Add a utility class for extracting details of the DOM

headless/public/util/dom_tree_extractor_browsertest.cc

+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "headless/public/util/dom_tree_extractor.h"
+
+#include <memory>
+#include "base/json/json_writer.h"
+#include "base/strings/string_util.h"
+#include "content/public/browser/render_widget_host_view.h"
+#include "content/public/browser/web_contents.h"
+#include "content/public/test/browser_test.h"
+#include "headless/lib/browser/headless_web_contents_impl.h"
+#include "headless/public/domains/browser.h"
+#include "headless/public/domains/emulation.h"
+#include "headless/public/domains/network.h"
+#include "headless/public/domains/page.h"
+#include "headless/public/headless_browser.h"
+#include "headless/public/headless_devtools_client.h"
+#include "headless/public/headless_devtools_target.h"
+#include "headless/test/headless_browser_test.h"
+#include "testing/gtest/include/gtest/gtest.h"
+#include "url/gurl.h"
+
+namespace headless {
+
+class DomTreeExtractorBrowserTest : public HeadlessAsyncDevTooledBrowserTest,
+                                    public page::Observer {
+ public:
+  void RunDevTooledTest() override {
+    EXPECT_TRUE(embedded_test_server()->Start());
+    devtools_client_->GetPage()->AddObserver(this);
+    devtools_client_->GetPage()->Enable();
+    devtools_client_->GetPage()->Navigate(
+        embedded_test_server()->GetURL("/dom_tree_test.html").spec());
+  }
+
+  void OnLoadEventFired(const page::LoadEventFiredParams& params) override {
+    devtools_client_->GetPage()->RemoveObserver(this);
+
+    extractor_.reset(new DomTreeExtractor(devtools_client_.get()));
+    extractor_->ExtractDom(
+        base::Bind(&DomTreeExtractorBrowserTest::OnDomTreeExtracted,
+                   base::Unretained(this)));
+  }
+
+  void OnDomTreeExtracted(
+      std::vector<std::unique_ptr<base::DictionaryValue>> dom_tree) {
+    std::vector<std::string> expected_results = {
+        "{'baseURL':'http://127.0.0.1/dom_tree_test.html','childIndicies':[1],"
+        "'documentURL':'http://127.0.0.1/dom_tree_test.html','localName':'',"
+        "'nodeId':1,'nodeName':'#document','nodeType':9,'nodeValue':'',"
+        "'xmlVersion':''}",
+
+        "{'attributes':[],'childIndicies':[2,5],'frameId':'?','localName'"
+        ":'html','nodeId':2,'nodeName':'HTML','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':[],'boundingBox':{'height':600.0,'width':800.0,'x':0.0,'"
+        "y':0.0},'childIndicies':[3],'localName':'head','nodeId':3,'nodeName':'"
+        "HEAD','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':[],'boundingBox':{'height':584.0,'width':784.0,'x':8.0,'"
+        "y':8.0},'childIndicies':[4],'localName':'title','nodeId':5,'nodeName':"
+        "'TITLE','nodeType':1,'nodeValue':''}",
+
+        "{'boundingBox':{'height':226.0,'width':784.0,'x':8.0,'y':8.0},'"
+        "localName':'','nodeId':6,'nodeName':'#text','nodeType':3,'nodeValue':'"
+        "Hello world!'}",
+
+        "{'attributes':[],'boundingBox':{'height':600.0,'width':800.0,'x':0.0,'"
+        "y':0.0},'childIndicies':[6],'localName':'body','nodeId':4,'nodeName':'"
+        "BODY','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':['id','id1'],'boundingBox':{'height':17.0,'width':73.0,'"
+        "x':8.0,'y':198.0},'childIndicies':[7,8,15],'inlineTextNodes':[{'"
+        "boundingBox':{'height':17.0,'width':72.4375,'x':8.0,'y':198.0},'"
+        "numCharacters':11,'startCharacterIndex':1}],'layoutText':'\\nSome "
+        "text.\\n','localName':'div','nodeId':7,'nodeName':'DIV','nodeType':1,'"
+        "nodeValue':''}",
+
+        "{'boundingBox':{'height':200.0,'width':400.0,'x':82.0,'y':10.0},'"
+        "localName':'','nodeId':8,'nodeName':'#text','nodeType':3,'nodeValue':'"
+        "\\nSome text.\\n'}",
+
+        "{'attributes':['src','/iframe.html','width','400','height','200'],"
+        "'boundingBox':{'height':171.0,'width':384.0,'x':90.0,'y':18.0},"
+        "'childIndicies':[],'contentDocumentIndex':9,'frameId':'?',"
+        "'localName':'iframe','nodeId':9,'nodeName':'IFRAME','nodeType':1,"
+        "'nodeValue':''}",
+
+        "{'baseURL':'http://127.0.0.1/iframe.html','boundingBox':{"
+        "'height':37.0,'width':384.0,'x':90.0,'y':18.0},'childIndicies':[10],"
+        "'documentURL':'http://127.0.0.1/iframe.html','localName':'',"
+        "'nodeId':10,'nodeName':'#document','nodeType':9,'nodeValue':'',"
+        "'xmlVersion':''}",
+
+        "{'attributes':[],'boundingBox':{'height':36.0,'width':308.0,'x':8.0,'"
+        "y':8.0},'childIndicies':[11,12],'frameId':'?','inlineTextNodes':"
+        "[{'boundingBox':{'height':36.0,'width':307.734375,'x':8.0,'y':8.0},'"
+        "numCharacters':22,'startCharacterIndex':0}],'layoutText':'Hello from "
+        "the iframe!','localName':'html','nodeId':11,'nodeName':'HTML',"
+        "'nodeType':1,'nodeValue':''}",
+
+        "{'attributes':[],'boundingBox':{'height':204.0,'width':405.0,'x':80.0,"
+        "'y':8.0},'childIndicies':[],'localName':'head','nodeId':12,'nodeName':"
+        "'HEAD','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':[],'boundingBox':{'height':0.0,'width':0.0,'x':0.0,'y':"
+        "0.0},'childIndicies':[13],'layoutText':'\\n','localName':'body','"
+        "nodeId':13,'nodeName':'BODY','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':[],'boundingBox':{'height':18.0,'width':784.0,'x':8.0,'"
+        "y':216.0},'childIndicies':[14],'localName':'h1','nodeId':14,'nodeName'"
+        ":'H1','nodeType':1,'nodeValue':''}",
+
+        "{'boundingBox':{'height':18.0,'width':784.0,'x':8.0,'y':216.0},'"
+        "localName':'','nodeId':15,'nodeName':'#text','nodeType':3,'nodeValue':"
+        "'Hello from the iframe!'}",
+
+        "{'attributes':['id','id2'],'boundingBox':{'height':18.0,'width':784.0,"
+        "'x':8.0,'y':216.0},'childIndicies':[16],'localName':'div','nodeId':16,"
+        "'nodeName':'DIV','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':['id','id3'],'boundingBox':{'height':17.0,'width':53.0,'"
+        "x':8.0,'y':216.0},'childIndicies':[17],'localName':'div','nodeId':17,'"
+        "nodeName':'DIV','nodeType':1,'nodeValue':''}",
+
+        "{'attributes':['id','id4'],'boundingBox':{'height':17.0,'width':53.0,'"
+        "x':8.0,'y':216.0},'childIndicies':[18],'inlineTextNodes':[{'"
+        "boundingBox':{'height':17.0,'width':52.421875,'x':8.0,'y':216.0},'"
+        "numCharacters':7,'startCharacterIndex':0}],'layoutText':'Google!','"
+        "localName':'div','nodeId':18,'nodeName':'DIV','nodeType':1,'nodeValue'"
+        ":''}",
+
+        "{'attributes':['href','https://www.google.com'],'boundingBox':{"
+        "'height':0.0,'width':0.0,'x':0.0,'y':0.0},'childIndicies':[19],"
+        "'layoutText':'\\n    ','localName':'a','nodeId':19,'nodeName':'A',"
+        "'nodeType':1,'nodeValue':''}",
+
+        "{'localName':'','nodeId':20,'nodeName':'#text','nodeType':3,'"
+        "nodeValue':'Google!'}"};
+
+    GURL::Replacements replace_port;
+    replace_port.SetPortStr("");
+    for (size_t i = 0; i < dom_tree.size(); i++) {
+      // Frame IDs are random.
+      if (dom_tree[i]->HasKey("frameId"))
+        dom_tree[i]->SetString("frameId", "?");
+
+      // Ports are random.
+      std::string url;
+      if (dom_tree[i]->GetString("baseURL", &url)) {
+        dom_tree[i]->SetString(
+            "baseURL", GURL(url).ReplaceComponents(replace_port).spec());
+      }
+
+      if (dom_tree[i]->GetString("documentURL", &url)) {
+        dom_tree[i]->SetString(
+            "documentURL", GURL(url).ReplaceComponents(replace_port).spec());
+      }
+
+      std::string result_json;
+      base::JSONWriter::Write(*dom_tree[i], &result_json);
+
+      // Not allowed to use C++11 string literals so we make do as best we can.
+      base::ReplaceChars(result_json, "\"", "'", &result_json);
+
+      ASSERT_LT(i, expected_results.size());
+      EXPECT_EQ(expected_results[i], result_json) << " Node # " << i;
+    }
+
+    FinishAsynchronousTest();
+  }
+
+  std::unique_ptr<DomTreeExtractor> extractor_;
+};
+
+HEADLESS_ASYNC_DEVTOOLED_TEST_F(DomTreeExtractorBrowserTest);
+
+}  // namespace headless