components/omnibox/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 "components/omnibox/history_quick_provider.h"
   6
   7 #include <vector>
   8
   9 #include "base/basictypes.h"
  10 #include "base/debug/crash_logging.h"
  11 #include "base/i18n/break_iterator.h"
  12 #include "base/logging.h"
  13 #include "base/metrics/field_trial.h"
  14 #include "base/prefs/pref_service.h"
  15 #include "base/strings/string_number_conversions.h"
  16 #include "base/strings/string_util.h"
  17 #include "base/strings/utf_string_conversions.h"
  18 #include "components/bookmarks/browser/bookmark_model.h"
  19 #include "components/history/core/browser/history_database.h"
  20 #include "components/history/core/browser/history_service.h"
  21 #include "components/metrics/proto/omnibox_input_type.pb.h"
  22 #include "components/omnibox/autocomplete_match_type.h"
  23 #include "components/omnibox/autocomplete_provider_client.h"
  24 #include "components/omnibox/autocomplete_result.h"
  25 #include "components/omnibox/history_url_provider.h"
  26 #include "components/omnibox/in_memory_url_index.h"
  27 #include "components/omnibox/in_memory_url_index_types.h"
  28 #include "components/omnibox/omnibox_field_trial.h"
  29 #include "components/search_engines/template_url.h"
  30 #include "components/search_engines/template_url_service.h"
  31 #include "net/base/escape.h"
  32 #include "net/base/net_util.h"
  33 #include "net/base/registry_controlled_domains/registry_controlled_domain.h"
  34 #include "url/third_party/mozilla/url_parse.h"
  35 #include "url/url_util.h"
  36
  37 namespace {
  38
  39 // Used to help investigate bug 464926.  NOTE: This value is defined multiple
  40 // places in the codebase due to layering issues. DO NOT change the value here
  41 // without changing it in all other places that it is defined in the codebase
  42 // (search for |kBug464926CrashKey|).
  43 const char kBug464926CrashKey[] = "bug-464926-info";
  44
  45 }  // namespace
  46
  47 bool HistoryQuickProvider::disabled_ = false;
  48
  49 HistoryQuickProvider::HistoryQuickProvider(AutocompleteProviderClient* client)
  50     : HistoryProvider(AutocompleteProvider::TYPE_HISTORY_QUICK, client),
  51       languages_(client->GetAcceptLanguages()),
  52       in_memory_url_index_(client->GetInMemoryURLIndex()) {
  53 }
  54
  55 void HistoryQuickProvider::Start(const AutocompleteInput& input,
  56                                  bool minimal_changes) {
  57   matches_.clear();
  58   if (disabled_ || input.from_omnibox_focus())
  59     return;
  60
  61   // Don't bother with INVALID and FORCED_QUERY.
  62   if ((input.type() == metrics::OmniboxInputType::INVALID) ||
  63       (input.type() == metrics::OmniboxInputType::FORCED_QUERY))
  64     return;
  65
  66   autocomplete_input_ = input;
  67
  68   // TODO(pkasting): We should just block here until this loads.  Any time
  69   // someone unloads the history backend, we'll get inconsistent inline
  70   // autocomplete behavior here.
  71   if (in_memory_url_index_) {
  72     DoAutocomplete();
  73   }
  74 }
  75
  76 HistoryQuickProvider::~HistoryQuickProvider() {
  77 }
  78
  79 void HistoryQuickProvider::DoAutocomplete() {
  80   // Get the matching URLs from the DB.
  81   ScoredHistoryMatches matches = in_memory_url_index_->HistoryItemsForTerms(
  82       autocomplete_input_.text(), autocomplete_input_.cursor_position(),
  83       AutocompleteProvider::kMaxMatches);
  84   if (matches.empty())
  85     return;
  86
  87   // Figure out if HistoryURL provider has a URL-what-you-typed match
  88   // that ought to go first and what its score will be.
  89   bool will_have_url_what_you_typed_match_first = false;
  90   int url_what_you_typed_match_score = -1;  // undefined
  91   // These are necessary (but not sufficient) conditions for the omnibox
  92   // input to be a URL-what-you-typed match.  The username test checks that
  93   // either the username does not exist (a regular URL such as http://site/)
  94   // or, if the username exists (http://user@site/), there must be either
  95   // a password or a port.  Together these exclude pure username@site
  96   // inputs because these are likely to be an e-mail address.  HistoryURL
  97   // provider won't promote the URL-what-you-typed match to first
  98   // for these inputs.
  99   const bool can_have_url_what_you_typed_match_first =
 100       (autocomplete_input_.type() != metrics::OmniboxInputType::QUERY) &&
 101       (!autocomplete_input_.parts().username.is_nonempty() ||
 102        autocomplete_input_.parts().password.is_nonempty() ||
 103        autocomplete_input_.parts().path.is_nonempty());
 104   if (can_have_url_what_you_typed_match_first) {
 105     history::HistoryService* const history_service =
 106         client()->GetHistoryService();
 107     // We expect HistoryService to be available.  In case it's not,
 108     // (e.g., due to Profile corruption) we let HistoryQuick provider
 109     // completions (which may be available because it's a different
 110     // data structure) compete with the URL-what-you-typed match as
 111     // normal.
 112     if (history_service) {
 113       history::URLDatabase* url_db = history_service->InMemoryDatabase();
 114       // url_db can be NULL if it hasn't finished initializing (or
 115       // failed to to initialize).  In this case, we let HistoryQuick
 116       // provider completions compete with the URL-what-you-typed
 117       // match as normal.
 118       if (url_db) {
 119         const std::string host(base::UTF16ToUTF8(
 120             autocomplete_input_.text().substr(
 121                 autocomplete_input_.parts().host.begin,
 122                 autocomplete_input_.parts().host.len)));
 123         // We want to put the URL-what-you-typed match first if either
 124         // * the user visited the URL before (intranet or internet).
 125         // * it's a URL on a host that user visited before and this
 126         //   is the root path of the host.  (If the user types some
 127         //   of a path--more than a simple "/"--we let autocomplete compete
 128         //   normally with the URL-what-you-typed match.)
 129         // TODO(mpearson): Remove this hacky code and simply score URL-what-
 130         // you-typed in some sane way relative to possible completions:
 131         // URL-what-you-typed should get some sort of a boost relative
 132         // to completions, but completions should naturally win if
 133         // they're a lot more popular.  In this process, if the input
 134         // is a bare intranet hostname that has been visited before, we
 135         // may want to enforce that the only completions that can outscore
 136         // the URL-what-you-typed match are on the same host (i.e., aren't
 137         // from a longer internet hostname for which the omnibox input is
 138         // a prefix).
 139         if (url_db->GetRowForURL(
 140             autocomplete_input_.canonicalized_url(), NULL) != 0) {
 141           // We visited this URL before.
 142           will_have_url_what_you_typed_match_first = true;
 143           // HistoryURLProvider gives visited what-you-typed URLs a high score.
 144           url_what_you_typed_match_score =
 145               HistoryURLProvider::kScoreForBestInlineableResult;
 146         } else if (url_db->IsTypedHost(host) &&
 147              (!autocomplete_input_.parts().path.is_nonempty() ||
 148               ((autocomplete_input_.parts().path.len == 1) &&
 149                (autocomplete_input_.text()[
 150                    autocomplete_input_.parts().path.begin] == '/'))) &&
 151              !autocomplete_input_.parts().query.is_nonempty() &&
 152              !autocomplete_input_.parts().ref.is_nonempty()) {
 153           // Not visited, but we've seen the host before.
 154           will_have_url_what_you_typed_match_first = true;
 155           const size_t registry_length =
 156               net::registry_controlled_domains::GetRegistryLength(
 157                   host,
 158                   net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
 159                   net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES);
 160           if (registry_length == 0) {
 161             // Known intranet hosts get one score.
 162             url_what_you_typed_match_score =
 163                 HistoryURLProvider::kScoreForUnvisitedIntranetResult;
 164           } else {
 165             // Known internet hosts get another.
 166             url_what_you_typed_match_score =
 167                 HistoryURLProvider::kScoreForWhatYouTypedResult;
 168           }
 169         }
 170       }
 171     }
 172   }
 173
 174   // Loop over every result and add it to matches_.  In the process,
 175   // guarantee that scores are decreasing.  |max_match_score| keeps
 176   // track of the highest score we can assign to any later results we
 177   // see.  Also, reduce |max_match_score| if we think there will be
 178   // a URL-what-you-typed match.  (We want URL-what-you-typed matches for
 179   // visited URLs to beat out any longer URLs, no matter how frequently
 180   // they're visited.)  The strength of this reduction depends on the
 181   // likely score for the URL-what-you-typed result.
 182
 183   // |template_url_service| or |template_url| can be NULL in unit tests.
 184   TemplateURLService* template_url_service = client()->GetTemplateURLService();
 185   TemplateURL* template_url = template_url_service ?
 186       template_url_service->GetDefaultSearchProvider() : NULL;
 187   int max_match_score = matches.begin()->raw_score;
 188   if (will_have_url_what_you_typed_match_first) {
 189     max_match_score = std::min(max_match_score,
 190         url_what_you_typed_match_score - 1);
 191   }
 192   for (ScoredHistoryMatches::const_iterator match_iter = matches.begin();
 193        match_iter != matches.end(); ++match_iter) {
 194     const ScoredHistoryMatch& history_match(*match_iter);
 195     // Culls results corresponding to queries from the default search engine.
 196     // These are low-quality, difficult-to-understand matches for users, and the
 197     // SearchProvider should surface past queries in a better way anyway.
 198     if (!template_url ||
 199         !template_url->IsSearchURL(history_match.url_info.url(),
 200                                    template_url_service->search_terms_data())) {
 201       // Set max_match_score to the score we'll assign this result:
 202       max_match_score = std::min(max_match_score, history_match.raw_score);
 203       matches_.push_back(QuickMatchToACMatch(history_match, max_match_score));
 204       // Mark this max_match_score as being used:
 205       max_match_score--;
 206     }
 207   }
 208 }
 209
 210 AutocompleteMatch HistoryQuickProvider::QuickMatchToACMatch(
 211     const ScoredHistoryMatch& history_match,
 212     int score) {
 213   const history::URLRow& info = history_match.url_info;
 214   AutocompleteMatch match(
 215       this, score, !!info.visit_count(),
 216       history_match.url_matches.empty() ?
 217           AutocompleteMatchType::HISTORY_TITLE :
 218           AutocompleteMatchType::HISTORY_URL);
 219   match.typed_count = info.typed_count();
 220   match.destination_url = info.url();
 221   DCHECK(match.destination_url.is_valid());
 222
 223   // Format the URL autocomplete presentation.
 224   const net::FormatUrlTypes format_types = net::kFormatUrlOmitAll &
 225       ~(!history_match.match_in_scheme ? 0 : net::kFormatUrlOmitHTTP);
 226   match.fill_into_edit =
 227       AutocompleteInput::FormattedStringWithEquivalentMeaning(
 228           info.url(),
 229           net::FormatUrl(info.url(), languages_, format_types,
 230                          net::UnescapeRule::SPACES, NULL, NULL, NULL),
 231           client()->GetSchemeClassifier());
 232   std::vector<size_t> offsets =
 233       OffsetsFromTermMatches(history_match.url_matches);
 234   base::OffsetAdjuster::Adjustments adjustments;
 235   match.contents = net::FormatUrlWithAdjustments(
 236       info.url(), languages_, format_types, net::UnescapeRule::SPACES, NULL,
 237       NULL, &adjustments);
 238   base::OffsetAdjuster::AdjustOffsets(adjustments, &offsets);
 239   TermMatches new_matches =
 240       ReplaceOffsetsInTermMatches(history_match.url_matches, offsets);
 241   match.contents_class =
 242       SpansFromTermMatch(new_matches, match.contents.length(), true);
 243
 244   // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
 245   if (history_match.can_inline) {
 246     base::debug::ScopedCrashKey crash_info(
 247         kBug464926CrashKey,
 248         info.url().spec().substr(0, 30) + " " +
 249             base::UTF16ToUTF8(autocomplete_input_.text()).substr(0, 20) + " " +
 250             base::SizeTToString(history_match.url_matches.size()) + " " +
 251             base::SizeTToString(offsets.size()));
 252     CHECK(!new_matches.empty());
 253     size_t inline_autocomplete_offset = new_matches[0].offset +
 254         new_matches[0].length;
 255     // |inline_autocomplete_offset| may be beyond the end of the
 256     // |fill_into_edit| if the user has typed an URL with a scheme and the
 257     // last character typed is a slash.  That slash is removed by the
 258     // FormatURLWithOffsets call above.
 259     if (inline_autocomplete_offset < match.fill_into_edit.length()) {
 260       match.inline_autocompletion =
 261           match.fill_into_edit.substr(inline_autocomplete_offset);
 262     }
 263     match.allowed_to_be_default_match = match.inline_autocompletion.empty() ||
 264         !PreventInlineAutocomplete(autocomplete_input_);
 265   }
 266   match.EnsureUWYTIsAllowedToBeDefault(autocomplete_input_.canonicalized_url(),
 267                                        client()->GetTemplateURLService());
 268
 269   // Format the description autocomplete presentation.
 270   match.description = info.title();
 271   match.description_class = SpansFromTermMatch(
 272       history_match.title_matches, match.description.length(), false);
 273
 274   match.RecordAdditionalInfo("typed count", info.typed_count());
 275   match.RecordAdditionalInfo("visit count", info.visit_count());
 276   match.RecordAdditionalInfo("last visit", info.last_visit());
 277
 278   return match;
 279 }