OLD | NEW |
1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "chrome/browser/history/in_memory_url_index.h" | 5 #include "chrome/browser/history/in_memory_url_index.h" |
6 | 6 |
7 #include <algorithm> | 7 #include <algorithm> |
8 #include <functional> | 8 #include <functional> |
9 #include <iterator> | 9 #include <iterator> |
10 #include <limits> | 10 #include <limits> |
11 #include <numeric> | 11 #include <numeric> |
12 | 12 |
13 #include "base/file_util.h" | 13 #include "base/file_util.h" |
14 #include "base/i18n/case_conversion.h" | 14 #include "base/i18n/case_conversion.h" |
15 #include "base/metrics/histogram.h" | 15 #include "base/metrics/histogram.h" |
| 16 #include "base/string_util.h" |
16 #include "base/threading/thread_restrictions.h" | 17 #include "base/threading/thread_restrictions.h" |
| 18 #include "base/time.h" |
17 #include "base/utf_string_conversions.h" | 19 #include "base/utf_string_conversions.h" |
18 #include "chrome/browser/autocomplete/autocomplete.h" | 20 #include "chrome/browser/autocomplete/autocomplete.h" |
19 #include "chrome/browser/history/history_notifications.h" | 21 #include "chrome/browser/autocomplete/history_provider_util.h" |
20 #include "chrome/browser/history/url_database.h" | 22 #include "chrome/browser/history/url_database.h" |
21 #include "chrome/common/chrome_notification_types.h" | 23 #include "chrome/browser/profiles/profile.h" |
22 #include "chrome/common/url_constants.h" | 24 #include "chrome/common/url_constants.h" |
23 #include "content/public/browser/notification_service.h" | 25 #include "googleurl/src/url_parse.h" |
| 26 #include "googleurl/src/url_util.h" |
| 27 #include "net/base/escape.h" |
24 #include "net/base/net_util.h" | 28 #include "net/base/net_util.h" |
| 29 #include "third_party/protobuf/src/google/protobuf/repeated_field.h" |
25 #include "ui/base/l10n/l10n_util.h" | 30 #include "ui/base/l10n/l10n_util.h" |
26 | 31 |
27 using google::protobuf::RepeatedField; | 32 using google::protobuf::RepeatedField; |
28 using google::protobuf::RepeatedPtrField; | 33 using google::protobuf::RepeatedPtrField; |
29 using in_memory_url_index::InMemoryURLIndexCacheItem; | 34 using in_memory_url_index::InMemoryURLIndexCacheItem; |
30 | 35 |
31 namespace history { | 36 namespace history { |
32 | 37 |
33 typedef imui::InMemoryURLIndexCacheItem_WordListItem WordListItem; | 38 typedef imui::InMemoryURLIndexCacheItem_WordListItem WordListItem; |
34 typedef imui::InMemoryURLIndexCacheItem_WordMapItem_WordMapEntry WordMapEntry; | 39 typedef imui::InMemoryURLIndexCacheItem_WordMapItem_WordMapEntry WordMapEntry; |
(...skipping 65 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
100 return 0; | 105 return 0; |
101 int score = kScoreRank[i]; | 106 int score = kScoreRank[i]; |
102 if (i > 0) { | 107 if (i > 0) { |
103 score += (value - value_ranks[i]) * | 108 score += (value - value_ranks[i]) * |
104 (kScoreRank[i - 1] - kScoreRank[i]) / | 109 (kScoreRank[i - 1] - kScoreRank[i]) / |
105 (value_ranks[i - 1] - value_ranks[i]); | 110 (value_ranks[i - 1] - value_ranks[i]); |
106 } | 111 } |
107 return score; | 112 return score; |
108 } | 113 } |
109 | 114 |
110 InMemoryURLIndex::InMemoryURLIndex(Profile* profile, | 115 InMemoryURLIndex::InMemoryURLIndex(const FilePath& history_dir) |
111 const FilePath& history_dir) | |
112 : history_dir_(history_dir), | 116 : history_dir_(history_dir), |
113 private_data_(new URLIndexPrivateData), | 117 private_data_(new URLIndexPrivateData), |
114 cached_at_shutdown_(false) { | 118 cached_at_shutdown_(false) { |
115 InMemoryURLIndex::InitializeSchemeWhitelist(&scheme_whitelist_); | 119 InMemoryURLIndex::InitializeSchemeWhitelist(&scheme_whitelist_); |
116 if (profile) { | |
117 content::Source<Profile> source(profile); | |
118 registrar_.Add(this, chrome::NOTIFICATION_HISTORY_URL_VISITED, source); | |
119 registrar_.Add(this, chrome::NOTIFICATION_HISTORY_TYPED_URLS_MODIFIED, | |
120 source); | |
121 registrar_.Add(this, chrome::NOTIFICATION_HISTORY_URLS_DELETED, source); | |
122 } | |
123 } | 120 } |
124 | 121 |
125 // Called only by unit tests. | 122 // Called only by unit tests. |
126 InMemoryURLIndex::InMemoryURLIndex() | 123 InMemoryURLIndex::InMemoryURLIndex() |
127 : private_data_(new URLIndexPrivateData), | 124 : private_data_(new URLIndexPrivateData), |
128 cached_at_shutdown_(false) { | 125 cached_at_shutdown_(false) { |
129 InMemoryURLIndex::InitializeSchemeWhitelist(&scheme_whitelist_); | 126 InMemoryURLIndex::InitializeSchemeWhitelist(&scheme_whitelist_); |
130 } | 127 } |
131 | 128 |
132 InMemoryURLIndex::~InMemoryURLIndex() { | 129 InMemoryURLIndex::~InMemoryURLIndex() { |
(...skipping 18 matching lines...) Expand all Loading... |
151 // Indexing | 148 // Indexing |
152 | 149 |
153 bool InMemoryURLIndex::Init(URLDatabase* history_db, | 150 bool InMemoryURLIndex::Init(URLDatabase* history_db, |
154 const std::string& languages) { | 151 const std::string& languages) { |
155 // TODO(mrossetti): Register for profile/language change notifications. | 152 // TODO(mrossetti): Register for profile/language change notifications. |
156 languages_ = languages; | 153 languages_ = languages; |
157 return ReloadFromHistory(history_db, false); | 154 return ReloadFromHistory(history_db, false); |
158 } | 155 } |
159 | 156 |
160 void InMemoryURLIndex::ShutDown() { | 157 void InMemoryURLIndex::ShutDown() { |
161 registrar_.RemoveAll(); | 158 // Write our cache. |
162 SaveToCacheFile(); | 159 SaveToCacheFile(); |
163 cached_at_shutdown_ = true; | 160 cached_at_shutdown_ = true; |
164 } | 161 } |
165 | 162 |
166 void InMemoryURLIndex::Observe(int type, | |
167 const content::NotificationSource& source, | |
168 const content::NotificationDetails& details) { | |
169 switch (type) { | |
170 case chrome::NOTIFICATION_HISTORY_URL_VISITED: | |
171 OnURLVisited(content::Details<URLVisitedDetails>(details).ptr()); | |
172 break; | |
173 case chrome::NOTIFICATION_HISTORY_TYPED_URLS_MODIFIED: | |
174 OnURLsModified( | |
175 content::Details<history::URLsModifiedDetails>(details).ptr()); | |
176 break; | |
177 case chrome::NOTIFICATION_HISTORY_URLS_DELETED: | |
178 OnURLsDeleted( | |
179 content::Details<history::URLsDeletedDetails>(details).ptr()); | |
180 break; | |
181 default: | |
182 // For simplicity, the unit tests send us all notifications, even when | |
183 // we haven't registered for them, so don't assert here. | |
184 break; | |
185 } | |
186 } | |
187 | |
188 void InMemoryURLIndex::OnURLVisited(const URLVisitedDetails* details) { | |
189 UpdateURL(details->row); | |
190 } | |
191 | |
192 void InMemoryURLIndex::OnURLsModified(const URLsModifiedDetails* details) { | |
193 for (std::vector<history::URLRow>::const_iterator row = | |
194 details->changed_urls.begin(); | |
195 row != details->changed_urls.end(); ++row) | |
196 UpdateURL(*row); | |
197 } | |
198 | |
199 void InMemoryURLIndex::OnURLsDeleted(const URLsDeletedDetails* details) { | |
200 if (details->all_history) { | |
201 ClearPrivateData(); | |
202 } else { | |
203 for (std::vector<URLRow>::const_iterator row = details->rows.begin(); | |
204 row != details->rows.end(); ++row) | |
205 DeleteURL(*row); | |
206 } | |
207 } | |
208 | |
209 void InMemoryURLIndex::IndexRow(const URLRow& row) { | 163 void InMemoryURLIndex::IndexRow(const URLRow& row) { |
210 const GURL& gurl(row.url()); | 164 const GURL& gurl(row.url()); |
211 | 165 |
212 // Index only URLs with a whitelisted scheme. | 166 // Index only URLs with a whitelisted scheme. |
213 if (!InMemoryURLIndex::URLSchemeIsWhitelisted(gurl)) | 167 if (!InMemoryURLIndex::URLSchemeIsWhitelisted(gurl)) |
214 return; | 168 return; |
215 | 169 |
216 URLID row_id = row.id(); | 170 URLID row_id = row.id(); |
217 // Strip out username and password before saving and indexing. | 171 // Strip out username and password before saving and indexing. |
218 string16 url(net::FormatUrl(gurl, languages_, | 172 string16 url(net::FormatUrl(gurl, languages_, |
(...skipping 169 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
388 int size = data.size(); | 342 int size = data.size(); |
389 if (file_util::WriteFile(file_path, data.c_str(), size) != size) { | 343 if (file_util::WriteFile(file_path, data.c_str(), size) != size) { |
390 LOG(WARNING) << "Failed to write " << file_path.value(); | 344 LOG(WARNING) << "Failed to write " << file_path.value(); |
391 return false; | 345 return false; |
392 } | 346 } |
393 UMA_HISTOGRAM_TIMES("History.InMemoryURLIndexSaveCacheTime", | 347 UMA_HISTOGRAM_TIMES("History.InMemoryURLIndexSaveCacheTime", |
394 base::TimeTicks::Now() - beginning_time); | 348 base::TimeTicks::Now() - beginning_time); |
395 return true; | 349 return true; |
396 } | 350 } |
397 | 351 |
398 void InMemoryURLIndex::UpdateURL(const URLRow& row) { | 352 void InMemoryURLIndex::UpdateURL(URLID row_id, const URLRow& row) { |
399 // The row may or may not already be in our index. If it is not already | 353 // The row may or may not already be in our index. If it is not already |
400 // indexed and it qualifies then it gets indexed. If it is already | 354 // indexed and it qualifies then it gets indexed. If it is already |
401 // indexed and still qualifies then it gets updated, otherwise it | 355 // indexed and still qualifies then it gets updated, otherwise it |
402 // is deleted from the index. | 356 // is deleted from the index. |
403 HistoryInfoMap::iterator row_pos = | 357 HistoryInfoMap::iterator row_pos = |
404 private_data_->history_info_map_.find(row.id()); | 358 private_data_->history_info_map_.find(row_id); |
405 if (row_pos == private_data_->history_info_map_.end()) { | 359 if (row_pos == private_data_->history_info_map_.end()) { |
406 // This new row should be indexed if it qualifies. | 360 // This new row should be indexed if it qualifies. |
407 if (RowQualifiesAsSignificant(row, base::Time())) | 361 URLRow new_row(row); |
408 IndexRow(row); | 362 new_row.set_id(row_id); |
| 363 if (RowQualifiesAsSignificant(new_row, base::Time())) |
| 364 IndexRow(new_row); |
409 } else if (RowQualifiesAsSignificant(row, base::Time())) { | 365 } else if (RowQualifiesAsSignificant(row, base::Time())) { |
410 // This indexed row still qualifies and will be re-indexed. | 366 // This indexed row still qualifies and will be re-indexed. |
411 // The url won't have changed but the title, visit count, etc. | 367 // The url won't have changed but the title, visit count, etc. |
412 // might have changed. | 368 // might have changed. |
413 URLRow& old_row = row_pos->second; | 369 URLRow& updated_row = row_pos->second; |
414 old_row.set_visit_count(row.visit_count()); | 370 updated_row.set_visit_count(row.visit_count()); |
415 old_row.set_typed_count(row.typed_count()); | 371 updated_row.set_typed_count(row.typed_count()); |
416 old_row.set_last_visit(row.last_visit()); | 372 updated_row.set_last_visit(row.last_visit()); |
417 // While the URL is guaranteed to remain stable, the title may have changed. | 373 // While the URL is guaranteed to remain stable, the title may have changed. |
418 // If so, then we need to update the index with the changed words. | 374 // If so, then we need to update the index with the changed words. |
419 if (old_row.title() != row.title()) { | 375 if (updated_row.title() != row.title()) { |
420 // Clear all words associated with this row and re-index both the | 376 // Clear all words associated with this row and re-index both the |
421 // URL and title. | 377 // URL and title. |
422 RemoveRowWordsFromIndex(row); | 378 RemoveRowWordsFromIndex(updated_row); |
423 old_row.set_title(row.title()); | 379 updated_row.set_title(row.title()); |
424 AddRowWordsToIndex(old_row); | 380 AddRowWordsToIndex(updated_row); |
425 } | 381 } |
426 } else { | 382 } else { |
427 // This indexed row no longer qualifies and will be de-indexed by | 383 // This indexed row no longer qualifies and will be de-indexed by |
428 // clearing all words associated with this row. | 384 // clearing all words associated with this row. |
429 RemoveRowFromIndex(row); | 385 URLRow& removed_row = row_pos->second; |
| 386 RemoveRowFromIndex(removed_row); |
430 } | 387 } |
431 // This invalidates the cache. | 388 // This invalidates the cache. |
432 search_term_cache_.clear(); | 389 search_term_cache_.clear(); |
433 } | 390 } |
434 | 391 |
435 void InMemoryURLIndex::DeleteURL(const URLRow& row) { | 392 void InMemoryURLIndex::DeleteURL(URLID row_id) { |
436 RemoveRowFromIndex(row); | 393 // Note that this does not remove any reference to this row from the |
437 search_term_cache_.clear(); // Invalidate the word cache. | 394 // word_id_history_map_. That map will continue to contain (and return) |
| 395 // hits against this row until that map is rebuilt, but since the |
| 396 // history_info_map_ no longer references the row no erroneous results |
| 397 // will propagate to the user. |
| 398 private_data_->history_info_map_.erase(row_id); |
| 399 // This invalidates the word cache. |
| 400 search_term_cache_.clear(); |
438 } | 401 } |
439 | 402 |
440 // Searching | 403 // Searching |
441 | 404 |
442 ScoredHistoryMatches InMemoryURLIndex::HistoryItemsForTerms( | 405 ScoredHistoryMatches InMemoryURLIndex::HistoryItemsForTerms( |
443 const String16Vector& terms) { | 406 const String16Vector& terms) { |
444 ScoredHistoryMatches scored_items; | 407 ScoredHistoryMatches scored_items; |
445 | 408 |
446 // Do nothing if we have indexed no words (probably because we've not been | 409 // Do nothing if we have indexed no words (probably because we've not been |
447 // initialized yet). | 410 // initialized yet). |
(...skipping 658 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
1106 if (iter->has_title()) { | 1069 if (iter->has_title()) { |
1107 string16 title(UTF8ToUTF16(iter->title())); | 1070 string16 title(UTF8ToUTF16(iter->title())); |
1108 url_row.set_title(title); | 1071 url_row.set_title(title); |
1109 } | 1072 } |
1110 private_data_->history_info_map_[history_id] = url_row; | 1073 private_data_->history_info_map_[history_id] = url_row; |
1111 } | 1074 } |
1112 return true; | 1075 return true; |
1113 } | 1076 } |
1114 | 1077 |
1115 } // namespace history | 1078 } // namespace history |
OLD | NEW |