src/chrome/browser/safe_browsing/safe_browsing_test.cc

   1 // Copyright (c) 2012 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 // This test uses the safebrowsing test server published at
   6 // http://code.google.com/p/google-safe-browsing/ to test the safebrowsing
   7 // protocol implemetation. Details of the safebrowsing testing flow is
   8 // documented at
   9 // http://code.google.com/p/google-safe-browsing/wiki/ProtocolTesting
  10 //
  11 // This test launches safebrowsing test server and issues several update
  12 // requests against that server. Each update would get different data and after
  13 // each update, the test will get a list of URLs from the test server to verify
  14 // its repository. The test will succeed only if all updates are performed and
  15 // URLs match what the server expected.
  16
  17 #include <vector>
  18
  19 #include "base/bind.h"
  20 #include "base/command_line.h"
  21 #include "base/environment.h"
  22 #include "base/path_service.h"
  23 #include "base/strings/string_number_conversions.h"
  24 #include "base/strings/string_split.h"
  25 #include "base/strings/stringprintf.h"
  26 #include "base/strings/utf_string_conversions.h"
  27 #include "base/synchronization/lock.h"
  28 #include "base/test/test_timeouts.h"
  29 #include "base/threading/platform_thread.h"
  30 #include "base/threading/thread.h"
  31 #include "base/time/time.h"
  32 #include "chrome/browser/browser_process.h"
  33 #include "chrome/browser/chrome_notification_types.h"
  34 #include "chrome/browser/profiles/profile.h"
  35 #include "chrome/browser/safe_browsing/database_manager.h"
  36 #include "chrome/browser/safe_browsing/local_safebrowsing_test_server.h"
  37 #include "chrome/browser/safe_browsing/protocol_manager.h"
  38 #include "chrome/browser/safe_browsing/safe_browsing_service.h"
  39 #include "chrome/browser/ui/browser.h"
  40 #include "chrome/common/chrome_switches.h"
  41 #include "chrome/common/url_constants.h"
  42 #include "chrome/test/base/in_process_browser_test.h"
  43 #include "chrome/test/base/ui_test_utils.h"
  44 #include "content/public/browser/browser_context.h"
  45 #include "content/public/test/test_browser_thread.h"
  46 #include "net/base/load_flags.h"
  47 #include "net/base/net_log.h"
  48 #include "net/dns/host_resolver.h"
  49 #include "net/test/python_utils.h"
  50 #include "net/url_request/url_fetcher.h"
  51 #include "net/url_request/url_fetcher_delegate.h"
  52 #include "net/url_request/url_request_status.h"
  53 #include "testing/gtest/include/gtest/gtest.h"
  54
  55 using content::BrowserThread;
  56
  57 namespace {
  58
  59 const base::FilePath::CharType kDataFile[] =
  60     FILE_PATH_LITERAL("testing_input_nomac.dat");
  61 const char kUrlVerifyPath[] = "safebrowsing/verify_urls";
  62 const char kDBVerifyPath[] = "safebrowsing/verify_database";
  63 const char kTestCompletePath[] = "test_complete";
  64
  65 struct PhishingUrl {
  66   std::string url;
  67   std::string list_name;
  68   bool is_phishing;
  69 };
  70
  71 // Parses server response for verify_urls. The expected format is:
  72 //
  73 // first.random.url.com/   internal-test-shavar   yes
  74 // second.random.url.com/  internal-test-shavar   yes
  75 // ...
  76 bool ParsePhishingUrls(const std::string& data,
  77                        std::vector<PhishingUrl>* phishing_urls) {
  78   if (data.empty())
  79     return false;
  80
  81   std::vector<std::string> urls;
  82   base::SplitString(data, '\n', &urls);
  83   for (size_t i = 0; i < urls.size(); ++i) {
  84     if (urls[i].empty())
  85       continue;
  86     PhishingUrl phishing_url;
  87     std::vector<std::string> record_parts;
  88     base::SplitString(urls[i], '\t', &record_parts);
  89     if (record_parts.size() != 3) {
  90       LOG(ERROR) << "Unexpected URL format in phishing URL list: "
  91                  << urls[i];
  92       return false;
  93     }
  94     phishing_url.url = std::string(content::kHttpScheme) +
  95         "://" + record_parts[0];
  96     phishing_url.list_name = record_parts[1];
  97     if (record_parts[2] == "yes") {
  98       phishing_url.is_phishing = true;
  99     } else if (record_parts[2] == "no") {
 100       phishing_url.is_phishing = false;
 101     } else {
 102       LOG(ERROR) << "Unrecognized expectation in " << urls[i]
 103                  << ": " << record_parts[2];
 104       return false;
 105     }
 106     phishing_urls->push_back(phishing_url);
 107   }
 108   return true;
 109 }
 110
 111 }  // namespace
 112
 113 // This starts the browser and keeps status of states related to SafeBrowsing.
 114 class SafeBrowsingServerTest : public InProcessBrowserTest {
 115  public:
 116   SafeBrowsingServerTest()
 117     : safe_browsing_service_(NULL),
 118       is_database_ready_(true),
 119       is_update_scheduled_(false),
 120       is_checked_url_in_db_(false),
 121       is_checked_url_safe_(false) {
 122   }
 123
 124   virtual ~SafeBrowsingServerTest() {
 125   }
 126
 127   void UpdateSafeBrowsingStatus() {
 128     ASSERT_TRUE(safe_browsing_service_);
 129     base::AutoLock lock(update_status_mutex_);
 130     last_update_ = safe_browsing_service_->protocol_manager_->last_update();
 131     is_update_scheduled_ =
 132         safe_browsing_service_->protocol_manager_->update_timer_.IsRunning();
 133   }
 134
 135   void ForceUpdate() {
 136     content::WindowedNotificationObserver observer(
 137         chrome::NOTIFICATION_SAFE_BROWSING_UPDATE_COMPLETE,
 138         content::Source<SafeBrowsingDatabaseManager>(database_manager()));
 139     BrowserThread::PostTask(BrowserThread::IO, FROM_HERE,
 140         base::Bind(&SafeBrowsingServerTest::ForceUpdateOnIOThread,
 141                    this));
 142     observer.Wait();
 143   }
 144
 145   void ForceUpdateOnIOThread() {
 146     EXPECT_TRUE(BrowserThread::CurrentlyOn(BrowserThread::IO));
 147     ASSERT_TRUE(safe_browsing_service_);
 148     safe_browsing_service_->protocol_manager_->ForceScheduleNextUpdate(
 149         base::TimeDelta::FromSeconds(0));
 150   }
 151
 152   void CheckIsDatabaseReady() {
 153     base::AutoLock lock(update_status_mutex_);
 154     is_database_ready_ = !database_manager()->database_update_in_progress_;
 155   }
 156
 157   void CheckUrl(SafeBrowsingDatabaseManager::Client* helper, const GURL& url) {
 158     ASSERT_TRUE(safe_browsing_service_);
 159     base::AutoLock lock(update_status_mutex_);
 160     if (database_manager()->CheckBrowseUrl(url, helper)) {
 161       is_checked_url_in_db_ = false;
 162       is_checked_url_safe_ = true;
 163     } else {
 164       // In this case, Safebrowsing service will fetch the full hash
 165       // from the server and examine that. Once it is done,
 166       // set_is_checked_url_safe() will be called via callback.
 167       is_checked_url_in_db_ = true;
 168     }
 169   }
 170
 171   SafeBrowsingDatabaseManager* database_manager() {
 172     return safe_browsing_service_->database_manager().get();
 173   }
 174
 175   bool is_checked_url_in_db() {
 176     base::AutoLock l(update_status_mutex_);
 177     return is_checked_url_in_db_;
 178   }
 179
 180   void set_is_checked_url_safe(bool safe) {
 181     base::AutoLock l(update_status_mutex_);
 182     is_checked_url_safe_ = safe;
 183   }
 184
 185   bool is_checked_url_safe() {
 186     base::AutoLock l(update_status_mutex_);
 187     return is_checked_url_safe_;
 188   }
 189
 190   bool is_database_ready() {
 191     base::AutoLock l(update_status_mutex_);
 192     return is_database_ready_;
 193   }
 194
 195   base::Time last_update() {
 196     base::AutoLock l(update_status_mutex_);
 197     return last_update_;
 198   }
 199
 200   bool is_update_scheduled() {
 201     base::AutoLock l(update_status_mutex_);
 202     return is_update_scheduled_;
 203   }
 204
 205   base::MessageLoop* SafeBrowsingMessageLoop() {
 206     return database_manager()->safe_browsing_thread_->message_loop();
 207   }
 208
 209   const net::SpawnedTestServer& test_server() const {
 210     return *test_server_;
 211   }
 212
 213  protected:
 214   bool InitSafeBrowsingService() {
 215     safe_browsing_service_ = g_browser_process->safe_browsing_service();
 216     return safe_browsing_service_ != NULL;
 217   }
 218
 219   virtual void SetUpCommandLine(CommandLine* command_line) OVERRIDE {
 220     base::FilePath datafile_path;
 221     ASSERT_TRUE(PathService::Get(base::DIR_SOURCE_ROOT, &datafile_path));
 222
 223     datafile_path = datafile_path.Append(FILE_PATH_LITERAL("third_party"))
 224           .Append(FILE_PATH_LITERAL("safe_browsing"))
 225           .Append(FILE_PATH_LITERAL("testing"))
 226           .Append(kDataFile);
 227     test_server_.reset(new LocalSafeBrowsingTestServer(datafile_path));
 228     ASSERT_TRUE(test_server_->Start());
 229     LOG(INFO) << "server is " << test_server_->host_port_pair().ToString();
 230
 231     // Makes sure the auto update is not triggered. This test will force the
 232     // update when needed.
 233     command_line->AppendSwitch(switches::kSbDisableAutoUpdate);
 234
 235     // This test uses loopback. No need to use IPv6 especially it makes
 236     // local requests slow on Windows trybot when ipv6 local address [::1]
 237     // is not setup.
 238     command_line->AppendSwitch(switches::kDisableIPv6);
 239
 240     // TODO(lzheng): The test server does not understand download related
 241     // requests. We need to fix the server.
 242     command_line->AppendSwitch(switches::kSbDisableDownloadProtection);
 243
 244     // TODO(gcasto): Generate new testing data that includes the
 245     // client-side phishing whitelist.
 246     command_line->AppendSwitch(
 247         switches::kDisableClientSidePhishingDetection);
 248
 249     // TODO(kalman): Generate new testing data that includes the extension
 250     // blacklist.
 251     command_line->AppendSwitch(switches::kSbDisableExtensionBlacklist);
 252
 253     // TODO(tburkard): Generate new testing data that includes the side-effect
 254     // free whitelist.
 255     command_line->AppendSwitch(switches::kSbDisableSideEffectFreeWhitelist);
 256
 257     // Point to the testing server for all SafeBrowsing requests.
 258     std::string url_prefix = test_server_->GetURL("safebrowsing").spec();
 259     command_line->AppendSwitchASCII(switches::kSbURLPrefix, url_prefix);
 260   }
 261
 262   void SetTestStep(int step) {
 263     std::string test_step = base::StringPrintf("test_step=%d", step);
 264     safe_browsing_service_->protocol_manager_->set_additional_query(test_step);
 265   }
 266
 267  private:
 268   SafeBrowsingService* safe_browsing_service_;
 269
 270   scoped_ptr<net::SpawnedTestServer> test_server_;
 271
 272   // Protects all variables below since they are read on UI thread
 273   // but updated on IO thread or safebrowsing thread.
 274   base::Lock update_status_mutex_;
 275
 276   // States associated with safebrowsing service updates.
 277   bool is_database_ready_;
 278   base::Time last_update_;
 279   bool is_update_scheduled_;
 280   // Indicates if there is a match between a URL's prefix and safebrowsing
 281   // database (thus potentially it is a phishing URL).
 282   bool is_checked_url_in_db_;
 283   // True if last verified URL is not a phishing URL and thus it is safe.
 284   bool is_checked_url_safe_;
 285
 286   DISALLOW_COPY_AND_ASSIGN(SafeBrowsingServerTest);
 287 };
 288
 289 // A ref counted helper class that handles callbacks between IO thread and UI
 290 // thread.
 291 class SafeBrowsingServerTestHelper
 292     : public base::RefCountedThreadSafe<SafeBrowsingServerTestHelper>,
 293       public SafeBrowsingDatabaseManager::Client,
 294       public net::URLFetcherDelegate {
 295  public:
 296   SafeBrowsingServerTestHelper(SafeBrowsingServerTest* safe_browsing_test,
 297                                net::URLRequestContextGetter* request_context)
 298       : safe_browsing_test_(safe_browsing_test),
 299         response_status_(net::URLRequestStatus::FAILED),
 300         request_context_(request_context) {
 301   }
 302
 303   // Callbacks for SafeBrowsingDatabaseManager::Client.
 304   virtual void OnCheckBrowseUrlResult(const GURL& url,
 305                                       SBThreatType threat_type) OVERRIDE {
 306     EXPECT_TRUE(BrowserThread::CurrentlyOn(BrowserThread::IO));
 307     EXPECT_TRUE(safe_browsing_test_->is_checked_url_in_db());
 308     safe_browsing_test_->set_is_checked_url_safe(
 309         threat_type == SB_THREAT_TYPE_SAFE);
 310     BrowserThread::PostTask(BrowserThread::UI, FROM_HERE,
 311         base::Bind(&SafeBrowsingServerTestHelper::OnCheckUrlDone, this));
 312   }
 313
 314   virtual void OnBlockingPageComplete(bool proceed) {
 315     NOTREACHED() << "Not implemented.";
 316   }
 317
 318   // Functions and callbacks related to CheckUrl. These are used to verify
 319   // phishing URLs.
 320   void CheckUrl(const GURL& url) {
 321     BrowserThread::PostTask(BrowserThread::IO, FROM_HERE,
 322         base::Bind(&SafeBrowsingServerTestHelper::CheckUrlOnIOThread,
 323                    this, url));
 324     content::RunMessageLoop();
 325   }
 326   void CheckUrlOnIOThread(const GURL& url) {
 327     EXPECT_TRUE(BrowserThread::CurrentlyOn(BrowserThread::IO));
 328     safe_browsing_test_->CheckUrl(this, url);
 329     if (!safe_browsing_test_->is_checked_url_in_db()) {
 330       // Ends the checking since this URL's prefix is not in database.
 331       BrowserThread::PostTask(BrowserThread::UI, FROM_HERE,
 332         base::Bind(&SafeBrowsingServerTestHelper::OnCheckUrlDone, this));
 333     }
 334     // Otherwise, OnCheckUrlDone is called in OnUrlCheckResult since
 335     // safebrowsing service further fetches hashes from safebrowsing server.
 336   }
 337
 338   void OnCheckUrlDone() {
 339     StopUILoop();
 340   }
 341
 342   // Updates status from IO Thread.
 343   void CheckStatusOnIOThread() {
 344     EXPECT_TRUE(BrowserThread::CurrentlyOn(BrowserThread::IO));
 345     safe_browsing_test_->UpdateSafeBrowsingStatus();
 346     safe_browsing_test_->SafeBrowsingMessageLoop()->PostTask(FROM_HERE,
 347         base::Bind(&SafeBrowsingServerTestHelper::CheckIsDatabaseReady, this));
 348   }
 349
 350   // Checks status in SafeBrowsing Thread.
 351   void CheckIsDatabaseReady() {
 352     EXPECT_EQ(base::MessageLoop::current(),
 353               safe_browsing_test_->SafeBrowsingMessageLoop());
 354     safe_browsing_test_->CheckIsDatabaseReady();
 355     BrowserThread::PostTask(BrowserThread::UI, FROM_HERE,
 356         base::Bind(&SafeBrowsingServerTestHelper::OnWaitForStatusUpdateDone,
 357                    this));
 358   }
 359
 360   void OnWaitForStatusUpdateDone() {
 361     StopUILoop();
 362   }
 363
 364   // Update safebrowsing status.
 365   void UpdateStatus() {
 366     BrowserThread::PostTask(
 367         BrowserThread::IO,
 368         FROM_HERE,
 369         base::Bind(&SafeBrowsingServerTestHelper::CheckStatusOnIOThread, this));
 370     // Will continue after OnWaitForStatusUpdateDone().
 371     content::RunMessageLoop();
 372   }
 373
 374   // Calls test server to fetch database for verification.
 375   net::URLRequestStatus::Status FetchDBToVerify(
 376       const net::SpawnedTestServer& test_server,
 377       int test_step) {
 378     // TODO(lzheng): Remove chunk_type=add once it is not needed by the server.
 379     std::string path = base::StringPrintf(
 380         "%s?client=chromium&appver=1.0&pver=2.2&test_step=%d&chunk_type=add",
 381         kDBVerifyPath, test_step);
 382     return FetchUrl(test_server.GetURL(path));
 383   }
 384
 385   // Calls test server to fetch URLs for verification.
 386   net::URLRequestStatus::Status FetchUrlsToVerify(
 387       const net::SpawnedTestServer& test_server,
 388       int test_step) {
 389     std::string path = base::StringPrintf(
 390         "%s?client=chromium&appver=1.0&pver=2.2&test_step=%d",
 391         kUrlVerifyPath, test_step);
 392     return FetchUrl(test_server.GetURL(path));
 393   }
 394
 395   // Calls test server to check if test data is done. E.g.: if there is a
 396   // bad URL that server expects test to fetch full hash but the test didn't,
 397   // this verification will fail.
 398   net::URLRequestStatus::Status VerifyTestComplete(
 399       const net::SpawnedTestServer& test_server,
 400       int test_step) {
 401     std::string path = base::StringPrintf(
 402         "%s?test_step=%d", kTestCompletePath, test_step);
 403     return FetchUrl(test_server.GetURL(path));
 404   }
 405
 406   // Callback for URLFetcher.
 407   virtual void OnURLFetchComplete(const net::URLFetcher* source) OVERRIDE {
 408     source->GetResponseAsString(&response_data_);
 409     response_status_ = source->GetStatus().status();
 410     StopUILoop();
 411   }
 412
 413   const std::string& response_data() {
 414     return response_data_;
 415   }
 416
 417  private:
 418   friend class base::RefCountedThreadSafe<SafeBrowsingServerTestHelper>;
 419   virtual ~SafeBrowsingServerTestHelper() {}
 420
 421   // Stops UI loop after desired status is updated.
 422   void StopUILoop() {
 423     EXPECT_TRUE(BrowserThread::CurrentlyOn(BrowserThread::UI));
 424     base::MessageLoopForUI::current()->Quit();
 425   }
 426
 427   // Fetch a URL. If message_loop_started is true, starts the message loop
 428   // so the caller could wait till OnURLFetchComplete is called.
 429   net::URLRequestStatus::Status FetchUrl(const GURL& url) {
 430     url_fetcher_.reset(net::URLFetcher::Create(
 431         url, net::URLFetcher::GET, this));
 432     url_fetcher_->SetLoadFlags(net::LOAD_DISABLE_CACHE);
 433     url_fetcher_->SetRequestContext(request_context_);
 434     url_fetcher_->Start();
 435     content::RunMessageLoop();
 436     return response_status_;
 437   }
 438
 439   base::OneShotTimer<SafeBrowsingServerTestHelper> check_update_timer_;
 440   SafeBrowsingServerTest* safe_browsing_test_;
 441   scoped_ptr<net::URLFetcher> url_fetcher_;
 442   std::string response_data_;
 443   net::URLRequestStatus::Status response_status_;
 444   net::URLRequestContextGetter* request_context_;
 445   DISALLOW_COPY_AND_ASSIGN(SafeBrowsingServerTestHelper);
 446 };
 447
 448 IN_PROC_BROWSER_TEST_F(SafeBrowsingServerTest,
 449                        SafeBrowsingServerTest) {
 450   LOG(INFO) << "Start test";
 451   ASSERT_TRUE(InitSafeBrowsingService());
 452
 453   net::URLRequestContextGetter* request_context =
 454       browser()->profile()->GetRequestContext();
 455   scoped_refptr<SafeBrowsingServerTestHelper> safe_browsing_helper(
 456       new SafeBrowsingServerTestHelper(this, request_context));
 457   int last_step = 0;
 458
 459   // Waits and makes sure safebrowsing update is not happening.
 460   // The wait will stop once OnWaitForStatusUpdateDone in
 461   // safe_browsing_helper is called and status from safe_browsing_service_
 462   // is checked.
 463   safe_browsing_helper->UpdateStatus();
 464   EXPECT_TRUE(is_database_ready());
 465   EXPECT_FALSE(is_update_scheduled());
 466   EXPECT_TRUE(last_update().is_null());
 467   // Starts updates. After each update, the test will fetch a list of URLs with
 468   // expected results to verify with safebrowsing service. If there is no error,
 469   // the test moves on to the next step to get more update chunks.
 470   // This repeats till there is no update data.
 471   for (int step = 1;; step++) {
 472     // Every step should be a fresh start.
 473     SCOPED_TRACE(base::StringPrintf("step=%d", step));
 474     EXPECT_TRUE(is_database_ready());
 475     EXPECT_FALSE(is_update_scheduled());
 476
 477     // Starts safebrowsing update on IO thread. Waits till scheduled
 478     // update finishes.
 479     base::Time now = base::Time::Now();
 480     SetTestStep(step);
 481     ForceUpdate();
 482
 483     safe_browsing_helper->UpdateStatus();
 484     EXPECT_TRUE(is_database_ready());
 485     EXPECT_FALSE(is_update_scheduled());
 486     EXPECT_FALSE(last_update().is_null());
 487     if (last_update() < now) {
 488       // This means no data available anymore.
 489       break;
 490     }
 491
 492     // Fetches URLs to verify and waits till server responses with data.
 493     EXPECT_EQ(net::URLRequestStatus::SUCCESS,
 494               safe_browsing_helper->FetchUrlsToVerify(test_server(), step));
 495
 496     std::vector<PhishingUrl> phishing_urls;
 497     EXPECT_TRUE(ParsePhishingUrls(safe_browsing_helper->response_data(),
 498                                   &phishing_urls));
 499     EXPECT_GT(phishing_urls.size(), 0U);
 500     for (size_t j = 0; j < phishing_urls.size(); ++j) {
 501       // Verifes with server if a URL is a phishing URL and waits till server
 502       // responses.
 503       safe_browsing_helper->CheckUrl(GURL(phishing_urls[j].url));
 504       if (phishing_urls[j].is_phishing) {
 505         EXPECT_TRUE(is_checked_url_in_db())
 506             << phishing_urls[j].url
 507             << " is_phishing: " << phishing_urls[j].is_phishing
 508             << " test step: " << step;
 509         EXPECT_FALSE(is_checked_url_safe())
 510             << phishing_urls[j].url
 511             << " is_phishing: " << phishing_urls[j].is_phishing
 512             << " test step: " << step;
 513       } else {
 514         EXPECT_TRUE(is_checked_url_safe())
 515             << phishing_urls[j].url
 516             << " is_phishing: " << phishing_urls[j].is_phishing
 517             << " test step: " << step;
 518       }
 519     }
 520     // TODO(lzheng): We should verify the fetched database with local
 521     // database to make sure they match.
 522     EXPECT_EQ(net::URLRequestStatus::SUCCESS,
 523               safe_browsing_helper->FetchDBToVerify(test_server(), step));
 524     EXPECT_GT(safe_browsing_helper->response_data().size(), 0U);
 525     last_step = step;
 526   }
 527
 528   // Verifies with server if test is done and waits till server responses.
 529   EXPECT_EQ(net::URLRequestStatus::SUCCESS,
 530             safe_browsing_helper->VerifyTestComplete(test_server(), last_step));
 531   EXPECT_EQ("yes", safe_browsing_helper->response_data());
 532 }