chrome/browser/spellchecker/feedback_sender.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 // The |FeedbackSender| object stores the user feedback to spellcheck
   6 // suggestions in a |Feedback| object.
   7 //
   8 // When spelling service returns spellcheck results, these results first arrive
   9 // in |FeedbackSender| to assign hash identifiers for each
  10 // misspelling-suggestion pair. If the spelling service identifies the same
  11 // misspelling as already displayed to the user, then |FeedbackSender| reuses
  12 // the same hash identifiers to avoid duplication. It detects the duplicates by
  13 // comparing misspelling offsets in text. Spelling service can return duplicates
  14 // because we request spellcheck for whole paragraphs, as context around a
  15 // misspelled word is important to the spellcheck algorithm.
  16 //
  17 // All feedback is initially pending. When a user acts upon a misspelling such
  18 // that the misspelling is no longer displayed (red squiggly line goes away),
  19 // then the feedback for this misspelling is finalized. All finalized feedback
  20 // is erased after being sent to the spelling service. Pending feedback is kept
  21 // around for |kSessionHours| hours and then finalized even if user did not act
  22 // on the misspellings.
  23 //
  24 // |FeedbackSender| periodically requests a list of hashes of all remaining
  25 // misspellings in renderers. When a renderer responds with a list of hashes,
  26 // |FeedbackSender| uses the list to determine which misspellings are no longer
  27 // displayed to the user and sends the current state of user feedback to the
  28 // spelling service.
  29
  30 #include "chrome/browser/spellchecker/feedback_sender.h"
  31
  32 #include <algorithm>
  33 #include <iterator>
  34
  35 #include "base/command_line.h"
  36 #include "base/hash.h"
  37 #include "base/json/json_writer.h"
  38 #include "base/location.h"
  39 #include "base/metrics/field_trial.h"
  40 #include "base/single_thread_task_runner.h"
  41 #include "base/stl_util.h"
  42 #include "base/strings/string_number_conversions.h"
  43 #include "base/strings/stringprintf.h"
  44 #include "base/thread_task_runner_handle.h"
  45 #include "base/values.h"
  46 #include "chrome/browser/spellchecker/word_trimmer.h"
  47 #include "chrome/common/chrome_switches.h"
  48 #include "chrome/common/spellcheck_common.h"
  49 #include "chrome/common/spellcheck_marker.h"
  50 #include "chrome/common/spellcheck_messages.h"
  51 #include "content/public/browser/render_process_host.h"
  52 #include "google_apis/google_api_keys.h"
  53 #include "net/base/load_flags.h"
  54 #include "net/url_request/url_fetcher.h"
  55 #include "net/url_request/url_request_context_getter.h"
  56
  57 namespace spellcheck {
  58
  59 namespace {
  60
  61 // The default URL where feedback data is sent.
  62 const char kFeedbackServiceURL[] = "https://www.googleapis.com/rpc";
  63
  64 // The minimum number of seconds between sending batches of feedback.
  65 const int kMinIntervalSeconds = 5;
  66
  67 // Returns a hash of |session_start|, the current timestamp, and
  68 // |suggestion_index|.
  69 uint32 BuildHash(const base::Time& session_start, size_t suggestion_index) {
  70   return base::Hash(
  71       base::StringPrintf("%" PRId64 "%" PRId64 "%" PRIuS,
  72                          session_start.ToInternalValue(),
  73                          base::Time::Now().ToInternalValue(),
  74                          suggestion_index));
  75 }
  76
  77 // Returns a pending feedback data structure for the spellcheck |result| and
  78 // |text|.
  79 Misspelling BuildFeedback(const SpellCheckResult& result,
  80                           const base::string16& text) {
  81   size_t start = result.location;
  82   base::string16 context = TrimWords(&start,
  83                                result.length,
  84                                text,
  85                                chrome::spellcheck_common::kContextWordCount);
  86   return Misspelling(context,
  87                      start,
  88                      result.length,
  89                      std::vector<base::string16>(1, result.replacement),
  90                      result.hash);
  91 }
  92
  93 // Builds suggestion info from |suggestions|. The caller owns the result.
  94 base::ListValue* BuildSuggestionInfo(
  95     const std::vector<Misspelling>& suggestions,
  96     bool is_first_feedback_batch) {
  97   base::ListValue* list = new base::ListValue;
  98   for (std::vector<Misspelling>::const_iterator suggestion_it =
  99            suggestions.begin();
 100        suggestion_it != suggestions.end();
 101        ++suggestion_it) {
 102     base::DictionaryValue* suggestion = suggestion_it->Serialize();
 103     suggestion->SetBoolean("isFirstInSession", is_first_feedback_batch);
 104     suggestion->SetBoolean("isAutoCorrection", false);
 105     list->Append(suggestion);
 106   }
 107   return list;
 108 }
 109
 110 // Builds feedback parameters from |suggestion_info|, |language|, and |country|.
 111 // Takes ownership of |suggestion_list|. The caller owns the result.
 112 base::DictionaryValue* BuildParams(base::ListValue* suggestion_info,
 113                                    const std::string& language,
 114                                    const std::string& country) {
 115   base::DictionaryValue* params = new base::DictionaryValue;
 116   params->Set("suggestionInfo", suggestion_info);
 117   params->SetString("key", google_apis::GetAPIKey());
 118   params->SetString("language", language);
 119   params->SetString("originCountry", country);
 120   params->SetString("clientName", "Chrome");
 121   return params;
 122 }
 123
 124 // Builds feedback data from |params|. Takes ownership of |params|. The caller
 125 // owns the result.
 126 base::Value* BuildFeedbackValue(base::DictionaryValue* params,
 127                                 const std::string& api_version) {
 128   base::DictionaryValue* result = new base::DictionaryValue;
 129   result->Set("params", params);
 130   result->SetString("method", "spelling.feedback");
 131   result->SetString("apiVersion", api_version);
 132   return result;
 133 }
 134
 135 // Returns true if the misspelling location is within text bounds.
 136 bool IsInBounds(int misspelling_location,
 137                 int misspelling_length,
 138                 size_t text_length) {
 139   return misspelling_location >= 0 && misspelling_length > 0 &&
 140          static_cast<size_t>(misspelling_location) < text_length &&
 141          static_cast<size_t>(misspelling_location + misspelling_length) <=
 142              text_length;
 143 }
 144
 145 // Returns the feedback API version.
 146 std::string GetApiVersion() {
 147   // This guard is temporary.
 148   // TODO(rouslan): Remove the guard. http://crbug.com/247726
 149   if (base::FieldTrialList::FindFullName(kFeedbackFieldTrialName) ==
 150           kFeedbackFieldTrialEnabledGroupName &&
 151       base::CommandLine::ForCurrentProcess()->HasSwitch(
 152           switches::kEnableSpellingFeedbackFieldTrial)) {
 153     return "v2-internal";
 154   }
 155   return "v2";
 156 }
 157
 158 }  // namespace
 159
 160 FeedbackSender::FeedbackSender(net::URLRequestContextGetter* request_context,
 161                                const std::string& language,
 162                                const std::string& country)
 163     : request_context_(request_context),
 164       api_version_(GetApiVersion()),
 165       language_(language),
 166       country_(country),
 167       misspelling_counter_(0),
 168       session_start_(base::Time::Now()),
 169       feedback_service_url_(kFeedbackServiceURL) {
 170   // The command-line switch is for testing and temporary.
 171   // TODO(rouslan): Remove the command-line switch when testing is complete.
 172   // http://crbug.com/247726
 173   if (base::CommandLine::ForCurrentProcess()->HasSwitch(
 174           switches::kSpellingServiceFeedbackUrl)) {
 175     feedback_service_url_ =
 176         GURL(base::CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
 177             switches::kSpellingServiceFeedbackUrl));
 178   }
 179 }
 180
 181 FeedbackSender::~FeedbackSender() {
 182 }
 183
 184 void FeedbackSender::SelectedSuggestion(uint32 hash, int suggestion_index) {
 185   Misspelling* misspelling = feedback_.GetMisspelling(hash);
 186   // GetMisspelling() returns null for flushed feedback. Feedback is flushed
 187   // when the session expires every |kSessionHours| hours.
 188   if (!misspelling)
 189     return;
 190   misspelling->action.type = SpellcheckAction::TYPE_SELECT;
 191   misspelling->action.index = suggestion_index;
 192   misspelling->timestamp = base::Time::Now();
 193 }
 194
 195 void FeedbackSender::AddedToDictionary(uint32 hash) {
 196   Misspelling* misspelling = feedback_.GetMisspelling(hash);
 197   // GetMisspelling() returns null for flushed feedback. Feedback is flushed
 198   // when the session expires every |kSessionHours| hours.
 199   if (!misspelling)
 200     return;
 201   misspelling->action.type = SpellcheckAction::TYPE_ADD_TO_DICT;
 202   misspelling->timestamp = base::Time::Now();
 203   const std::set<uint32>& hashes =
 204       feedback_.FindMisspellings(misspelling->GetMisspelledString());
 205   for (std::set<uint32>::const_iterator hash_it = hashes.begin();
 206        hash_it != hashes.end();
 207        ++hash_it) {
 208     Misspelling* duplicate_misspelling = feedback_.GetMisspelling(*hash_it);
 209     if (!duplicate_misspelling || duplicate_misspelling->action.IsFinal())
 210       continue;
 211     duplicate_misspelling->action.type = SpellcheckAction::TYPE_ADD_TO_DICT;
 212     duplicate_misspelling->timestamp = misspelling->timestamp;
 213   }
 214 }
 215
 216 void FeedbackSender::RecordInDictionary(uint32 hash) {
 217   Misspelling* misspelling = feedback_.GetMisspelling(hash);
 218   // GetMisspelling() returns null for flushed feedback. Feedback is flushed
 219   // when the session expires every |kSessionHours| hours.
 220   if (!misspelling)
 221     return;
 222   misspelling->action.type = SpellcheckAction::TYPE_IN_DICTIONARY;
 223 }
 224
 225 void FeedbackSender::IgnoredSuggestions(uint32 hash) {
 226   Misspelling* misspelling = feedback_.GetMisspelling(hash);
 227   // GetMisspelling() returns null for flushed feedback. Feedback is flushed
 228   // when the session expires every |kSessionHours| hours.
 229   if (!misspelling)
 230     return;
 231   misspelling->action.type = SpellcheckAction::TYPE_PENDING_IGNORE;
 232   misspelling->timestamp = base::Time::Now();
 233 }
 234
 235 void FeedbackSender::ManuallyCorrected(uint32 hash,
 236                                        const base::string16& correction) {
 237   Misspelling* misspelling = feedback_.GetMisspelling(hash);
 238   // GetMisspelling() returns null for flushed feedback. Feedback is flushed
 239   // when the session expires every |kSessionHours| hours.
 240   if (!misspelling)
 241     return;
 242   misspelling->action.type = SpellcheckAction::TYPE_MANUALLY_CORRECTED;
 243   misspelling->action.value = correction;
 244   misspelling->timestamp = base::Time::Now();
 245 }
 246
 247 void FeedbackSender::OnReceiveDocumentMarkers(
 248     int renderer_process_id,
 249     const std::vector<uint32>& markers) {
 250   if ((base::Time::Now() - session_start_).InHours() >=
 251       chrome::spellcheck_common::kSessionHours) {
 252     FlushFeedback();
 253     return;
 254   }
 255
 256   if (!feedback_.RendererHasMisspellings(renderer_process_id))
 257     return;
 258
 259   feedback_.FinalizeRemovedMisspellings(renderer_process_id, markers);
 260   SendFeedback(feedback_.GetMisspellingsInRenderer(renderer_process_id),
 261                !renderers_sent_feedback_.count(renderer_process_id));
 262   renderers_sent_feedback_.insert(renderer_process_id);
 263   feedback_.EraseFinalizedMisspellings(renderer_process_id);
 264 }
 265
 266 void FeedbackSender::OnSpellcheckResults(
 267     int renderer_process_id,
 268     const base::string16& text,
 269     const std::vector<SpellCheckMarker>& markers,
 270     std::vector<SpellCheckResult>* results) {
 271   // Don't collect feedback if not going to send it.
 272   if (!timer_.IsRunning())
 273     return;
 274
 275   // Generate a map of marker offsets to marker hashes. This map helps to
 276   // efficiently lookup feedback data based on the position of the misspelling
 277   // in text.
 278   typedef std::map<size_t, uint32> MarkerMap;
 279   MarkerMap marker_map;
 280   for (size_t i = 0; i < markers.size(); ++i)
 281     marker_map[markers[i].offset] = markers[i].hash;
 282
 283   for (std::vector<SpellCheckResult>::iterator result_it = results->begin();
 284        result_it != results->end();
 285        ++result_it) {
 286     if (!IsInBounds(result_it->location, result_it->length, text.length()))
 287       continue;
 288     MarkerMap::const_iterator marker_it = marker_map.find(result_it->location);
 289     if (marker_it != marker_map.end() &&
 290         feedback_.HasMisspelling(marker_it->second)) {
 291       // If the renderer already has a marker for this spellcheck result, then
 292       // set the hash of the spellcheck result to be the same as the marker.
 293       result_it->hash = marker_it->second;
 294     } else {
 295       // If the renderer does not yet have a marker for this spellcheck result,
 296       // then generate a new hash for the spellcheck result.
 297       result_it->hash = BuildHash(session_start_, ++misspelling_counter_);
 298     }
 299     // Save the feedback data for the spellcheck result.
 300     feedback_.AddMisspelling(renderer_process_id,
 301                              BuildFeedback(*result_it, text));
 302   }
 303 }
 304
 305 void FeedbackSender::OnLanguageCountryChange(const std::string& language,
 306                                              const std::string& country) {
 307   FlushFeedback();
 308   language_ = language;
 309   country_ = country;
 310 }
 311
 312 void FeedbackSender::StartFeedbackCollection() {
 313   if (timer_.IsRunning())
 314     return;
 315
 316   int interval_seconds = chrome::spellcheck_common::kFeedbackIntervalSeconds;
 317   // This command-line switch is for testing and temporary.
 318   // TODO(rouslan): Remove the command-line switch when testing is complete.
 319   // http://crbug.com/247726
 320   if (base::CommandLine::ForCurrentProcess()->HasSwitch(
 321           switches::kSpellingServiceFeedbackIntervalSeconds)) {
 322     base::StringToInt(
 323         base::CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
 324             switches::kSpellingServiceFeedbackIntervalSeconds),
 325         &interval_seconds);
 326     if (interval_seconds < kMinIntervalSeconds)
 327       interval_seconds = kMinIntervalSeconds;
 328     static const int kSessionSeconds =
 329         chrome::spellcheck_common::kSessionHours * 60 * 60;
 330     if (interval_seconds >  kSessionSeconds)
 331       interval_seconds = kSessionSeconds;
 332   }
 333   timer_.Start(FROM_HERE,
 334                base::TimeDelta::FromSeconds(interval_seconds),
 335                this,
 336                &FeedbackSender::RequestDocumentMarkers);
 337 }
 338
 339 void FeedbackSender::StopFeedbackCollection() {
 340   if (!timer_.IsRunning())
 341     return;
 342
 343   FlushFeedback();
 344   timer_.Stop();
 345 }
 346
 347 void FeedbackSender::OnURLFetchComplete(const net::URLFetcher* source) {
 348   for (ScopedVector<net::URLFetcher>::iterator sender_it = senders_.begin();
 349        sender_it != senders_.end();
 350        ++sender_it) {
 351     if (*sender_it == source) {
 352       senders_.erase(sender_it);
 353       return;
 354     }
 355   }
 356   delete source;
 357 }
 358
 359 void FeedbackSender::RequestDocumentMarkers() {
 360   // Request document markers from all the renderers that are still alive.
 361   std::set<int> alive_renderers;
 362   for (content::RenderProcessHost::iterator it(
 363            content::RenderProcessHost::AllHostsIterator());
 364        !it.IsAtEnd();
 365        it.Advance()) {
 366     alive_renderers.insert(it.GetCurrentValue()->GetID());
 367     it.GetCurrentValue()->Send(new SpellCheckMsg_RequestDocumentMarkers());
 368   }
 369
 370   // Asynchronously send out the feedback for all the renderers that are no
 371   // longer alive.
 372   std::vector<int> known_renderers = feedback_.GetRendersWithMisspellings();
 373   std::sort(known_renderers.begin(), known_renderers.end());
 374   std::vector<int> dead_renderers =
 375       base::STLSetDifference<std::vector<int> >(known_renderers,
 376                                                 alive_renderers);
 377   for (std::vector<int>::const_iterator it = dead_renderers.begin();
 378        it != dead_renderers.end();
 379        ++it) {
 380     base::ThreadTaskRunnerHandle::Get()->PostTask(
 381         FROM_HERE, base::Bind(&FeedbackSender::OnReceiveDocumentMarkers,
 382                               AsWeakPtr(), *it, std::vector<uint32>()));
 383   }
 384 }
 385
 386 void FeedbackSender::FlushFeedback() {
 387   if (feedback_.Empty())
 388     return;
 389   feedback_.FinalizeAllMisspellings();
 390   SendFeedback(feedback_.GetAllMisspellings(),
 391                renderers_sent_feedback_.empty());
 392   feedback_.Clear();
 393   renderers_sent_feedback_.clear();
 394   session_start_ = base::Time::Now();
 395   timer_.Reset();
 396 }
 397
 398 void FeedbackSender::SendFeedback(const std::vector<Misspelling>& feedback_data,
 399                                   bool is_first_feedback_batch) {
 400   scoped_ptr<base::Value> feedback_value(BuildFeedbackValue(
 401       BuildParams(BuildSuggestionInfo(feedback_data, is_first_feedback_batch),
 402                   language_,
 403                   country_),
 404       api_version_));
 405   std::string feedback;
 406   base::JSONWriter::Write(*feedback_value, &feedback);
 407
 408   // The tests use this identifier to mock the URL fetcher.
 409   static const int kUrlFetcherId = 0;
 410   net::URLFetcher* sender =
 411       net::URLFetcher::Create(kUrlFetcherId, feedback_service_url_,
 412                               net::URLFetcher::POST, this).release();
 413   sender->SetLoadFlags(net::LOAD_DO_NOT_SEND_COOKIES |
 414                        net::LOAD_DO_NOT_SAVE_COOKIES);
 415   sender->SetUploadData("application/json", feedback);
 416   senders_.push_back(sender);
 417
 418   // Request context is NULL in testing.
 419   if (request_context_.get()) {
 420     sender->SetRequestContext(request_context_.get());
 421     sender->Start();
 422   }
 423 }
 424
 425 }  // namespace spellcheck