| 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 #ifndef CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ | 5 #ifndef CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ |
| 6 #define CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ | 6 #define CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ |
| 7 #pragma once | 7 #pragma once |
| 8 | 8 |
| 9 #include <functional> | 9 #include <functional> |
| 10 #include <map> | 10 #include <map> |
| 11 #include <set> | 11 #include <set> |
| 12 #include <string> | 12 #include <string> |
| 13 #include <vector> | 13 #include <vector> |
| 14 | 14 |
| 15 #include "base/basictypes.h" | 15 #include "base/basictypes.h" |
| 16 #include "base/file_path.h" | 16 #include "base/file_path.h" |
| 17 #include "base/gtest_prod_util.h" | 17 #include "base/gtest_prod_util.h" |
| 18 #include "base/memory/linked_ptr.h" | 18 #include "base/memory/linked_ptr.h" |
| 19 #include "base/memory/scoped_ptr.h" | 19 #include "base/memory/scoped_ptr.h" |
| 20 #include "base/string16.h" | 20 #include "base/string16.h" |
| 21 #include "chrome/browser/autocomplete/autocomplete_match.h" | 21 #include "chrome/browser/autocomplete/autocomplete_match.h" |
| 22 #include "chrome/browser/autocomplete/history_provider_util.h" | 22 #include "chrome/browser/autocomplete/history_provider_util.h" |
| 23 #include "chrome/browser/history/history_types.h" | 23 #include "chrome/browser/history/history_types.h" |
| 24 #include "chrome/browser/history/in_memory_url_index_types.h" | 24 #include "chrome/browser/history/in_memory_url_index_types.h" |
| 25 #include "chrome/browser/history/in_memory_url_index_cache.pb.h" | 25 #include "chrome/browser/history/in_memory_url_index_cache.pb.h" |
| 26 #include "content/public/browser/notification_observer.h" | |
| 27 #include "content/public/browser/notification_registrar.h" | |
| 28 #include "sql/connection.h" | 26 #include "sql/connection.h" |
| 29 #include "testing/gtest/include/gtest/gtest_prod.h" | 27 #include "testing/gtest/include/gtest/gtest_prod.h" |
| 30 | 28 |
| 31 class Profile; | 29 class Profile; |
| 32 | 30 |
| 31 namespace base { |
| 32 class Time; |
| 33 } |
| 34 |
| 33 namespace in_memory_url_index { | 35 namespace in_memory_url_index { |
| 34 class InMemoryURLIndexCacheItem; | 36 class InMemoryURLIndexCacheItem; |
| 35 } | 37 } |
| 36 | 38 |
| 37 namespace history { | 39 namespace history { |
| 38 | 40 |
| 39 namespace imui = in_memory_url_index; | 41 namespace imui = in_memory_url_index; |
| 40 | 42 |
| 41 class URLDatabase; | 43 class URLDatabase; |
| 42 struct URLsDeletedDetails; | |
| 43 struct URLsModifiedDetails; | |
| 44 struct URLVisitedDetails; | |
| 45 | 44 |
| 46 // The URL history source. | 45 // The URL history source. |
| 47 // Holds portions of the URL database in memory in an indexed form. Used to | 46 // Holds portions of the URL database in memory in an indexed form. Used to |
| 48 // quickly look up matching URLs for a given query string. Used by | 47 // quickly look up matching URLs for a given query string. Used by |
| 49 // the HistoryURLProvider for inline autocomplete and to provide URL | 48 // the HistoryURLProvider for inline autocomplete and to provide URL |
| 50 // matches to the omnibox. | 49 // matches to the omnibox. |
| 51 // | 50 // |
| 52 // Note about multi-byte codepoints and the data structures in the | 51 // Note about multi-byte codepoints and the data structures in the |
| 53 // InMemoryURLIndex class: One will quickly notice that no effort is made to | 52 // InMemoryURLIndex class: One will quickly notice that no effort is made to |
| 54 // insure that multi-byte character boundaries are detected when indexing the | 53 // insure that multi-byte character boundaries are detected when indexing the |
| 55 // words and characters in the URL history database except when converting | 54 // words and characters in the URL history database except when converting |
| 56 // URL strings to lowercase. Multi-byte-edness makes no difference when | 55 // URL strings to lowercase. Multi-byte-edness makes no difference when |
| 57 // indexing or when searching the index as the final filtering of results | 56 // indexing or when searching the index as the final filtering of results |
| 58 // is dependent on the comparison of a string of bytes, not individual | 57 // is dependent on the comparison of a string of bytes, not individual |
| 59 // characters. While the lookup of those bytes during a search in the | 58 // characters. While the lookup of those bytes during a search in the |
| 60 // |char_word_map_| could serve up words in which the individual char16 | 59 // |char_word_map_| could serve up words in which the individual char16 |
| 61 // occurs as a portion of a composite character the next filtering step | 60 // occurs as a portion of a composite character the next filtering step |
| 62 // will eliminate such words except in the case where a single character | 61 // will eliminate such words except in the case where a single character |
| 63 // is being searched on and which character occurs as the second char16 of a | 62 // is being searched on and which character occurs as the second char16 of a |
| 64 // multi-char16 instance. | 63 // multi-char16 instance. |
| 65 class InMemoryURLIndex : public content::NotificationObserver { | 64 class InMemoryURLIndex { |
| 66 public: | 65 public: |
| 67 // |history_dir| is a path to the directory containing the history database | 66 // |history_dir| is a path to the directory containing the history database |
| 68 // within the profile wherein the cache and transaction journals will be | 67 // within the profile wherein the cache and transaction journals will be |
| 69 // stored. | 68 // stored. |
| 70 explicit InMemoryURLIndex(Profile* profile, | 69 explicit InMemoryURLIndex(const FilePath& history_dir); |
| 71 const FilePath& history_dir); | |
| 72 virtual ~InMemoryURLIndex(); | 70 virtual ~InMemoryURLIndex(); |
| 73 | 71 |
| 74 // Opens and indexes the URL history database. | 72 // Opens and indexes the URL history database. |
| 75 // |languages| gives a list of language encodings with which the history | 73 // |languages| gives a list of language encodings with which the history |
| 76 // URLs and omnibox searches are interpreted, i.e. when each is broken | 74 // URLs and omnibox searches are interpreted, i.e. when each is broken |
| 77 // down into words and each word is broken down into characters. | 75 // down into words and each word is broken down into characters. |
| 78 bool Init(URLDatabase* history_db, const std::string& languages); | 76 bool Init(URLDatabase* history_db, const std::string& languages); |
| 79 | 77 |
| 80 // Reloads the history index. Attempts to reload from the cache unless | 78 // Reloads the history index. Attempts to reload from the cache unless |
| 81 // |clear_cache| is true. If the cache is unavailable then reload the | 79 // |clear_cache| is true. If the cache is unavailable then reload the |
| (...skipping 20 matching lines...) Expand all Loading... |
| 102 // from |terms| may contain punctuation but should not contain spaces. | 100 // from |terms| may contain punctuation but should not contain spaces. |
| 103 // A search request which results in more than |kItemsToScoreLimit| total | 101 // A search request which results in more than |kItemsToScoreLimit| total |
| 104 // candidate items returns no matches (though the results set will be | 102 // candidate items returns no matches (though the results set will be |
| 105 // retained and used for subsequent calls to this function) as the scoring | 103 // retained and used for subsequent calls to this function) as the scoring |
| 106 // of such a large number of candidates may cause perceptible typing response | 104 // of such a large number of candidates may cause perceptible typing response |
| 107 // delays in the omnibox. This is likely to occur for short omnibox terms | 105 // delays in the omnibox. This is likely to occur for short omnibox terms |
| 108 // such as 'h' and 'w' which will be found in nearly all history candidates. | 106 // such as 'h' and 'w' which will be found in nearly all history candidates. |
| 109 ScoredHistoryMatches HistoryItemsForTerms(const String16Vector& terms); | 107 ScoredHistoryMatches HistoryItemsForTerms(const String16Vector& terms); |
| 110 | 108 |
| 111 // Updates or adds an history item to the index if it meets the minimum | 109 // Updates or adds an history item to the index if it meets the minimum |
| 112 // selection criteria. | 110 // 'quick' criteria. |
| 113 void UpdateURL(const URLRow& row); | 111 void UpdateURL(URLID row_id, const URLRow& row); |
| 114 | 112 |
| 115 // Deletes indexing data for an history item. The item may not have actually | 113 // Deletes indexing data for an history item. The item may not have actually |
| 116 // been indexed (which is the case if it did not previously meet minimum | 114 // been indexed (which is the case if it did not previously meet minimum |
| 117 // 'quick' criteria). | 115 // 'quick' criteria). |
| 118 void DeleteURL(const URLRow& row); | 116 void DeleteURL(URLID row_id); |
| 119 | |
| 120 // Notification callback. | |
| 121 virtual void Observe(int type, | |
| 122 const content::NotificationSource& source, | |
| 123 const content::NotificationDetails& details); | |
| 124 | 117 |
| 125 private: | 118 private: |
| 126 friend class AddHistoryMatch; | 119 friend class AddHistoryMatch; |
| 127 friend class InMemoryURLIndexTest; | 120 friend class InMemoryURLIndexTest; |
| 128 FRIEND_TEST_ALL_PREFIXES(LimitedInMemoryURLIndexTest, Initialization); | 121 FRIEND_TEST_ALL_PREFIXES(LimitedInMemoryURLIndexTest, Initialization); |
| 129 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheFilePath); | 122 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheFilePath); |
| 130 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheSaveRestore); | 123 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheSaveRestore); |
| 131 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Char16Utilities); | 124 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Char16Utilities); |
| 132 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, NonUniqueTermCharacterSets); | 125 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, NonUniqueTermCharacterSets); |
| 133 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Scoring); | 126 FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Scoring); |
| (...skipping 119 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 253 | 246 |
| 254 // Calculates a component score based on position, ordering and total | 247 // Calculates a component score based on position, ordering and total |
| 255 // substring match size using metrics recorded in |matches|. |max_length| | 248 // substring match size using metrics recorded in |matches|. |max_length| |
| 256 // is the length of the string against which the terms are being searched. | 249 // is the length of the string against which the terms are being searched. |
| 257 static int ScoreComponentForMatches(const TermMatches& matches, | 250 static int ScoreComponentForMatches(const TermMatches& matches, |
| 258 size_t max_length); | 251 size_t max_length); |
| 259 | 252 |
| 260 // Determines if |gurl| has a whitelisted scheme and returns true if so. | 253 // Determines if |gurl| has a whitelisted scheme and returns true if so. |
| 261 bool URLSchemeIsWhitelisted(const GURL& gurl) const; | 254 bool URLSchemeIsWhitelisted(const GURL& gurl) const; |
| 262 | 255 |
| 263 // Notification handlers. | |
| 264 void OnURLVisited(const URLVisitedDetails* details); | |
| 265 void OnURLsModified(const URLsModifiedDetails* details); | |
| 266 void OnURLsDeleted(const URLsDeletedDetails* details); | |
| 267 | |
| 268 // Utility functions supporting RestoreFromCache and SaveToCache. | 256 // Utility functions supporting RestoreFromCache and SaveToCache. |
| 269 | 257 |
| 270 // Construct a file path for the cache file within the same directory where | 258 // Construct a file path for the cache file within the same directory where |
| 271 // the history database is kept and saves that path to |file_path|. Returns | 259 // the history database is kept and saves that path to |file_path|. Returns |
| 272 // true if |file_path| can be successfully constructed. (This function | 260 // true if |file_path| can be successfully constructed. (This function |
| 273 // provided as a hook for unit testing.) | 261 // provided as a hook for unit testing.) |
| 274 bool GetCacheFilePath(FilePath* file_path); | 262 bool GetCacheFilePath(FilePath* file_path); |
| 275 | 263 |
| 276 // Encode a data structure into the protobuf |cache|. | 264 // Encode a data structure into the protobuf |cache|. |
| 277 void SavePrivateData(imui::InMemoryURLIndexCacheItem* cache) const; | 265 void SavePrivateData(imui::InMemoryURLIndexCacheItem* cache) const; |
| 278 void SaveWordList(imui::InMemoryURLIndexCacheItem* cache) const; | 266 void SaveWordList(imui::InMemoryURLIndexCacheItem* cache) const; |
| 279 void SaveWordMap(imui::InMemoryURLIndexCacheItem* cache) const; | 267 void SaveWordMap(imui::InMemoryURLIndexCacheItem* cache) const; |
| 280 void SaveCharWordMap(imui::InMemoryURLIndexCacheItem* cache) const; | 268 void SaveCharWordMap(imui::InMemoryURLIndexCacheItem* cache) const; |
| 281 void SaveWordIDHistoryMap(imui::InMemoryURLIndexCacheItem* cache) const; | 269 void SaveWordIDHistoryMap(imui::InMemoryURLIndexCacheItem* cache) const; |
| 282 void SaveHistoryInfoMap(imui::InMemoryURLIndexCacheItem* cache) const; | 270 void SaveHistoryInfoMap(imui::InMemoryURLIndexCacheItem* cache) const; |
| 283 | 271 |
| 284 // Decode a data structure from the protobuf |cache|. Return false if there | 272 // Decode a data structure from the protobuf |cache|. Return false if there |
| 285 // is any kind of failure. | 273 // is any kind of failure. |
| 286 bool RestorePrivateData(const imui::InMemoryURLIndexCacheItem& cache); | 274 bool RestorePrivateData(const imui::InMemoryURLIndexCacheItem& cache); |
| 287 bool RestoreWordList(const imui::InMemoryURLIndexCacheItem& cache); | 275 bool RestoreWordList(const imui::InMemoryURLIndexCacheItem& cache); |
| 288 bool RestoreWordMap(const imui::InMemoryURLIndexCacheItem& cache); | 276 bool RestoreWordMap(const imui::InMemoryURLIndexCacheItem& cache); |
| 289 bool RestoreCharWordMap(const imui::InMemoryURLIndexCacheItem& cache); | 277 bool RestoreCharWordMap(const imui::InMemoryURLIndexCacheItem& cache); |
| 290 bool RestoreWordIDHistoryMap(const imui::InMemoryURLIndexCacheItem& cache); | 278 bool RestoreWordIDHistoryMap(const imui::InMemoryURLIndexCacheItem& cache); |
| 291 bool RestoreHistoryInfoMap(const imui::InMemoryURLIndexCacheItem& cache); | 279 bool RestoreHistoryInfoMap(const imui::InMemoryURLIndexCacheItem& cache); |
| 292 | 280 |
| 293 content::NotificationRegistrar registrar_; | |
| 294 | |
| 295 // Directory where cache file resides. This is, except when unit testing, | 281 // Directory where cache file resides. This is, except when unit testing, |
| 296 // the same directory in which the profile's history database is found. It | 282 // the same directory in which the profile's history database is found. It |
| 297 // should never be empty. | 283 // should never be empty. |
| 298 FilePath history_dir_; | 284 FilePath history_dir_; |
| 299 | 285 |
| 300 // The timestamp of when the cache was last saved. This is used to validate | 286 // The timestamp of when the cache was last saved. This is used to validate |
| 301 // the transaction journal's applicability to the cache. The timestamp is | 287 // the transaction journal's applicability to the cache. The timestamp is |
| 302 // initialized to the NULL time, indicating that the cache was not used with | 288 // initialized to the NULL time, indicating that the cache was not used with |
| 303 // the InMemoryURLIndex was last populated. | 289 // the InMemoryURLIndex was last populated. |
| 304 base::Time last_saved_; | 290 base::Time last_saved_; |
| (...skipping 16 matching lines...) Expand all Loading... |
| 321 // TODO(mrossetti): Eliminate once the transition to SQLite has been done. | 307 // TODO(mrossetti): Eliminate once the transition to SQLite has been done. |
| 322 // http://crbug.com/83659 | 308 // http://crbug.com/83659 |
| 323 bool cached_at_shutdown_; | 309 bool cached_at_shutdown_; |
| 324 | 310 |
| 325 DISALLOW_COPY_AND_ASSIGN(InMemoryURLIndex); | 311 DISALLOW_COPY_AND_ASSIGN(InMemoryURLIndex); |
| 326 }; | 312 }; |
| 327 | 313 |
| 328 } // namespace history | 314 } // namespace history |
| 329 | 315 |
| 330 #endif // CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ | 316 #endif // CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_ |
| OLD | NEW |