components/omnibox/shortcuts_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/shortcuts_provider.h"
   6
   7 #include <algorithm>
   8 #include <cmath>
   9 #include <map>
  10 #include <vector>
  11
  12 #include "base/i18n/break_iterator.h"
  13 #include "base/i18n/case_conversion.h"
  14 #include "base/logging.h"
  15 #include "base/metrics/histogram.h"
  16 #include "base/prefs/pref_service.h"
  17 #include "base/strings/string_number_conversions.h"
  18 #include "base/strings/string_util.h"
  19 #include "base/strings/utf_string_conversions.h"
  20 #include "base/time/time.h"
  21 #include "components/history/core/browser/history_service.h"
  22 #include "components/metrics/proto/omnibox_input_type.pb.h"
  23 #include "components/omnibox/autocomplete_input.h"
  24 #include "components/omnibox/autocomplete_match.h"
  25 #include "components/omnibox/autocomplete_provider_client.h"
  26 #include "components/omnibox/autocomplete_result.h"
  27 #include "components/omnibox/history_provider.h"
  28 #include "components/omnibox/omnibox_field_trial.h"
  29 #include "components/omnibox/url_prefix.h"
  30 #include "components/url_fixer/url_fixer.h"
  31 #include "url/third_party/mozilla/url_parse.h"
  32
  33 namespace {
  34
  35 class DestinationURLEqualsURL {
  36  public:
  37   explicit DestinationURLEqualsURL(const GURL& url) : url_(url) {}
  38   bool operator()(const AutocompleteMatch& match) const {
  39     return match.destination_url == url_;
  40   }
  41
  42  private:
  43   const GURL url_;
  44 };
  45
  46 }  // namespace
  47
  48 const int ShortcutsProvider::kShortcutsProviderDefaultMaxRelevance = 1199;
  49
  50 ShortcutsProvider::ShortcutsProvider(AutocompleteProviderClient* client)
  51     : AutocompleteProvider(AutocompleteProvider::TYPE_SHORTCUTS),
  52       client_(client),
  53       languages_(client_->GetAcceptLanguages()),
  54       initialized_(false) {
  55   scoped_refptr<ShortcutsBackend> backend = client_->GetShortcutsBackend();
  56   if (backend.get()) {
  57     backend->AddObserver(this);
  58     if (backend->initialized())
  59       initialized_ = true;
  60   }
  61 }
  62
  63 void ShortcutsProvider::Start(const AutocompleteInput& input,
  64                               bool minimal_changes) {
  65   matches_.clear();
  66
  67   if (input.from_omnibox_focus() ||
  68       (input.type() == metrics::OmniboxInputType::INVALID) ||
  69       (input.type() == metrics::OmniboxInputType::FORCED_QUERY) ||
  70       input.text().empty() || !initialized_)
  71     return;
  72
  73   base::TimeTicks start_time = base::TimeTicks::Now();
  74   GetMatches(input);
  75   if (input.text().length() < 6) {
  76     base::TimeTicks end_time = base::TimeTicks::Now();
  77     std::string name = "ShortcutsProvider.QueryIndexTime." +
  78                        base::IntToString(input.text().size());
  79     base::HistogramBase* counter = base::Histogram::FactoryGet(
  80         name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
  81     counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
  82   }
  83 }
  84
  85 void ShortcutsProvider::DeleteMatch(const AutocompleteMatch& match) {
  86   // Copy the URL since deleting from |matches_| will invalidate |match|.
  87   GURL url(match.destination_url);
  88   DCHECK(url.is_valid());
  89
  90   // When a user deletes a match, he probably means for the URL to disappear out
  91   // of history entirely. So nuke all shortcuts that map to this URL.
  92   scoped_refptr<ShortcutsBackend> backend =
  93       client_->GetShortcutsBackendIfExists();
  94   if (backend.get())  // Can be NULL in Incognito.
  95     backend->DeleteShortcutsWithURL(url);
  96
  97   matches_.erase(std::remove_if(matches_.begin(), matches_.end(),
  98                                 DestinationURLEqualsURL(url)),
  99                  matches_.end());
 100   // NOTE: |match| is now dead!
 101
 102   // Delete the match from the history DB. This will eventually result in a
 103   // second call to DeleteShortcutsWithURL(), which is harmless.
 104   history::HistoryService* const history_service = client_->GetHistoryService();
 105   DCHECK(history_service);
 106   history_service->DeleteURL(url);
 107 }
 108
 109 ShortcutsProvider::~ShortcutsProvider() {
 110   scoped_refptr<ShortcutsBackend> backend =
 111       client_->GetShortcutsBackendIfExists();
 112   if (backend.get())
 113     backend->RemoveObserver(this);
 114 }
 115
 116 void ShortcutsProvider::OnShortcutsLoaded() {
 117   initialized_ = true;
 118 }
 119
 120 void ShortcutsProvider::GetMatches(const AutocompleteInput& input) {
 121   scoped_refptr<ShortcutsBackend> backend =
 122       client_->GetShortcutsBackendIfExists();
 123   if (!backend.get())
 124     return;
 125   // Get the URLs from the shortcuts database with keys that partially or
 126   // completely match the search term.
 127   base::string16 term_string(base::i18n::ToLower(input.text()));
 128   DCHECK(!term_string.empty());
 129
 130   int max_relevance;
 131   if (!OmniboxFieldTrial::ShortcutsScoringMaxRelevance(
 132           input.current_page_classification(), &max_relevance))
 133     max_relevance = kShortcutsProviderDefaultMaxRelevance;
 134   TemplateURLService* template_url_service = client_->GetTemplateURLService();
 135   const base::string16 fixed_up_input(FixupUserInput(input).second);
 136   for (ShortcutsBackend::ShortcutMap::const_iterator it =
 137            FindFirstMatch(term_string, backend.get());
 138        it != backend->shortcuts_map().end() &&
 139            base::StartsWith(it->first, term_string, true);
 140        ++it) {
 141     // Don't return shortcuts with zero relevance.
 142     int relevance = CalculateScore(term_string, it->second, max_relevance);
 143     if (relevance) {
 144       matches_.push_back(
 145           ShortcutToACMatch(it->second, relevance, input, fixed_up_input));
 146       matches_.back().ComputeStrippedDestinationURL(template_url_service);
 147     }
 148   }
 149   // Remove duplicates.  Duplicates don't need to be preserved in the matches
 150   // because they are only used for deletions, and shortcuts deletes matches
 151   // based on the URL.
 152   AutocompleteResult::DedupMatchesByDestination(
 153       input.current_page_classification(), false, &matches_);
 154   // Find best matches.
 155   std::partial_sort(
 156       matches_.begin(),
 157       matches_.begin() +
 158           std::min(AutocompleteProvider::kMaxMatches, matches_.size()),
 159       matches_.end(), &AutocompleteMatch::MoreRelevant);
 160   if (matches_.size() > AutocompleteProvider::kMaxMatches) {
 161     matches_.erase(matches_.begin() + AutocompleteProvider::kMaxMatches,
 162                    matches_.end());
 163   }
 164   // Guarantee that all scores are decreasing (but do not assign any scores
 165   // below 1).
 166   for (ACMatches::iterator it = matches_.begin(); it != matches_.end(); ++it) {
 167     max_relevance = std::min(max_relevance, it->relevance);
 168     it->relevance = max_relevance;
 169     if (max_relevance > 1)
 170       --max_relevance;
 171   }
 172 }
 173
 174 AutocompleteMatch ShortcutsProvider::ShortcutToACMatch(
 175     const ShortcutsDatabase::Shortcut& shortcut,
 176     int relevance,
 177     const AutocompleteInput& input,
 178     const base::string16& fixed_up_input_text) {
 179   DCHECK(!input.text().empty());
 180   AutocompleteMatch match;
 181   match.provider = this;
 182   match.relevance = relevance;
 183   match.deletable = true;
 184   match.fill_into_edit = shortcut.match_core.fill_into_edit;
 185   match.destination_url = shortcut.match_core.destination_url;
 186   DCHECK(match.destination_url.is_valid());
 187   match.contents = shortcut.match_core.contents;
 188   match.contents_class = AutocompleteMatch::ClassificationsFromString(
 189       shortcut.match_core.contents_class);
 190   match.description = shortcut.match_core.description;
 191   match.description_class = AutocompleteMatch::ClassificationsFromString(
 192       shortcut.match_core.description_class);
 193   match.transition = ui::PageTransitionFromInt(shortcut.match_core.transition);
 194   match.type = static_cast<AutocompleteMatch::Type>(shortcut.match_core.type);
 195   match.keyword = shortcut.match_core.keyword;
 196   match.RecordAdditionalInfo("number of hits", shortcut.number_of_hits);
 197   match.RecordAdditionalInfo("last access time", shortcut.last_access_time);
 198   match.RecordAdditionalInfo("original input text",
 199                              base::UTF16ToUTF8(shortcut.text));
 200
 201   // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
 202   // If the match is a search query this is easy: simply check whether the
 203   // user text is a prefix of the query.  If the match is a navigation, we
 204   // assume the fill_into_edit looks something like a URL, so we use
 205   // URLPrefix::GetInlineAutocompleteOffset() to try and strip off any prefixes
 206   // that the user might not think would change the meaning, but would
 207   // otherwise prevent inline autocompletion.  This allows, for example, the
 208   // input of "foo.c" to autocomplete to "foo.com" for a fill_into_edit of
 209   // "http://foo.com".
 210   if (AutocompleteMatch::IsSearchType(match.type)) {
 211     if (base::StartsWith(match.fill_into_edit, input.text(), false)) {
 212       match.inline_autocompletion =
 213           match.fill_into_edit.substr(input.text().length());
 214       match.allowed_to_be_default_match =
 215           !input.prevent_inline_autocomplete() ||
 216           match.inline_autocompletion.empty();
 217     }
 218   } else {
 219     const size_t inline_autocomplete_offset =
 220         URLPrefix::GetInlineAutocompleteOffset(
 221             input.text(), fixed_up_input_text, true, match.fill_into_edit);
 222     if (inline_autocomplete_offset != base::string16::npos) {
 223       match.inline_autocompletion =
 224           match.fill_into_edit.substr(inline_autocomplete_offset);
 225       match.allowed_to_be_default_match =
 226           !HistoryProvider::PreventInlineAutocomplete(input) ||
 227           match.inline_autocompletion.empty();
 228     }
 229   }
 230   match.EnsureUWYTIsAllowedToBeDefault(input.canonicalized_url(),
 231                                        client_->GetTemplateURLService());
 232
 233   // Try to mark pieces of the contents and description as matches if they
 234   // appear in |input.text()|.
 235   const base::string16 term_string = base::i18n::ToLower(input.text());
 236   WordMap terms_map(CreateWordMapForString(term_string));
 237   if (!terms_map.empty()) {
 238     match.contents_class = ClassifyAllMatchesInString(
 239         term_string, terms_map, match.contents, match.contents_class);
 240     match.description_class = ClassifyAllMatchesInString(
 241         term_string, terms_map, match.description, match.description_class);
 242   }
 243   return match;
 244 }
 245
 246 // static
 247 ShortcutsProvider::WordMap ShortcutsProvider::CreateWordMapForString(
 248     const base::string16& text) {
 249   // First, convert |text| to a vector of the unique words in it.
 250   WordMap word_map;
 251   base::i18n::BreakIterator word_iter(text,
 252                                       base::i18n::BreakIterator::BREAK_WORD);
 253   if (!word_iter.Init())
 254     return word_map;
 255   std::vector<base::string16> words;
 256   while (word_iter.Advance()) {
 257     if (word_iter.IsWord())
 258       words.push_back(word_iter.GetString());
 259   }
 260   if (words.empty())
 261     return word_map;
 262   std::sort(words.begin(), words.end());
 263   words.erase(std::unique(words.begin(), words.end()), words.end());
 264
 265   // Now create a map from (first character) to (words beginning with that
 266   // character).  We insert in reverse lexicographical order and rely on the
 267   // multimap preserving insertion order for values with the same key.  (This
 268   // is mandated in C++11, and part of that decision was based on a survey of
 269   // existing implementations that found that it was already true everywhere.)
 270   std::reverse(words.begin(), words.end());
 271   for (std::vector<base::string16>::const_iterator i(words.begin());
 272        i != words.end(); ++i)
 273     word_map.insert(std::make_pair((*i)[0], *i));
 274   return word_map;
 275 }
 276
 277 // static
 278 ACMatchClassifications ShortcutsProvider::ClassifyAllMatchesInString(
 279     const base::string16& find_text,
 280     const WordMap& find_words,
 281     const base::string16& text,
 282     const ACMatchClassifications& original_class) {
 283   DCHECK(!find_text.empty());
 284   DCHECK(!find_words.empty());
 285
 286   // The code below assumes |text| is nonempty and therefore the resulting
 287   // classification vector should always be nonempty as well.  Returning early
 288   // if |text| is empty assures we'll return the (correct) empty vector rather
 289   // than a vector with a single (0, NONE) match.
 290   if (text.empty())
 291     return original_class;
 292
 293   // First check whether |text| begins with |find_text| and mark that whole
 294   // section as a match if so.
 295   base::string16 text_lowercase(base::i18n::ToLower(text));
 296   ACMatchClassifications match_class;
 297   size_t last_position = 0;
 298   if (base::StartsWith(text_lowercase, find_text, true)) {
 299     match_class.push_back(
 300         ACMatchClassification(0, ACMatchClassification::MATCH));
 301     last_position = find_text.length();
 302     // If |text_lowercase| is actually equal to |find_text|, we don't need to
 303     // (and in fact shouldn't) put a trailing NONE classification after the end
 304     // of the string.
 305     if (last_position < text_lowercase.length()) {
 306       match_class.push_back(
 307           ACMatchClassification(last_position, ACMatchClassification::NONE));
 308     }
 309   } else {
 310     // |match_class| should start at position 0.  If the first matching word is
 311     // found at position 0, this will be popped from the vector further down.
 312     match_class.push_back(
 313         ACMatchClassification(0, ACMatchClassification::NONE));
 314   }
 315
 316   // Now, starting with |last_position|, check each character in
 317   // |text_lowercase| to see if we have words starting with that character in
 318   // |find_words|.  If so, check each of them to see if they match the portion
 319   // of |text_lowercase| beginning with |last_position|.  Accept the first
 320   // matching word found (which should be the longest possible match at this
 321   // location, given the construction of |find_words|) and add a MATCH region to
 322   // |match_class|, moving |last_position| to be after the matching word.  If we
 323   // found no matching words, move to the next character and repeat.
 324   while (last_position < text_lowercase.length()) {
 325     std::pair<WordMap::const_iterator, WordMap::const_iterator> range(
 326         find_words.equal_range(text_lowercase[last_position]));
 327     size_t next_character = last_position + 1;
 328     for (WordMap::const_iterator i(range.first); i != range.second; ++i) {
 329       const base::string16& word = i->second;
 330       size_t word_end = last_position + word.length();
 331       if ((word_end <= text_lowercase.length()) &&
 332           !text_lowercase.compare(last_position, word.length(), word)) {
 333         // Collapse adjacent ranges into one.
 334         if (match_class.back().offset == last_position)
 335           match_class.pop_back();
 336
 337         AutocompleteMatch::AddLastClassificationIfNecessary(
 338             &match_class, last_position, ACMatchClassification::MATCH);
 339         if (word_end < text_lowercase.length()) {
 340           match_class.push_back(
 341               ACMatchClassification(word_end, ACMatchClassification::NONE));
 342         }
 343         last_position = word_end;
 344         break;
 345       }
 346     }
 347     last_position = std::max(last_position, next_character);
 348   }
 349
 350   return AutocompleteMatch::MergeClassifications(original_class, match_class);
 351 }
 352
 353 ShortcutsBackend::ShortcutMap::const_iterator ShortcutsProvider::FindFirstMatch(
 354     const base::string16& keyword,
 355     ShortcutsBackend* backend) {
 356   DCHECK(backend);
 357   ShortcutsBackend::ShortcutMap::const_iterator it =
 358       backend->shortcuts_map().lower_bound(keyword);
 359   // Lower bound not necessarily matches the keyword, check for item pointed by
 360   // the lower bound iterator to at least start with keyword.
 361   return ((it == backend->shortcuts_map().end()) ||
 362           base::StartsWith(it->first, keyword, true))
 363              ? it
 364              : backend->shortcuts_map().end();
 365 }
 366
 367 int ShortcutsProvider::CalculateScore(
 368     const base::string16& terms,
 369     const ShortcutsDatabase::Shortcut& shortcut,
 370     int max_relevance) {
 371   DCHECK(!terms.empty());
 372   DCHECK_LE(terms.length(), shortcut.text.length());
 373
 374   // The initial score is based on how much of the shortcut the user has typed.
 375   // Using the square root of the typed fraction boosts the base score rapidly
 376   // as characters are typed, compared with simply using the typed fraction
 377   // directly. This makes sense since the first characters typed are much more
 378   // important for determining how likely it is a user wants a particular
 379   // shortcut than are the remaining continued characters.
 380   double base_score = max_relevance * sqrt(static_cast<double>(terms.length()) /
 381                                            shortcut.text.length());
 382
 383   // Then we decay this by half each week.
 384   const double kLn2 = 0.6931471805599453;
 385   base::TimeDelta time_passed = base::Time::Now() - shortcut.last_access_time;
 386   // Clamp to 0 in case time jumps backwards (e.g. due to DST).
 387   double decay_exponent =
 388       std::max(0.0, kLn2 * static_cast<double>(time_passed.InMicroseconds()) /
 389                         base::Time::kMicrosecondsPerWeek);
 390
 391   // We modulate the decay factor based on how many times the shortcut has been
 392   // used. Newly created shortcuts decay at full speed; otherwise, decaying by
 393   // half takes |n| times as much time, where n increases by
 394   // (1.0 / each 5 additional hits), up to a maximum of 5x as long.
 395   const double kMaxDecaySpeedDivisor = 5.0;
 396   const double kNumUsesPerDecaySpeedDivisorIncrement = 5.0;
 397   double decay_divisor = std::min(
 398       kMaxDecaySpeedDivisor,
 399       (shortcut.number_of_hits + kNumUsesPerDecaySpeedDivisorIncrement - 1) /
 400           kNumUsesPerDecaySpeedDivisorIncrement);
 401
 402   return static_cast<int>((base_score / exp(decay_exponent / decay_divisor)) +
 403                           0.5);
 404 }