zero_suggest_provider.cc revision a36e5920737c6adbddd3e43b760e5de8431db6e0
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#include "chrome/browser/autocomplete/zero_suggest_provider.h"
6
7#include "base/callback.h"
8#include "base/i18n/case_conversion.h"
9#include "base/json/json_string_value_serializer.h"
10#include "base/metrics/histogram.h"
11#include "base/prefs/pref_service.h"
12#include "base/strings/string16.h"
13#include "base/strings/string_util.h"
14#include "base/strings/utf_string_conversions.h"
15#include "base/time/time.h"
16#include "chrome/browser/autocomplete/autocomplete_input.h"
17#include "chrome/browser/autocomplete/autocomplete_match.h"
18#include "chrome/browser/autocomplete/autocomplete_provider_listener.h"
19#include "chrome/browser/autocomplete/history_url_provider.h"
20#include "chrome/browser/autocomplete/search_provider.h"
21#include "chrome/browser/autocomplete/url_prefix.h"
22#include "chrome/browser/metrics/variations/variations_http_header_provider.h"
23#include "chrome/browser/omnibox/omnibox_field_trial.h"
24#include "chrome/browser/profiles/profile.h"
25#include "chrome/browser/search/search.h"
26#include "chrome/browser/search_engines/template_url_service.h"
27#include "chrome/browser/search_engines/template_url_service_factory.h"
28#include "chrome/browser/sync/profile_sync_service.h"
29#include "chrome/browser/sync/profile_sync_service_factory.h"
30#include "chrome/common/net/url_fixer_upper.h"
31#include "chrome/common/pref_names.h"
32#include "chrome/common/url_constants.h"
33#include "net/base/escape.h"
34#include "net/base/load_flags.h"
35#include "net/base/net_util.h"
36#include "net/http/http_request_headers.h"
37#include "net/http/http_response_headers.h"
38#include "net/url_request/url_fetcher.h"
39#include "net/url_request/url_request_status.h"
40#include "url/gurl.h"
41
42namespace {
43
44// TODO(hfung): The histogram code was copied and modified from
45// search_provider.cc.  Refactor and consolidate the code.
46// We keep track in a histogram how many suggest requests we send, how
47// many suggest requests we invalidate (e.g., due to a user typing
48// another character), and how many replies we receive.
49// *** ADD NEW ENUMS AFTER ALL PREVIOUSLY DEFINED ONES! ***
50//     (excluding the end-of-list enum value)
51// We do not want values of existing enums to change or else it screws
52// up the statistics.
53enum ZeroSuggestRequestsHistogramValue {
54  ZERO_SUGGEST_REQUEST_SENT = 1,
55  ZERO_SUGGEST_REQUEST_INVALIDATED,
56  ZERO_SUGGEST_REPLY_RECEIVED,
57  ZERO_SUGGEST_MAX_REQUEST_HISTOGRAM_VALUE
58};
59
60void LogOmniboxZeroSuggestRequest(
61    ZeroSuggestRequestsHistogramValue request_value) {
62  UMA_HISTOGRAM_ENUMERATION("Omnibox.ZeroSuggestRequests", request_value,
63                            ZERO_SUGGEST_MAX_REQUEST_HISTOGRAM_VALUE);
64}
65
66// The maximum relevance of the top match from this provider.
67const int kDefaultVerbatimZeroSuggestRelevance = 1300;
68
69// Relevance value to use if it was not set explicitly by the server.
70const int kDefaultZeroSuggestRelevance = 100;
71
72}  // namespace
73
74// static
75ZeroSuggestProvider* ZeroSuggestProvider::Create(
76    AutocompleteProviderListener* listener,
77    Profile* profile) {
78  return new ZeroSuggestProvider(listener, profile);
79}
80
81void ZeroSuggestProvider::Start(const AutocompleteInput& input,
82                                bool /*minimal_changes*/) {
83}
84
85void ZeroSuggestProvider::Stop(bool clear_cached_results) {
86  if (have_pending_request_)
87    LogOmniboxZeroSuggestRequest(ZERO_SUGGEST_REQUEST_INVALIDATED);
88  have_pending_request_ = false;
89  fetcher_.reset();
90  done_ = true;
91  if (clear_cached_results) {
92    query_matches_map_.clear();
93    navigation_results_.clear();
94    current_query_.clear();
95    matches_.clear();
96  }
97}
98
99void ZeroSuggestProvider::AddProviderInfo(ProvidersInfo* provider_info) const {
100  provider_info->push_back(metrics::OmniboxEventProto_ProviderInfo());
101  metrics::OmniboxEventProto_ProviderInfo& new_entry = provider_info->back();
102  new_entry.set_provider(AsOmniboxEventProviderType());
103  new_entry.set_provider_done(done_);
104  std::vector<uint32> field_trial_hashes;
105  OmniboxFieldTrial::GetActiveSuggestFieldTrialHashes(&field_trial_hashes);
106  for (size_t i = 0; i < field_trial_hashes.size(); ++i) {
107    if (field_trial_triggered_)
108      new_entry.mutable_field_trial_triggered()->Add(field_trial_hashes[i]);
109    if (field_trial_triggered_in_session_) {
110      new_entry.mutable_field_trial_triggered_in_session()->Add(
111          field_trial_hashes[i]);
112     }
113  }
114}
115
116void ZeroSuggestProvider::ResetSession() {
117  // The user has started editing in the omnibox, so leave
118  // |field_trial_triggered_in_session_| unchanged and set
119  // |field_trial_triggered_| to false since zero suggest is inactive now.
120  field_trial_triggered_ = false;
121  Stop(true);
122}
123
124void ZeroSuggestProvider::OnURLFetchComplete(const net::URLFetcher* source) {
125  have_pending_request_ = false;
126  LogOmniboxZeroSuggestRequest(ZERO_SUGGEST_REPLY_RECEIVED);
127
128  std::string json_data;
129  source->GetResponseAsString(&json_data);
130  const bool request_succeeded =
131      source->GetStatus().is_success() && source->GetResponseCode() == 200;
132
133  bool have_results = false;
134  if (request_succeeded) {
135    JSONStringValueSerializer deserializer(json_data);
136    deserializer.set_allow_trailing_comma(true);
137    scoped_ptr<Value> data(deserializer.Deserialize(NULL, NULL));
138    if (data.get()) {
139      ParseSuggestResults(*data.get());
140      have_results = !query_matches_map_.empty() ||
141          !navigation_results_.empty();
142    }
143  }
144  done_ = true;
145
146  if (have_results) {
147    ConvertResultsToAutocompleteMatches();
148    listener_->OnProviderUpdate(true);
149  }
150}
151
152void ZeroSuggestProvider::StartZeroSuggest(
153    const GURL& url,
154    AutocompleteInput::PageClassification page_classification,
155    const string16& permanent_text) {
156  Stop(true);
157  field_trial_triggered_ = false;
158  field_trial_triggered_in_session_ = false;
159  if (!ShouldRunZeroSuggest(url))
160    return;
161  verbatim_relevance_ = kDefaultVerbatimZeroSuggestRelevance;
162  done_ = false;
163  permanent_text_ = permanent_text;
164  current_query_ = url.spec();
165  current_page_classification_ = page_classification;
166  current_url_match_ = MatchForCurrentURL();
167  // TODO(jered): Consider adding locally-sourced zero-suggestions here too.
168  // These may be useful on the NTP or more relevant to the user than server
169  // suggestions, if based on local browsing history.
170  Run();
171}
172
173ZeroSuggestProvider::ZeroSuggestProvider(
174  AutocompleteProviderListener* listener,
175  Profile* profile)
176    : AutocompleteProvider(listener, profile,
177          AutocompleteProvider::TYPE_ZERO_SUGGEST),
178      template_url_service_(TemplateURLServiceFactory::GetForProfile(profile)),
179      have_pending_request_(false),
180      verbatim_relevance_(kDefaultVerbatimZeroSuggestRelevance),
181      field_trial_triggered_(false),
182      field_trial_triggered_in_session_(false) {
183}
184
185ZeroSuggestProvider::~ZeroSuggestProvider() {
186}
187
188bool ZeroSuggestProvider::ShouldRunZeroSuggest(const GURL& url) const {
189  if (!url.is_valid())
190    return false;
191
192  // Do not query non-http URLs. There will be no useful suggestions for https
193  // or chrome URLs.
194  if (url.scheme() != chrome::kHttpScheme)
195    return false;
196
197  // Don't run if there's no profile or in incognito mode.
198  if (profile_ == NULL || profile_->IsOffTheRecord())
199    return false;
200
201  // Don't run if we can't get preferences or search suggest is not enabled.
202  PrefService* prefs = profile_->GetPrefs();
203  if (prefs == NULL || !prefs->GetBoolean(prefs::kSearchSuggestEnabled))
204    return false;
205
206  ProfileSyncService* service =
207      ProfileSyncServiceFactory::GetInstance()->GetForProfile(profile_);
208  browser_sync::SyncPrefs sync_prefs(prefs);
209  // The user has needs to have Chrome Sync enabled (for permissions to
210  // transmit their current URL) and be in the field trial.
211  if (!OmniboxFieldTrial::InZeroSuggestFieldTrial() ||
212      service == NULL ||
213      !service->IsSyncEnabledAndLoggedIn() ||
214      !sync_prefs.HasKeepEverythingSynced()) {
215    return false;
216  }
217  return true;
218}
219
220void ZeroSuggestProvider::FillResults(
221    const Value& root_val,
222    int* verbatim_relevance,
223    SearchProvider::SuggestResults* suggest_results,
224    SearchProvider::NavigationResults* navigation_results) {
225  string16 query;
226  const ListValue* root_list = NULL;
227  const ListValue* results = NULL;
228  const ListValue* relevances = NULL;
229  // The response includes the query, which should be empty for ZeroSuggest
230  // responses.
231  if (!root_val.GetAsList(&root_list) || !root_list->GetString(0, &query) ||
232      (!query.empty()) || !root_list->GetList(1, &results))
233    return;
234
235  // 3rd element: Description list.
236  const ListValue* descriptions = NULL;
237  root_list->GetList(2, &descriptions);
238
239  // 4th element: Disregard the query URL list for now.
240
241  // Reset suggested relevance information from the provider.
242  *verbatim_relevance = kDefaultVerbatimZeroSuggestRelevance;
243
244  // 5th element: Optional key-value pairs from the Suggest server.
245  const ListValue* types = NULL;
246  const DictionaryValue* extras = NULL;
247  if (root_list->GetDictionary(4, &extras)) {
248    extras->GetList("google:suggesttype", &types);
249
250    // Discard this list if its size does not match that of the suggestions.
251    if (extras->GetList("google:suggestrelevance", &relevances) &&
252        relevances->GetSize() != results->GetSize())
253      relevances = NULL;
254    extras->GetInteger("google:verbatimrelevance", verbatim_relevance);
255
256    // Check if the active suggest field trial (if any) has triggered.
257    bool triggered = false;
258    extras->GetBoolean("google:fieldtrialtriggered", &triggered);
259    field_trial_triggered_ |= triggered;
260    field_trial_triggered_in_session_ |= triggered;
261  }
262
263  // Clear the previous results now that new results are available.
264  suggest_results->clear();
265  navigation_results->clear();
266
267  string16 result, title;
268  std::string type;
269  for (size_t index = 0; results->GetString(index, &result); ++index) {
270    // Google search may return empty suggestions for weird input characters,
271    // they make no sense at all and can cause problems in our code.
272    if (result.empty())
273      continue;
274
275    int relevance = kDefaultZeroSuggestRelevance;
276
277    // Apply valid suggested relevance scores; discard invalid lists.
278    if (relevances != NULL && !relevances->GetInteger(index, &relevance))
279      relevances = NULL;
280    if (types && types->GetString(index, &type) && (type == "NAVIGATION")) {
281      // Do not blindly trust the URL coming from the server to be valid.
282      GURL url(URLFixerUpper::FixupURL(UTF16ToUTF8(result), std::string()));
283      if (url.is_valid()) {
284        if (descriptions != NULL)
285          descriptions->GetString(index, &title);
286        navigation_results->push_back(SearchProvider::NavigationResult(
287            *this, url, title, false, relevance, relevances != NULL));
288      }
289    } else {
290      suggest_results->push_back(SearchProvider::SuggestResult(
291          result, false, relevance, relevances != NULL));
292    }
293  }
294}
295
296void ZeroSuggestProvider::AddSuggestResultsToMap(
297    const SearchProvider::SuggestResults& results,
298    const TemplateURL* template_url,
299    SearchProvider::MatchMap* map) {
300  for (size_t i = 0; i < results.size(); ++i) {
301    AddMatchToMap(results[i].relevance(), AutocompleteMatchType::SEARCH_SUGGEST,
302                  template_url, results[i].suggestion(), i, map);
303  }
304}
305
306void ZeroSuggestProvider::AddMatchToMap(int relevance,
307                                        AutocompleteMatch::Type type,
308                                        const TemplateURL* template_url,
309                                        const string16& query_string,
310                                        int accepted_suggestion,
311                                        SearchProvider::MatchMap* map) {
312  // Pass in query_string as the input_text since we don't want any bolding.
313  // TODO(samarth|melevin): use the actual omnibox margin here as well instead
314  // of passing in -1.
315  AutocompleteMatch match = SearchProvider::CreateSearchSuggestion(
316      this, relevance, type, template_url, query_string, query_string,
317      AutocompleteInput(), false, accepted_suggestion, -1, true);
318  if (!match.destination_url.is_valid())
319    return;
320
321  // Try to add |match| to |map|.  If a match for |query_string| is already in
322  // |map|, replace it if |match| is more relevant.
323  // NOTE: Keep this ToLower() call in sync with url_database.cc.
324  const std::pair<SearchProvider::MatchMap::iterator, bool> i(map->insert(
325      std::make_pair(base::i18n::ToLower(query_string), match)));
326  // NOTE: We purposefully do a direct relevance comparison here instead of
327  // using AutocompleteMatch::MoreRelevant(), so that we'll prefer "items added
328  // first" rather than "items alphabetically first" when the scores are equal.
329  // The only case this matters is when a user has results with the same score
330  // that differ only by capitalization; because the history system returns
331  // results sorted by recency, this means we'll pick the most recent such
332  // result even if the precision of our relevance score is too low to
333  // distinguish the two.
334  if (!i.second && (match.relevance > i.first->second.relevance))
335    i.first->second = match;
336}
337
338AutocompleteMatch ZeroSuggestProvider::NavigationToMatch(
339    const SearchProvider::NavigationResult& navigation) {
340  AutocompleteMatch match(this, navigation.relevance(), false,
341                          AutocompleteMatchType::NAVSUGGEST);
342  match.destination_url = navigation.url();
343
344  const std::string languages(
345      profile_->GetPrefs()->GetString(prefs::kAcceptLanguages));
346  match.contents = net::FormatUrl(navigation.url(), languages,
347      net::kFormatUrlOmitAll, net::UnescapeRule::SPACES, NULL, NULL, NULL);
348  match.fill_into_edit +=
349      AutocompleteInput::FormattedStringWithEquivalentMeaning(navigation.url(),
350          match.contents);
351
352  AutocompleteMatch::ClassifyLocationInString(string16::npos, 0,
353      match.contents.length(), ACMatchClassification::URL,
354      &match.contents_class);
355
356  match.description =
357      AutocompleteMatch::SanitizeString(navigation.description());
358  AutocompleteMatch::ClassifyLocationInString(string16::npos, 0,
359      match.description.length(), ACMatchClassification::NONE,
360      &match.description_class);
361  return match;
362}
363
364void ZeroSuggestProvider::Run() {
365  have_pending_request_ = false;
366  const int kFetcherID = 1;
367
368  const TemplateURL* default_provider =
369     template_url_service_->GetDefaultSearchProvider();
370  // TODO(hfung): Generalize if the default provider supports zero suggest.
371  // Only make the request if we know that the provider supports zero suggest
372  // (currently only the prepopulated Google provider).
373  if (default_provider == NULL || !default_provider->SupportsReplacement() ||
374      default_provider->prepopulate_id() != 1) {
375    Stop(true);
376    return;
377  }
378  string16 prefix;
379  TemplateURLRef::SearchTermsArgs search_term_args(prefix);
380  search_term_args.zero_prefix_url = current_query_;
381  std::string req_url = default_provider->suggestions_url_ref().
382      ReplaceSearchTerms(search_term_args);
383  GURL suggest_url(req_url);
384  // Make sure we are sending the suggest request through HTTPS.
385  if (!suggest_url.SchemeIs(chrome::kHttpsScheme)) {
386    Stop(true);
387    return;
388  }
389
390  fetcher_.reset(
391      net::URLFetcher::Create(kFetcherID,
392          suggest_url,
393          net::URLFetcher::GET, this));
394  fetcher_->SetRequestContext(profile_->GetRequestContext());
395  fetcher_->SetLoadFlags(net::LOAD_DO_NOT_SAVE_COOKIES);
396  // Add Chrome experiment state to the request headers.
397  net::HttpRequestHeaders headers;
398  chrome_variations::VariationsHttpHeaderProvider::GetInstance()->AppendHeaders(
399      fetcher_->GetOriginalURL(), profile_->IsOffTheRecord(), false, &headers);
400  fetcher_->SetExtraRequestHeaders(headers.ToString());
401
402  fetcher_->Start();
403  have_pending_request_ = true;
404  LogOmniboxZeroSuggestRequest(ZERO_SUGGEST_REQUEST_SENT);
405}
406
407void ZeroSuggestProvider::ParseSuggestResults(const Value& root_val) {
408  SearchProvider::SuggestResults suggest_results;
409  FillResults(root_val, &verbatim_relevance_,
410              &suggest_results, &navigation_results_);
411
412  query_matches_map_.clear();
413  AddSuggestResultsToMap(suggest_results,
414                         template_url_service_->GetDefaultSearchProvider(),
415                         &query_matches_map_);
416}
417
418void ZeroSuggestProvider::ConvertResultsToAutocompleteMatches() {
419  matches_.clear();
420
421  const TemplateURL* default_provider =
422      template_url_service_->GetDefaultSearchProvider();
423  // Fail if we can't set the clickthrough URL for query suggestions.
424  if (default_provider == NULL || !default_provider->SupportsReplacement())
425    return;
426
427  const int num_query_results = query_matches_map_.size();
428  const int num_nav_results = navigation_results_.size();
429  const int num_results = num_query_results + num_nav_results;
430  UMA_HISTOGRAM_COUNTS("ZeroSuggest.QueryResults", num_query_results);
431  UMA_HISTOGRAM_COUNTS("ZeroSuggest.URLResults",  num_nav_results);
432  UMA_HISTOGRAM_COUNTS("ZeroSuggest.AllResults", num_results);
433
434  if (num_results == 0)
435    return;
436
437  // TODO(jered): Rip this out once the first match is decoupled from the
438  // current typing in the omnibox.
439  matches_.push_back(current_url_match_);
440
441  for (SearchProvider::MatchMap::const_iterator it(query_matches_map_.begin());
442       it != query_matches_map_.end(); ++it)
443    matches_.push_back(it->second);
444
445  for (SearchProvider::NavigationResults::const_iterator it(
446       navigation_results_.begin()); it != navigation_results_.end(); ++it)
447    matches_.push_back(NavigationToMatch(*it));
448}
449
450AutocompleteMatch ZeroSuggestProvider::MatchForCurrentURL() {
451  AutocompleteInput input(permanent_text_, string16::npos, string16(),
452                          GURL(current_query_), current_page_classification_,
453                          false, false, true, AutocompleteInput::ALL_MATCHES);
454
455  AutocompleteMatch match(
456      HistoryURLProvider::SuggestExactInput(this, input,
457                                            !HasHTTPScheme(input.text())));
458  match.is_history_what_you_typed_match = false;
459
460  // The placeholder suggestion for the current URL has high relevance so
461  // that it is in the first suggestion slot and inline autocompleted. It
462  // gets dropped as soon as the user types something.
463  match.relevance = verbatim_relevance_;
464
465  return match;
466}
467