history_quick_provider.cc revision c5cede9ae108bb15f6b7a8aea21c7e1fefa2834c 
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/net/url_fixer_upper.h"
37#include "chrome/common/pref_names.h"
38#include "chrome/common/url_constants.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() == AutocompleteInput::INVALID) ||
69      (input.type() == AutocompleteInput::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
92void HistoryQuickProvider::DeleteMatch(const AutocompleteMatch& match) {
93  DCHECK(match.deletable);
94  DCHECK(match.destination_url.is_valid());
95  // Delete the match from the InMemoryURLIndex.
96  GetIndex()->DeleteURL(match.destination_url);
97  DeleteMatchFromMatches(match);
98}
99
100HistoryQuickProvider::~HistoryQuickProvider() {}
101
102void HistoryQuickProvider::DoAutocomplete() {
103  // Get the matching URLs from the DB.
104  ScoredHistoryMatches matches = GetIndex()->HistoryItemsForTerms(
105      autocomplete_input_.text(),
106      autocomplete_input_.cursor_position());
107  if (matches.empty())
108    return;
109
110  // Figure out if HistoryURL provider has a URL-what-you-typed match
111  // that ought to go first and what its score will be.
112  bool will_have_url_what_you_typed_match_first = false;
113  int url_what_you_typed_match_score = -1;  // undefined
114  // These are necessary (but not sufficient) conditions for the omnibox
115  // input to be a URL-what-you-typed match.  The username test checks that
116  // either the username does not exist (a regular URL such as http://site/)
117  // or, if the username exists (http://user@site/), there must be either
118  // a password or a port.  Together these exclude pure username@site
119  // inputs because these are likely to be an e-mail address.  HistoryURL
120  // provider won't promote the URL-what-you-typed match to first
121  // for these inputs.
122  const bool can_have_url_what_you_typed_match_first =
123      autocomplete_input_.canonicalized_url().is_valid() &&
124      (autocomplete_input_.type() != AutocompleteInput::QUERY) &&
125      (autocomplete_input_.type() != AutocompleteInput::FORCED_QUERY) &&
126      (!autocomplete_input_.parts().username.is_nonempty() ||
127       autocomplete_input_.parts().password.is_nonempty() ||
128       autocomplete_input_.parts().path.is_nonempty());
129  if (can_have_url_what_you_typed_match_first) {
130    HistoryService* const history_service =
131        HistoryServiceFactory::GetForProfile(profile_,
132                                             Profile::EXPLICIT_ACCESS);
133    // We expect HistoryService to be available.  In case it's not,
134    // (e.g., due to Profile corruption) we let HistoryQuick provider
135    // completions (which may be available because it's a different
136    // data structure) compete with the URL-what-you-typed match as
137    // normal.
138    if (history_service) {
139      history::URLDatabase* url_db = history_service->InMemoryDatabase();
140      // url_db can be NULL if it hasn't finished initializing (or
141      // failed to to initialize).  In this case, we let HistoryQuick
142      // provider completions compete with the URL-what-you-typed
143      // match as normal.
144      if (url_db) {
145        const std::string host(base::UTF16ToUTF8(
146            autocomplete_input_.text().substr(
147                autocomplete_input_.parts().host.begin,
148                autocomplete_input_.parts().host.len)));
149        // We want to put the URL-what-you-typed match first if either
150        // * the user visited the URL before (intranet or internet).
151        // * it's a URL on a host that user visited before and this
152        //   is the root path of the host.  (If the user types some
153        //   of a path--more than a simple "/"--we let autocomplete compete
154        //   normally with the URL-what-you-typed match.)
155        // TODO(mpearson): Remove this hacky code and simply score URL-what-
156        // you-typed in some sane way relative to possible completions:
157        // URL-what-you-typed should get some sort of a boost relative
158        // to completions, but completions should naturally win if
159        // they're a lot more popular.  In this process, if the input
160        // is a bare intranet hostname that has been visited before, we
161        // may want to enforce that the only completions that can outscore
162        // the URL-what-you-typed match are on the same host (i.e., aren't
163        // from a longer internet hostname for which the omnibox input is
164        // a prefix).
165        if (url_db->GetRowForURL(
166            autocomplete_input_.canonicalized_url(), NULL) != 0) {
167          // We visited this URL before.
168          will_have_url_what_you_typed_match_first = true;
169          // HistoryURLProvider gives visited what-you-typed URLs a high score.
170          url_what_you_typed_match_score =
171              HistoryURLProvider::kScoreForBestInlineableResult;
172        } else if (url_db->IsTypedHost(host) &&
173             (!autocomplete_input_.parts().path.is_nonempty() ||
174              ((autocomplete_input_.parts().path.len == 1) &&
175               (autocomplete_input_.text()[
176                   autocomplete_input_.parts().path.begin] == '/'))) &&
177             !autocomplete_input_.parts().query.is_nonempty() &&
178             !autocomplete_input_.parts().ref.is_nonempty()) {
179          // Not visited, but we've seen the host before.
180          will_have_url_what_you_typed_match_first = true;
181          const size_t registry_length =
182              net::registry_controlled_domains::GetRegistryLength(
183                  host,
184                  net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
185                  net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES);
186          if (registry_length == 0) {
187            // Known intranet hosts get one score.
188            url_what_you_typed_match_score =
189                HistoryURLProvider::kScoreForUnvisitedIntranetResult;
190          } else {
191            // Known internet hosts get another.
192            url_what_you_typed_match_score =
193                HistoryURLProvider::kScoreForWhatYouTypedResult;
194          }
195        }
196      }
197    }
198  }
199
200  // Loop over every result and add it to matches_.  In the process,
201  // guarantee that scores are decreasing.  |max_match_score| keeps
202  // track of the highest score we can assign to any later results we
203  // see.  Also, if we're not allowing inline autocompletions in
204  // general or the current best suggestion isn't inlineable,
205  // artificially reduce the starting |max_match_score| (which
206  // therefore applies to all results) to something low enough that
207  // guarantees no result will be offered as an inline autocomplete
208  // suggestion.  Also do a similar reduction if we think there will be
209  // a URL-what-you-typed match.  (We want URL-what-you-typed matches for
210  // visited URLs to beat out any longer URLs, no matter how frequently
211  // they're visited.)  The strength of this last reduction depends on the
212  // likely score for the URL-what-you-typed result.
213
214  // |template_url_service| or |template_url| can be NULL in unit tests.
215  TemplateURLService* template_url_service =
216      TemplateURLServiceFactory::GetForProfile(profile_);
217  TemplateURL* template_url = template_url_service ?
218      template_url_service->GetDefaultSearchProvider() : NULL;
219  int max_match_score =
220      (OmniboxFieldTrial::ReorderForLegalDefaultMatch(
221         autocomplete_input_.current_page_classification()) ||
222       (!PreventInlineAutocomplete(autocomplete_input_) &&
223        matches.begin()->can_inline())) ?
224      matches.begin()->raw_score() :
225      (AutocompleteResult::kLowestDefaultScore - 1);
226  if (will_have_url_what_you_typed_match_first) {
227    max_match_score = std::min(max_match_score,
228        url_what_you_typed_match_score - 1);
229  }
230  for (ScoredHistoryMatches::const_iterator match_iter = matches.begin();
231       match_iter != matches.end(); ++match_iter) {
232    const ScoredHistoryMatch& history_match(*match_iter);
233    // Culls results corresponding to queries from the default search engine.
234    // These are low-quality, difficult-to-understand matches for users, and the
235    // SearchProvider should surface past queries in a better way anyway.
236    if (!template_url ||
237        !template_url->IsSearchURL(history_match.url_info.url())) {
238      // Set max_match_score to the score we'll assign this result:
239      max_match_score = std::min(max_match_score, history_match.raw_score());
240      matches_.push_back(QuickMatchToACMatch(history_match, max_match_score));
241      // Mark this max_match_score as being used:
242      max_match_score--;
243    }
244  }
245}
246
247AutocompleteMatch HistoryQuickProvider::QuickMatchToACMatch(
248    const ScoredHistoryMatch& history_match,
249    int score) {
250  const history::URLRow& info = history_match.url_info;
251  AutocompleteMatch match(
252      this, score, !!info.visit_count(),
253      history_match.url_matches().empty() ?
254          AutocompleteMatchType::HISTORY_TITLE :
255          AutocompleteMatchType::HISTORY_URL);
256  match.typed_count = info.typed_count();
257  match.destination_url = info.url();
258  DCHECK(match.destination_url.is_valid());
259
260  // Format the URL autocomplete presentation.
261  std::vector<size_t> offsets =
262      OffsetsFromTermMatches(history_match.url_matches());
263  const net::FormatUrlTypes format_types = net::kFormatUrlOmitAll &
264      ~(!history_match.match_in_scheme ? 0 : net::kFormatUrlOmitHTTP);
265  match.fill_into_edit =
266      AutocompleteInput::FormattedStringWithEquivalentMeaning(info.url(),
267          net::FormatUrlWithOffsets(info.url(), languages_, format_types,
268              net::UnescapeRule::SPACES, NULL, NULL, &offsets));
269  history::TermMatches new_matches =
270      ReplaceOffsetsInTermMatches(history_match.url_matches(), offsets);
271  match.contents = net::FormatUrl(info.url(), languages_, format_types,
272              net::UnescapeRule::SPACES, NULL, NULL, NULL);
273  match.contents_class =
274      SpansFromTermMatch(new_matches, match.contents.length(), true);
275
276  if (history_match.can_inline()) {
277    DCHECK(!new_matches.empty());
278    size_t inline_autocomplete_offset = new_matches[0].offset +
279        new_matches[0].length;
280    // |inline_autocomplete_offset| may be beyond the end of the
281    // |fill_into_edit| if the user has typed an URL with a scheme and the
282    // last character typed is a slash.  That slash is removed by the
283    // FormatURLWithOffsets call above.
284    if (inline_autocomplete_offset < match.fill_into_edit.length()) {
285      match.inline_autocompletion =
286          match.fill_into_edit.substr(inline_autocomplete_offset);
287    }
288    match.allowed_to_be_default_match = match.inline_autocompletion.empty() ||
289        !PreventInlineAutocomplete(autocomplete_input_);
290  }
291
292  // Format the description autocomplete presentation.
293  match.description = info.title();
294  match.description_class = SpansFromTermMatch(
295      history_match.title_matches(), match.description.length(), false);
296
297  match.RecordAdditionalInfo("typed count", info.typed_count());
298  match.RecordAdditionalInfo("visit count", info.visit_count());
299  match.RecordAdditionalInfo("last visit", info.last_visit());
300
301  return match;
302}
303
304history::InMemoryURLIndex* HistoryQuickProvider::GetIndex() {
305  if (index_for_testing_.get())
306    return index_for_testing_.get();
307
308  HistoryService* const history_service =
309      HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
310  if (!history_service)
311    return NULL;
312
313  return history_service->InMemoryIndex();
314}
315