chrome/browser/supervised_user/experimental/supervised_user_async_url_checker.cc

   1 // Copyright 2014 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 "chrome/browser/supervised_user/experimental/supervised_user_async_url_checker.h"
   6
   7 #include <string>
   8
   9 #include "base/callback.h"
  10 #include "base/json/json_reader.h"
  11 #include "base/metrics/histogram.h"
  12 #include "base/stl_util.h"
  13 #include "base/strings/string_piece.h"
  14 #include "base/strings/string_util.h"
  15 #include "base/strings/stringprintf.h"
  16 #include "base/time/time.h"
  17 #include "base/values.h"
  18 #include "components/google/core/browser/google_util.h"
  19 #include "google_apis/google_api_keys.h"
  20 #include "net/base/escape.h"
  21 #include "net/base/load_flags.h"
  22 #include "net/url_request/url_fetcher.h"
  23 #include "net/url_request/url_request_context.h"
  24 #include "url/url_constants.h"
  25
  26 using net::URLFetcher;
  27 using net::URLFetcherDelegate;
  28 using net::URLRequestContextGetter;
  29 using net::URLRequestStatus;
  30
  31 namespace {
  32
  33 const char kApiUrl[] = "https://safesearch.googleapis.com/v1:classify";
  34 const char kDataContentType[] = "application/x-www-form-urlencoded";
  35 const char kDataFormat[] = "key=%s&urls=%s";
  36
  37 const size_t kDefaultCacheSize = 1000;
  38
  39 // Builds the POST data for SafeSearch API requests.
  40 std::string BuildRequestData(const std::string& api_key, const GURL& url) {
  41   std::string query = net::EscapeQueryParamValue(url.spec(), true);
  42   return base::StringPrintf(kDataFormat, api_key.c_str(), query.c_str());
  43 }
  44
  45 // Creates a URLFetcher to call the SafeSearch API for |url|.
  46 scoped_ptr<net::URLFetcher> CreateFetcher(URLFetcherDelegate* delegate,
  47                                           URLRequestContextGetter* context,
  48                                           const std::string& api_key,
  49                                           const GURL& url) {
  50   scoped_ptr<net::URLFetcher> fetcher = URLFetcher::Create(
  51       0, GURL(kApiUrl), URLFetcher::POST, delegate);
  52   fetcher->SetUploadData(kDataContentType, BuildRequestData(api_key, url));
  53   fetcher->SetRequestContext(context);
  54   fetcher->SetLoadFlags(net::LOAD_DO_NOT_SEND_COOKIES |
  55                         net::LOAD_DO_NOT_SAVE_COOKIES);
  56   return fetcher.Pass();
  57 }
  58
  59 // Parses a SafeSearch API |response| and stores the result in |is_porn|.
  60 // On errors, returns false and doesn't set |is_porn|.
  61 bool ParseResponse(const std::string& response, bool* is_porn) {
  62   scoped_ptr<base::Value> value = base::JSONReader::Read(response);
  63   const base::DictionaryValue* dict = nullptr;
  64   if (!value || !value->GetAsDictionary(&dict)) {
  65     DLOG(WARNING) << "ParseResponse failed to parse global dictionary";
  66     return false;
  67   }
  68   const base::ListValue* classifications_list = nullptr;
  69   if (!dict->GetList("classifications", &classifications_list)) {
  70     DLOG(WARNING) << "ParseResponse failed to parse classifications list";
  71     return false;
  72   }
  73   if (classifications_list->GetSize() != 1) {
  74     DLOG(WARNING) << "ParseResponse expected exactly one result";
  75     return false;
  76   }
  77   const base::DictionaryValue* classification_dict = nullptr;
  78   if (!classifications_list->GetDictionary(0, &classification_dict)) {
  79     DLOG(WARNING) << "ParseResponse failed to parse classification dict";
  80     return false;
  81   }
  82   classification_dict->GetBoolean("pornography", is_porn);
  83   return true;
  84 }
  85
  86 }  // namespace
  87
  88 struct SupervisedUserAsyncURLChecker::Check {
  89   Check(const GURL& url,
  90         scoped_ptr<net::URLFetcher> fetcher,
  91         const CheckCallback& callback);
  92   ~Check();
  93
  94   GURL url;
  95   scoped_ptr<net::URLFetcher> fetcher;
  96   std::vector<CheckCallback> callbacks;
  97   base::Time start_time;
  98 };
  99
 100 SupervisedUserAsyncURLChecker::Check::Check(
 101     const GURL& url,
 102     scoped_ptr<net::URLFetcher> fetcher,
 103     const CheckCallback& callback)
 104     : url(url),
 105       fetcher(fetcher.Pass()),
 106       callbacks(1, callback),
 107       start_time(base::Time::Now()) {
 108 }
 109
 110 SupervisedUserAsyncURLChecker::Check::~Check() {}
 111
 112 SupervisedUserAsyncURLChecker::CheckResult::CheckResult(
 113     SupervisedUserURLFilter::FilteringBehavior behavior, bool uncertain)
 114     : behavior(behavior), uncertain(uncertain) {
 115 }
 116
 117 SupervisedUserAsyncURLChecker::SupervisedUserAsyncURLChecker(
 118     URLRequestContextGetter* context)
 119     : context_(context), cache_(kDefaultCacheSize) {
 120 }
 121
 122 SupervisedUserAsyncURLChecker::SupervisedUserAsyncURLChecker(
 123     URLRequestContextGetter* context,
 124     size_t cache_size)
 125     : context_(context), cache_(cache_size) {
 126 }
 127
 128 SupervisedUserAsyncURLChecker::~SupervisedUserAsyncURLChecker() {}
 129
 130 bool SupervisedUserAsyncURLChecker::CheckURL(const GURL& url,
 131                                              const CheckCallback& callback) {
 132   // TODO(treib): Hack: For now, allow all Google URLs to save search QPS. If we
 133   // ever remove this, we should find a way to allow at least the NTP.
 134   if (google_util::IsGoogleDomainUrl(url,
 135                                      google_util::ALLOW_SUBDOMAIN,
 136                                      google_util::ALLOW_NON_STANDARD_PORTS)) {
 137     callback.Run(url, SupervisedUserURLFilter::ALLOW, false);
 138     return true;
 139   }
 140   // TODO(treib): Hack: For now, allow all YouTube URLs since YouTube has its
 141   // own Safety Mode anyway.
 142   if (google_util::IsYoutubeDomainUrl(url,
 143                                       google_util::ALLOW_SUBDOMAIN,
 144                                       google_util::ALLOW_NON_STANDARD_PORTS)) {
 145     callback.Run(url, SupervisedUserURLFilter::ALLOW, false);
 146     return true;
 147   }
 148
 149   auto cache_it = cache_.Get(url);
 150   if (cache_it != cache_.end()) {
 151     const CheckResult& result = cache_it->second;
 152     DVLOG(1) << "Cache hit! " << url.spec() << " is "
 153              << (result.behavior == SupervisedUserURLFilter::BLOCK ? "NOT" : "")
 154              << " safe; certain: " << !result.uncertain;
 155     callback.Run(url, result.behavior, result.uncertain);
 156     return true;
 157   }
 158
 159   // See if we already have a check in progress for this URL.
 160   for (Check* check : checks_in_progress_) {
 161     if (check->url == url) {
 162       DVLOG(1) << "Adding to pending check for " << url.spec();
 163       check->callbacks.push_back(callback);
 164       return false;
 165     }
 166   }
 167
 168   DVLOG(1) << "Checking URL " << url;
 169   std::string api_key = google_apis::GetSafeSitesAPIKey();
 170   scoped_ptr<URLFetcher> fetcher(CreateFetcher(this, context_, api_key, url));
 171   fetcher->Start();
 172   checks_in_progress_.push_back(new Check(url, fetcher.Pass(), callback));
 173   return false;
 174 }
 175
 176 void SupervisedUserAsyncURLChecker::OnURLFetchComplete(
 177     const net::URLFetcher* source) {
 178   ScopedVector<Check>::iterator it = checks_in_progress_.begin();
 179   while (it != checks_in_progress_.end()) {
 180     if (source == (*it)->fetcher.get())
 181       break;
 182     ++it;
 183   }
 184   DCHECK(it != checks_in_progress_.end());
 185   Check* check = *it;
 186
 187   const URLRequestStatus& status = source->GetStatus();
 188   if (!status.is_success()) {
 189     DLOG(WARNING) << "URL request failed! Letting through...";
 190     for (size_t i = 0; i < check->callbacks.size(); i++)
 191       check->callbacks[i].Run(check->url, SupervisedUserURLFilter::ALLOW, true);
 192     checks_in_progress_.erase(it);
 193     return;
 194   }
 195
 196   std::string response_body;
 197   source->GetResponseAsString(&response_body);
 198   bool is_porn = false;
 199   bool uncertain = !ParseResponse(response_body, &is_porn);
 200   SupervisedUserURLFilter::FilteringBehavior behavior =
 201       is_porn ? SupervisedUserURLFilter::BLOCK : SupervisedUserURLFilter::ALLOW;
 202
 203   UMA_HISTOGRAM_TIMES("ManagedUsers.SafeSitesDelay",
 204                       base::Time::Now() - check->start_time);
 205
 206   cache_.Put(check->url, CheckResult(behavior, uncertain));
 207
 208   for (size_t i = 0; i < check->callbacks.size(); i++)
 209     check->callbacks[i].Run(check->url, behavior, uncertain);
 210   checks_in_progress_.erase(it);
 211 }