content/renderer/savable_resources.cc

   1 // Copyright (c) 2013 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "content/renderer/savable_resources.h"
   6
   7 #include <set>
   8
   9 #include "base/compiler_specific.h"
  10 #include "base/logging.h"
  11 #include "base/strings/string_util.h"
  12 #include "third_party/WebKit/public/platform/WebString.h"
  13 #include "third_party/WebKit/public/platform/WebVector.h"
  14 #include "third_party/WebKit/public/web/WebDocument.h"
  15 #include "third_party/WebKit/public/web/WebElement.h"
  16 #include "third_party/WebKit/public/web/WebElementCollection.h"
  17 #include "third_party/WebKit/public/web/WebInputElement.h"
  18 #include "third_party/WebKit/public/web/WebLocalFrame.h"
  19 #include "third_party/WebKit/public/web/WebNode.h"
  20 #include "third_party/WebKit/public/web/WebNodeList.h"
  21 #include "third_party/WebKit/public/web/WebView.h"
  22
  23 using blink::WebDocument;
  24 using blink::WebElement;
  25 using blink::WebElementCollection;
  26 using blink::WebFrame;
  27 using blink::WebInputElement;
  28 using blink::WebLocalFrame;
  29 using blink::WebNode;
  30 using blink::WebNodeList;
  31 using blink::WebString;
  32 using blink::WebVector;
  33 using blink::WebView;
  34
  35 namespace content {
  36 namespace {
  37
  38 // Get all savable resource links from current element. One element might
  39 // have more than one resource link. It is possible to have some links
  40 // in one CSS stylesheet.
  41 void GetSavableResourceLinkForElement(
  42     const WebElement& element,
  43     const WebDocument& current_doc,
  44     SavableResourcesResult* result) {
  45   // Skipping frame and iframe tag.
  46   if (element.hasHTMLTagName("iframe") || element.hasHTMLTagName("frame"))
  47     return;
  48
  49   // Check whether the node has sub resource URL or not.
  50   WebString value = GetSubResourceLinkFromElement(element);
  51   if (value.isNull())
  52     return;
  53   // Get absolute URL.
  54   GURL u = current_doc.completeURL(value);
  55   // ignore invalid URL
  56   if (!u.is_valid())
  57     return;
  58   // Ignore those URLs which are not standard protocols. Because FTP
  59   // protocol does no have cache mechanism, we will skip all
  60   // sub-resources if they use FTP protocol.
  61   if (!u.SchemeIsHTTPOrHTTPS() && !u.SchemeIs(url::kFileScheme))
  62     return;
  63   // Ignore duplicated resource link.
  64   result->resources_list->push_back(u);
  65   // Insert referrer for above new resource link.
  66   result->referrer_urls_list->push_back(GURL());
  67   result->referrer_policies_list->push_back(blink::WebReferrerPolicyDefault);
  68 }
  69
  70 }  // namespace
  71
  72 bool GetSavableResourceLinksForFrame(WebFrame* current_frame,
  73                                      SavableResourcesResult* result,
  74                                      const char** savable_schemes) {
  75   // Get current frame's URL.
  76   GURL current_frame_url = current_frame->document().url();
  77
  78   // If url of current frame is invalid, ignore it.
  79   if (!current_frame_url.is_valid())
  80     return false;
  81
  82   // If url of current frame is not a savable protocol, ignore it.
  83   bool is_valid_protocol = false;
  84   for (int i = 0; savable_schemes[i] != NULL; ++i) {
  85     if (current_frame_url.SchemeIs(savable_schemes[i])) {
  86       is_valid_protocol = true;
  87       break;
  88     }
  89   }
  90   if (!is_valid_protocol)
  91     return false;
  92
  93   // Get current using document.
  94   WebDocument current_doc = current_frame->document();
  95   // Go through all descent nodes.
  96   WebElementCollection all = current_doc.all();
  97   // Go through all elements in this frame.
  98   for (WebElement element = all.firstItem(); !element.isNull();
  99        element = all.nextItem()) {
 100     GetSavableResourceLinkForElement(element,
 101                                      current_doc,
 102                                      result);
 103   }
 104
 105   return true;
 106 }
 107
 108 WebString GetSubResourceLinkFromElement(const WebElement& element) {
 109   const char* attribute_name = NULL;
 110   if (element.hasHTMLTagName("img") ||
 111       element.hasHTMLTagName("script")) {
 112     attribute_name = "src";
 113   } else if (element.hasHTMLTagName("input")) {
 114     const WebInputElement input = element.toConst<WebInputElement>();
 115     if (input.isImageButton()) {
 116       attribute_name = "src";
 117     }
 118   } else if (element.hasHTMLTagName("body") ||
 119              element.hasHTMLTagName("table") ||
 120              element.hasHTMLTagName("tr") ||
 121              element.hasHTMLTagName("td")) {
 122     attribute_name = "background";
 123   } else if (element.hasHTMLTagName("blockquote") ||
 124              element.hasHTMLTagName("q") ||
 125              element.hasHTMLTagName("del") ||
 126              element.hasHTMLTagName("ins")) {
 127     attribute_name = "cite";
 128   } else if (element.hasHTMLTagName("link")) {
 129     // If the link element is not linked to css, ignore it.
 130     if (base::LowerCaseEqualsASCII(
 131             base::StringPiece16(element.getAttribute("type")), "text/css") ||
 132         base::LowerCaseEqualsASCII(
 133             base::StringPiece16(element.getAttribute("rel")), "stylesheet")) {
 134       // TODO(jnd): Add support for extracting links of sub-resources which
 135       // are inside style-sheet such as @import, url(), etc.
 136       // See bug: http://b/issue?id=1111667.
 137       attribute_name = "href";
 138     }
 139   }
 140   if (!attribute_name)
 141     return WebString();
 142   WebString value = element.getAttribute(WebString::fromUTF8(attribute_name));
 143   // If value has content and not start with "javascript:" then return it,
 144   // otherwise return NULL.
 145   if (!value.isNull() && !value.isEmpty() &&
 146       !base::StartsWith(value.utf8(), "javascript:",
 147                         base::CompareCase::INSENSITIVE_ASCII))
 148     return value;
 149
 150   return WebString();
 151 }
 152
 153 }  // namespace content