shortcuts_provider.cc revision ba5b9a6411cb1792fd21f0a078d7a25cd1ceec16 
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_provider_listener.h"
23#include "chrome/browser/autocomplete/autocomplete_result.h"
24#include "chrome/browser/history/history_notifications.h"
25#include "chrome/browser/history/history_service.h"
26#include "chrome/browser/history/history_service_factory.h"
27#include "chrome/browser/history/shortcuts_backend_factory.h"
28#include "chrome/browser/omnibox/omnibox_field_trial.h"
29#include "chrome/browser/profiles/profile.h"
30#include "chrome/common/pref_names.h"
31#include "chrome/common/url_constants.h"
32#include "url/url_parse.h"
33
34namespace {
35
36class RemoveMatchPredicate {
37 public:
38  explicit RemoveMatchPredicate(const std::set<GURL>& urls)
39      : urls_(urls) {
40  }
41  bool operator()(const AutocompleteMatch& match) {
42    return urls_.find(match.destination_url) != urls_.end();
43  }
44 private:
45  // Lifetime of the object is less than the lifetime of passed |urls|, so
46  // it is safe to store reference.
47  const std::set<GURL>& urls_;
48};
49
50}  // namespace
51
52ShortcutsProvider::ShortcutsProvider(AutocompleteProviderListener* listener,
53                                     Profile* profile)
54    : AutocompleteProvider(listener, profile,
55          AutocompleteProvider::TYPE_SHORTCUTS),
56      languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)),
57      initialized_(false) {
58  scoped_refptr<history::ShortcutsBackend> backend =
59      ShortcutsBackendFactory::GetForProfile(profile_);
60  if (backend.get()) {
61    backend->AddObserver(this);
62    if (backend->initialized())
63      initialized_ = true;
64  }
65}
66
67void ShortcutsProvider::Start(const AutocompleteInput& input,
68                              bool minimal_changes) {
69  matches_.clear();
70
71  if ((input.type() == AutocompleteInput::INVALID) ||
72      (input.type() == AutocompleteInput::FORCED_QUERY))
73    return;
74
75  // None of our results are applicable for best match.
76  if (input.matches_requested() == AutocompleteInput::BEST_MATCH)
77    return;
78
79  if (input.text().empty())
80    return;
81
82  if (!initialized_)
83    return;
84
85  base::TimeTicks start_time = base::TimeTicks::Now();
86  GetMatches(input);
87  if (input.text().length() < 6) {
88    base::TimeTicks end_time = base::TimeTicks::Now();
89    std::string name = "ShortcutsProvider.QueryIndexTime." +
90        base::IntToString(input.text().size());
91    base::HistogramBase* counter = base::Histogram::FactoryGet(
92        name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
93    counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
94  }
95  UpdateStarredStateOfMatches();
96}
97
98void ShortcutsProvider::DeleteMatch(const AutocompleteMatch& match) {
99  // Copy the URL since DeleteMatchesWithURLs() will invalidate |match|.
100  GURL url(match.destination_url);
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  std::set<GURL> urls;
105  urls.insert(url);
106  // Immediately delete matches and shortcuts with the URL, so we can update the
107  // controller synchronously.
108  DeleteShortcutsWithURLs(urls);
109  DeleteMatchesWithURLs(urls);  // NOTE: |match| is now dead!
110
111  // Delete the match from the history DB. This will eventually result in a
112  // second call to DeleteShortcutsWithURLs(), which is harmless.
113  HistoryService* const history_service =
114      HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
115
116  DCHECK(history_service && url.is_valid());
117  history_service->DeleteURL(url);
118}
119
120ShortcutsProvider::~ShortcutsProvider() {
121  scoped_refptr<history::ShortcutsBackend> backend =
122      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
123  if (backend.get())
124    backend->RemoveObserver(this);
125}
126
127void ShortcutsProvider::OnShortcutsLoaded() {
128  initialized_ = true;
129}
130
131void ShortcutsProvider::DeleteMatchesWithURLs(const std::set<GURL>& urls) {
132  std::remove_if(matches_.begin(), matches_.end(), RemoveMatchPredicate(urls));
133  listener_->OnProviderUpdate(true);
134}
135
136void ShortcutsProvider::DeleteShortcutsWithURLs(const std::set<GURL>& urls) {
137  scoped_refptr<history::ShortcutsBackend> backend =
138      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
139  if (!backend.get())
140    return;  // We are off the record.
141  for (std::set<GURL>::const_iterator url = urls.begin(); url != urls.end();
142       ++url)
143    backend->DeleteShortcutsWithUrl(*url);
144}
145
146void ShortcutsProvider::GetMatches(const AutocompleteInput& input) {
147  scoped_refptr<history::ShortcutsBackend> backend =
148      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
149  if (!backend.get())
150    return;
151  // Get the URLs from the shortcuts database with keys that partially or
152  // completely match the search term.
153  string16 term_string(base::i18n::ToLower(input.text()));
154  DCHECK(!term_string.empty());
155
156  int max_relevance;
157  if (!OmniboxFieldTrial::ShortcutsScoringMaxRelevance(
158      input.current_page_classification(), &max_relevance))
159    max_relevance = AutocompleteResult::kLowestDefaultScore - 1;
160
161  for (history::ShortcutsBackend::ShortcutMap::const_iterator it =
162           FindFirstMatch(term_string, backend.get());
163       it != backend->shortcuts_map().end() &&
164           StartsWith(it->first, term_string, true); ++it) {
165    // Don't return shortcuts with zero relevance.
166    int relevance = CalculateScore(term_string, it->second, max_relevance);
167    if (relevance)
168      matches_.push_back(ShortcutToACMatch(relevance, term_string, it->second));
169  }
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  // Reset relevance scores to guarantee no results are given an
179  // inlineable score and all scores are decreasing (but not do assign
180  // any scores below 1).
181  max_relevance = AutocompleteResult::kLowestDefaultScore - 1;
182  for (ACMatches::iterator it = matches_.begin(); it != matches_.end(); ++it) {
183    max_relevance = std::min(max_relevance, it->relevance);
184    it->relevance = max_relevance;
185    if (max_relevance > 1)
186      --max_relevance;
187  }
188}
189
190AutocompleteMatch ShortcutsProvider::ShortcutToACMatch(
191    int relevance,
192    const string16& term_string,
193    const history::ShortcutsBackend::Shortcut& shortcut) {
194  DCHECK(!term_string.empty());
195  AutocompleteMatch match(this, relevance, true,
196                          AutocompleteMatchType::HISTORY_TITLE);
197  match.destination_url = shortcut.url;
198  DCHECK(match.destination_url.is_valid());
199  match.fill_into_edit = UTF8ToUTF16(shortcut.url.spec());
200  match.contents = shortcut.contents;
201  match.contents_class = shortcut.contents_class;
202  match.description = shortcut.description;
203  match.description_class = shortcut.description_class;
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", UTF16ToUTF8(shortcut.text));
207
208  // Try to mark pieces of the contents and description as matches if they
209  // appear in |term_string|.
210  WordMap terms_map(CreateWordMapForString(term_string));
211  if (!terms_map.empty()) {
212    match.contents_class = ClassifyAllMatchesInString(term_string, terms_map,
213        match.contents, match.contents_class);
214    match.description_class = ClassifyAllMatchesInString(term_string, terms_map,
215        match.description, match.description_class);
216  }
217  return match;
218}
219
220// static
221ShortcutsProvider::WordMap ShortcutsProvider::CreateWordMapForString(
222    const string16& text) {
223  // First, convert |text| to a vector of the unique words in it.
224  WordMap word_map;
225  base::i18n::BreakIterator word_iter(text,
226                                      base::i18n::BreakIterator::BREAK_WORD);
227  if (!word_iter.Init())
228    return word_map;
229  std::vector<string16> words;
230  while (word_iter.Advance()) {
231    if (word_iter.IsWord())
232      words.push_back(word_iter.GetString());
233  }
234  if (words.empty())
235    return word_map;
236  std::sort(words.begin(), words.end());
237  words.erase(std::unique(words.begin(), words.end()), words.end());
238
239  // Now create a map from (first character) to (words beginning with that
240  // character).  We insert in reverse lexicographical order and rely on the
241  // multimap preserving insertion order for values with the same key.  (This
242  // is mandated in C++11, and part of that decision was based on a survey of
243  // existing implementations that found that it was already true everywhere.)
244  std::reverse(words.begin(), words.end());
245  for (std::vector<string16>::const_iterator i(words.begin()); i != words.end();
246       ++i)
247    word_map.insert(std::make_pair((*i)[0], *i));
248  return word_map;
249}
250
251// static
252ACMatchClassifications ShortcutsProvider::ClassifyAllMatchesInString(
253    const string16& find_text,
254    const WordMap& find_words,
255    const string16& text,
256    const ACMatchClassifications& original_class) {
257  DCHECK(!find_text.empty());
258  DCHECK(!find_words.empty());
259
260  // The code below assumes |text| is nonempty and therefore the resulting
261  // classification vector should always be nonempty as well.  Returning early
262  // if |text| is empty assures we'll return the (correct) empty vector rather
263  // than a vector with a single (0, NONE) match.
264  if (text.empty())
265    return original_class;
266
267  // First check whether |text| begins with |find_text| and mark that whole
268  // section as a match if so.
269  string16 text_lowercase(base::i18n::ToLower(text));
270  ACMatchClassifications match_class;
271  size_t last_position = 0;
272  if (StartsWith(text_lowercase, find_text, true)) {
273    match_class.push_back(
274        ACMatchClassification(0, ACMatchClassification::MATCH));
275    last_position = find_text.length();
276    // If |text_lowercase| is actually equal to |find_text|, we don't need to
277    // (and in fact shouldn't) put a trailing NONE classification after the end
278    // of the string.
279    if (last_position < text_lowercase.length()) {
280      match_class.push_back(
281          ACMatchClassification(last_position, ACMatchClassification::NONE));
282    }
283  } else {
284    // |match_class| should start at position 0.  If the first matching word is
285    // found at position 0, this will be popped from the vector further down.
286    match_class.push_back(
287        ACMatchClassification(0, ACMatchClassification::NONE));
288  }
289
290  // Now, starting with |last_position|, check each character in
291  // |text_lowercase| to see if we have words starting with that character in
292  // |find_words|.  If so, check each of them to see if they match the portion
293  // of |text_lowercase| beginning with |last_position|.  Accept the first
294  // matching word found (which should be the longest possible match at this
295  // location, given the construction of |find_words|) and add a MATCH region to
296  // |match_class|, moving |last_position| to be after the matching word.  If we
297  // found no matching words, move to the next character and repeat.
298  while (last_position < text_lowercase.length()) {
299    std::pair<WordMap::const_iterator, WordMap::const_iterator> range(
300        find_words.equal_range(text_lowercase[last_position]));
301    size_t next_character = last_position + 1;
302    for (WordMap::const_iterator i(range.first); i != range.second; ++i) {
303      const string16& word = i->second;
304      size_t word_end = last_position + word.length();
305      if ((word_end <= text_lowercase.length()) &&
306          !text_lowercase.compare(last_position, word.length(), word)) {
307        // Collapse adjacent ranges into one.
308        if (match_class.back().offset == last_position)
309          match_class.pop_back();
310
311        AutocompleteMatch::AddLastClassificationIfNecessary(&match_class,
312            last_position, ACMatchClassification::MATCH);
313        if (word_end < text_lowercase.length()) {
314          match_class.push_back(
315              ACMatchClassification(word_end, ACMatchClassification::NONE));
316        }
317        last_position = word_end;
318        break;
319      }
320    }
321    last_position = std::max(last_position, next_character);
322  }
323
324  return AutocompleteMatch::MergeClassifications(original_class, match_class);
325}
326
327history::ShortcutsBackend::ShortcutMap::const_iterator
328    ShortcutsProvider::FindFirstMatch(const string16& keyword,
329                                      history::ShortcutsBackend* backend) {
330  DCHECK(backend);
331  history::ShortcutsBackend::ShortcutMap::const_iterator it =
332      backend->shortcuts_map().lower_bound(keyword);
333  // Lower bound not necessarily matches the keyword, check for item pointed by
334  // the lower bound iterator to at least start with keyword.
335  return ((it == backend->shortcuts_map().end()) ||
336    StartsWith(it->first, keyword, true)) ? it :
337    backend->shortcuts_map().end();
338}
339
340int ShortcutsProvider::CalculateScore(
341    const string16& terms,
342    const history::ShortcutsBackend::Shortcut& shortcut,
343    int max_relevance) {
344  DCHECK(!terms.empty());
345  DCHECK_LE(terms.length(), shortcut.text.length());
346
347  // The initial score is based on how much of the shortcut the user has typed.
348  // Using the square root of the typed fraction boosts the base score rapidly
349  // as characters are typed, compared with simply using the typed fraction
350  // directly. This makes sense since the first characters typed are much more
351  // important for determining how likely it is a user wants a particular
352  // shortcut than are the remaining continued characters.
353  double base_score = max_relevance *
354      sqrt(static_cast<double>(terms.length()) / shortcut.text.length());
355
356  // Then we decay this by half each week.
357  const double kLn2 = 0.6931471805599453;
358  base::TimeDelta time_passed = base::Time::Now() - shortcut.last_access_time;
359  // Clamp to 0 in case time jumps backwards (e.g. due to DST).
360  double decay_exponent = std::max(0.0, kLn2 * static_cast<double>(
361      time_passed.InMicroseconds()) / base::Time::kMicrosecondsPerWeek);
362
363  // We modulate the decay factor based on how many times the shortcut has been
364  // used. Newly created shortcuts decay at full speed; otherwise, decaying by
365  // half takes |n| times as much time, where n increases by
366  // (1.0 / each 5 additional hits), up to a maximum of 5x as long.
367  const double kMaxDecaySpeedDivisor = 5.0;
368  const double kNumUsesPerDecaySpeedDivisorIncrement = 5.0;
369  double decay_divisor = std::min(kMaxDecaySpeedDivisor,
370      (shortcut.number_of_hits + kNumUsesPerDecaySpeedDivisorIncrement - 1) /
371      kNumUsesPerDecaySpeedDivisorIncrement);
372
373  return static_cast<int>((base_score / exp(decay_exponent / decay_divisor)) +
374      0.5);
375}
376