src/chrome/browser/extensions/activity_log/uma_policy.cc

   1 // Copyright 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 "chrome/browser/extensions/activity_log/uma_policy.h"
   6
   7 #include "base/metrics/histogram.h"
   8 #include "base/strings/stringprintf.h"
   9 #include "chrome/browser/browser_process.h"
  10 #include "chrome/browser/extensions/activity_log/activity_action_constants.h"
  11 #include "chrome/browser/extensions/activity_log/ad_network_database.h"
  12 #include "chrome/browser/sessions/session_id.h"
  13 #include "chrome/browser/ui/browser.h"
  14 #include "chrome/browser/ui/browser_list.h"
  15 #include "chrome/browser/ui/tabs/tab_strip_model.h"
  16 #include "chrome/common/url_constants.h"
  17 #include "content/public/browser/web_contents.h"
  18 #include "content/public/common/url_constants.h"
  19 #include "extensions/common/dom_action_types.h"
  20
  21 namespace extensions {
  22
  23 namespace {
  24
  25 // For convenience.
  26 const int kNoStatus           = UmaPolicy::NONE;
  27 const int kContentScript      = 1 << UmaPolicy::CONTENT_SCRIPT;
  28 const int kReadDom            = 1 << UmaPolicy::READ_DOM;
  29 const int kModifiedDom        = 1 << UmaPolicy::MODIFIED_DOM;
  30 const int kDomMethod          = 1 << UmaPolicy::DOM_METHOD;
  31 const int kDocumentWrite      = 1 << UmaPolicy::DOCUMENT_WRITE;
  32 const int kInnerHtml          = 1 << UmaPolicy::INNER_HTML;
  33 const int kCreatedScript      = 1 << UmaPolicy::CREATED_SCRIPT;
  34 const int kCreatedIframe      = 1 << UmaPolicy::CREATED_IFRAME;
  35 const int kCreatedDiv         = 1 << UmaPolicy::CREATED_DIV;
  36 const int kCreatedLink        = 1 << UmaPolicy::CREATED_LINK;
  37 const int kCreatedInput       = 1 << UmaPolicy::CREATED_INPUT;
  38 const int kCreatedEmbed       = 1 << UmaPolicy::CREATED_EMBED;
  39 const int kCreatedObject      = 1 << UmaPolicy::CREATED_OBJECT;
  40 const int kAdInjected         = 1 << UmaPolicy::AD_INJECTED;
  41 const int kAdRemoved          = 1 << UmaPolicy::AD_REMOVED;
  42 const int kAdReplaced         = 1 << UmaPolicy::AD_REPLACED;
  43
  44 }  // namespace
  45
  46 // Class constants, also used in testing. --------------------------------------
  47
  48 const char UmaPolicy::kNumberOfTabs[]       = "num_tabs";
  49 const size_t UmaPolicy::kMaxTabsTracked     = 50;
  50
  51 // Setup and shutdown. ---------------------------------------------------------
  52
  53 UmaPolicy::UmaPolicy(Profile* profile)
  54     : ActivityLogPolicy(profile), profile_(profile) {
  55   DCHECK(!profile->IsOffTheRecord());
  56   BrowserList::AddObserver(this);
  57 }
  58
  59 UmaPolicy::~UmaPolicy() {
  60   BrowserList::RemoveObserver(this);
  61 }
  62
  63 // Unlike the other policies, UmaPolicy can commit suicide directly because it
  64 // doesn't have a dependency on a database.
  65 void UmaPolicy::Close() {
  66   delete this;
  67 }
  68
  69 // Process actions. ------------------------------------------------------------
  70
  71 void UmaPolicy::ProcessAction(scoped_refptr<Action> action) {
  72   if (!action->page_url().is_valid() && !action->arg_url().is_valid())
  73     return;
  74   if (action->page_incognito() || action->arg_incognito())
  75     return;
  76   std::string url;
  77   int status = MatchActionToStatus(action);
  78   if (action->page_url().is_valid()) {
  79     url = CleanURL(action->page_url());
  80   } else if (status & kContentScript) {
  81     // This is for the tabs.executeScript case.
  82     url = CleanURL(action->arg_url());
  83   }
  84   if (url.empty())
  85     return;
  86
  87   SiteMap::iterator site_lookup = url_status_.find(url);
  88   if (site_lookup != url_status_.end())
  89     site_lookup->second[action->extension_id()] |= status;
  90 }
  91
  92 int UmaPolicy::MatchActionToStatus(scoped_refptr<Action> action) {
  93   if (action->action_type() == Action::ACTION_CONTENT_SCRIPT) {
  94     return kContentScript;
  95   } else if (action->action_type() == Action::ACTION_API_CALL &&
  96              action->api_name() == "tabs.executeScript") {
  97     return kContentScript;
  98   } else if (action->action_type() != Action::ACTION_DOM_ACCESS) {
  99     return kNoStatus;
 100   }
 101
 102   int dom_verb;
 103   if (!action->other() ||
 104       !action->other()->GetIntegerWithoutPathExpansion(
 105           activity_log_constants::kActionDomVerb, &dom_verb)) {
 106     return kNoStatus;
 107   }
 108
 109   int ret_bit = kNoStatus;
 110   DomActionType::Type dom_type = static_cast<DomActionType::Type>(dom_verb);
 111   if (dom_type == DomActionType::GETTER)
 112     return kReadDom;
 113   if (dom_type == DomActionType::SETTER) {
 114     ret_bit |= kModifiedDom;
 115   } else if (dom_type == DomActionType::METHOD) {
 116     ret_bit |= kDomMethod;
 117   } else {
 118     return kNoStatus;
 119   }
 120
 121   if (action->api_name() == "HTMLDocument.write" ||
 122       action->api_name() == "HTMLDocument.writeln") {
 123     ret_bit |= kDocumentWrite;
 124   } else if (action->api_name() == "Element.innerHTML") {
 125     ret_bit |= kInnerHtml;
 126   } else if (action->api_name() == "Document.createElement") {
 127     std::string arg;
 128     action->args()->GetString(0, &arg);
 129     if (arg == "script") {
 130       ret_bit |= kCreatedScript;
 131     } else if (arg == "iframe") {
 132       ret_bit |= kCreatedIframe;
 133     } else if (arg == "div") {
 134       ret_bit |= kCreatedDiv;
 135     } else if (arg == "a") {
 136       ret_bit |= kCreatedLink;
 137     } else if (arg == "input") {
 138       ret_bit |= kCreatedInput;
 139     } else if (arg == "embed") {
 140       ret_bit |= kCreatedEmbed;
 141     } else if (arg == "object") {
 142       ret_bit |= kCreatedObject;
 143     }
 144   }
 145
 146   const Action::InjectionType ad_injection =
 147       action->DidInjectAd(g_browser_process->rappor_service());
 148   switch (ad_injection) {
 149     case Action::INJECTION_NEW_AD:
 150       ret_bit |= kAdInjected;
 151       break;
 152     case Action::INJECTION_REMOVED_AD:
 153       ret_bit |= kAdRemoved;
 154       break;
 155     case Action::INJECTION_REPLACED_AD:
 156       ret_bit |= kAdReplaced;
 157       break;
 158     case Action::NO_AD_INJECTION:
 159       break;
 160     case Action::NUM_INJECTION_TYPES:
 161       NOTREACHED();
 162   };
 163
 164   return ret_bit;
 165 }
 166
 167 void UmaPolicy::HistogramOnClose(const std::string& url) {
 168   // Let's try to avoid histogramming useless URLs.
 169   if (url.empty() || url == content::kAboutBlankURL ||
 170       url == chrome::kChromeUINewTabURL)
 171     return;
 172
 173   int statuses[MAX_STATUS - 1];
 174   std::memset(statuses, 0, sizeof(statuses));
 175
 176   SiteMap::iterator site_lookup = url_status_.find(url);
 177   ExtensionMap exts = site_lookup->second;
 178   ExtensionMap::iterator ext_iter;
 179   for (ext_iter = exts.begin(); ext_iter != exts.end(); ++ext_iter) {
 180     if (ext_iter->first == kNumberOfTabs)
 181       continue;
 182     for (int i = NONE + 1; i < MAX_STATUS; ++i) {
 183       if (ext_iter->second & (1 << i))
 184         statuses[i-1]++;
 185     }
 186   }
 187
 188   std::string prefix = "ExtensionActivity.";
 189   if (GURL(url).host() != "www.google.com") {
 190     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CONTENT_SCRIPT),
 191                              statuses[CONTENT_SCRIPT - 1]);
 192     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(READ_DOM),
 193                              statuses[READ_DOM - 1]);
 194     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(MODIFIED_DOM),
 195                              statuses[MODIFIED_DOM - 1]);
 196     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(DOM_METHOD),
 197                              statuses[DOM_METHOD - 1]);
 198     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(DOCUMENT_WRITE),
 199                              statuses[DOCUMENT_WRITE - 1]);
 200     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(INNER_HTML),
 201                              statuses[INNER_HTML - 1]);
 202     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_SCRIPT),
 203                              statuses[CREATED_SCRIPT - 1]);
 204     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_IFRAME),
 205                              statuses[CREATED_IFRAME - 1]);
 206     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_DIV),
 207                              statuses[CREATED_DIV - 1]);
 208     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_LINK),
 209                              statuses[CREATED_LINK - 1]);
 210     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_INPUT),
 211                              statuses[CREATED_INPUT - 1]);
 212     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_EMBED),
 213                              statuses[CREATED_EMBED - 1]);
 214     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_OBJECT),
 215                              statuses[CREATED_OBJECT - 1]);
 216     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_INJECTED),
 217                              statuses[AD_INJECTED - 1]);
 218     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_REMOVED),
 219                              statuses[AD_REMOVED - 1]);
 220     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_REPLACED),
 221                              statuses[AD_REPLACED - 1]);
 222   } else {
 223     prefix += "Google.";
 224     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CONTENT_SCRIPT),
 225                              statuses[CONTENT_SCRIPT - 1]);
 226     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(READ_DOM),
 227                              statuses[READ_DOM - 1]);
 228     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(MODIFIED_DOM),
 229                              statuses[MODIFIED_DOM - 1]);
 230     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(DOM_METHOD),
 231                              statuses[DOM_METHOD - 1]);
 232     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(DOCUMENT_WRITE),
 233                              statuses[DOCUMENT_WRITE - 1]);
 234     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(INNER_HTML),
 235                              statuses[INNER_HTML - 1]);
 236     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_SCRIPT),
 237                              statuses[CREATED_SCRIPT - 1]);
 238     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_IFRAME),
 239                              statuses[CREATED_IFRAME - 1]);
 240     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_DIV),
 241                              statuses[CREATED_DIV - 1]);
 242     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_LINK),
 243                              statuses[CREATED_LINK - 1]);
 244     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_INPUT),
 245                              statuses[CREATED_INPUT - 1]);
 246     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_EMBED),
 247                              statuses[CREATED_EMBED - 1]);
 248     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(CREATED_OBJECT),
 249                              statuses[CREATED_OBJECT - 1]);
 250     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_INJECTED),
 251                              statuses[AD_INJECTED - 1]);
 252     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_REMOVED),
 253                              statuses[AD_REMOVED - 1]);
 254     UMA_HISTOGRAM_COUNTS_100(prefix + GetHistogramName(AD_REPLACED),
 255                              statuses[AD_REPLACED - 1]);
 256   }
 257 }
 258
 259 // Handle tab tracking. --------------------------------------------------------
 260
 261 void UmaPolicy::OnBrowserAdded(Browser* browser) {
 262   if (!profile_->IsSameProfile(browser->profile()))
 263     return;
 264   browser->tab_strip_model()->AddObserver(this);
 265 }
 266
 267 void UmaPolicy::OnBrowserRemoved(Browser* browser) {
 268   if (!profile_->IsSameProfile(browser->profile()))
 269     return;
 270   browser->tab_strip_model()->RemoveObserver(this);
 271 }
 272
 273 // Use the value from SessionID::IdForTab, *not* |index|. |index| will be
 274 // duplicated across tabs in a session, whereas IdForTab uniquely identifies
 275 // each tab.
 276 void UmaPolicy::TabChangedAt(content::WebContents* contents,
 277                              int index,
 278                              TabChangeType change_type) {
 279   if (change_type != TabStripModelObserver::LOADING_ONLY)
 280     return;
 281   if (!contents)
 282     return;
 283
 284   std::string url = CleanURL(contents->GetLastCommittedURL());
 285   int32 tab_id = SessionID::IdForTab(contents);
 286
 287   std::map<int32, std::string>::iterator tab_it = tab_list_.find(tab_id);
 288
 289   // Ignore tabs that haven't changed status.
 290   if (tab_it != tab_list_.end() && tab_it->second == url)
 291     return;
 292
 293   // Is this an existing tab whose URL has changed.
 294   if (tab_it != tab_list_.end()) {
 295     CleanupClosedPage(tab_it->second);
 296     tab_list_.erase(tab_id);
 297   }
 298
 299   // Check that tab_list_ isn't over the kMaxTabsTracked budget.
 300   if (tab_list_.size() >= kMaxTabsTracked)
 301     return;
 302
 303   // Set up the new entries.
 304   tab_list_[tab_id] = url;
 305   SetupOpenedPage(url);
 306 }
 307
 308 // Use the value from SessionID::IdForTab, *not* |index|. |index| will be
 309 // duplicated across tabs in a session, whereas IdForTab uniquely identifies
 310 // each tab.
 311 void UmaPolicy::TabClosingAt(TabStripModel* tab_strip_model,
 312                              content::WebContents* contents,
 313                              int index) {
 314   if (!contents)
 315     return;
 316   std::string url = CleanURL(contents->GetLastCommittedURL());
 317   int32 tab_id = SessionID::IdForTab(contents);
 318   std::map<int, std::string>::iterator tab_it = tab_list_.find(tab_id);
 319   if (tab_it != tab_list_.end())
 320     tab_list_.erase(tab_id);
 321
 322   CleanupClosedPage(url);
 323 }
 324
 325 void UmaPolicy::SetupOpenedPage(const std::string& url) {
 326   url_status_[url][kNumberOfTabs]++;
 327 }
 328
 329 void UmaPolicy::CleanupClosedPage(const std::string& url) {
 330   SiteMap::iterator old_site_lookup = url_status_.find(url);
 331   if (old_site_lookup == url_status_.end())
 332     return;
 333   old_site_lookup->second[kNumberOfTabs]--;
 334   if (old_site_lookup->second[kNumberOfTabs] == 0) {
 335     HistogramOnClose(url);
 336     url_status_.erase(url);
 337   }
 338 }
 339
 340 // Helpers. --------------------------------------------------------------------
 341
 342 // We don't want to treat # ref navigations as if they were new pageloads.
 343 // So we get rid of the ref if it has it.
 344 // We convert to a string in the hopes that this is faster than Replacements.
 345 std::string UmaPolicy::CleanURL(const GURL& gurl) {
 346   if (gurl.spec().empty())
 347     return GURL(content::kAboutBlankURL).spec();
 348   if (!gurl.is_valid())
 349     return gurl.spec();
 350   if (!gurl.has_ref())
 351     return gurl.spec();
 352   std::string port = "";
 353   if (gurl.has_port())
 354     port = ":" + gurl.port();
 355   std::string query = "";
 356   if (gurl.has_query())
 357     query = "?" + gurl.query();
 358   return base::StringPrintf("%s://%s%s%s%s",
 359                             gurl.scheme().c_str(),
 360                             gurl.host().c_str(),
 361                             port.c_str(),
 362                             gurl.path().c_str(),
 363                             query.c_str());
 364 }
 365
 366 const char* UmaPolicy::GetHistogramName(PageStatus status) {
 367   switch (status) {
 368     case CONTENT_SCRIPT:
 369       return "ContentScript";
 370     case READ_DOM:
 371       return "ReadDom";
 372     case MODIFIED_DOM:
 373       return "ModifiedDom";
 374     case DOM_METHOD:
 375       return "InvokedDomMethod";
 376     case DOCUMENT_WRITE:
 377       return "DocumentWrite";
 378     case INNER_HTML:
 379       return "InnerHtml";
 380     case CREATED_SCRIPT:
 381       return "CreatedScript";
 382     case CREATED_IFRAME:
 383       return "CreatedIframe";
 384     case CREATED_DIV:
 385       return "CreatedDiv";
 386     case CREATED_LINK:
 387       return "CreatedLink";
 388     case CREATED_INPUT:
 389       return "CreatedInput";
 390     case CREATED_EMBED:
 391       return "CreatedEmbed";
 392     case CREATED_OBJECT:
 393       return "CreatedObject";
 394     case AD_INJECTED:
 395       return "AdInjected";
 396     case AD_REMOVED:
 397       return "AdRemoved";
 398     case AD_REPLACED:
 399       return "AdReplaced";
 400     case NONE:
 401     case MAX_STATUS:
 402     default:
 403       NOTREACHED();
 404       return "";
 405   }
 406 }
 407
 408 }  // namespace extensions