history_quick_provider.cc revision f8ee788a64d60abd8f2d742a5fdedde054ecd910 
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/history_quick_provider.h"
6
7#include <vector>
8
9#include "base/basictypes.h"
10#include "base/command_line.h"
11#include "base/i18n/break_iterator.h"
12#include "base/logging.h"
13#include "base/metrics/field_trial.h"
14#include "base/metrics/histogram.h"
15#include "base/prefs/pref_service.h"
16#include "base/strings/string_number_conversions.h"
17#include "base/strings/string_util.h"
18#include "base/strings/utf_string_conversions.h"
19#include "base/time/time.h"
20#include "chrome/browser/autocomplete/autocomplete_result.h"
21#include "chrome/browser/autocomplete/history_url_provider.h"
22#include "chrome/browser/history/history_database.h"
23#include "chrome/browser/history/history_service.h"
24#include "chrome/browser/history/history_service_factory.h"
25#include "chrome/browser/history/in_memory_url_index.h"
26#include "chrome/browser/history/in_memory_url_index_types.h"
27#include "chrome/browser/history/scored_history_match.h"
28#include "chrome/browser/omnibox/omnibox_field_trial.h"
29#include "chrome/browser/profiles/profile.h"
30#include "chrome/browser/search/search.h"
31#include "chrome/browser/search_engines/template_url.h"
32#include "chrome/browser/search_engines/template_url_service.h"
33#include "chrome/browser/search_engines/template_url_service_factory.h"
34#include "chrome/common/autocomplete_match_type.h"
35#include "chrome/common/chrome_switches.h"
36#include "chrome/common/pref_names.h"
37#include "chrome/common/url_constants.h"
38#include "components/metrics/proto/omnibox_input_type.pb.h"
39#include "content/public/browser/notification_source.h"
40#include "content/public/browser/notification_types.h"
41#include "net/base/escape.h"
42#include "net/base/net_util.h"
43#include "net/base/registry_controlled_domains/registry_controlled_domain.h"
44#include "url/url_parse.h"
45#include "url/url_util.h"
46
47using history::InMemoryURLIndex;
48using history::ScoredHistoryMatch;
49using history::ScoredHistoryMatches;
50
51bool HistoryQuickProvider::disabled_ = false;
52
53HistoryQuickProvider::HistoryQuickProvider(
54    AutocompleteProviderListener* listener,
55    Profile* profile)
56    : HistoryProvider(listener, profile,
57          AutocompleteProvider::TYPE_HISTORY_QUICK),
58      languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)) {
59}
60
61void HistoryQuickProvider::Start(const AutocompleteInput& input,
62                                 bool minimal_changes) {
63  matches_.clear();
64  if (disabled_)
65    return;
66
67  // Don't bother with INVALID and FORCED_QUERY.
68  if ((input.type() == metrics::OmniboxInputType::INVALID) ||
69      (input.type() == metrics::OmniboxInputType::FORCED_QUERY))
70    return;
71
72  autocomplete_input_ = input;
73
74  // TODO(pkasting): We should just block here until this loads.  Any time
75  // someone unloads the history backend, we'll get inconsistent inline
76  // autocomplete behavior here.
77  if (GetIndex()) {
78    base::TimeTicks start_time = base::TimeTicks::Now();
79    DoAutocomplete();
80    if (input.text().length() < 6) {
81      base::TimeTicks end_time = base::TimeTicks::Now();
82      std::string name = "HistoryQuickProvider.QueryIndexTime." +
83          base::IntToString(input.text().length());
84      base::HistogramBase* counter = base::Histogram::FactoryGet(
85          name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
86      counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
87    }
88    UpdateStarredStateOfMatches();
89  }
90}
91
92HistoryQuickProvider::~HistoryQuickProvider() {}
93
94void HistoryQuickProvider::DoAutocomplete() {
95  // Get the matching URLs from the DB.
96  ScoredHistoryMatches matches = GetIndex()->HistoryItemsForTerms(
97      autocomplete_input_.text(),
98      autocomplete_input_.cursor_position(),
99      AutocompleteProvider::kMaxMatches);
100  if (matches.empty())
101    return;
102
103  // Figure out if HistoryURL provider has a URL-what-you-typed match
104  // that ought to go first and what its score will be.
105  bool will_have_url_what_you_typed_match_first = false;
106  int url_what_you_typed_match_score = -1;  // undefined
107  // These are necessary (but not sufficient) conditions for the omnibox
108  // input to be a URL-what-you-typed match.  The username test checks that
109  // either the username does not exist (a regular URL such as http://site/)
110  // or, if the username exists (http://user@site/), there must be either
111  // a password or a port.  Together these exclude pure username@site
112  // inputs because these are likely to be an e-mail address.  HistoryURL
113  // provider won't promote the URL-what-you-typed match to first
114  // for these inputs.
115  const bool can_have_url_what_you_typed_match_first =
116      (autocomplete_input_.type() != metrics::OmniboxInputType::QUERY) &&
117      (!autocomplete_input_.parts().username.is_nonempty() ||
118       autocomplete_input_.parts().password.is_nonempty() ||
119       autocomplete_input_.parts().path.is_nonempty());
120  if (can_have_url_what_you_typed_match_first) {
121    HistoryService* const history_service =
122        HistoryServiceFactory::GetForProfile(profile_,
123                                             Profile::EXPLICIT_ACCESS);
124    // We expect HistoryService to be available.  In case it's not,
125    // (e.g., due to Profile corruption) we let HistoryQuick provider
126    // completions (which may be available because it's a different
127    // data structure) compete with the URL-what-you-typed match as
128    // normal.
129    if (history_service) {
130      history::URLDatabase* url_db = history_service->InMemoryDatabase();
131      // url_db can be NULL if it hasn't finished initializing (or
132      // failed to to initialize).  In this case, we let HistoryQuick
133      // provider completions compete with the URL-what-you-typed
134      // match as normal.
135      if (url_db) {
136        const std::string host(base::UTF16ToUTF8(
137            autocomplete_input_.text().substr(
138                autocomplete_input_.parts().host.begin,
139                autocomplete_input_.parts().host.len)));
140        // We want to put the URL-what-you-typed match first if either
141        // * the user visited the URL before (intranet or internet).
142        // * it's a URL on a host that user visited before and this
143        //   is the root path of the host.  (If the user types some
144        //   of a path--more than a simple "/"--we let autocomplete compete
145        //   normally with the URL-what-you-typed match.)
146        // TODO(mpearson): Remove this hacky code and simply score URL-what-
147        // you-typed in some sane way relative to possible completions:
148        // URL-what-you-typed should get some sort of a boost relative
149        // to completions, but completions should naturally win if
150        // they're a lot more popular.  In this process, if the input
151        // is a bare intranet hostname that has been visited before, we
152        // may want to enforce that the only completions that can outscore
153        // the URL-what-you-typed match are on the same host (i.e., aren't
154        // from a longer internet hostname for which the omnibox input is
155        // a prefix).
156        if (url_db->GetRowForURL(
157            autocomplete_input_.canonicalized_url(), NULL) != 0) {
158          // We visited this URL before.
159          will_have_url_what_you_typed_match_first = true;
160          // HistoryURLProvider gives visited what-you-typed URLs a high score.
161          url_what_you_typed_match_score =
162              HistoryURLProvider::kScoreForBestInlineableResult;
163        } else if (url_db->IsTypedHost(host) &&
164             (!autocomplete_input_.parts().path.is_nonempty() ||
165              ((autocomplete_input_.parts().path.len == 1) &&
166               (autocomplete_input_.text()[
167                   autocomplete_input_.parts().path.begin] == '/'))) &&
168             !autocomplete_input_.parts().query.is_nonempty() &&
169             !autocomplete_input_.parts().ref.is_nonempty()) {
170          // Not visited, but we've seen the host before.
171          will_have_url_what_you_typed_match_first = true;
172          const size_t registry_length =
173              net::registry_controlled_domains::GetRegistryLength(
174                  host,
175                  net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
176                  net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES);
177          if (registry_length == 0) {
178            // Known intranet hosts get one score.
179            url_what_you_typed_match_score =
180                HistoryURLProvider::kScoreForUnvisitedIntranetResult;
181          } else {
182            // Known internet hosts get another.
183            url_what_you_typed_match_score =
184                HistoryURLProvider::kScoreForWhatYouTypedResult;
185          }
186        }
187      }
188    }
189  }
190
191  // Loop over every result and add it to matches_.  In the process,
192  // guarantee that scores are decreasing.  |max_match_score| keeps
193  // track of the highest score we can assign to any later results we
194  // see.  Also, reduce |max_match_score| if we think there will be
195  // a URL-what-you-typed match.  (We want URL-what-you-typed matches for
196  // visited URLs to beat out any longer URLs, no matter how frequently
197  // they're visited.)  The strength of this reduction depends on the
198  // likely score for the URL-what-you-typed result.
199
200  // |template_url_service| or |template_url| can be NULL in unit tests.
201  TemplateURLService* template_url_service =
202      TemplateURLServiceFactory::GetForProfile(profile_);
203  TemplateURL* template_url = template_url_service ?
204      template_url_service->GetDefaultSearchProvider() : NULL;
205  int max_match_score = matches.begin()->raw_score();
206  if (will_have_url_what_you_typed_match_first) {
207    max_match_score = std::min(max_match_score,
208        url_what_you_typed_match_score - 1);
209  }
210  for (ScoredHistoryMatches::const_iterator match_iter = matches.begin();
211       match_iter != matches.end(); ++match_iter) {
212    const ScoredHistoryMatch& history_match(*match_iter);
213    // Culls results corresponding to queries from the default search engine.
214    // These are low-quality, difficult-to-understand matches for users, and the
215    // SearchProvider should surface past queries in a better way anyway.
216    if (!template_url ||
217        !template_url->IsSearchURL(history_match.url_info.url(),
218                                   template_url_service->search_terms_data())) {
219      // Set max_match_score to the score we'll assign this result:
220      max_match_score = std::min(max_match_score, history_match.raw_score());
221      matches_.push_back(QuickMatchToACMatch(history_match, max_match_score));
222      // Mark this max_match_score as being used:
223      max_match_score--;
224    }
225  }
226}
227
228AutocompleteMatch HistoryQuickProvider::QuickMatchToACMatch(
229    const ScoredHistoryMatch& history_match,
230    int score) {
231  const history::URLRow& info = history_match.url_info;
232  AutocompleteMatch match(
233      this, score, !!info.visit_count(),
234      history_match.url_matches().empty() ?
235          AutocompleteMatchType::HISTORY_TITLE :
236          AutocompleteMatchType::HISTORY_URL);
237  match.typed_count = info.typed_count();
238  match.destination_url = info.url();
239  DCHECK(match.destination_url.is_valid());
240
241  // Format the URL autocomplete presentation.
242  const net::FormatUrlTypes format_types = net::kFormatUrlOmitAll &
243      ~(!history_match.match_in_scheme ? 0 : net::kFormatUrlOmitHTTP);
244  match.fill_into_edit =
245      AutocompleteInput::FormattedStringWithEquivalentMeaning(
246          info.url(),
247          net::FormatUrl(info.url(), languages_, format_types,
248                         net::UnescapeRule::SPACES, NULL, NULL, NULL));
249  std::vector<size_t> offsets =
250      OffsetsFromTermMatches(history_match.url_matches());
251  base::OffsetAdjuster::Adjustments adjustments;
252  match.contents = net::FormatUrlWithAdjustments(
253      info.url(), languages_, format_types, net::UnescapeRule::SPACES, NULL,
254      NULL, &adjustments);
255  base::OffsetAdjuster::AdjustOffsets(adjustments, &offsets);
256  history::TermMatches new_matches =
257      ReplaceOffsetsInTermMatches(history_match.url_matches(), offsets);
258  match.contents_class =
259      SpansFromTermMatch(new_matches, match.contents.length(), true);
260
261  // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
262  if (history_match.can_inline()) {
263    DCHECK(!new_matches.empty());
264    size_t inline_autocomplete_offset = new_matches[0].offset +
265        new_matches[0].length;
266    // |inline_autocomplete_offset| may be beyond the end of the
267    // |fill_into_edit| if the user has typed an URL with a scheme and the
268    // last character typed is a slash.  That slash is removed by the
269    // FormatURLWithOffsets call above.
270    if (inline_autocomplete_offset < match.fill_into_edit.length()) {
271      match.inline_autocompletion =
272          match.fill_into_edit.substr(inline_autocomplete_offset);
273    }
274    match.allowed_to_be_default_match = match.inline_autocompletion.empty() ||
275        !PreventInlineAutocomplete(autocomplete_input_);
276  }
277
278  // Format the description autocomplete presentation.
279  match.description = info.title();
280  match.description_class = SpansFromTermMatch(
281      history_match.title_matches(), match.description.length(), false);
282
283  match.RecordAdditionalInfo("typed count", info.typed_count());
284  match.RecordAdditionalInfo("visit count", info.visit_count());
285  match.RecordAdditionalInfo("last visit", info.last_visit());
286
287  return match;
288}
289
290history::InMemoryURLIndex* HistoryQuickProvider::GetIndex() {
291  if (index_for_testing_.get())
292    return index_for_testing_.get();
293
294  HistoryService* const history_service =
295      HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
296  if (!history_service)
297    return NULL;
298
299  return history_service->InMemoryIndex();
300}
301