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/browser/render_view_host.h"
  14 #include "content/public/browser/web_contents.h"
  15 #include "content/public/common/content_switches.h"
  16 #include "content/public/renderer/render_view.h"
  17 #include "content/public/renderer/render_view_observer.h"
  18 #include "content/public/test/content_browser_test.h"
  19 #include "content/public/test/content_browser_test_utils.h"
  20 #include "content/public/test/test_utils.h"
  21 #include "content/renderer/savable_resources.h"
  22 #include "content/shell/browser/shell.h"
  23 #include "net/base/filename_util.h"
  24 #include "net/url_request/url_request_context.h"
  25 #include "third_party/WebKit/public/platform/WebCString.h"
  26 #include "third_party/WebKit/public/platform/WebData.h"
  27 #include "third_party/WebKit/public/platform/WebString.h"
  28 #include "third_party/WebKit/public/platform/WebURL.h"
  29 #include "third_party/WebKit/public/platform/WebVector.h"
  30 #include "third_party/WebKit/public/web/WebDocument.h"
  31 #include "third_party/WebKit/public/web/WebElement.h"
  32 #include "third_party/WebKit/public/web/WebElementCollection.h"
  33 #include "third_party/WebKit/public/web/WebLocalFrame.h"
  34 #include "third_party/WebKit/public/web/WebNode.h"
  35 #include "third_party/WebKit/public/web/WebNodeList.h"
  36 #include "third_party/WebKit/public/web/WebPageSerializer.h"
  37 #include "third_party/WebKit/public/web/WebPageSerializerClient.h"
  38 #include "third_party/WebKit/public/web/WebView.h"
  39
  40 using blink::WebCString;
  41 using blink::WebData;
  42 using blink::WebDocument;
  43 using blink::WebElement;
  44 using blink::WebElementCollection;
  45 using blink::WebFrame;
  46 using blink::WebLocalFrame;
  47 using blink::WebNode;
  48 using blink::WebNodeList;
  49 using blink::WebPageSerializer;
  50 using blink::WebPageSerializerClient;
  51 using blink::WebString;
  52 using blink::WebURL;
  53 using blink::WebView;
  54 using blink::WebVector;
  55
  56 namespace content {
  57
  58 // Iterate recursively over sub-frames to find one with with a given url.
  59 WebFrame* FindSubFrameByURL(WebView* web_view, const GURL& url) {
  60   if (!web_view->mainFrame())
  61     return NULL;
  62
  63   std::vector<WebFrame*> stack;
  64   stack.push_back(web_view->mainFrame());
  65
  66   while (!stack.empty()) {
  67     WebFrame* current_frame = stack.back();
  68     stack.pop_back();
  69     if (GURL(current_frame->document().url()) == url)
  70       return current_frame;
  71     WebElementCollection all = current_frame->document().all();
  72     for (WebElement element = all.firstItem();
  73          !element.isNull(); element = all.nextItem()) {
  74       // Check frame tag and iframe tag
  75       if (!element.hasHTMLTagName("frame") && !element.hasHTMLTagName("iframe"))
  76         continue;
  77       WebFrame* sub_frame = WebLocalFrame::fromFrameOwnerElement(element);
  78       if (sub_frame)
  79         stack.push_back(sub_frame);
  80     }
  81   }
  82   return NULL;
  83 }
  84
  85 // Helper function that test whether the first node in the doc is a doc type
  86 // node.
  87 bool HasDocType(const WebDocument& doc) {
  88   WebNode node = doc.firstChild();
  89   if (node.isNull())
  90     return false;
  91   return node.nodeType() == WebNode::DocumentTypeNode;
  92 }
  93
  94   // Helper function for checking whether input node is META tag. Return true
  95 // means it is META element, otherwise return false. The parameter charset_info
  96 // return actual charset info if the META tag has charset declaration.
  97 bool IsMetaElement(const WebNode& node, std::string& charset_info) {
  98   if (!node.isElementNode())
  99     return false;
 100   const WebElement meta = node.toConst<WebElement>();
 101   if (!meta.hasHTMLTagName("meta"))
 102     return false;
 103   charset_info.erase(0, charset_info.length());
 104   // Check the META charset declaration.
 105   WebString httpEquiv = meta.getAttribute("http-equiv");
 106   if (base::LowerCaseEqualsASCII(base::StringPiece16(httpEquiv),
 107                                  "content-type")) {
 108     std::string content = meta.getAttribute("content").utf8();
 109     int pos = content.find("charset", 0);
 110     if (pos > -1) {
 111       // Add a dummy charset declaration to charset_info, which indicates this
 112       // META tag has charset declaration although we do not get correct value
 113       // yet.
 114       charset_info.append("has-charset-declaration");
 115       int remaining_length = content.length() - pos - 7;
 116       if (!remaining_length)
 117         return true;
 118       int start_pos = pos + 7;
 119       // Find "=" symbol.
 120       while (remaining_length--)
 121         if (content[start_pos++] == L'=')
 122           break;
 123       // Skip beginning space.
 124       while (remaining_length) {
 125         if (content[start_pos] > 0x0020)
 126           break;
 127         ++start_pos;
 128         --remaining_length;
 129       }
 130       if (!remaining_length)
 131         return true;
 132       int end_pos = start_pos;
 133       // Now we find out the start point of charset info. Search the end point.
 134       while (remaining_length--) {
 135         if (content[end_pos] <= 0x0020 || content[end_pos] == L';')
 136           break;
 137         ++end_pos;
 138       }
 139       // Get actual charset info.
 140       charset_info = content.substr(start_pos, end_pos - start_pos);
 141       return true;
 142     }
 143   }
 144   return true;
 145 }
 146
 147 class LoadObserver : public RenderViewObserver {
 148  public:
 149   LoadObserver(RenderView* render_view, const base::Closure& quit_closure)
 150       : RenderViewObserver(render_view),
 151         quit_closure_(quit_closure) {}
 152
 153   void DidFinishLoad(blink::WebLocalFrame* frame) override {
 154     if (frame == render_view()->GetWebView()->mainFrame())
 155       quit_closure_.Run();
 156   }
 157
 158  private:
 159   base::Closure quit_closure_;
 160 };
 161
 162 class DomSerializerTests : public ContentBrowserTest,
 163                            public WebPageSerializerClient {
 164  public:
 165   DomSerializerTests()
 166     : serialized_(false),
 167       local_directory_name_(FILE_PATH_LITERAL("./dummy_files/")) {}
 168
 169   void SetUpCommandLine(base::CommandLine* command_line) override {
 170     command_line->AppendSwitch(switches::kSingleProcess);
 171 #if defined(OS_WIN)
 172     // Don't want to try to create a GPU process.
 173     command_line->AppendSwitch(switches::kDisableGpu);
 174 #endif
 175   }
 176
 177   void SetUpOnMainThread() override {
 178     render_view_routing_id_ =
 179         shell()->web_contents()->GetRenderViewHost()->GetRoutingID();
 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     return RenderView::FromRoutingID(render_view_routing_id_);
 230   }
 231
 232   WebView* GetWebView() {
 233     return GetRenderView()->GetWebView();
 234   }
 235
 236   WebFrame* GetMainFrame() {
 237     return GetWebView()->mainFrame();
 238   }
 239
 240   // Load web page according to input content and relative URLs within
 241   // the document.
 242   void LoadContents(const std::string& contents,
 243                     const GURL& base_url,
 244                     const WebString encoding_info) {
 245     scoped_refptr<MessageLoopRunner> runner = new MessageLoopRunner;
 246     LoadObserver observer(GetRenderView(), runner->QuitClosure());
 247
 248     // If input encoding is empty, use UTF-8 as default encoding.
 249     if (encoding_info.isEmpty()) {
 250       GetMainFrame()->loadHTMLString(contents, base_url);
 251     } else {
 252       WebData data(contents.data(), contents.length());
 253
 254       // Do not use WebFrame.LoadHTMLString because it assumes that input
 255       // html contents use UTF-8 encoding.
 256       // TODO(darin): This should use WebFrame::loadData.
 257       WebFrame* web_frame = GetMainFrame();
 258
 259       ASSERT_TRUE(web_frame != NULL);
 260
 261       web_frame->loadData(data, "text/html", encoding_info, base_url);
 262     }
 263
 264     runner->Run();
 265   }
 266
 267   // Serialize page DOM according to specific page URL. The parameter
 268   // recursive_serialization indicates whether we will serialize all
 269   // sub-frames.
 270   void SerializeDomForURL(const GURL& page_url,
 271                           bool recursive_serialization) {
 272     // Find corresponding WebFrame according to page_url.
 273     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), page_url);
 274     ASSERT_TRUE(web_frame != NULL);
 275     WebVector<WebURL> links;
 276     links.assign(&page_url, 1);
 277     WebString file_path =
 278         base::FilePath(FILE_PATH_LITERAL("c:\\dummy.htm")).AsUTF16Unsafe();
 279     WebVector<WebString> local_paths;
 280     local_paths.assign(&file_path, 1);
 281     // Start serializing DOM.
 282     bool result = WebPageSerializer::serialize(web_frame->toWebLocalFrame(),
 283        recursive_serialization,
 284        static_cast<WebPageSerializerClient*>(this),
 285        links,
 286        local_paths,
 287        local_directory_name_.AsUTF16Unsafe());
 288     ASSERT_TRUE(result);
 289     ASSERT_TRUE(serialized_);
 290   }
 291
 292   void SerializeHTMLDOMWithDocTypeOnRenderer(const GURL& file_url) {
 293     // Make sure original contents have document type.
 294     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 295     ASSERT_TRUE(web_frame != NULL);
 296     WebDocument doc = web_frame->document();
 297     ASSERT_TRUE(HasDocType(doc));
 298     // Do serialization.
 299     SerializeDomForURL(file_url, false);
 300     // Load the serialized contents.
 301     ASSERT_TRUE(HasSerializedFrame(file_url));
 302     const std::string& serialized_contents =
 303         GetSerializedContentForFrame(file_url);
 304     LoadContents(serialized_contents, file_url,
 305                  web_frame->document().encoding());
 306     // Make sure serialized contents still have document type.
 307     web_frame = GetMainFrame();
 308     doc = web_frame->document();
 309     ASSERT_TRUE(HasDocType(doc));
 310   }
 311
 312   void SerializeHTMLDOMWithoutDocTypeOnRenderer(const GURL& file_url) {
 313     // Make sure original contents do not have document type.
 314     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 315     ASSERT_TRUE(web_frame != NULL);
 316     WebDocument doc = web_frame->document();
 317     ASSERT_TRUE(!HasDocType(doc));
 318     // Do serialization.
 319     SerializeDomForURL(file_url, false);
 320     // Load the serialized contents.
 321     ASSERT_TRUE(HasSerializedFrame(file_url));
 322     const std::string& serialized_contents =
 323         GetSerializedContentForFrame(file_url);
 324     LoadContents(serialized_contents, file_url,
 325                  web_frame->document().encoding());
 326     // Make sure serialized contents do not have document type.
 327     web_frame = GetMainFrame();
 328     doc = web_frame->document();
 329     ASSERT_TRUE(!HasDocType(doc));
 330   }
 331
 332   void SerializeXMLDocWithBuiltInEntitiesOnRenderer(
 333       const GURL& xml_file_url, const std::string& original_contents) {
 334     // Do serialization.
 335     SerializeDomForURL(xml_file_url, false);
 336     // Compare the serialized contents with original contents.
 337     ASSERT_TRUE(HasSerializedFrame(xml_file_url));
 338     const std::string& serialized_contents =
 339         GetSerializedContentForFrame(xml_file_url);
 340     ASSERT_EQ(original_contents, serialized_contents);
 341   }
 342
 343   void SerializeHTMLDOMWithAddingMOTWOnRenderer(
 344       const GURL& file_url, const std::string& original_contents) {
 345     // Make sure original contents does not have MOTW;
 346     std::string motw_declaration =
 347        WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 348     ASSERT_FALSE(motw_declaration.empty());
 349     // The encoding of original contents is ISO-8859-1, so we convert the MOTW
 350     // declaration to ASCII and search whether original contents has it or not.
 351     ASSERT_TRUE(std::string::npos == original_contents.find(motw_declaration));
 352
 353     // Do serialization.
 354     SerializeDomForURL(file_url, false);
 355     // Make sure the serialized contents have MOTW ;
 356     ASSERT_TRUE(HasSerializedFrame(file_url));
 357     const std::string& serialized_contents =
 358         GetSerializedContentForFrame(file_url);
 359     ASSERT_FALSE(std::string::npos ==
 360         serialized_contents.find(motw_declaration));
 361   }
 362
 363   void SerializeHTMLDOMWithNoMetaCharsetInOriginalDocOnRenderer(
 364       const GURL& file_url) {
 365     // Make sure there is no META charset declaration in original document.
 366     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 367     ASSERT_TRUE(web_frame != NULL);
 368     WebDocument doc = web_frame->document();
 369     ASSERT_TRUE(doc.isHTMLDocument());
 370     WebElement head_element = doc.head();
 371     ASSERT_TRUE(!head_element.isNull());
 372     // Go through all children of HEAD element.
 373     for (WebNode child = head_element.firstChild(); !child.isNull();
 374          child = child.nextSibling()) {
 375       std::string charset_info;
 376       if (IsMetaElement(child, charset_info))
 377         ASSERT_TRUE(charset_info.empty());
 378     }
 379     // Do serialization.
 380     SerializeDomForURL(file_url, false);
 381
 382     // Load the serialized contents.
 383     ASSERT_TRUE(HasSerializedFrame(file_url));
 384     const std::string& serialized_contents =
 385         GetSerializedContentForFrame(file_url);
 386     LoadContents(serialized_contents, file_url,
 387                  web_frame->document().encoding());
 388     // Make sure the first child of HEAD element is META which has charset
 389     // declaration in serialized contents.
 390     web_frame = GetMainFrame();
 391     ASSERT_TRUE(web_frame != NULL);
 392     doc = web_frame->document();
 393     ASSERT_TRUE(doc.isHTMLDocument());
 394     head_element = doc.head();
 395     ASSERT_TRUE(!head_element.isNull());
 396     WebNode meta_node = head_element.firstChild();
 397     ASSERT_TRUE(!meta_node.isNull());
 398     // Get meta charset info.
 399     std::string charset_info2;
 400     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 401     ASSERT_TRUE(!charset_info2.empty());
 402     ASSERT_EQ(charset_info2,
 403               std::string(web_frame->document().encoding().utf8()));
 404
 405     // Make sure no more additional META tags which have charset declaration.
 406     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 407          child = child.nextSibling()) {
 408       std::string charset_info;
 409       if (IsMetaElement(child, charset_info))
 410         ASSERT_TRUE(charset_info.empty());
 411     }
 412   }
 413
 414   void SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDocOnRenderer(
 415       const GURL& file_url) {
 416     // Make sure there are multiple META charset declarations in original
 417     // document.
 418     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 419     ASSERT_TRUE(web_frame != NULL);
 420     WebDocument doc = web_frame->document();
 421     ASSERT_TRUE(doc.isHTMLDocument());
 422     WebElement head_ele = doc.head();
 423     ASSERT_TRUE(!head_ele.isNull());
 424     // Go through all children of HEAD element.
 425     int charset_declaration_count = 0;
 426     for (WebNode child = head_ele.firstChild(); !child.isNull();
 427          child = child.nextSibling()) {
 428       std::string charset_info;
 429       if (IsMetaElement(child, charset_info) && !charset_info.empty())
 430         charset_declaration_count++;
 431     }
 432     // The original doc has more than META tags which have charset declaration.
 433     ASSERT_TRUE(charset_declaration_count > 1);
 434
 435     // Do serialization.
 436     SerializeDomForURL(file_url, false);
 437
 438     // Load the serialized contents.
 439     ASSERT_TRUE(HasSerializedFrame(file_url));
 440     const std::string& serialized_contents =
 441         GetSerializedContentForFrame(file_url);
 442     LoadContents(serialized_contents, file_url,
 443                  web_frame->document().encoding());
 444     // Make sure only first child of HEAD element is META which has charset
 445     // declaration in serialized contents.
 446     web_frame = GetMainFrame();
 447     ASSERT_TRUE(web_frame != NULL);
 448     doc = web_frame->document();
 449     ASSERT_TRUE(doc.isHTMLDocument());
 450     head_ele = doc.head();
 451     ASSERT_TRUE(!head_ele.isNull());
 452     WebNode meta_node = head_ele.firstChild();
 453     ASSERT_TRUE(!meta_node.isNull());
 454     // Get meta charset info.
 455     std::string charset_info2;
 456     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 457     ASSERT_TRUE(!charset_info2.empty());
 458     ASSERT_EQ(charset_info2,
 459               std::string(web_frame->document().encoding().utf8()));
 460
 461     // Make sure no more additional META tags which have charset declaration.
 462     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 463          child = child.nextSibling()) {
 464       std::string charset_info;
 465       if (IsMetaElement(child, charset_info))
 466         ASSERT_TRUE(charset_info.empty());
 467     }
 468   }
 469
 470   void SerializeHTMLDOMWithEntitiesInTextOnRenderer() {
 471     base::FilePath page_file_path = GetTestFilePath(
 472         "dom_serializer", "dom_serializer/htmlentities_in_text.htm");
 473     // Get file URL. The URL is dummy URL to identify the following loading
 474     // actions. The test content is in constant:original_contents.
 475     GURL file_url = net::FilePathToFileURL(page_file_path);
 476     ASSERT_TRUE(file_url.SchemeIsFile());
 477     // Test contents.
 478     static const char* const original_contents =
 479         "<html><body>&amp;&lt;&gt;\"\'</body></html>";
 480     // Load the test contents.
 481     LoadContents(original_contents, file_url, WebString());
 482
 483     // Get BODY's text content in DOM.
 484     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 485     ASSERT_TRUE(web_frame != NULL);
 486     WebDocument doc = web_frame->document();
 487     ASSERT_TRUE(doc.isHTMLDocument());
 488     WebElement body_ele = doc.body();
 489     ASSERT_TRUE(!body_ele.isNull());
 490     WebNode text_node = body_ele.firstChild();
 491     ASSERT_TRUE(text_node.isTextNode());
 492     ASSERT_TRUE(std::string(text_node.createMarkup().utf8()) ==
 493                 "&amp;&lt;&gt;\"\'");
 494     // Do serialization.
 495     SerializeDomForURL(file_url, false);
 496     // Compare the serialized contents with original contents.
 497     ASSERT_TRUE(HasSerializedFrame(file_url));
 498     const std::string& serialized_contents =
 499         GetSerializedContentForFrame(file_url);
 500     // Compare the serialized contents with original contents to make sure
 501     // they are same.
 502     // Because we add MOTW when serializing DOM, so before comparison, we also
 503     // need to add MOTW to original_contents.
 504     std::string original_str =
 505       WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 506     original_str += original_contents;
 507     // Since WebCore now inserts a new HEAD element if there is no HEAD element
 508     // when creating BODY element. (Please see
 509     // HTMLParser::bodyCreateErrorCheck.) We need to append the HEAD content and
 510     // corresponding META content if we find WebCore-generated HEAD element.
 511     if (!doc.head().isNull()) {
 512       WebString encoding = web_frame->document().encoding();
 513       std::string htmlTag("<html>");
 514       std::string::size_type pos = original_str.find(htmlTag);
 515       ASSERT_NE(std::string::npos, pos);
 516       pos += htmlTag.length();
 517       std::string head_part("<head>");
 518       head_part +=
 519           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 520       head_part += "</head>";
 521       original_str.insert(pos, head_part);
 522     }
 523     ASSERT_EQ(original_str, serialized_contents);
 524   }
 525
 526   void SerializeHTMLDOMWithEntitiesInAttributeValueOnRenderer() {
 527     base::FilePath page_file_path = GetTestFilePath(
 528         "dom_serializer", "dom_serializer/htmlentities_in_attribute_value.htm");
 529     // Get file URL. The URL is dummy URL to identify the following loading
 530     // actions. The test content is in constant:original_contents.
 531     GURL file_url = net::FilePathToFileURL(page_file_path);
 532     ASSERT_TRUE(file_url.SchemeIsFile());
 533     // Test contents.
 534     static const char* const original_contents =
 535         "<html><body title=\"&amp;&lt;&gt;&quot;&#39;\"></body></html>";
 536     // Load the test contents.
 537     LoadContents(original_contents, file_url, WebString());
 538     // Get value of BODY's title attribute in DOM.
 539     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 540     ASSERT_TRUE(web_frame != NULL);
 541     WebDocument doc = web_frame->document();
 542     ASSERT_TRUE(doc.isHTMLDocument());
 543     WebElement body_ele = doc.body();
 544     ASSERT_TRUE(!body_ele.isNull());
 545     WebString value = body_ele.getAttribute("title");
 546     ASSERT_TRUE(std::string(value.utf8()) == "&<>\"\'");
 547     // Do serialization.
 548     SerializeDomForURL(file_url, false);
 549     // Compare the serialized contents with original contents.
 550     ASSERT_TRUE(HasSerializedFrame(file_url));
 551     const std::string& serialized_contents =
 552         GetSerializedContentForFrame(file_url);
 553     // Compare the serialized contents with original contents to make sure
 554     // they are same.
 555     std::string original_str =
 556         WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 557     original_str += original_contents;
 558     if (!doc.isNull()) {
 559       WebString encoding = web_frame->document().encoding();
 560       std::string htmlTag("<html>");
 561       std::string::size_type pos = original_str.find(htmlTag);
 562       ASSERT_NE(std::string::npos, pos);
 563       pos += htmlTag.length();
 564       std::string head_part("<head>");
 565       head_part +=
 566           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 567       head_part += "</head>";
 568       original_str.insert(pos, head_part);
 569     }
 570     ASSERT_EQ(original_str, serialized_contents);
 571   }
 572
 573   void SerializeHTMLDOMWithNonStandardEntitiesOnRenderer(const GURL& file_url) {
 574     // Get value of BODY's title attribute in DOM.
 575     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 576     WebDocument doc = web_frame->document();
 577     ASSERT_TRUE(doc.isHTMLDocument());
 578     WebElement body_element = doc.body();
 579     // Unescaped string for "&percnt;&nsup;&sup1;&apos;".
 580     static const wchar_t parsed_value[] = {
 581       '%', 0x2285, 0x00b9, '\'', 0
 582     };
 583     WebString value = body_element.getAttribute("title");
 584     WebString content = doc.contentAsTextForTesting();
 585     ASSERT_TRUE(base::UTF16ToWide(value) == parsed_value);
 586     ASSERT_TRUE(base::UTF16ToWide(content) == parsed_value);
 587
 588     // Do serialization.
 589     SerializeDomForURL(file_url, false);
 590     // Check the serialized string.
 591     ASSERT_TRUE(HasSerializedFrame(file_url));
 592     const std::string& serialized_contents =
 593         GetSerializedContentForFrame(file_url);
 594     // Confirm that the serialized string has no non-standard HTML entities.
 595     ASSERT_EQ(std::string::npos, serialized_contents.find("&percnt;"));
 596     ASSERT_EQ(std::string::npos, serialized_contents.find("&nsup;"));
 597     ASSERT_EQ(std::string::npos, serialized_contents.find("&sup1;"));
 598     ASSERT_EQ(std::string::npos, serialized_contents.find("&apos;"));
 599   }
 600
 601   void SerializeHTMLDOMWithBaseTagOnRenderer(const GURL& file_url,
 602                                              const GURL& path_dir_url) {
 603     // There are total 2 available base tags in this test file.
 604     const int kTotalBaseTagCountInTestFile = 2;
 605
 606     // Since for this test, we assume there is no savable sub-resource links for
 607     // this test file, also all links are relative URLs in this test file, so we
 608     // need to check those relative URLs and make sure document has BASE tag.
 609     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 610     ASSERT_TRUE(web_frame != NULL);
 611     WebDocument doc = web_frame->document();
 612     ASSERT_TRUE(doc.isHTMLDocument());
 613     // Go through all descent nodes.
 614     WebElementCollection all = doc.all();
 615     int original_base_tag_count = 0;
 616     for (WebElement element = all.firstItem(); !element.isNull();
 617          element = all.nextItem()) {
 618       if (element.hasHTMLTagName("base")) {
 619         original_base_tag_count++;
 620       } else {
 621         // Get link.
 622         WebString value = GetSubResourceLinkFromElement(element);
 623         if (value.isNull() && element.hasHTMLTagName("a")) {
 624           value = element.getAttribute("href");
 625           if (value.isEmpty())
 626             value = WebString();
 627         }
 628         // Each link is relative link.
 629         if (!value.isNull()) {
 630           GURL link(value.utf8());
 631           ASSERT_TRUE(link.scheme().empty());
 632         }
 633       }
 634     }
 635     ASSERT_EQ(original_base_tag_count, kTotalBaseTagCountInTestFile);
 636     // Make sure in original document, the base URL is not equal with the
 637     // |path_dir_url|.
 638     GURL original_base_url(doc.baseURL());
 639     ASSERT_NE(original_base_url, path_dir_url);
 640
 641     // Do serialization.
 642     SerializeDomForURL(file_url, false);
 643
 644     // Load the serialized contents.
 645     ASSERT_TRUE(HasSerializedFrame(file_url));
 646     const std::string& serialized_contents =
 647         GetSerializedContentForFrame(file_url);
 648     LoadContents(serialized_contents, file_url,
 649                  web_frame->document().encoding());
 650
 651     // Make sure all links are absolute URLs and doc there are some number of
 652     // BASE tags in serialized HTML data. Each of those BASE tags have same base
 653     // URL which is as same as URL of current test file.
 654     web_frame = GetMainFrame();
 655     ASSERT_TRUE(web_frame != NULL);
 656     doc = web_frame->document();
 657     ASSERT_TRUE(doc.isHTMLDocument());
 658     // Go through all descent nodes.
 659     all = doc.all();
 660     int new_base_tag_count = 0;
 661     for (WebNode node = all.firstItem(); !node.isNull();
 662          node = all.nextItem()) {
 663       if (!node.isElementNode())
 664         continue;
 665       WebElement element = node.to<WebElement>();
 666       if (element.hasHTMLTagName("base")) {
 667         new_base_tag_count++;
 668       } else {
 669         // Get link.
 670         WebString value = GetSubResourceLinkFromElement(element);
 671         if (value.isNull() && element.hasHTMLTagName("a")) {
 672           value = element.getAttribute("href");
 673           if (value.isEmpty())
 674             value = WebString();
 675         }
 676         // Each link is absolute link.
 677         if (!value.isNull()) {
 678           GURL link(std::string(value.utf8()));
 679           ASSERT_FALSE(link.scheme().empty());
 680         }
 681       }
 682     }
 683     // We have one more added BASE tag which is generated by JavaScript.
 684     ASSERT_EQ(new_base_tag_count, original_base_tag_count + 1);
 685     // Make sure in new document, the base URL is equal with the |path_dir_url|.
 686     GURL new_base_url(doc.baseURL());
 687     ASSERT_EQ(new_base_url, path_dir_url);
 688   }
 689
 690   void SerializeHTMLDOMWithEmptyHeadOnRenderer() {
 691     base::FilePath page_file_path = GetTestFilePath(
 692         "dom_serializer", "empty_head.htm");
 693     GURL file_url = net::FilePathToFileURL(page_file_path);
 694     ASSERT_TRUE(file_url.SchemeIsFile());
 695
 696     // Load the test html content.
 697     static const char* const empty_head_contents =
 698       "<html><head></head><body>hello world</body></html>";
 699     LoadContents(empty_head_contents, file_url, WebString());
 700
 701     // Make sure the head tag is empty.
 702     WebFrame* web_frame = GetMainFrame();
 703     ASSERT_TRUE(web_frame != NULL);
 704     WebDocument doc = web_frame->document();
 705     ASSERT_TRUE(doc.isHTMLDocument());
 706     WebElement head_element = doc.head();
 707     ASSERT_TRUE(!head_element.isNull());
 708     ASSERT_TRUE(!head_element.hasChildNodes());
 709     ASSERT_TRUE(head_element.childNodes().length() == 0);
 710
 711     // Do serialization.
 712     SerializeDomForURL(file_url, false);
 713     // Make sure the serialized contents have META ;
 714     ASSERT_TRUE(HasSerializedFrame(file_url));
 715     const std::string& serialized_contents =
 716         GetSerializedContentForFrame(file_url);
 717
 718     // Reload serialized contents and make sure there is only one META tag.
 719     LoadContents(serialized_contents, file_url,
 720                  web_frame->document().encoding());
 721     web_frame = GetMainFrame();
 722     ASSERT_TRUE(web_frame != NULL);
 723     doc = web_frame->document();
 724     ASSERT_TRUE(doc.isHTMLDocument());
 725     head_element = doc.head();
 726     ASSERT_TRUE(!head_element.isNull());
 727     ASSERT_TRUE(head_element.hasChildNodes());
 728     ASSERT_TRUE(head_element.childNodes().length() == 1);
 729     WebNode meta_node = head_element.firstChild();
 730     ASSERT_TRUE(!meta_node.isNull());
 731     // Get meta charset info.
 732     std::string charset_info;
 733     ASSERT_TRUE(IsMetaElement(meta_node, charset_info));
 734     ASSERT_TRUE(!charset_info.empty());
 735     ASSERT_EQ(charset_info,
 736               std::string(web_frame->document().encoding().utf8()));
 737
 738     // Check the body's first node is text node and its contents are
 739     // "hello world"
 740     WebElement body_element = doc.body();
 741     ASSERT_TRUE(!body_element.isNull());
 742     WebNode text_node = body_element.firstChild();
 743     ASSERT_TRUE(text_node.isTextNode());
 744     WebString text_node_contents = text_node.nodeValue();
 745     ASSERT_TRUE(std::string(text_node_contents.utf8()) == "hello world");
 746   }
 747
 748   void SerializeDocumentWithDownloadedIFrameOnRenderer(const GURL& file_url) {
 749     // Do a recursive serialization. We pass if we don't crash.
 750     SerializeDomForURL(file_url, true);
 751   }
 752
 753   void SubResourceForElementsInNonHTMLNamespaceOnRenderer(
 754       const GURL& file_url) {
 755     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 756     ASSERT_TRUE(web_frame != NULL);
 757     WebDocument doc = web_frame->document();
 758     WebNode lastNodeInBody = doc.body().lastChild();
 759     ASSERT_EQ(WebNode::ElementNode, lastNodeInBody.nodeType());
 760     WebString uri = GetSubResourceLinkFromElement(
 761         lastNodeInBody.to<WebElement>());
 762     EXPECT_TRUE(uri.isNull());
 763   }
 764
 765  private:
 766   int32 render_view_routing_id_;
 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
 780 // If original contents have document type, the serialized contents also have
 781 // document type.
 782 // Disabled by ellyjones@ on 2015-05-18, see https://crbug.com/488495.
 783 #if defined(OS_MACOSX)
 784 #define MAYBE_SerializeHTMLDOMWithDocType DISABLED_SerializeHTMLDOMWithDocType
 785 #else
 786 #define MAYBE_SerializeHTMLDOMWithDocType SerializeHTMLDOMWithDocType
 787 #endif
 788
 789 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 790                        MAYBE_SerializeHTMLDOMWithDocType) {
 791   base::FilePath page_file_path =
 792       GetTestFilePath("dom_serializer", "youtube_1.htm");
 793   GURL file_url = net::FilePathToFileURL(page_file_path);
 794   ASSERT_TRUE(file_url.SchemeIsFile());
 795   // Load the test file.
 796   NavigateToURL(shell(), file_url);
 797
 798   PostTaskToInProcessRendererAndWait(
 799         base::Bind(&DomSerializerTests::SerializeHTMLDOMWithDocTypeOnRenderer,
 800                    base::Unretained(this), file_url));
 801 }
 802
 803 // If original contents do not have document type, the serialized contents
 804 // also do not have document type.
 805 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithoutDocType) {
 806   base::FilePath page_file_path =
 807       GetTestFilePath("dom_serializer", "youtube_2.htm");
 808   GURL file_url = net::FilePathToFileURL(page_file_path);
 809   ASSERT_TRUE(file_url.SchemeIsFile());
 810   // Load the test file.
 811   NavigateToURL(shell(), file_url);
 812
 813   PostTaskToInProcessRendererAndWait(
 814         base::Bind(
 815             &DomSerializerTests::SerializeHTMLDOMWithoutDocTypeOnRenderer,
 816             base::Unretained(this), file_url));
 817 }
 818
 819 // Serialize XML document which has all 5 built-in entities. After
 820 // finishing serialization, the serialized contents should be same
 821 // with original XML document.
 822 //
 823 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 824 // XML headers are handled differently in the merged serializer.
 825 // Bug: http://crbug.com/328354
 826 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 827                        DISABLED_SerializeXMLDocWithBuiltInEntities) {
 828   base::FilePath page_file_path =
 829       GetTestFilePath("dom_serializer", "note.html");
 830   base::FilePath xml_file_path = GetTestFilePath("dom_serializer", "note.xml");
 831   // Read original contents for later comparison.
 832   std::string original_contents;
 833   ASSERT_TRUE(base::ReadFileToString(xml_file_path, &original_contents));
 834   // Get file URL.
 835   GURL file_url = net::FilePathToFileURL(page_file_path);
 836   GURL xml_file_url = net::FilePathToFileURL(xml_file_path);
 837   ASSERT_TRUE(file_url.SchemeIsFile());
 838   // Load the test file.
 839   NavigateToURL(shell(), file_url);
 840
 841   PostTaskToInProcessRendererAndWait(
 842         base::Bind(
 843             &DomSerializerTests::SerializeXMLDocWithBuiltInEntitiesOnRenderer,
 844             base::Unretained(this), xml_file_url, original_contents));
 845 }
 846
 847 // When serializing DOM, we add MOTW declaration before html tag.
 848 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithAddingMOTW) {
 849   base::FilePath page_file_path =
 850       GetTestFilePath("dom_serializer", "youtube_2.htm");
 851   // Read original contents for later comparison .
 852   std::string original_contents;
 853   ASSERT_TRUE(base::ReadFileToString(page_file_path, &original_contents));
 854   // Get file URL.
 855   GURL file_url = net::FilePathToFileURL(page_file_path);
 856   ASSERT_TRUE(file_url.SchemeIsFile());
 857
 858   // Load the test file.
 859   NavigateToURL(shell(), file_url);
 860
 861   PostTaskToInProcessRendererAndWait(
 862         base::Bind(
 863             &DomSerializerTests::SerializeHTMLDOMWithAddingMOTWOnRenderer,
 864             base::Unretained(this), file_url, original_contents));
 865 }
 866
 867 // When serializing DOM, we will add the META which have correct charset
 868 // declaration as first child of HEAD element for resolving WebKit bug:
 869 // http://bugs.webkit.org/show_bug.cgi?id=16621 even the original document
 870 // does not have META charset declaration.
 871 // Disabled by battre@ on 2015-05-21, see https://crbug.com/488495.
 872 #if defined(OS_MACOSX)
 873 #define MAYBE_SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc \
 874   DISABLED_SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc
 875 #else
 876 #define MAYBE_SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc \
 877   SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc
 878 #endif
 879 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 880                        MAYBE_SerializeHTMLDOMWithNoMetaCharsetInOriginalDoc) {
 881   base::FilePath page_file_path =
 882       GetTestFilePath("dom_serializer", "youtube_1.htm");
 883   // Get file URL.
 884   GURL file_url = net::FilePathToFileURL(page_file_path);
 885   ASSERT_TRUE(file_url.SchemeIsFile());
 886   // Load the test file.
 887   NavigateToURL(shell(), file_url);
 888
 889   PostTaskToInProcessRendererAndWait(
 890         base::Bind(
 891             &DomSerializerTests::
 892                 SerializeHTMLDOMWithNoMetaCharsetInOriginalDocOnRenderer,
 893             base::Unretained(this), file_url));
 894 }
 895
 896 // When serializing DOM, if the original document has multiple META charset
 897 // declaration, we will add the META which have correct charset declaration
 898 // as first child of HEAD element and remove all original META charset
 899 // declarations.
 900 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 901                        SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDoc) {
 902   base::FilePath page_file_path =
 903       GetTestFilePath("dom_serializer", "youtube_2.htm");
 904   // Get file URL.
 905   GURL file_url = net::FilePathToFileURL(page_file_path);
 906   ASSERT_TRUE(file_url.SchemeIsFile());
 907   // Load the test file.
 908   NavigateToURL(shell(), file_url);
 909
 910   PostTaskToInProcessRendererAndWait(
 911         base::Bind(
 912             &DomSerializerTests::
 913                 SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDocOnRenderer,
 914             base::Unretained(this), file_url));
 915 }
 916
 917 // Test situation of html entities in text when serializing HTML DOM.
 918 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithEntitiesInText) {
 919   // Need to spin up the renderer and also navigate to a file url so that the
 920   // renderer code doesn't attempt a fork when it sees a load to file scheme
 921   // from non-file scheme.
 922   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
 923
 924   PostTaskToInProcessRendererAndWait(
 925         base::Bind(
 926             &DomSerializerTests::SerializeHTMLDOMWithEntitiesInTextOnRenderer,
 927             base::Unretained(this)));
 928 }
 929
 930 // Test situation of html entities in attribute value when serializing
 931 // HTML DOM.
 932 // This test started to fail at WebKit r65388. See http://crbug.com/52279.
 933 //
 934 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 935 // Some attributes are handled differently in the merged serializer.
 936 // Bug: http://crbug.com/328354
 937 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 938                        DISABLED_SerializeHTMLDOMWithEntitiesInAttributeValue) {
 939   // Need to spin up the renderer and also navigate to a file url so that the
 940   // renderer code doesn't attempt a fork when it sees a load to file scheme
 941   // from non-file scheme.
 942   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
 943
 944   PostTaskToInProcessRendererAndWait(
 945         base::Bind(
 946             &DomSerializerTests::
 947                 SerializeHTMLDOMWithEntitiesInAttributeValueOnRenderer,
 948             base::Unretained(this)));
 949 }
 950
 951 // Test situation of non-standard HTML entities when serializing HTML DOM.
 952 // This test started to fail at WebKit r65351. See http://crbug.com/52279.
 953 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 954                        SerializeHTMLDOMWithNonStandardEntities) {
 955   // Make a test file URL and load it.
 956   base::FilePath page_file_path = GetTestFilePath(
 957       "dom_serializer", "nonstandard_htmlentities.htm");
 958   GURL file_url = net::FilePathToFileURL(page_file_path);
 959   NavigateToURL(shell(), file_url);
 960
 961   PostTaskToInProcessRendererAndWait(
 962         base::Bind(
 963             &DomSerializerTests::
 964                 SerializeHTMLDOMWithNonStandardEntitiesOnRenderer,
 965             base::Unretained(this), file_url));
 966 }
 967
 968 // Test situation of BASE tag in original document when serializing HTML DOM.
 969 // When serializing, we should comment the BASE tag, append a new BASE tag.
 970 // rewrite all the savable URLs to relative local path, and change other URLs
 971 // to absolute URLs.
 972 //
 973 // TODO(tiger@opera.com): Disabled in preparation of page serializer merge --
 974 // Base tags are handled a bit different in merged version.
 975 // Bug: http://crbug.com/328354
 976 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
 977                        DISABLED_SerializeHTMLDOMWithBaseTag) {
 978   base::FilePath page_file_path = GetTestFilePath(
 979       "dom_serializer", "html_doc_has_base_tag.htm");
 980
 981   // Get page dir URL which is base URL of this file.
 982   base::FilePath dir_name = page_file_path.DirName();
 983   dir_name = dir_name.Append(
 984       base::FilePath::StringType(base::FilePath::kSeparators[0], 1));
 985   GURL path_dir_url = net::FilePathToFileURL(dir_name);
 986
 987   // Get file URL.
 988   GURL file_url = net::FilePathToFileURL(page_file_path);
 989   ASSERT_TRUE(file_url.SchemeIsFile());
 990   // Load the test file.
 991   NavigateToURL(shell(), file_url);
 992
 993   PostTaskToInProcessRendererAndWait(
 994         base::Bind(
 995             &DomSerializerTests::SerializeHTMLDOMWithBaseTagOnRenderer,
 996             base::Unretained(this), file_url, path_dir_url));
 997 }
 998
 999 // Serializing page which has an empty HEAD tag.
1000 IN_PROC_BROWSER_TEST_F(DomSerializerTests, SerializeHTMLDOMWithEmptyHead) {
1001   // Need to spin up the renderer and also navigate to a file url so that the
1002   // renderer code doesn't attempt a fork when it sees a load to file scheme
1003   // from non-file scheme.
1004   NavigateToURL(shell(), GetTestUrl(".", "simple_page.html"));
1005
1006   PostTaskToInProcessRendererAndWait(
1007         base::Bind(&DomSerializerTests::SerializeHTMLDOMWithEmptyHeadOnRenderer,
1008                    base::Unretained(this)));
1009 }
1010
1011 // Test that we don't crash when the page contains an iframe that
1012 // was handled as a download (http://crbug.com/42212).
1013 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
1014                        SerializeDocumentWithDownloadedIFrame) {
1015   base::FilePath page_file_path = GetTestFilePath(
1016       "dom_serializer", "iframe-src-is-exe.htm");
1017   GURL file_url = net::FilePathToFileURL(page_file_path);
1018   ASSERT_TRUE(file_url.SchemeIsFile());
1019   // Load the test file.
1020   NavigateToURL(shell(), file_url);
1021
1022   PostTaskToInProcessRendererAndWait(
1023         base::Bind(
1024             &DomSerializerTests::
1025                 SerializeDocumentWithDownloadedIFrameOnRenderer,
1026             base::Unretained(this), file_url));
1027 }
1028
1029 IN_PROC_BROWSER_TEST_F(DomSerializerTests,
1030                        SubResourceForElementsInNonHTMLNamespace) {
1031   base::FilePath page_file_path = GetTestFilePath(
1032       "dom_serializer", "non_html_namespace.htm");
1033   GURL file_url = net::FilePathToFileURL(page_file_path);
1034   NavigateToURL(shell(), file_url);
1035
1036   PostTaskToInProcessRendererAndWait(
1037         base::Bind(
1038             &DomSerializerTests::
1039                 SubResourceForElementsInNonHTMLNamespaceOnRenderer,
1040             base::Unretained(this), file_url));
1041 }
1042
1043 }  // namespace content