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/test_utils.h"
  19 #include "content/renderer/savable_resources.h"
  20 #include "content/shell/browser/shell.h"
  21 #include "net/base/filename_util.h"
  22 #include "net/url_request/url_request_context.h"
  23 #include "third_party/WebKit/public/platform/WebCString.h"
  24 #include "third_party/WebKit/public/platform/WebData.h"
  25 #include "third_party/WebKit/public/platform/WebString.h"
  26 #include "third_party/WebKit/public/platform/WebURL.h"
  27 #include "third_party/WebKit/public/platform/WebVector.h"
  28 #include "third_party/WebKit/public/web/WebDocument.h"
  29 #include "third_party/WebKit/public/web/WebElement.h"
  30 #include "third_party/WebKit/public/web/WebElementCollection.h"
  31 #include "third_party/WebKit/public/web/WebLocalFrame.h"
  32 #include "third_party/WebKit/public/web/WebNode.h"
  33 #include "third_party/WebKit/public/web/WebNodeList.h"
  34 #include "third_party/WebKit/public/web/WebPageSerializer.h"
  35 #include "third_party/WebKit/public/web/WebPageSerializerClient.h"
  36 #include "third_party/WebKit/public/web/WebView.h"
  37
  38 using blink::WebCString;
  39 using blink::WebData;
  40 using blink::WebDocument;
  41 using blink::WebElement;
  42 using blink::WebElementCollection;
  43 using blink::WebFrame;
  44 using blink::WebLocalFrame;
  45 using blink::WebNode;
  46 using blink::WebNodeList;
  47 using blink::WebPageSerializer;
  48 using blink::WebPageSerializerClient;
  49 using blink::WebString;
  50 using blink::WebURL;
  51 using blink::WebView;
  52 using blink::WebVector;
  53
  54 namespace {
  55
  56 // The first RenderFrame is routing ID 1, and the first RenderView is 2.
  57 const int kRenderViewRoutingId = 2;
  58
  59 }
  60
  61 namespace content {
  62
  63 // Iterate recursively over sub-frames to find one with with a given url.
  64 WebFrame* FindSubFrameByURL(WebView* web_view, const GURL& url) {
  65   if (!web_view->mainFrame())
  66     return NULL;
  67
  68   std::vector<WebFrame*> stack;
  69   stack.push_back(web_view->mainFrame());
  70
  71   while (!stack.empty()) {
  72     WebFrame* current_frame = stack.back();
  73     stack.pop_back();
  74     if (GURL(current_frame->document().url()) == url)
  75       return current_frame;
  76     WebElementCollection all = current_frame->document().all();
  77     for (WebElement element = all.firstItem();
  78          !element.isNull(); element = all.nextItem()) {
  79       // Check frame tag and iframe tag
  80       if (!element.hasHTMLTagName("frame") && !element.hasHTMLTagName("iframe"))
  81         continue;
  82       WebFrame* sub_frame = WebLocalFrame::fromFrameOwnerElement(element);
  83       if (sub_frame)
  84         stack.push_back(sub_frame);
  85     }
  86   }
  87   return NULL;
  88 }
  89
  90 // Helper function that test whether the first node in the doc is a doc type
  91 // node.
  92 bool HasDocType(const WebDocument& doc) {
  93   WebNode node = doc.firstChild();
  94   if (node.isNull())
  95     return false;
  96   return node.nodeType() == WebNode::DocumentTypeNode;
  97 }
  98
  99   // Helper function for checking whether input node is META tag. Return true
 100 // means it is META element, otherwise return false. The parameter charset_info
 101 // return actual charset info if the META tag has charset declaration.
 102 bool IsMetaElement(const WebNode& node, std::string& charset_info) {
 103   if (!node.isElementNode())
 104     return false;
 105   const WebElement meta = node.toConst<WebElement>();
 106   if (!meta.hasHTMLTagName("meta"))
 107     return false;
 108   charset_info.erase(0, charset_info.length());
 109   // Check the META charset declaration.
 110   WebString httpEquiv = meta.getAttribute("http-equiv");
 111   if (base::LowerCaseEqualsASCII(httpEquiv, "content-type")) {
 112     std::string content = meta.getAttribute("content").utf8();
 113     int pos = content.find("charset", 0);
 114     if (pos > -1) {
 115       // Add a dummy charset declaration to charset_info, which indicates this
 116       // META tag has charset declaration although we do not get correct value
 117       // yet.
 118       charset_info.append("has-charset-declaration");
 119       int remaining_length = content.length() - pos - 7;
 120       if (!remaining_length)
 121         return true;
 122       int start_pos = pos + 7;
 123       // Find "=" symbol.
 124       while (remaining_length--)
 125         if (content[start_pos++] == L'=')
 126           break;
 127       // Skip beginning space.
 128       while (remaining_length) {
 129         if (content[start_pos] > 0x0020)
 130           break;
 131         ++start_pos;
 132         --remaining_length;
 133       }
 134       if (!remaining_length)
 135         return true;
 136       int end_pos = start_pos;
 137       // Now we find out the start point of charset info. Search the end point.
 138       while (remaining_length--) {
 139         if (content[end_pos] <= 0x0020 || content[end_pos] == L';')
 140           break;
 141         ++end_pos;
 142       }
 143       // Get actual charset info.
 144       charset_info = content.substr(start_pos, end_pos - start_pos);
 145       return true;
 146     }
 147   }
 148   return true;
 149 }
 150
 151 class LoadObserver : public RenderViewObserver {
 152  public:
 153   LoadObserver(RenderView* render_view, const base::Closure& quit_closure)
 154       : RenderViewObserver(render_view),
 155         quit_closure_(quit_closure) {}
 156
 157   void DidFinishLoad(blink::WebLocalFrame* frame) override {
 158     if (frame == render_view()->GetWebView()->mainFrame())
 159       quit_closure_.Run();
 160   }
 161
 162  private:
 163   base::Closure quit_closure_;
 164 };
 165
 166 class DomSerializerTests : public ContentBrowserTest,
 167                            public WebPageSerializerClient {
 168  public:
 169   DomSerializerTests()
 170     : serialized_(false),
 171       local_directory_name_(FILE_PATH_LITERAL("./dummy_files/")) {}
 172
 173   void SetUpCommandLine(base::CommandLine* command_line) override {
 174     command_line->AppendSwitch(switches::kSingleProcess);
 175 #if defined(OS_WIN)
 176     // Don't want to try to create a GPU process.
 177     command_line->AppendSwitch(switches::kDisableGpu);
 178 #endif
 179   }
 180
 181   // DomSerializerDelegate.
 182   virtual void didSerializeDataForFrame(const WebURL& frame_web_url,
 183                                         const WebCString& data,
 184                                         PageSerializationStatus status) {
 185
 186     GURL frame_url(frame_web_url);
 187     // If the all frames are finished saving, check all finish status
 188     if (status == WebPageSerializerClient::AllFramesAreFinished) {
 189       SerializationFinishStatusMap::iterator it =
 190           serialization_finish_status_.begin();
 191       for (; it != serialization_finish_status_.end(); ++it)
 192         ASSERT_TRUE(it->second);
 193       serialized_ = true;
 194       return;
 195     }
 196
 197     // Check finish status of current frame.
 198     SerializationFinishStatusMap::iterator it =
 199         serialization_finish_status_.find(frame_url.spec());
 200     // New frame, set initial status as false.
 201     if (it == serialization_finish_status_.end())
 202       serialization_finish_status_[frame_url.spec()] = false;
 203
 204     it = serialization_finish_status_.find(frame_url.spec());
 205     ASSERT_TRUE(it != serialization_finish_status_.end());
 206     // In process frame, finish status should be false.
 207     ASSERT_FALSE(it->second);
 208
 209     // Add data to corresponding frame's content.
 210     serialized_frame_map_[frame_url.spec()] += data.data();
 211
 212     // Current frame is completed saving, change the finish status.
 213     if (status == WebPageSerializerClient::CurrentFrameIsFinished)
 214       it->second = true;
 215   }
 216
 217   bool HasSerializedFrame(const GURL& frame_url) {
 218     return serialized_frame_map_.find(frame_url.spec()) !=
 219            serialized_frame_map_.end();
 220   }
 221
 222   const std::string& GetSerializedContentForFrame(
 223       const GURL& frame_url) {
 224     return serialized_frame_map_[frame_url.spec()];
 225   }
 226
 227   RenderView* GetRenderView() {
 228     // We could have the test on the UI thread get the WebContent's routing ID,
 229     // but we know this will be the first RV so skip that and just hardcode it.
 230     return RenderView::FromRoutingID(kRenderViewRoutingId);
 231   }
 232
 233   WebView* GetWebView() {
 234     return GetRenderView()->GetWebView();
 235   }
 236
 237   WebFrame* GetMainFrame() {
 238     return GetWebView()->mainFrame();
 239   }
 240
 241   // Load web page according to input content and relative URLs within
 242   // the document.
 243   void LoadContents(const std::string& contents,
 244                     const GURL& base_url,
 245                     const WebString encoding_info) {
 246     scoped_refptr<MessageLoopRunner> runner = new MessageLoopRunner;
 247     LoadObserver observer(GetRenderView(), runner->QuitClosure());
 248
 249     // If input encoding is empty, use UTF-8 as default encoding.
 250     if (encoding_info.isEmpty()) {
 251       GetMainFrame()->loadHTMLString(contents, base_url);
 252     } else {
 253       WebData data(contents.data(), contents.length());
 254
 255       // Do not use WebFrame.LoadHTMLString because it assumes that input
 256       // html contents use UTF-8 encoding.
 257       // TODO(darin): This should use WebFrame::loadData.
 258       WebFrame* web_frame = GetMainFrame();
 259
 260       ASSERT_TRUE(web_frame != NULL);
 261
 262       web_frame->loadData(data, "text/html", encoding_info, base_url);
 263     }
 264
 265     runner->Run();
 266   }
 267
 268   // Serialize page DOM according to specific page URL. The parameter
 269   // recursive_serialization indicates whether we will serialize all
 270   // sub-frames.
 271   void SerializeDomForURL(const GURL& page_url,
 272                           bool recursive_serialization) {
 273     // Find corresponding WebFrame according to page_url.
 274     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), page_url);
 275     ASSERT_TRUE(web_frame != NULL);
 276     WebVector<WebURL> links;
 277     links.assign(&page_url, 1);
 278     WebString file_path =
 279         base::FilePath(FILE_PATH_LITERAL("c:\\dummy.htm")).AsUTF16Unsafe();
 280     WebVector<WebString> local_paths;
 281     local_paths.assign(&file_path, 1);
 282     // Start serializing DOM.
 283     bool result = WebPageSerializer::serialize(web_frame->toWebLocalFrame(),
 284        recursive_serialization,
 285        static_cast<WebPageSerializerClient*>(this),
 286        links,
 287        local_paths,
 288        local_directory_name_.AsUTF16Unsafe());
 289     ASSERT_TRUE(result);
 290     ASSERT_TRUE(serialized_);
 291   }
 292
 293   void SerializeHTMLDOMWithDocTypeOnRenderer(const GURL& file_url) {
 294     // Make sure original contents have document type.
 295     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 296     ASSERT_TRUE(web_frame != NULL);
 297     WebDocument doc = web_frame->document();
 298     ASSERT_TRUE(HasDocType(doc));
 299     // Do serialization.
 300     SerializeDomForURL(file_url, false);
 301     // Load the serialized contents.
 302     ASSERT_TRUE(HasSerializedFrame(file_url));
 303     const std::string& serialized_contents =
 304         GetSerializedContentForFrame(file_url);
 305     LoadContents(serialized_contents, file_url,
 306                  web_frame->document().encoding());
 307     // Make sure serialized contents still have document type.
 308     web_frame = GetMainFrame();
 309     doc = web_frame->document();
 310     ASSERT_TRUE(HasDocType(doc));
 311   }
 312
 313   void SerializeHTMLDOMWithoutDocTypeOnRenderer(const GURL& file_url) {
 314     // Make sure original contents do not have document type.
 315     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 316     ASSERT_TRUE(web_frame != NULL);
 317     WebDocument doc = web_frame->document();
 318     ASSERT_TRUE(!HasDocType(doc));
 319     // Do serialization.
 320     SerializeDomForURL(file_url, false);
 321     // Load the serialized contents.
 322     ASSERT_TRUE(HasSerializedFrame(file_url));
 323     const std::string& serialized_contents =
 324         GetSerializedContentForFrame(file_url);
 325     LoadContents(serialized_contents, file_url,
 326                  web_frame->document().encoding());
 327     // Make sure serialized contents do not have document type.
 328     web_frame = GetMainFrame();
 329     doc = web_frame->document();
 330     ASSERT_TRUE(!HasDocType(doc));
 331   }
 332
 333   void SerializeXMLDocWithBuiltInEntitiesOnRenderer(
 334       const GURL& xml_file_url, const std::string& original_contents) {
 335     // Do serialization.
 336     SerializeDomForURL(xml_file_url, false);
 337     // Compare the serialized contents with original contents.
 338     ASSERT_TRUE(HasSerializedFrame(xml_file_url));
 339     const std::string& serialized_contents =
 340         GetSerializedContentForFrame(xml_file_url);
 341     ASSERT_EQ(original_contents, serialized_contents);
 342   }
 343
 344   void SerializeHTMLDOMWithAddingMOTWOnRenderer(
 345       const GURL& file_url, const std::string& original_contents) {
 346     // Make sure original contents does not have MOTW;
 347     std::string motw_declaration =
 348        WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 349     ASSERT_FALSE(motw_declaration.empty());
 350     // The encoding of original contents is ISO-8859-1, so we convert the MOTW
 351     // declaration to ASCII and search whether original contents has it or not.
 352     ASSERT_TRUE(std::string::npos == original_contents.find(motw_declaration));
 353
 354     // Do serialization.
 355     SerializeDomForURL(file_url, false);
 356     // Make sure the serialized contents have MOTW ;
 357     ASSERT_TRUE(HasSerializedFrame(file_url));
 358     const std::string& serialized_contents =
 359         GetSerializedContentForFrame(file_url);
 360     ASSERT_FALSE(std::string::npos ==
 361         serialized_contents.find(motw_declaration));
 362   }
 363
 364   void SerializeHTMLDOMWithNoMetaCharsetInOriginalDocOnRenderer(
 365       const GURL& file_url) {
 366     // Make sure there is no META charset declaration in original document.
 367     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 368     ASSERT_TRUE(web_frame != NULL);
 369     WebDocument doc = web_frame->document();
 370     ASSERT_TRUE(doc.isHTMLDocument());
 371     WebElement head_element = doc.head();
 372     ASSERT_TRUE(!head_element.isNull());
 373     // Go through all children of HEAD element.
 374     for (WebNode child = head_element.firstChild(); !child.isNull();
 375          child = child.nextSibling()) {
 376       std::string charset_info;
 377       if (IsMetaElement(child, charset_info))
 378         ASSERT_TRUE(charset_info.empty());
 379     }
 380     // Do serialization.
 381     SerializeDomForURL(file_url, false);
 382
 383     // Load the serialized contents.
 384     ASSERT_TRUE(HasSerializedFrame(file_url));
 385     const std::string& serialized_contents =
 386         GetSerializedContentForFrame(file_url);
 387     LoadContents(serialized_contents, file_url,
 388                  web_frame->document().encoding());
 389     // Make sure the first child of HEAD element is META which has charset
 390     // declaration in serialized contents.
 391     web_frame = GetMainFrame();
 392     ASSERT_TRUE(web_frame != NULL);
 393     doc = web_frame->document();
 394     ASSERT_TRUE(doc.isHTMLDocument());
 395     head_element = doc.head();
 396     ASSERT_TRUE(!head_element.isNull());
 397     WebNode meta_node = head_element.firstChild();
 398     ASSERT_TRUE(!meta_node.isNull());
 399     // Get meta charset info.
 400     std::string charset_info2;
 401     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 402     ASSERT_TRUE(!charset_info2.empty());
 403     ASSERT_EQ(charset_info2,
 404               std::string(web_frame->document().encoding().utf8()));
 405
 406     // Make sure no more additional META tags which have charset declaration.
 407     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 408          child = child.nextSibling()) {
 409       std::string charset_info;
 410       if (IsMetaElement(child, charset_info))
 411         ASSERT_TRUE(charset_info.empty());
 412     }
 413   }
 414
 415   void SerializeHTMLDOMWithMultipleMetaCharsetInOriginalDocOnRenderer(
 416       const GURL& file_url) {
 417     // Make sure there are multiple META charset declarations in original
 418     // document.
 419     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 420     ASSERT_TRUE(web_frame != NULL);
 421     WebDocument doc = web_frame->document();
 422     ASSERT_TRUE(doc.isHTMLDocument());
 423     WebElement head_ele = doc.head();
 424     ASSERT_TRUE(!head_ele.isNull());
 425     // Go through all children of HEAD element.
 426     int charset_declaration_count = 0;
 427     for (WebNode child = head_ele.firstChild(); !child.isNull();
 428          child = child.nextSibling()) {
 429       std::string charset_info;
 430       if (IsMetaElement(child, charset_info) && !charset_info.empty())
 431         charset_declaration_count++;
 432     }
 433     // The original doc has more than META tags which have charset declaration.
 434     ASSERT_TRUE(charset_declaration_count > 1);
 435
 436     // Do serialization.
 437     SerializeDomForURL(file_url, false);
 438
 439     // Load the serialized contents.
 440     ASSERT_TRUE(HasSerializedFrame(file_url));
 441     const std::string& serialized_contents =
 442         GetSerializedContentForFrame(file_url);
 443     LoadContents(serialized_contents, file_url,
 444                  web_frame->document().encoding());
 445     // Make sure only first child of HEAD element is META which has charset
 446     // declaration in serialized contents.
 447     web_frame = GetMainFrame();
 448     ASSERT_TRUE(web_frame != NULL);
 449     doc = web_frame->document();
 450     ASSERT_TRUE(doc.isHTMLDocument());
 451     head_ele = doc.head();
 452     ASSERT_TRUE(!head_ele.isNull());
 453     WebNode meta_node = head_ele.firstChild();
 454     ASSERT_TRUE(!meta_node.isNull());
 455     // Get meta charset info.
 456     std::string charset_info2;
 457     ASSERT_TRUE(IsMetaElement(meta_node, charset_info2));
 458     ASSERT_TRUE(!charset_info2.empty());
 459     ASSERT_EQ(charset_info2,
 460               std::string(web_frame->document().encoding().utf8()));
 461
 462     // Make sure no more additional META tags which have charset declaration.
 463     for (WebNode child = meta_node.nextSibling(); !child.isNull();
 464          child = child.nextSibling()) {
 465       std::string charset_info;
 466       if (IsMetaElement(child, charset_info))
 467         ASSERT_TRUE(charset_info.empty());
 468     }
 469   }
 470
 471   void SerializeHTMLDOMWithEntitiesInTextOnRenderer() {
 472     base::FilePath page_file_path = GetTestFilePath(
 473         "dom_serializer", "dom_serializer/htmlentities_in_text.htm");
 474     // Get file URL. The URL is dummy URL to identify the following loading
 475     // actions. The test content is in constant:original_contents.
 476     GURL file_url = net::FilePathToFileURL(page_file_path);
 477     ASSERT_TRUE(file_url.SchemeIsFile());
 478     // Test contents.
 479     static const char* const original_contents =
 480         "<html><body>&amp;&lt;&gt;\"\'</body></html>";
 481     // Load the test contents.
 482     LoadContents(original_contents, file_url, WebString());
 483
 484     // Get BODY's text content in DOM.
 485     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 486     ASSERT_TRUE(web_frame != NULL);
 487     WebDocument doc = web_frame->document();
 488     ASSERT_TRUE(doc.isHTMLDocument());
 489     WebElement body_ele = doc.body();
 490     ASSERT_TRUE(!body_ele.isNull());
 491     WebNode text_node = body_ele.firstChild();
 492     ASSERT_TRUE(text_node.isTextNode());
 493     ASSERT_TRUE(std::string(text_node.createMarkup().utf8()) ==
 494                 "&amp;&lt;&gt;\"\'");
 495     // Do serialization.
 496     SerializeDomForURL(file_url, false);
 497     // Compare the serialized contents with original contents.
 498     ASSERT_TRUE(HasSerializedFrame(file_url));
 499     const std::string& serialized_contents =
 500         GetSerializedContentForFrame(file_url);
 501     // Compare the serialized contents with original contents to make sure
 502     // they are same.
 503     // Because we add MOTW when serializing DOM, so before comparison, we also
 504     // need to add MOTW to original_contents.
 505     std::string original_str =
 506       WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 507     original_str += original_contents;
 508     // Since WebCore now inserts a new HEAD element if there is no HEAD element
 509     // when creating BODY element. (Please see
 510     // HTMLParser::bodyCreateErrorCheck.) We need to append the HEAD content and
 511     // corresponding META content if we find WebCore-generated HEAD element.
 512     if (!doc.head().isNull()) {
 513       WebString encoding = web_frame->document().encoding();
 514       std::string htmlTag("<html>");
 515       std::string::size_type pos = original_str.find(htmlTag);
 516       ASSERT_NE(std::string::npos, pos);
 517       pos += htmlTag.length();
 518       std::string head_part("<head>");
 519       head_part +=
 520           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 521       head_part += "</head>";
 522       original_str.insert(pos, head_part);
 523     }
 524     ASSERT_EQ(original_str, serialized_contents);
 525   }
 526
 527   void SerializeHTMLDOMWithEntitiesInAttributeValueOnRenderer() {
 528     base::FilePath page_file_path = GetTestFilePath(
 529         "dom_serializer", "dom_serializer/htmlentities_in_attribute_value.htm");
 530     // Get file URL. The URL is dummy URL to identify the following loading
 531     // actions. The test content is in constant:original_contents.
 532     GURL file_url = net::FilePathToFileURL(page_file_path);
 533     ASSERT_TRUE(file_url.SchemeIsFile());
 534     // Test contents.
 535     static const char* const original_contents =
 536         "<html><body title=\"&amp;&lt;&gt;&quot;&#39;\"></body></html>";
 537     // Load the test contents.
 538     LoadContents(original_contents, file_url, WebString());
 539     // Get value of BODY's title attribute in DOM.
 540     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 541     ASSERT_TRUE(web_frame != NULL);
 542     WebDocument doc = web_frame->document();
 543     ASSERT_TRUE(doc.isHTMLDocument());
 544     WebElement body_ele = doc.body();
 545     ASSERT_TRUE(!body_ele.isNull());
 546     WebString value = body_ele.getAttribute("title");
 547     ASSERT_TRUE(std::string(value.utf8()) == "&<>\"\'");
 548     // Do serialization.
 549     SerializeDomForURL(file_url, false);
 550     // Compare the serialized contents with original contents.
 551     ASSERT_TRUE(HasSerializedFrame(file_url));
 552     const std::string& serialized_contents =
 553         GetSerializedContentForFrame(file_url);
 554     // Compare the serialized contents with original contents to make sure
 555     // they are same.
 556     std::string original_str =
 557         WebPageSerializer::generateMarkOfTheWebDeclaration(file_url).utf8();
 558     original_str += original_contents;
 559     if (!doc.isNull()) {
 560       WebString encoding = web_frame->document().encoding();
 561       std::string htmlTag("<html>");
 562       std::string::size_type pos = original_str.find(htmlTag);
 563       ASSERT_NE(std::string::npos, pos);
 564       pos += htmlTag.length();
 565       std::string head_part("<head>");
 566       head_part +=
 567           WebPageSerializer::generateMetaCharsetDeclaration(encoding).utf8();
 568       head_part += "</head>";
 569       original_str.insert(pos, head_part);
 570     }
 571     ASSERT_EQ(original_str, serialized_contents);
 572   }
 573
 574   void SerializeHTMLDOMWithNonStandardEntitiesOnRenderer(const GURL& file_url) {
 575     // Get value of BODY's title attribute in DOM.
 576     WebFrame* web_frame = FindSubFrameByURL(GetWebView(), file_url);
 577     WebDocument doc = web_frame->document();
 578     ASSERT_TRUE(doc.isHTMLDocument());
 579     WebElement body_element = doc.body();
 580     // Unescaped string for "&percnt;&nsup;&sup1;&apos;".
 581     static const wchar_t parsed_value[] = {
 582       '%', 0x2285, 0x00b9, '\'', 0
 583     };
 584     WebString value = body_element.getAttribute("title");
 585     WebString content = doc.contentAsTextForTesting();
 586     ASSERT_TRUE(base::UTF16ToWide(value) == parsed_value);
 587     ASSERT_TRUE(base::UTF16ToWide(content) == 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
 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