chrome/browser/autocomplete/history_quick_provider.cc

   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
  47 using history::InMemoryURLIndex;
  48 using history::ScoredHistoryMatch;
  49 using history::ScoredHistoryMatches;
  50
  51 bool HistoryQuickProvider::disabled_ = false;
  52
  53 HistoryQuickProvider::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
  61 void 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
  92 void 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
 100 HistoryQuickProvider::~HistoryQuickProvider() {}
 101
 102 void 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       AutocompleteProvider::kMaxMatches);
 108   if (matches.empty())
 109     return;
 110
 111   // Figure out if HistoryURL provider has a URL-what-you-typed match
 112   // that ought to go first and what its score will be.
 113   bool will_have_url_what_you_typed_match_first = false;
 114   int url_what_you_typed_match_score = -1;  // undefined
 115   // These are necessary (but not sufficient) conditions for the omnibox
 116   // input to be a URL-what-you-typed match.  The username test checks that
 117   // either the username does not exist (a regular URL such as http://site/)
 118   // or, if the username exists (http://user@site/), there must be either
 119   // a password or a port.  Together these exclude pure username@site
 120   // inputs because these are likely to be an e-mail address.  HistoryURL
 121   // provider won't promote the URL-what-you-typed match to first
 122   // for these inputs.
 123   const bool can_have_url_what_you_typed_match_first =
 124       (autocomplete_input_.type() != AutocompleteInput::QUERY) &&
 125       (!autocomplete_input_.parts().username.is_nonempty() ||
 126        autocomplete_input_.parts().password.is_nonempty() ||
 127        autocomplete_input_.parts().path.is_nonempty());
 128   if (can_have_url_what_you_typed_match_first) {
 129     HistoryService* const history_service =
 130         HistoryServiceFactory::GetForProfile(profile_,
 131                                              Profile::EXPLICIT_ACCESS);
 132     // We expect HistoryService to be available.  In case it's not,
 133     // (e.g., due to Profile corruption) we let HistoryQuick provider
 134     // completions (which may be available because it's a different
 135     // data structure) compete with the URL-what-you-typed match as
 136     // normal.
 137     if (history_service) {
 138       history::URLDatabase* url_db = history_service->InMemoryDatabase();
 139       // url_db can be NULL if it hasn't finished initializing (or
 140       // failed to to initialize).  In this case, we let HistoryQuick
 141       // provider completions compete with the URL-what-you-typed
 142       // match as normal.
 143       if (url_db) {
 144         const std::string host(base::UTF16ToUTF8(
 145             autocomplete_input_.text().substr(
 146                 autocomplete_input_.parts().host.begin,
 147                 autocomplete_input_.parts().host.len)));
 148         // We want to put the URL-what-you-typed match first if either
 149         // * the user visited the URL before (intranet or internet).
 150         // * it's a URL on a host that user visited before and this
 151         //   is the root path of the host.  (If the user types some
 152         //   of a path--more than a simple "/"--we let autocomplete compete
 153         //   normally with the URL-what-you-typed match.)
 154         // TODO(mpearson): Remove this hacky code and simply score URL-what-
 155         // you-typed in some sane way relative to possible completions:
 156         // URL-what-you-typed should get some sort of a boost relative
 157         // to completions, but completions should naturally win if
 158         // they're a lot more popular.  In this process, if the input
 159         // is a bare intranet hostname that has been visited before, we
 160         // may want to enforce that the only completions that can outscore
 161         // the URL-what-you-typed match are on the same host (i.e., aren't
 162         // from a longer internet hostname for which the omnibox input is
 163         // a prefix).
 164         if (url_db->GetRowForURL(
 165             autocomplete_input_.canonicalized_url(), NULL) != 0) {
 166           // We visited this URL before.
 167           will_have_url_what_you_typed_match_first = true;
 168           // HistoryURLProvider gives visited what-you-typed URLs a high score.
 169           url_what_you_typed_match_score =
 170               HistoryURLProvider::kScoreForBestInlineableResult;
 171         } else if (url_db->IsTypedHost(host) &&
 172              (!autocomplete_input_.parts().path.is_nonempty() ||
 173               ((autocomplete_input_.parts().path.len == 1) &&
 174                (autocomplete_input_.text()[
 175                    autocomplete_input_.parts().path.begin] == '/'))) &&
 176              !autocomplete_input_.parts().query.is_nonempty() &&
 177              !autocomplete_input_.parts().ref.is_nonempty()) {
 178           // Not visited, but we've seen the host before.
 179           will_have_url_what_you_typed_match_first = true;
 180           const size_t registry_length =
 181               net::registry_controlled_domains::GetRegistryLength(
 182                   host,
 183                   net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
 184                   net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES);
 185           if (registry_length == 0) {
 186             // Known intranet hosts get one score.
 187             url_what_you_typed_match_score =
 188                 HistoryURLProvider::kScoreForUnvisitedIntranetResult;
 189           } else {
 190             // Known internet hosts get another.
 191             url_what_you_typed_match_score =
 192                 HistoryURLProvider::kScoreForWhatYouTypedResult;
 193           }
 194         }
 195       }
 196     }
 197   }
 198
 199   // Loop over every result and add it to matches_.  In the process,
 200   // guarantee that scores are decreasing.  |max_match_score| keeps
 201   // track of the highest score we can assign to any later results we
 202   // see.  Also, reduce |max_match_score| if we think there will be
 203   // a URL-what-you-typed match.  (We want URL-what-you-typed matches for
 204   // visited URLs to beat out any longer URLs, no matter how frequently
 205   // they're visited.)  The strength of this reduction depends on the
 206   // likely score for the URL-what-you-typed result.
 207
 208   // |template_url_service| or |template_url| can be NULL in unit tests.
 209   TemplateURLService* template_url_service =
 210       TemplateURLServiceFactory::GetForProfile(profile_);
 211   TemplateURL* template_url = template_url_service ?
 212       template_url_service->GetDefaultSearchProvider() : NULL;
 213   int max_match_score = matches.begin()->raw_score();
 214   if (will_have_url_what_you_typed_match_first) {
 215     max_match_score = std::min(max_match_score,
 216         url_what_you_typed_match_score - 1);
 217   }
 218   for (ScoredHistoryMatches::const_iterator match_iter = matches.begin();
 219        match_iter != matches.end(); ++match_iter) {
 220     const ScoredHistoryMatch& history_match(*match_iter);
 221     // Culls results corresponding to queries from the default search engine.
 222     // These are low-quality, difficult-to-understand matches for users, and the
 223     // SearchProvider should surface past queries in a better way anyway.
 224     if (!template_url ||
 225         !template_url->IsSearchURL(history_match.url_info.url())) {
 226       // Set max_match_score to the score we'll assign this result:
 227       max_match_score = std::min(max_match_score, history_match.raw_score());
 228       matches_.push_back(QuickMatchToACMatch(history_match, max_match_score));
 229       // Mark this max_match_score as being used:
 230       max_match_score--;
 231     }
 232   }
 233 }
 234
 235 AutocompleteMatch HistoryQuickProvider::QuickMatchToACMatch(
 236     const ScoredHistoryMatch& history_match,
 237     int score) {
 238   const history::URLRow& info = history_match.url_info;
 239   AutocompleteMatch match(
 240       this, score, !!info.visit_count(),
 241       history_match.url_matches().empty() ?
 242           AutocompleteMatchType::HISTORY_TITLE :
 243           AutocompleteMatchType::HISTORY_URL);
 244   match.typed_count = info.typed_count();
 245   match.destination_url = info.url();
 246   DCHECK(match.destination_url.is_valid());
 247
 248   // Format the URL autocomplete presentation.
 249   const net::FormatUrlTypes format_types = net::kFormatUrlOmitAll &
 250       ~(!history_match.match_in_scheme ? 0 : net::kFormatUrlOmitHTTP);
 251   match.fill_into_edit =
 252       AutocompleteInput::FormattedStringWithEquivalentMeaning(
 253           info.url(),
 254           net::FormatUrl(info.url(), languages_, format_types,
 255                          net::UnescapeRule::SPACES, NULL, NULL, NULL));
 256   std::vector<size_t> offsets =
 257       OffsetsFromTermMatches(history_match.url_matches());
 258   base::OffsetAdjuster::Adjustments adjustments;
 259   match.contents = net::FormatUrlWithAdjustments(
 260       info.url(), languages_, format_types, net::UnescapeRule::SPACES, NULL,
 261       NULL, &adjustments);
 262   base::OffsetAdjuster::AdjustOffsets(adjustments, &offsets);
 263   history::TermMatches new_matches =
 264       ReplaceOffsetsInTermMatches(history_match.url_matches(), offsets);
 265   match.contents_class =
 266       SpansFromTermMatch(new_matches, match.contents.length(), true);
 267
 268   // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
 269   if (history_match.can_inline()) {
 270     DCHECK(!new_matches.empty());
 271     size_t inline_autocomplete_offset = new_matches[0].offset +
 272         new_matches[0].length;
 273     // |inline_autocomplete_offset| may be beyond the end of the
 274     // |fill_into_edit| if the user has typed an URL with a scheme and the
 275     // last character typed is a slash.  That slash is removed by the
 276     // FormatURLWithOffsets call above.
 277     if (inline_autocomplete_offset < match.fill_into_edit.length()) {
 278       match.inline_autocompletion =
 279           match.fill_into_edit.substr(inline_autocomplete_offset);
 280     }
 281     match.allowed_to_be_default_match = match.inline_autocompletion.empty() ||
 282         !PreventInlineAutocomplete(autocomplete_input_);
 283   }
 284
 285   // Format the description autocomplete presentation.
 286   match.description = info.title();
 287   match.description_class = SpansFromTermMatch(
 288       history_match.title_matches(), match.description.length(), false);
 289
 290   match.RecordAdditionalInfo("typed count", info.typed_count());
 291   match.RecordAdditionalInfo("visit count", info.visit_count());
 292   match.RecordAdditionalInfo("last visit", info.last_visit());
 293
 294   return match;
 295 }
 296
 297 history::InMemoryURLIndex* HistoryQuickProvider::GetIndex() {
 298   if (index_for_testing_.get())
 299     return index_for_testing_.get();
 300
 301   HistoryService* const history_service =
 302       HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
 303   if (!history_service)
 304     return NULL;
 305
 306   return history_service->InMemoryIndex();
 307 }