shortcuts_provider.cc revision 4e180b6a0b4720a9b8e9e959a882386f690f08ff 
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 DestinationURLEqualsURL {
37 public:
38  explicit DestinationURLEqualsURL(const GURL& url) : url_(url) {}
39  bool operator()(const AutocompleteMatch& match) const {
40    return match.destination_url == url_;
41  }
42 private:
43  const GURL url_;
44};
45
46}  // namespace
47
48ShortcutsProvider::ShortcutsProvider(AutocompleteProviderListener* listener,
49                                     Profile* profile)
50    : AutocompleteProvider(listener, profile,
51          AutocompleteProvider::TYPE_SHORTCUTS),
52      languages_(profile_->GetPrefs()->GetString(prefs::kAcceptLanguages)),
53      initialized_(false) {
54  scoped_refptr<history::ShortcutsBackend> backend =
55      ShortcutsBackendFactory::GetForProfile(profile_);
56  if (backend.get()) {
57    backend->AddObserver(this);
58    if (backend->initialized())
59      initialized_ = true;
60  }
61}
62
63void ShortcutsProvider::Start(const AutocompleteInput& input,
64                              bool minimal_changes) {
65  matches_.clear();
66
67  if ((input.type() == AutocompleteInput::INVALID) ||
68      (input.type() == AutocompleteInput::FORCED_QUERY))
69    return;
70
71  // None of our results are applicable for best match.
72  if (input.matches_requested() == AutocompleteInput::BEST_MATCH)
73    return;
74
75  if (input.text().empty())
76    return;
77
78  if (!initialized_)
79    return;
80
81  base::TimeTicks start_time = base::TimeTicks::Now();
82  GetMatches(input);
83  if (input.text().length() < 6) {
84    base::TimeTicks end_time = base::TimeTicks::Now();
85    std::string name = "ShortcutsProvider.QueryIndexTime." +
86        base::IntToString(input.text().size());
87    base::HistogramBase* counter = base::Histogram::FactoryGet(
88        name, 1, 1000, 50, base::Histogram::kUmaTargetedHistogramFlag);
89    counter->Add(static_cast<int>((end_time - start_time).InMilliseconds()));
90  }
91  UpdateStarredStateOfMatches();
92}
93
94void ShortcutsProvider::DeleteMatch(const AutocompleteMatch& match) {
95  // Copy the URL since DeleteMatchesWithURLs() will invalidate |match|.
96  GURL url(match.destination_url);
97
98  // When a user deletes a match, he probably means for the URL to disappear out
99  // of history entirely. So nuke all shortcuts that map to this URL.
100  scoped_refptr<history::ShortcutsBackend> backend =
101      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
102  if (backend)  // Can be NULL in Incognito.
103    backend->DeleteShortcutsWithUrl(url);
104  matches_.erase(std::remove_if(matches_.begin(), matches_.end(),
105                                DestinationURLEqualsURL(url)),
106                 matches_.end());
107  // NOTE: |match| is now dead!
108  listener_->OnProviderUpdate(true);
109
110  // Delete the match from the history DB. This will eventually result in a
111  // second call to DeleteShortcutsWithURLs(), which is harmless.
112  HistoryService* const history_service =
113      HistoryServiceFactory::GetForProfile(profile_, Profile::EXPLICIT_ACCESS);
114
115  DCHECK(history_service && url.is_valid());
116  history_service->DeleteURL(url);
117}
118
119ShortcutsProvider::~ShortcutsProvider() {
120  scoped_refptr<history::ShortcutsBackend> backend =
121      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
122  if (backend.get())
123    backend->RemoveObserver(this);
124}
125
126void ShortcutsProvider::OnShortcutsLoaded() {
127  initialized_ = true;
128}
129
130void ShortcutsProvider::GetMatches(const AutocompleteInput& input) {
131  scoped_refptr<history::ShortcutsBackend> backend =
132      ShortcutsBackendFactory::GetForProfileIfExists(profile_);
133  if (!backend.get())
134    return;
135  // Get the URLs from the shortcuts database with keys that partially or
136  // completely match the search term.
137  string16 term_string(base::i18n::ToLower(input.text()));
138  DCHECK(!term_string.empty());
139
140  int max_relevance;
141  if (!OmniboxFieldTrial::ShortcutsScoringMaxRelevance(
142      input.current_page_classification(), &max_relevance))
143    max_relevance = AutocompleteResult::kLowestDefaultScore - 1;
144
145  for (history::ShortcutsBackend::ShortcutMap::const_iterator it =
146           FindFirstMatch(term_string, backend.get());
147       it != backend->shortcuts_map().end() &&
148           StartsWith(it->first, term_string, true); ++it) {
149    // Don't return shortcuts with zero relevance.
150    int relevance = CalculateScore(term_string, it->second, max_relevance);
151    if (relevance)
152      matches_.push_back(ShortcutToACMatch(relevance, term_string, it->second));
153  }
154  std::partial_sort(matches_.begin(),
155      matches_.begin() +
156          std::min(AutocompleteProvider::kMaxMatches, matches_.size()),
157      matches_.end(), &AutocompleteMatch::MoreRelevant);
158  if (matches_.size() > AutocompleteProvider::kMaxMatches) {
159    matches_.erase(matches_.begin() + AutocompleteProvider::kMaxMatches,
160                   matches_.end());
161  }
162  // Reset relevance scores to guarantee no match is given a score that may
163  // allow it to become the highest ranked match (i.e., the default match)
164  // unless the omnibox will reorder matches as necessary to correct the
165  // problem.  (Shortcuts matches are sometimes not inline-autocompletable
166  // and, even when they are, the ShortcutsProvider does not bother to set
167  // |inline_autocompletion|.  Hence these matches can never be displayed
168  // corectly in the omnibox as the default match.)  In the process of
169  // resetting scores, guarantee that all scores are decreasing (but do
170  // not assign any scores below 1).
171  if (!OmniboxFieldTrial::ReorderForLegalDefaultMatch(
172      input.current_page_classification())) {
173    int max_relevance = AutocompleteResult::kLowestDefaultScore - 1;
174    for (ACMatches::iterator it = matches_.begin(); it != matches_.end();
175        ++it) {
176      max_relevance = std::min(max_relevance, it->relevance);
177      it->relevance = max_relevance;
178      if (max_relevance > 1)
179        --max_relevance;
180    }
181  }
182}
183
184AutocompleteMatch ShortcutsProvider::ShortcutToACMatch(
185    int relevance,
186    const string16& term_string,
187    const history::ShortcutsBackend::Shortcut& shortcut) {
188  DCHECK(!term_string.empty());
189  AutocompleteMatch match(shortcut.match_core.ToMatch());
190  match.provider = this;
191  match.relevance = relevance;
192  match.deletable = true;
193  DCHECK(match.destination_url.is_valid());
194  match.RecordAdditionalInfo("number of hits", shortcut.number_of_hits);
195  match.RecordAdditionalInfo("last access time", shortcut.last_access_time);
196  match.RecordAdditionalInfo("original input text", UTF16ToUTF8(shortcut.text));
197
198  // Try to mark pieces of the contents and description as matches if they
199  // appear in |term_string|.
200  WordMap terms_map(CreateWordMapForString(term_string));
201  if (!terms_map.empty()) {
202    match.contents_class = ClassifyAllMatchesInString(term_string, terms_map,
203        match.contents, match.contents_class);
204    match.description_class = ClassifyAllMatchesInString(term_string, terms_map,
205        match.description, match.description_class);
206  }
207  return match;
208}
209
210// static
211ShortcutsProvider::WordMap ShortcutsProvider::CreateWordMapForString(
212    const string16& text) {
213  // First, convert |text| to a vector of the unique words in it.
214  WordMap word_map;
215  base::i18n::BreakIterator word_iter(text,
216                                      base::i18n::BreakIterator::BREAK_WORD);
217  if (!word_iter.Init())
218    return word_map;
219  std::vector<string16> words;
220  while (word_iter.Advance()) {
221    if (word_iter.IsWord())
222      words.push_back(word_iter.GetString());
223  }
224  if (words.empty())
225    return word_map;
226  std::sort(words.begin(), words.end());
227  words.erase(std::unique(words.begin(), words.end()), words.end());
228
229  // Now create a map from (first character) to (words beginning with that
230  // character).  We insert in reverse lexicographical order and rely on the
231  // multimap preserving insertion order for values with the same key.  (This
232  // is mandated in C++11, and part of that decision was based on a survey of
233  // existing implementations that found that it was already true everywhere.)
234  std::reverse(words.begin(), words.end());
235  for (std::vector<string16>::const_iterator i(words.begin()); i != words.end();
236       ++i)
237    word_map.insert(std::make_pair((*i)[0], *i));
238  return word_map;
239}
240
241// static
242ACMatchClassifications ShortcutsProvider::ClassifyAllMatchesInString(
243    const string16& find_text,
244    const WordMap& find_words,
245    const string16& text,
246    const ACMatchClassifications& original_class) {
247  DCHECK(!find_text.empty());
248  DCHECK(!find_words.empty());
249
250  // The code below assumes |text| is nonempty and therefore the resulting
251  // classification vector should always be nonempty as well.  Returning early
252  // if |text| is empty assures we'll return the (correct) empty vector rather
253  // than a vector with a single (0, NONE) match.
254  if (text.empty())
255    return original_class;
256
257  // First check whether |text| begins with |find_text| and mark that whole
258  // section as a match if so.
259  string16 text_lowercase(base::i18n::ToLower(text));
260  ACMatchClassifications match_class;
261  size_t last_position = 0;
262  if (StartsWith(text_lowercase, find_text, true)) {
263    match_class.push_back(
264        ACMatchClassification(0, ACMatchClassification::MATCH));
265    last_position = find_text.length();
266    // If |text_lowercase| is actually equal to |find_text|, we don't need to
267    // (and in fact shouldn't) put a trailing NONE classification after the end
268    // of the string.
269    if (last_position < text_lowercase.length()) {
270      match_class.push_back(
271          ACMatchClassification(last_position, ACMatchClassification::NONE));
272    }
273  } else {
274    // |match_class| should start at position 0.  If the first matching word is
275    // found at position 0, this will be popped from the vector further down.
276    match_class.push_back(
277        ACMatchClassification(0, ACMatchClassification::NONE));
278  }
279
280  // Now, starting with |last_position|, check each character in
281  // |text_lowercase| to see if we have words starting with that character in
282  // |find_words|.  If so, check each of them to see if they match the portion
283  // of |text_lowercase| beginning with |last_position|.  Accept the first
284  // matching word found (which should be the longest possible match at this
285  // location, given the construction of |find_words|) and add a MATCH region to
286  // |match_class|, moving |last_position| to be after the matching word.  If we
287  // found no matching words, move to the next character and repeat.
288  while (last_position < text_lowercase.length()) {
289    std::pair<WordMap::const_iterator, WordMap::const_iterator> range(
290        find_words.equal_range(text_lowercase[last_position]));
291    size_t next_character = last_position + 1;
292    for (WordMap::const_iterator i(range.first); i != range.second; ++i) {
293      const string16& word = i->second;
294      size_t word_end = last_position + word.length();
295      if ((word_end <= text_lowercase.length()) &&
296          !text_lowercase.compare(last_position, word.length(), word)) {
297        // Collapse adjacent ranges into one.
298        if (match_class.back().offset == last_position)
299          match_class.pop_back();
300
301        AutocompleteMatch::AddLastClassificationIfNecessary(&match_class,
302            last_position, ACMatchClassification::MATCH);
303        if (word_end < text_lowercase.length()) {
304          match_class.push_back(
305              ACMatchClassification(word_end, ACMatchClassification::NONE));
306        }
307        last_position = word_end;
308        break;
309      }
310    }
311    last_position = std::max(last_position, next_character);
312  }
313
314  return AutocompleteMatch::MergeClassifications(original_class, match_class);
315}
316
317history::ShortcutsBackend::ShortcutMap::const_iterator
318    ShortcutsProvider::FindFirstMatch(const string16& keyword,
319                                      history::ShortcutsBackend* backend) {
320  DCHECK(backend);
321  history::ShortcutsBackend::ShortcutMap::const_iterator it =
322      backend->shortcuts_map().lower_bound(keyword);
323  // Lower bound not necessarily matches the keyword, check for item pointed by
324  // the lower bound iterator to at least start with keyword.
325  return ((it == backend->shortcuts_map().end()) ||
326    StartsWith(it->first, keyword, true)) ? it :
327    backend->shortcuts_map().end();
328}
329
330int ShortcutsProvider::CalculateScore(
331    const string16& terms,
332    const history::ShortcutsBackend::Shortcut& shortcut,
333    int max_relevance) {
334  DCHECK(!terms.empty());
335  DCHECK_LE(terms.length(), shortcut.text.length());
336
337  // The initial score is based on how much of the shortcut the user has typed.
338  // Using the square root of the typed fraction boosts the base score rapidly
339  // as characters are typed, compared with simply using the typed fraction
340  // directly. This makes sense since the first characters typed are much more
341  // important for determining how likely it is a user wants a particular
342  // shortcut than are the remaining continued characters.
343  double base_score = max_relevance *
344      sqrt(static_cast<double>(terms.length()) / shortcut.text.length());
345
346  // Then we decay this by half each week.
347  const double kLn2 = 0.6931471805599453;
348  base::TimeDelta time_passed = base::Time::Now() - shortcut.last_access_time;
349  // Clamp to 0 in case time jumps backwards (e.g. due to DST).
350  double decay_exponent = std::max(0.0, kLn2 * static_cast<double>(
351      time_passed.InMicroseconds()) / base::Time::kMicrosecondsPerWeek);
352
353  // We modulate the decay factor based on how many times the shortcut has been
354  // used. Newly created shortcuts decay at full speed; otherwise, decaying by
355  // half takes |n| times as much time, where n increases by
356  // (1.0 / each 5 additional hits), up to a maximum of 5x as long.
357  const double kMaxDecaySpeedDivisor = 5.0;
358  const double kNumUsesPerDecaySpeedDivisorIncrement = 5.0;
359  double decay_divisor = std::min(kMaxDecaySpeedDivisor,
360      (shortcut.number_of_hits + kNumUsesPerDecaySpeedDivisorIncrement - 1) /
361      kNumUsesPerDecaySpeedDivisorIncrement);
362
363  return static_cast<int>((base_score / exp(decay_exponent / decay_divisor)) +
364      0.5);
365}
366