shortcuts_provider.cc revision f8ee788a64d60abd8f2d742a5fdedde054ecd910 
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/autocomplete_input.h"
22#include "chrome/browser/autocomplete/autocomplete_match.h"
23#include "chrome/browser/autocomplete/autocomplete_provider_listener.h"
24#include "chrome/browser/autocomplete/autocomplete_result.h"
25#include "chrome/browser/autocomplete/history_provider.h"
26#include "chrome/browser/autocomplete/shortcuts_backend_factory.h"
27#include "chrome/browser/history/history_notifications.h"
28#include "chrome/browser/history/history_service.h"
29#include "chrome/browser/history/history_service_factory.h"
30#include "chrome/browser/omnibox/omnibox_field_trial.h"
31#include "chrome/browser/profiles/profile.h"
32#include "chrome/common/pref_names.h"
33#include "chrome/common/url_constants.h"
34#include "components/autocomplete/url_prefix.h"
35#include "components/metrics/proto/omnibox_input_type.pb.h"
36#include "components/url_fixer/url_fixer.h"
37#include "url/url_parse.h"
38
39namespace {
40
41class DestinationURLEqualsURL {
42 public:
43  explicit DestinationURLEqualsURL(const GURL& url) : url_(url) {}
44  bool operator()(const AutocompleteMatch& match) const {
45    return match.destination_url == url_;
46  }
47 private:
48  const GURL url_;
49};
50
51}  // namespace
52
53const int ShortcutsProvider::kShortcutsProviderDefaultMaxRelevance = 1199;
54
55ShortcutsProvider::ShortcutsProvider(AutocompleteProviderListener* listener,
56                                     Profile* profile)
57    : AutocompleteProvider(listener, profile,
58          AutocompleteProvider::TYPE_SHORTCUTS),
59      languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)),
60      initialized_(false) {
61  scoped_refptr<ShortcutsBackend> backend =
62      ShortcutsBackendFactory::GetForProfile(profile_);
63  if (backend.get()) {
64    backend->AddObserver(this);
65    if (backend->initialized())
66      initialized_ = true;
67  }
68}
69
70void ShortcutsProvider::Start(const AutocompleteInput& input,
71                              bool minimal_changes) {
72  matches_.clear();
73
74  if ((input.type() == metrics::OmniboxInputType::INVALID) ||
75      (input.type() == metrics::OmniboxInputType::FORCED_QUERY))
76    return;
77
78  if (input.text().empty())
79    return;
80
81  if (!initialized_)
82    return;
83
84  base::TimeTicks start_time = base::TimeTicks::Now();
85  GetMatches(input);
86  if (input.text().length() < 6) {
87    base::TimeTicks end_time = base::TimeTicks::Now();
88    std::string name = "ShortcutsProvider.QueryIndexTime." +
89        base::IntToString(input.text().size());
90    base::HistogramBase* counter = base::Histogram::FactoryGet(
91        name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
92    counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
93  }
94  UpdateStarredStateOfMatches();
95}
96
97void ShortcutsProvider::DeleteMatch(const AutocompleteMatch& match) {
98  // Copy the URL since deleting from |matches_| will invalidate |match|.
99  GURL url(match.destination_url);
100  DCHECK(url.is_valid());
101
102  // When a user deletes a match, he probably means for the URL to disappear out
103  // of history entirely. So nuke all shortcuts that map to this URL.
104  scoped_refptr<ShortcutsBackend> backend =
105      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
106  if (backend) // Can be NULL in Incognito.
107    backend->DeleteShortcutsWithURL(url);
108
109  matches_.erase(std::remove_if(matches_.begin(), matches_.end(),
110                                DestinationURLEqualsURL(url)),
111                 matches_.end());
112  // NOTE: |match| is now dead!
113
114  // Delete the match from the history DB. This will eventually result in a
115  // second call to DeleteShortcutsWithURL(), which is harmless.
116  HistoryService* const history_service =
117      HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
118  DCHECK(history_service);
119  history_service->DeleteURL(url);
120}
121
122ShortcutsProvider::~ShortcutsProvider() {
123  scoped_refptr<ShortcutsBackend> backend =
124      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
125  if (backend.get())
126    backend->RemoveObserver(this);
127}
128
129void ShortcutsProvider::OnShortcutsLoaded() {
130  initialized_ = true;
131}
132
133void ShortcutsProvider::GetMatches(const AutocompleteInput& input) {
134  scoped_refptr<ShortcutsBackend> backend =
135      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
136  if (!backend.get())
137    return;
138  // Get the URLs from the shortcuts database with keys that partially or
139  // completely match the search term.
140  base::string16 term_string(base::i18n::ToLower(input.text()));
141  DCHECK(!term_string.empty());
142
143  const GURL& input_as_gurl =
144      url_fixer::FixupURL(base::UTF16ToUTF8(input.text()), std::string());
145  const base::string16 fixed_up_input(FixupUserInput(input).second);
146
147  int max_relevance;
148  if (!OmniboxFieldTrial::ShortcutsScoringMaxRelevance(
149      input.current_page_classification(), &max_relevance))
150    max_relevance = kShortcutsProviderDefaultMaxRelevance;
151
152  for (ShortcutsBackend::ShortcutMap::const_iterator it =
153           FindFirstMatch(term_string, backend.get());
154       it != backend->shortcuts_map().end() &&
155           StartsWith(it->first, term_string, true); ++it) {
156    // Don't return shortcuts with zero relevance.
157    int relevance = CalculateScore(term_string, it->second, max_relevance);
158    if (relevance) {
159      matches_.push_back(ShortcutToACMatch(it->second, relevance, input,
160                                           fixed_up_input, input_as_gurl));
161      matches_.back().ComputeStrippedDestinationURL(profile_);
162    }
163  }
164  // Remove duplicates.  Duplicates don't need to be preserved in the matches
165  // because they are only used for deletions, and shortcuts deletes matches
166  // based on the URL.
167  AutocompleteResult::DedupMatchesByDestination(
168      input.current_page_classification(), false, &matches_);
169  // Find best matches.
170  std::partial_sort(matches_.begin(),
171      matches_.begin() +
172          std::min(AutocompleteProvider::kMaxMatches, matches_.size()),
173      matches_.end(), &AutocompleteMatch::MoreRelevant);
174  if (matches_.size() > AutocompleteProvider::kMaxMatches) {
175    matches_.erase(matches_.begin() + AutocompleteProvider::kMaxMatches,
176                   matches_.end());
177  }
178  // Guarantee that all scores are decreasing (but do not assign any scores
179  // below 1).
180  for (ACMatches::iterator it = matches_.begin(); it != matches_.end(); ++it) {
181    max_relevance = std::min(max_relevance, it->relevance);
182    it->relevance = max_relevance;
183    if (max_relevance > 1)
184      --max_relevance;
185  }
186}
187
188AutocompleteMatch ShortcutsProvider::ShortcutToACMatch(
189    const history::ShortcutsDatabase::Shortcut& shortcut,
190    int relevance,
191    const AutocompleteInput& input,
192    const base::string16& fixed_up_input_text,
193    const GURL& input_as_gurl) {
194  DCHECK(!input.text().empty());
195  AutocompleteMatch match;
196  match.provider = this;
197  match.relevance = relevance;
198  match.deletable = true;
199  match.fill_into_edit = shortcut.match_core.fill_into_edit;
200  match.destination_url = shortcut.match_core.destination_url;
201  DCHECK(match.destination_url.is_valid());
202  match.contents = shortcut.match_core.contents;
203  match.contents_class = AutocompleteMatch::ClassificationsFromString(
204      shortcut.match_core.contents_class);
205  match.description = shortcut.match_core.description;
206  match.description_class = AutocompleteMatch::ClassificationsFromString(
207      shortcut.match_core.description_class);
208  match.transition =
209      static_cast<content::PageTransition>(shortcut.match_core.transition);
210  match.type = static_cast<AutocompleteMatch::Type>(shortcut.match_core.type);
211  match.keyword = shortcut.match_core.keyword;
212  match.RecordAdditionalInfo("number of hits", shortcut.number_of_hits);
213  match.RecordAdditionalInfo("last access time", shortcut.last_access_time);
214  match.RecordAdditionalInfo("original input text",
215                             base::UTF16ToUTF8(shortcut.text));
216
217  // Set |inline_autocompletion| and |allowed_to_be_default_match| if possible.
218  // If the match is a search query this is easy: simply check whether the
219  // user text is a prefix of the query.  If the match is a navigation, we
220  // assume the fill_into_edit looks something like a URL, so we use
221  // URLPrefix::GetInlineAutocompleteOffset() to try and strip off any prefixes
222  // that the user might not think would change the meaning, but would
223  // otherwise prevent inline autocompletion.  This allows, for example, the
224  // input of "foo.c" to autocomplete to "foo.com" for a fill_into_edit of
225  // "http://foo.com".
226  if (AutocompleteMatch::IsSearchType(match.type)) {
227    if (StartsWith(match.fill_into_edit, input.text(), false)) {
228      match.inline_autocompletion =
229          match.fill_into_edit.substr(input.text().length());
230      match.allowed_to_be_default_match =
231          !input.prevent_inline_autocomplete() ||
232          match.inline_autocompletion.empty();
233    }
234  } else {
235    const size_t inline_autocomplete_offset =
236        URLPrefix::GetInlineAutocompleteOffset(
237            input.text(), fixed_up_input_text, true, match.fill_into_edit);
238    if (inline_autocomplete_offset != base::string16::npos) {
239      match.inline_autocompletion =
240          match.fill_into_edit.substr(inline_autocomplete_offset);
241      match.allowed_to_be_default_match =
242          !HistoryProvider::PreventInlineAutocomplete(input) ||
243          match.inline_autocompletion.empty();
244    } else {
245      // Also allow a user's input to be marked as default if it would be fixed
246      // up to the same thing as the fill_into_edit.  This handles cases like
247      // the user input containing a trailing slash absent in fill_into_edit.
248      match.allowed_to_be_default_match =
249          (input_as_gurl ==
250           url_fixer::FixupURL(base::UTF16ToUTF8(match.fill_into_edit),
251                               std::string()));
252    }
253  }
254
255  // Try to mark pieces of the contents and description as matches if they
256  // appear in |input.text()|.
257  const base::string16 term_string = base::i18n::ToLower(input.text());
258  WordMap terms_map(CreateWordMapForString(term_string));
259  if (!terms_map.empty()) {
260    match.contents_class = ClassifyAllMatchesInString(term_string, terms_map,
261        match.contents, match.contents_class);
262    match.description_class = ClassifyAllMatchesInString(term_string, terms_map,
263        match.description, match.description_class);
264  }
265  return match;
266}
267
268// static
269ShortcutsProvider::WordMap ShortcutsProvider::CreateWordMapForString(
270    const base::string16& text) {
271  // First, convert |text| to a vector of the unique words in it.
272  WordMap word_map;
273  base::i18n::BreakIterator word_iter(text,
274                                      base::i18n::BreakIterator::BREAK_WORD);
275  if (!word_iter.Init())
276    return word_map;
277  std::vector<base::string16> words;
278  while (word_iter.Advance()) {
279    if (word_iter.IsWord())
280      words.push_back(word_iter.GetString());
281  }
282  if (words.empty())
283    return word_map;
284  std::sort(words.begin(), words.end());
285  words.erase(std::unique(words.begin(), words.end()), words.end());
286
287  // Now create a map from (first character) to (words beginning with that
288  // character).  We insert in reverse lexicographical order and rely on the
289  // multimap preserving insertion order for values with the same key.  (This
290  // is mandated in C++11, and part of that decision was based on a survey of
291  // existing implementations that found that it was already true everywhere.)
292  std::reverse(words.begin(), words.end());
293  for (std::vector<base::string16>::const_iterator i(words.begin());
294       i != words.end(); ++i)
295    word_map.insert(std::make_pair((*i)[0], *i));
296  return word_map;
297}
298
299// static
300ACMatchClassifications ShortcutsProvider::ClassifyAllMatchesInString(
301    const base::string16& find_text,
302    const WordMap& find_words,
303    const base::string16& text,
304    const ACMatchClassifications& original_class) {
305  DCHECK(!find_text.empty());
306  DCHECK(!find_words.empty());
307
308  // The code below assumes |text| is nonempty and therefore the resulting
309  // classification vector should always be nonempty as well.  Returning early
310  // if |text| is empty assures we'll return the (correct) empty vector rather
311  // than a vector with a single (0, NONE) match.
312  if (text.empty())
313    return original_class;
314
315  // First check whether |text| begins with |find_text| and mark that whole
316  // section as a match if so.
317  base::string16 text_lowercase(base::i18n::ToLower(text));
318  ACMatchClassifications match_class;
319  size_t last_position = 0;
320  if (StartsWith(text_lowercase, find_text, true)) {
321    match_class.push_back(
322        ACMatchClassification(0, ACMatchClassification::MATCH));
323    last_position = find_text.length();
324    // If |text_lowercase| is actually equal to |find_text|, we don't need to
325    // (and in fact shouldn't) put a trailing NONE classification after the end
326    // of the string.
327    if (last_position < text_lowercase.length()) {
328      match_class.push_back(
329          ACMatchClassification(last_position, ACMatchClassification::NONE));
330    }
331  } else {
332    // |match_class| should start at position 0.  If the first matching word is
333    // found at position 0, this will be popped from the vector further down.
334    match_class.push_back(
335        ACMatchClassification(0, ACMatchClassification::NONE));
336  }
337
338  // Now, starting with |last_position|, check each character in
339  // |text_lowercase| to see if we have words starting with that character in
340  // |find_words|.  If so, check each of them to see if they match the portion
341  // of |text_lowercase| beginning with |last_position|.  Accept the first
342  // matching word found (which should be the longest possible match at this
343  // location, given the construction of |find_words|) and add a MATCH region to
344  // |match_class|, moving |last_position| to be after the matching word.  If we
345  // found no matching words, move to the next character and repeat.
346  while (last_position < text_lowercase.length()) {
347    std::pair<WordMap::const_iterator, WordMap::const_iterator> range(
348        find_words.equal_range(text_lowercase[last_position]));
349    size_t next_character = last_position + 1;
350    for (WordMap::const_iterator i(range.first); i != range.second; ++i) {
351      const base::string16& word = i->second;
352      size_t word_end = last_position + word.length();
353      if ((word_end <= text_lowercase.length()) &&
354          !text_lowercase.compare(last_position, word.length(), word)) {
355        // Collapse adjacent ranges into one.
356        if (match_class.back().offset == last_position)
357          match_class.pop_back();
358
359        AutocompleteMatch::AddLastClassificationIfNecessary(&match_class,
360            last_position, ACMatchClassification::MATCH);
361        if (word_end < text_lowercase.length()) {
362          match_class.push_back(
363              ACMatchClassification(word_end, ACMatchClassification::NONE));
364        }
365        last_position = word_end;
366        break;
367      }
368    }
369    last_position = std::max(last_position, next_character);
370  }
371
372  return AutocompleteMatch::MergeClassifications(original_class, match_class);
373}
374
375ShortcutsBackend::ShortcutMap::const_iterator
376    ShortcutsProvider::FindFirstMatch(const base::string16& keyword,
377                                      ShortcutsBackend* backend) {
378  DCHECK(backend);
379  ShortcutsBackend::ShortcutMap::const_iterator it =
380      backend->shortcuts_map().lower_bound(keyword);
381  // Lower bound not necessarily matches the keyword, check for item pointed by
382  // the lower bound iterator to at least start with keyword.
383  return ((it == backend->shortcuts_map().end()) ||
384    StartsWith(it->first, keyword, true)) ? it :
385    backend->shortcuts_map().end();
386}
387
388int ShortcutsProvider::CalculateScore(
389    const base::string16& terms,
390    const history::ShortcutsDatabase::Shortcut& shortcut,
391    int max_relevance) {
392  DCHECK(!terms.empty());
393  DCHECK_LE(terms.length(), shortcut.text.length());
394
395  // The initial score is based on how much of the shortcut the user has typed.
396  // Using the square root of the typed fraction boosts the base score rapidly
397  // as characters are typed, compared with simply using the typed fraction
398  // directly. This makes sense since the first characters typed are much more
399  // important for determining how likely it is a user wants a particular
400  // shortcut than are the remaining continued characters.
401  double base_score = max_relevance *
402      sqrt(static_cast<double>(terms.length()) / shortcut.text.length());
403
404  // Then we decay this by half each week.
405  const double kLn2 = 0.6931471805599453;
406  base::TimeDelta time_passed = base::Time::Now() - shortcut.last_access_time;
407  // Clamp to 0 in case time jumps backwards (e.g. due to DST).
408  double decay_exponent = std::max(0.0, kLn2 * static_cast<double>(
409      time_passed.InMicroseconds()) / base::Time::kMicrosecondsPerWeek);
410
411  // We modulate the decay factor based on how many times the shortcut has been
412  // used. Newly created shortcuts decay at full speed; otherwise, decaying by
413  // half takes |n| times as much time, where n increases by
414  // (1.0 / each 5 additional hits), up to a maximum of 5x as long.
415  const double kMaxDecaySpeedDivisor = 5.0;
416  const double kNumUsesPerDecaySpeedDivisorIncrement = 5.0;
417  double decay_divisor = std::min(kMaxDecaySpeedDivisor,
418      (shortcut.number_of_hits + kNumUsesPerDecaySpeedDivisorIncrement - 1) /
419      kNumUsesPerDecaySpeedDivisorIncrement);
420
421  return static_cast<int>((base_score / exp(decay_exponent / decay_divisor)) +
422      0.5);
423}
424