chrome/browser/autocomplete/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 "chrome/browser/autocomplete/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 "chrome/browser/autocomplete/history_provider.h"
  22 #include "chrome/browser/autocomplete/shortcuts_backend_factory.h"
  23 #include "chrome/browser/history/history_service_factory.h"
  24 #include "chrome/browser/profiles/profile.h"
  25 #include "chrome/browser/search_engines/template_url_service_factory.h"
  26 #include "chrome/common/pref_names.h"
  27 #include "chrome/common/url_constants.h"
  28 #include "components/history/core/browser/history_service.h"
  29 #include "components/metrics/proto/omnibox_input_type.pb.h"
  30 #include "components/omnibox/autocomplete_input.h"
  31 #include "components/omnibox/autocomplete_match.h"
  32 #include "components/omnibox/autocomplete_result.h"
  33 #include "components/omnibox/omnibox_field_trial.h"
  34 #include "components/omnibox/url_prefix.h"
  35 #include "components/url_fixer/url_fixer.h"
  36 #include "url/url_parse.h"
  37
  38 namespace {
  39
  40 class DestinationURLEqualsURL {
  41  public:
  42   explicit DestinationURLEqualsURL(const GURL& url) : url_(url) {}
  43   bool operator()(const AutocompleteMatch& match) const {
  44     return match.destination_url == url_;
  45   }
  46  private:
  47   const GURL url_;
  48 };
  49
  50 }  // namespace
  51
  52 const int ShortcutsProvider::kShortcutsProviderDefaultMaxRelevance = 1199;
  53
  54 ShortcutsProvider::ShortcutsProvider(Profile* profile)
  55     : AutocompleteProvider(AutocompleteProvider::TYPE_SHORTCUTS),
  56       profile_(profile),
  57       languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)),
  58       initialized_(false) {
  59   scoped_refptr<ShortcutsBackend> backend =
  60       ShortcutsBackendFactory::GetForProfile(profile_);
  61   if (backend.get()) {
  62     backend->AddObserver(this);
  63     if (backend->initialized())
  64       initialized_ = true;
  65   }
  66 }
  67
  68 void ShortcutsProvider::Start(const AutocompleteInput& input,
  69                               bool minimal_changes,
  70                               bool called_due_to_focus) {
  71   matches_.clear();
  72
  73   if (called_due_to_focus ||
  74       (input.type() == metrics::OmniboxInputType::INVALID) ||
  75       (input.type() == metrics::OmniboxInputType::FORCED_QUERY) ||
  76       input.text().empty() ||
  77       !initialized_)
  78     return;
  79
  80   base::TimeTicks start_time = base::TimeTicks::Now();
  81   GetMatches(input);
  82   if (input.text().length() < 6) {
  83     base::TimeTicks end_time = base::TimeTicks::Now();
  84     std::string name = "ShortcutsProvider.QueryIndexTime." +
  85         base::IntToString(input.text().size());
  86     base::HistogramBase* counter = base::Histogram::FactoryGet(
  87         name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
  88     counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
  89   }
  90 }
  91
  92 void ShortcutsProvider::DeleteMatch(const AutocompleteMatch& match) {
  93   // Copy the URL since deleting from |matches_| will invalidate |match|.
  94   GURL url(match.destination_url);
  95   DCHECK(url.is_valid());
  96
  97   // When a user deletes a match, he probably means for the URL to disappear out
  98   // of history entirely. So nuke all shortcuts that map to this URL.
  99   scoped_refptr<ShortcutsBackend> backend =
 100       ShortcutsBackendFactory::GetForProfileIfExists(profile_);
 101   if (backend.get())  // Can be NULL in Incognito.
 102     backend->DeleteShortcutsWithURL(url);
 103
 104   matches_.erase(std::remove_if(matches_.begin(), matches_.end(),
 105                                 DestinationURLEqualsURL(url)),
 106                  matches_.end());
 107   // NOTE: |match| is now dead!
 108
 109   // Delete the match from the history DB. This will eventually result in a
 110   // second call to DeleteShortcutsWithURL(), which is harmless.
 111   history::HistoryService* const history_service =
 112       HistoryServiceFactory::GetForProfile(profile_,
 113                                            ServiceAccessType::EXPLICIT_ACCESS);
 114   DCHECK(history_service);
 115   history_service->DeleteURL(url);
 116 }
 117
 118 ShortcutsProvider::~ShortcutsProvider() {
 119   scoped_refptr<ShortcutsBackend> backend =
 120       ShortcutsBackendFactory::GetForProfileIfExists(profile_);
 121   if (backend.get())
 122     backend->RemoveObserver(this);
 123 }
 124
 125 void ShortcutsProvider::OnShortcutsLoaded() {
 126   initialized_ = true;
 127 }
 128
 129 void ShortcutsProvider::GetMatches(const AutocompleteInput& input) {
 130   scoped_refptr<ShortcutsBackend> backend =
 131       ShortcutsBackendFactory::GetForProfileIfExists(profile_);
 132   if (!backend.get())
 133     return;
 134   // Get the URLs from the shortcuts database with keys that partially or
 135   // completely match the search term.
 136   base::string16 term_string(base::i18n::ToLower(input.text()));
 137   DCHECK(!term_string.empty());
 138
 139   int max_relevance;
 140   if (!OmniboxFieldTrial::ShortcutsScoringMaxRelevance(
 141       input.current_page_classification(), &max_relevance))
 142     max_relevance = kShortcutsProviderDefaultMaxRelevance;
 143   TemplateURLService* template_url_service =
 144       TemplateURLServiceFactory::GetForProfile(profile_);
 145   const base::string16 fixed_up_input(FixupUserInput(input).second);
 146   for (ShortcutsBackend::ShortcutMap::const_iterator it =
 147            FindFirstMatch(term_string, backend.get());
 148        it != backend->shortcuts_map().end() &&
 149            StartsWith(it->first, term_string, true); ++it) {
 150     // Don't return shortcuts with zero relevance.
 151     int relevance = CalculateScore(term_string, it->second, max_relevance);
 152     if (relevance) {
 153       matches_.push_back(ShortcutToACMatch(it->second, relevance, input,
 154                                            fixed_up_input));
 155       matches_.back().ComputeStrippedDestinationURL(template_url_service);
 156     }
 157   }
 158   // Remove duplicates.  Duplicates don't need to be preserved in the matches
 159   // because they are only used for deletions, and shortcuts deletes matches
 160   // based on the URL.
 161   AutocompleteResult::DedupMatchesByDestination(
 162       input.current_page_classification(), false, &matches_);
 163   // Find best matches.
 164   std::partial_sort(matches_.begin(),
 165       matches_.begin() +
 166           std::min(AutocompleteProvider::kMaxMatches, matches_.size()),
 167       matches_.end(), &AutocompleteMatch::MoreRelevant);
 168   if (matches_.size() > AutocompleteProvider::kMaxMatches) {
 169     matches_.erase(matches_.begin() + AutocompleteProvider::kMaxMatches,
 170                    matches_.end());
 171   }
 172   // Guarantee that all scores are decreasing (but do not assign any scores
 173   // below 1).
 174   for (ACMatches::iterator it = matches_.begin(); it != matches_.end(); ++it) {
 175     max_relevance = std::min(max_relevance, it->relevance);
 176     it->relevance = max_relevance;
 177     if (max_relevance > 1)
 178       --max_relevance;
 179   }
 180 }
 181
 182 AutocompleteMatch ShortcutsProvider::ShortcutToACMatch(
 183     const ShortcutsDatabase::Shortcut& shortcut,
 184     int relevance,
 185     const AutocompleteInput& input,
 186     const base::string16& fixed_up_input_text) {
 187   DCHECK(!input.text().empty());
 188   AutocompleteMatch match;
 189   match.provider = this;
 190   match.relevance = relevance;
 191   match.deletable = true;
 192   match.fill_into_edit = shortcut.match_core.fill_into_edit;
 193   match.destination_url = shortcut.match_core.destination_url;
 194   DCHECK(match.destination_url.is_valid());
 195   match.contents = shortcut.match_core.contents;
 196   match.contents_class = AutocompleteMatch::ClassificationsFromString(
 197       shortcut.match_core.contents_class);
 198   match.description = shortcut.match_core.description;
 199   match.description_class = AutocompleteMatch::ClassificationsFromString(
 200       shortcut.match_core.description_class);
 201   match.transition = ui::PageTransitionFromInt(shortcut.match_core.transition);
 202   match.type = static_cast<AutocompleteMatch::Type>(shortcut.match_core.type);
 203   match.keyword = shortcut.match_core.keyword;
 204   match.RecordAdditionalInfo("number of hits", shortcut.number_of_hits);
 205   match.RecordAdditionalInfo("last access time", shortcut.last_access_time);
 206   match.RecordAdditionalInfo("original input text",
 207                              base::UTF16ToUTF8(shortcut.text));
 208
 209   // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
 210   // If the match is a search query this is easy: simply check whether the
 211   // user text is a prefix of the query.  If the match is a navigation, we
 212   // assume the fill_into_edit looks something like a URL, so we use
 213   // URLPrefix::GetInlineAutocompleteOffset() to try and strip off any prefixes
 214   // that the user might not think would change the meaning, but would
 215   // otherwise prevent inline autocompletion.  This allows, for example, the
 216   // input of "foo.c" to autocomplete to "foo.com" for a fill_into_edit of
 217   // "http://foo.com".
 218   if (AutocompleteMatch::IsSearchType(match.type)) {
 219     if (StartsWith(match.fill_into_edit, input.text(), false)) {
 220       match.inline_autocompletion =
 221           match.fill_into_edit.substr(input.text().length());
 222       match.allowed_to_be_default_match =
 223           !input.prevent_inline_autocomplete() ||
 224           match.inline_autocompletion.empty();
 225     }
 226   } else {
 227     const size_t inline_autocomplete_offset =
 228         URLPrefix::GetInlineAutocompleteOffset(
 229             input.text(), fixed_up_input_text, true, match.fill_into_edit);
 230     if (inline_autocomplete_offset != base::string16::npos) {
 231       match.inline_autocompletion =
 232           match.fill_into_edit.substr(inline_autocomplete_offset);
 233       match.allowed_to_be_default_match =
 234           !HistoryProvider::PreventInlineAutocomplete(input) ||
 235           match.inline_autocompletion.empty();
 236     }
 237   }
 238   match.EnsureUWYTIsAllowedToBeDefault(
 239       input.canonicalized_url(),
 240       TemplateURLServiceFactory::GetForProfile(profile_));
 241
 242   // Try to mark pieces of the contents and description as matches if they
 243   // appear in |input.text()|.
 244   const base::string16 term_string = base::i18n::ToLower(input.text());
 245   WordMap terms_map(CreateWordMapForString(term_string));
 246   if (!terms_map.empty()) {
 247     match.contents_class = ClassifyAllMatchesInString(term_string, terms_map,
 248         match.contents, match.contents_class);
 249     match.description_class = ClassifyAllMatchesInString(term_string, terms_map,
 250         match.description, match.description_class);
 251   }
 252   return match;
 253 }
 254
 255 // static
 256 ShortcutsProvider::WordMap ShortcutsProvider::CreateWordMapForString(
 257     const base::string16& text) {
 258   // First, convert |text| to a vector of the unique words in it.
 259   WordMap word_map;
 260   base::i18n::BreakIterator word_iter(text,
 261                                       base::i18n::BreakIterator::BREAK_WORD);
 262   if (!word_iter.Init())
 263     return word_map;
 264   std::vector<base::string16> words;
 265   while (word_iter.Advance()) {
 266     if (word_iter.IsWord())
 267       words.push_back(word_iter.GetString());
 268   }
 269   if (words.empty())
 270     return word_map;
 271   std::sort(words.begin(), words.end());
 272   words.erase(std::unique(words.begin(), words.end()), words.end());
 273
 274   // Now create a map from (first character) to (words beginning with that
 275   // character).  We insert in reverse lexicographical order and rely on the
 276   // multimap preserving insertion order for values with the same key.  (This
 277   // is mandated in C++11, and part of that decision was based on a survey of
 278   // existing implementations that found that it was already true everywhere.)
 279   std::reverse(words.begin(), words.end());
 280   for (std::vector<base::string16>::const_iterator i(words.begin());
 281        i != words.end(); ++i)
 282     word_map.insert(std::make_pair((*i)[0], *i));
 283   return word_map;
 284 }
 285
 286 // static
 287 ACMatchClassifications ShortcutsProvider::ClassifyAllMatchesInString(
 288     const base::string16& find_text,
 289     const WordMap& find_words,
 290     const base::string16& text,
 291     const ACMatchClassifications& original_class) {
 292   DCHECK(!find_text.empty());
 293   DCHECK(!find_words.empty());
 294
 295   // The code below assumes |text| is nonempty and therefore the resulting
 296   // classification vector should always be nonempty as well.  Returning early
 297   // if |text| is empty assures we'll return the (correct) empty vector rather
 298   // than a vector with a single (0, NONE) match.
 299   if (text.empty())
 300     return original_class;
 301
 302   // First check whether |text| begins with |find_text| and mark that whole
 303   // section as a match if so.
 304   base::string16 text_lowercase(base::i18n::ToLower(text));
 305   ACMatchClassifications match_class;
 306   size_t last_position = 0;
 307   if (StartsWith(text_lowercase, find_text, true)) {
 308     match_class.push_back(
 309         ACMatchClassification(0, ACMatchClassification::MATCH));
 310     last_position = find_text.length();
 311     // If |text_lowercase| is actually equal to |find_text|, we don't need to
 312     // (and in fact shouldn't) put a trailing NONE classification after the end
 313     // of the string.
 314     if (last_position < text_lowercase.length()) {
 315       match_class.push_back(
 316           ACMatchClassification(last_position, ACMatchClassification::NONE));
 317     }
 318   } else {
 319     // |match_class| should start at position 0.  If the first matching word is
 320     // found at position 0, this will be popped from the vector further down.
 321     match_class.push_back(
 322         ACMatchClassification(0, ACMatchClassification::NONE));
 323   }
 324
 325   // Now, starting with |last_position|, check each character in
 326   // |text_lowercase| to see if we have words starting with that character in
 327   // |find_words|.  If so, check each of them to see if they match the portion
 328   // of |text_lowercase| beginning with |last_position|.  Accept the first
 329   // matching word found (which should be the longest possible match at this
 330   // location, given the construction of |find_words|) and add a MATCH region to
 331   // |match_class|, moving |last_position| to be after the matching word.  If we
 332   // found no matching words, move to the next character and repeat.
 333   while (last_position < text_lowercase.length()) {
 334     std::pair<WordMap::const_iterator, WordMap::const_iterator> range(
 335         find_words.equal_range(text_lowercase[last_position]));
 336     size_t next_character = last_position + 1;
 337     for (WordMap::const_iterator i(range.first); i != range.second; ++i) {
 338       const base::string16& word = i->second;
 339       size_t word_end = last_position + word.length();
 340       if ((word_end <= text_lowercase.length()) &&
 341           !text_lowercase.compare(last_position, word.length(), word)) {
 342         // Collapse adjacent ranges into one.
 343         if (match_class.back().offset == last_position)
 344           match_class.pop_back();
 345
 346         AutocompleteMatch::AddLastClassificationIfNecessary(&match_class,
 347             last_position, ACMatchClassification::MATCH);
 348         if (word_end < text_lowercase.length()) {
 349           match_class.push_back(
 350               ACMatchClassification(word_end, ACMatchClassification::NONE));
 351         }
 352         last_position = word_end;
 353         break;
 354       }
 355     }
 356     last_position = std::max(last_position, next_character);
 357   }
 358
 359   return AutocompleteMatch::MergeClassifications(original_class, match_class);
 360 }
 361
 362 ShortcutsBackend::ShortcutMap::const_iterator
 363     ShortcutsProvider::FindFirstMatch(const base::string16& keyword,
 364                                       ShortcutsBackend* backend) {
 365   DCHECK(backend);
 366   ShortcutsBackend::ShortcutMap::const_iterator it =
 367       backend->shortcuts_map().lower_bound(keyword);
 368   // Lower bound not necessarily matches the keyword, check for item pointed by
 369   // the lower bound iterator to at least start with keyword.
 370   return ((it == backend->shortcuts_map().end()) ||
 371     StartsWith(it->first, keyword, true)) ? it :
 372     backend->shortcuts_map().end();
 373 }
 374
 375 int ShortcutsProvider::CalculateScore(
 376     const base::string16& terms,
 377     const ShortcutsDatabase::Shortcut& shortcut,
 378     int max_relevance) {
 379   DCHECK(!terms.empty());
 380   DCHECK_LE(terms.length(), shortcut.text.length());
 381
 382   // The initial score is based on how much of the shortcut the user has typed.
 383   // Using the square root of the typed fraction boosts the base score rapidly
 384   // as characters are typed, compared with simply using the typed fraction
 385   // directly. This makes sense since the first characters typed are much more
 386   // important for determining how likely it is a user wants a particular
 387   // shortcut than are the remaining continued characters.
 388   double base_score = max_relevance *
 389       sqrt(static_cast<double>(terms.length()) / shortcut.text.length());
 390
 391   // Then we decay this by half each week.
 392   const double kLn2 = 0.6931471805599453;
 393   base::TimeDelta time_passed = base::Time::Now() - shortcut.last_access_time;
 394   // Clamp to 0 in case time jumps backwards (e.g. due to DST).
 395   double decay_exponent = std::max(0.0, kLn2 * static_cast<double>(
 396       time_passed.InMicroseconds()) / base::Time::kMicrosecondsPerWeek);
 397
 398   // We modulate the decay factor based on how many times the shortcut has been
 399   // used. Newly created shortcuts decay at full speed; otherwise, decaying by
 400   // half takes |n| times as much time, where n increases by
 401   // (1.0 / each 5 additional hits), up to a maximum of 5x as long.
 402   const double kMaxDecaySpeedDivisor = 5.0;
 403   const double kNumUsesPerDecaySpeedDivisorIncrement = 5.0;
 404   double decay_divisor = std::min(kMaxDecaySpeedDivisor,
 405       (shortcut.number_of_hits + kNumUsesPerDecaySpeedDivisorIncrement - 1) /
 406       kNumUsesPerDecaySpeedDivisorIncrement);
 407
 408   return static_cast<int>((base_score / exp(decay_exponent / decay_divisor)) +
 409       0.5);
 410 }