WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 559     original_str += original_contents;
 560     if (!doc.isNull()) {
 561       WebString encoding = web_frame->document().encoding();
 562       std::string htmlTag("<html>");
 563       std::string::size_type pos = original_str.find(htmlTag);
 564       ASSERT_NE(std::string::npos, pos);
 565       pos += htmlTag.length();
 566       std::string head_part("<head>");
 567       head_part +=
 568           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 569       head_part += "</head>";
 570       original_str.insert(pos, head_part);
 571     }
 572     ASSERT_EQ(original_str, serialized_contents);
 573   }
 574
 575   void SerializeHTMLDOMWithNonStandardEntitiesOnRenderer(const GURL& file_url) {
 576     // Get value of BODY's title attribute in DOM.
 577     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 578     WebDocument doc = web_frame->document();
 579     ASSERT_TRUE(doc.isHTMLDocument());
 580     WebElement body_element = doc.body();
 581     // Unescaped string for "&percnt;&nsup;&sup1;&apos;".
 582     static const wchar_t parsed_value[] = {
 583       '%', 0x2285, 0x00b9, '\'', 0
 584     };
 585     WebString value = body_element.getAttribute("title");
 586     ASSERT_TRUE(base::UTF16ToWide(value) == parsed_value);
 587     ASSERT_TRUE(base::UTF16ToWide(body_element.innerText()) == parsed_value);
 588
 589     // Do serialization.
 590     SerializeDomForURL(file_url, false);
 591     // Check the serialized string.
 592     ASSERT_TRUE(HasSerializedFrame(file_url));
 593     const std::string& serialized_contents =
 594         GetSerializedContentForFrame(file_url);
 595     // Confirm that the serialized string has no non-standard HTML entities.
 596     ASSERT_EQ(std::string::npos, serialized_contents.find("&percnt;"));
 597     ASSERT_EQ(std::string::npos, serialized_contents.find("&nsup;"));
 598     ASSERT_EQ(std::string::npos, serialized_contents.find("&sup1;"));
 599     ASSERT_EQ(std::string::npos, serialized_contents.find("&apos;"));
 600   }
 601
 602   void SerializeHTMLDOMWithBaseTagOnRenderer(const GURL& file_url,
 603                                              const GURL& path_dir_url) {
 604     // There are total 2 available base tags in this test file.
 605     const int kTotalBaseTagCountInTestFile = 2;
 606
 607     // Since for this test, we assume there is no savable sub-resource links for
 608     // this test file, also all links are relative URLs in this test file, so we
 609     // need to check those relative URLs and make sure document has BASE tag.
 610     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 611     ASSERT_TRUE(web_frame != NULL);
 612     WebDocument doc = web_frame->document();
 613     ASSERT_TRUE(doc.isHTMLDocument());
 614     // Go through all descent nodes.
 615     WebElementCollection all = doc.all();
 616     int original_base_tag_count = 0;
 617     for (WebElement element = all.firstItem(); !element.isNull();
 618          element = all.nextItem()) {
 619       if (element.hasHTMLTagName("base")) {
 620         original_base_tag_count++;
 621       } else {
 622         // Get link.
 623         WebString value = GetSubResourceLinkFromElement(element);
 624         if (value.isNull() && element.hasHTMLTagName("a")) {
 625           value = element.getAttribute("href");
 626           if (value.isEmpty())
 627             value = WebString();
 628         }
 629         // Each link is relative link.
 630         if (!value.isNull()) {
 631           GURL link(value.utf8());
 632           ASSERT_TRUE(link.scheme().empty());
 633         }
 634       }
 635     }
 636     ASSERT_EQ(original_base_tag_count, kTotalBaseTagCountInTestFile);
 637     // Make sure in original document, the base URL is not equal with the
 638     // |path_dir_url|.
 639     GURL original_base_url(doc.baseURL());
 640     ASSERT_NE(original_base_url, path_dir_url);
 641
 642     // Do serialization.
 643     SerializeDomForURL(file_url, false);
 644
 645     // Load the serialized contents.
 646     ASSERT_TRUE(HasSerializedFrame(file_url));
 647     const std::string& serialized_contents =
 648         GetSerializedContentForFrame(file_url);
 649     LoadContents(serialized_contents, file_url,
 650                  web_frame->document().encoding());
 651
 652     // Make sure all links are absolute URLs and doc there are some number of
 653     // BASE tags in serialized HTML data. Each of those BASE tags have same base
 654     // URL which is as same as URL of current test file.
 655     web_frame = GetMainFrame();
 656     ASSERT_TRUE(web_frame != NULL);
 657     doc = web_frame->document();
 658     ASSERT_TRUE(doc.isHTMLDocument());
 659     // Go through all descent nodes.
 660     all = doc.all();
 661     int new_base_tag_count = 0;
 662     for (WebNode node = all.firstItem(); !node.isNull();
 663          node = all.nextItem()) {
 664       if (!node.isElementNode())
 665         continue;
 666       WebElement element = node.to<WebElement>();
 667       if (element.hasHTMLTagName("base")) {
 668         new_base_tag_count++;
 669       } else {
 670         // Get link.
 671         WebString value = GetSubResourceLinkFromElement(element);
 672         if (value.isNull() && element.hasHTMLTagName("a")) {
 673           value = element.getAttribute("href");
 674           if (value.isEmpty())
 675             value = WebString();
 676         }
 677         // Each link is absolute link.
 678         if (!value.isNull()) {
 679           GURL link(std::string(value.utf8()));
 680           ASSERT_FALSE(link.scheme().empty());
 681         }
 682       }
 683     }
 684     // We have one more added BASE tag which is generated by JavaScript.
 685     ASSERT_EQ(new_base_tag_count, original_base_tag_count + 1);
 686     // Make sure in new document, the base URL is equal with the |path_dir_url|.
 687     GURL new_base_url(doc.baseURL());
 688     ASSERT_EQ(new_base_url, path_dir_url);
 689   }
 690
 691   void SerializeHTMLDOMWithEmptyHeadOnRenderer() {
 692     base::FilePath page_file_path = GetTestFilePath(
 693         "dom_serializer", "empty_head.htm");
 694     GURL file_url = net::FilePathToFileURL(page_file_path);
 695     ASSERT_TRUE(file_url.SchemeIsFile());
 696
 697     // Load the test html content.
 698     static const char* const empty_head_contents =
 699       "<html><head></head><body>hello world</body></html>";
 700     LoadContents(empty_head_contents, file_url, WebString());
 701
 702     // Make sure the head tag is empty.
 703     WebFrame* web_frame = GetMainFrame();
 704     ASSERT_TRUE(web_frame != NULL);
 705     WebDocument doc = web_frame->document();
 706     ASSERT_TRUE(doc.isHTMLDocument());
 707     WebElement head_element = doc.head();
 708     ASSERT_TRUE(!head_element.isNull());
 709     ASSERT_TRUE(!head_element.hasChildNodes());
 710     ASSERT_TRUE(head_element.childNodes().length() == 0);
 711
 712     // Do serialization.
 713     SerializeDomForURL(file_url, false);
 714     // Make sure the serialized contents have META ;
 715     ASSERT_TRUE(HasSerializedFrame(file_url));
 716     const std::string& serialized_contents =
 717         GetSerializedContentForFrame(file_url);
 718
 719     // Reload serialized contents and make sure there is only one META tag.
 720     LoadContents(serialized_contents, file_url,
 721                  web_frame->document().encoding());
 722     web_frame = GetMainFrame();
 723     ASSERT_TRUE(web_frame != NULL);
 724     doc = web_frame->document();
 725     ASSERT_TRUE(doc.isHTMLDocument());
 726     head_element = doc.head();
 727     ASSERT_TRUE(!head_element.isNull());
 728     ASSERT_TRUE(head_element.hasChildNodes());
 729     ASSERT_TRUE(head_element.childNodes().length() == 1);
 730     WebNode meta_node = head_element.firstChild();
 731     ASSERT_TRUE(!meta_node.isNull());
 732     // Get meta charset info.
 733     std::string charset_info;
 734     ASSERT_TRUE(IsMetaElement(meta_node, charset_info));
 735     ASSERT_TRUE(!charset_info.empty());
 736     ASSERT_EQ(charset_info,
 737               std::string(web_frame->document().encoding().utf8()));
 738
 739     // Check the body's first node is text node and its contents are
 740     // "hello world"
 741     WebElement body_element = doc.body();
 742     ASSERT_TRUE(!body_element.isNull());
 743     WebNode text_node = body_element.firstChild();
 744     ASSERT_TRUE(text_node.isTextNode());
 745     WebString text_node_contents = text_node.nodeValue();
 746     ASSERT_TRUE(std::string(text_node_contents.utf8()) == "hello world");
 747   }
 748
 749   void SerializeDocumentWithDownloadedIFrameOnRenderer(const GURL& file_url) {
 750     // Do a recursive serialization. We pass if we don't crash.
 751     SerializeDomForURL(file_url, true);
 752   }
 753
 754   void SubResourceForElementsInNonHTMLNamespaceOnRenderer(
 755       const GURL& file_url) {
 756     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 757     ASSERT_TRUE(web_frame != NULL);
 758     WebDocument doc = web_frame->document();
 759     WebNode lastNodeInBody = doc.body().lastChild();
 760     ASSERT_EQ(WebNode::ElementNode, lastNodeInBody.nodeType());
 761     WebString uri = GetSubResourceLinkFromElement(
 762         lastNodeInBody.to<WebElement>());
 763     EXPECT_TRUE(uri.isNull());
 764   }
 765
 766  private:
 767   // Map frame_url to corresponding serialized_content.
 768   typedef base::hash_map<std::string, std::string> SerializedFrameContentMap;
 769   SerializedFrameContentMap serialized_frame_map_;
 770   // Map frame_url to corresponding status of serialization finish.
 771   typedef base::hash_map<std::string, bool> SerializationFinishStatusMap;
 772   SerializationFinishStatusMap serialization_finish_status_;
 773   // Flag indicates whether the process of serializing DOM is finished or not.
 774   bool serialized_;
 775   // The local_directory_name_ is dummy relative path of directory which
 776   // contain all saved auxiliary files included all sub frames and resources.
 777   const base::FilePath local_directory_name_;
 778
 779   content::RoutingIDManglingDisabler mangling_disabler_;
 780 };
 781
 782 // If original contents have document type, the serialized contents also have
 783 // document type.
 784 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithDocType) {
 785   base::FilePath page_file_path =
 786       GetTestFilePath("dom_serializer", "youtube_1.htm");
 787   GURL file_url = net::FilePathToFileURL(page_file_path);
 788   ASSERT_TRUE(file_url.SchemeIsFile());
 789   // Load the test file.
 790   NavigateToURL(shell(), file_url);
 791
 792   PostTaskToInProcessRendererAndWait(
 793         base::Bind(&DomSerializerTests::SerializeHTMLDOMWithDocTypeOnRenderer,
 794                    base::Unretained(this), file_url));
 795 }
 796
 797 // If original contents do not have document type, the serialized contents
 798 // also do not have document type.
 799 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithoutDocType) {
 800   base::FilePath page_file_path =
 801       GetTestFilePath("dom_serializer", "youtube_2.htm");
 802   GURL file_url = net::FilePathToFileURL(page_file_path);
 803   ASSERT_TRUE(file_url.SchemeIsFile());
 804   // Load the test file.
 805   NavigateToURL(shell(), file_url);
 806
 807   PostTaskToInProcessRendererAndWait(
 808         base::Bind(
 809             &DomSerializerTests::SerializeHTMLDOMWithoutDocTypeOnRenderer,
 810             base::Unretained(this), file_url));
 811 }
 812
 813 // Serialize XML document which has all 5 built-in entities. After
 814 // finishing serialization, the serialized contents should be same
 815 // with original XML document.
 816 //
 817 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 818 // XML headers are handled differently in the merged serializer.
 819 // Bug: http://crbug.com/328354
 820 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 821                        DISABLED_SerializeXMLDocWithBuiltInEntities) {
 822   base::FilePath page_file_path =
 823       GetTestFilePath("dom_serializer", "note.html");
 824   base::FilePath xml_file_path = GetTestFilePath("dom_serializer", "note.xml");
 825   // Read original contents for later comparison.
 826   std::string original_contents;
 827   ASSERT_TRUE(base::ReadFileToString(xml_file_path, &original_contents));
 828   // Get file URL.
 829   GURL file_url = net::FilePathToFileURL(page_file_path);
 830   GURL xml_file_url = net::FilePathToFileURL(xml_file_path);
 831   ASSERT_TRUE(file_url.SchemeIsFile());
 832   // Load the test file.
 833   NavigateToURL(shell(), file_url);
 834
 835   PostTaskToInProcessRendererAndWait(
 836         base::Bind(
 837             &DomSerializerTests::SerializeXMLDocWithBuiltInEntitiesOnRenderer,
 838             base::Unretained(this), xml_file_url, original_contents));
 839 }
 840
 841 // When serializing DOM, we add MOTW declaration before html tag.
 842 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithAddingMOTW) {
 843   base::FilePath page_file_path =
 844       GetTestFilePath("dom_serializer", "youtube_2.htm");
 845   // Read original contents for later comparison .
 846   std::string original_contents;
 847   ASSERT_TRUE(base::ReadFileToString(page_file_path, &original_contents));
 848   // Get file URL.
 849   GURL file_url = net::FilePathToFileURL(page_file_path);
 850   ASSERT_TRUE(file_url.SchemeIsFile());
 851
 852   // Load the test file.
 853   NavigateToURL(shell(), file_url);
 854
 855   PostTaskToInProcessRendererAndWait(
 856         base::Bind(
 857             &DomSerializerTests::SerializeHTMLDOMWithAddingMOTWOnRenderer,
 858             base::Unretained(this), file_url, original_contents));
 859 }
 860
 861 // When serializing DOM, we will add the META which have correct charset
 862 // declaration as first child of HEAD element for resolving WebKit bug:
 863 // http://bugs.webkit.org/show_bug.cgi?id=16621 even the original document
 864 // does not have META charset declaration.
 865 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 866                        SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc) {
 867   base::FilePath page_file_path =
 868       GetTestFilePath("dom_serializer", "youtube_1.htm");
 869   // Get file URL.
 870   GURL file_url = net::FilePathToFileURL(page_file_path);
 871   ASSERT_TRUE(file_url.SchemeIsFile());
 872   // Load the test file.
 873   NavigateToURL(shell(), file_url);
 874
 875   PostTaskToInProcessRendererAndWait(
 876         base::Bind(
 877             &DomSerializerTests::
 878                 SerializeHTMLDOMWithNoMetaCharsetInOriginalDocOnRenderer,
 879             base::Unretained(this), file_url));
 880 }
 881
 882 // When serializing DOM, if the original document has multiple META charset
 883 // declaration, we will add the META which have correct charset declaration
 884 // as first child of HEAD element and remove all original META charset
 885 // declarations.
 886 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 887                        SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDoc) {
 888   base::FilePath page_file_path =
 889       GetTestFilePath("dom_serializer", "youtube_2.htm");
 890   // Get file URL.
 891   GURL file_url = net::FilePathToFileURL(page_file_path);
 892   ASSERT_TRUE(file_url.SchemeIsFile());
 893   // Load the test file.
 894   NavigateToURL(shell(), file_url);
 895
 896   PostTaskToInProcessRendererAndWait(
 897         base::Bind(
 898             &DomSerializerTests::
 899                 SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDocOnRenderer,
 900             base::Unretained(this), file_url));
 901 }
 902
 903 // Test situation of html entities in text when serializing HTML DOM.
 904 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithEntitiesInText) {
 905   // Need to spin up the renderer and also navigate to a file url so that the
 906   // renderer code doesn't attempt a fork when it sees a load to file scheme
 907   // from non-file scheme.
 908   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
 909
 910   PostTaskToInProcessRendererAndWait(
 911         base::Bind(
 912             &DomSerializerTests::SerializeHTMLDOMWithEntitiesInTextOnRenderer,
 913             base::Unretained(this)));
 914 }
 915
 916 // Test situation of html entities in attribute value when serializing
 917 // HTML DOM.
 918 // This test started to fail at WebKit r65388. See http://crbug.com/52279.
 919 //
 920 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 921 // Some attributes are handled differently in the merged serializer.
 922 // Bug: http://crbug.com/328354
 923 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 924                        DISABLED_SerializeHTMLDOMWithEntitiesInAttributeValue) {
 925   // Need to spin up the renderer and also navigate to a file url so that the
 926   // renderer code doesn't attempt a fork when it sees a load to file scheme
 927   // from non-file scheme.
 928   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
 929
 930   PostTaskToInProcessRendererAndWait(
 931         base::Bind(
 932             &DomSerializerTests::
 933                 SerializeHTMLDOMWithEntitiesInAttributeValueOnRenderer,
 934             base::Unretained(this)));
 935 }
 936
 937 // Test situation of non-standard HTML entities when serializing HTML DOM.
 938 // This test started to fail at WebKit r65351. See http://crbug.com/52279.
 939 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 940                        SerializeHTMLDOMWithNonStandardEntities) {
 941   // Make a test file URL and load it.
 942   base::FilePath page_file_path = GetTestFilePath(
 943       "dom_serializer", "nonstandard_htmlentities.htm");
 944   GURL file_url = net::FilePathToFileURL(page_file_path);
 945   NavigateToURL(shell(), file_url);
 946
 947   PostTaskToInProcessRendererAndWait(
 948         base::Bind(
 949             &DomSerializerTests::
 950                 SerializeHTMLDOMWithNonStandardEntitiesOnRenderer,
 951             base::Unretained(this), file_url));
 952 }
 953
 954 // Test situation of BASE tag in original document when serializing HTML DOM.
 955 // When serializing, we should comment the BASE tag, append a new BASE tag.
 956 // rewrite all the savable URLs to relative local path, and change other URLs
 957 // to absolute URLs.
 958 //
 959 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 960 // Base tags are handled a bit different in merged version.
 961 // Bug: http://crbug.com/328354
 962 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 963                        DISABLED_SerializeHTMLDOMWithBaseTag) {
 964   base::FilePath page_file_path = GetTestFilePath(
 965       "dom_serializer", "html_doc_has_base_tag.htm");
 966
 967   // Get page dir URL which is base URL of this file.
 968   base::FilePath dir_name = page_file_path.DirName();
 969   dir_name = dir_name.Append(
 970       base::FilePath::StringType(base::FilePath::kSeparators[0], 1));
 971   GURL path_dir_url = net::FilePathToFileURL(dir_name);
 972
 973   // Get file URL.
 974   GURL file_url = net::FilePathToFileURL(page_file_path);
 975   ASSERT_TRUE(file_url.SchemeIsFile());
 976   // Load the test file.
 977   NavigateToURL(shell(), file_url);
 978
 979   PostTaskToInProcessRendererAndWait(
 980         base::Bind(
 981             &DomSerializerTests::SerializeHTMLDOMWithBaseTagOnRenderer,
 982             base::Unretained(this), file_url, path_dir_url));
 983 }
 984
 985 // Serializing page which has an empty HEAD tag.
 986 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithEmptyHead) {
 987   // Need to spin up the renderer and also navigate to a file url so that the
 988   // renderer code doesn't attempt a fork when it sees a load to file scheme
 989   // from non-file scheme.
 990   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
 991
 992   PostTaskToInProcessRendererAndWait(
 993         base::Bind(&DomSerializerTests::SerializeHTMLDOMWithEmptyHeadOnRenderer,
 994                    base::Unretained(this)));
 995 }
 996
 997 // Test that we don't crash when the page contains an iframe that
 998 // was handled as a download (http://crbug.com/42212).
 999 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
1000                        SerializeDocumentWithDownloadedIFrame) {
1001   base::FilePath page_file_path = GetTestFilePath(
1002       "dom_serializer", "iframe-src-is-exe.htm");
1003   GURL file_url = net::FilePathToFileURL(page_file_path);
1004   ASSERT_TRUE(file_url.SchemeIsFile());
1005   // Load the test file.
1006   NavigateToURL(shell(), file_url);
1007
1008   PostTaskToInProcessRendererAndWait(
1009         base::Bind(
1010             &DomSerializerTests::
1011                 SerializeDocumentWithDownloadedIFrameOnRenderer,
1012             base::Unretained(this), file_url));
1013 }
1014
1015 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
1016                        SubResourceForElementsInNonHTMLNamespace) {
1017   base::FilePath page_file_path = GetTestFilePath(
1018       "dom_serializer", "non_html_namespace.htm");
1019   GURL file_url = net::FilePathToFileURL(page_file_path);
1020   NavigateToURL(shell(), file_url);
1021
1022   PostTaskToInProcessRendererAndWait(
1023         base::Bind(
1024             &DomSerializerTests::
1025                 SubResourceForElementsInNonHTMLNamespaceOnRenderer,
1026             base::Unretained(this), file_url));
1027 }
1028
1029 }  // namespace content