Move InMemoryURLIndex Caching Operations to FILE Thread

chrome/browser/history/url_index_private_data.h

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #ifndef CHROME_BROWSER_HISTORY_URL_INDEX_PRIVATE_DATA_H_
 #define CHROME_BROWSER_HISTORY_URL_INDEX_PRIVATE_DATA_H_
 #pragma once
 
 #include "base/file_path.h"
 #include "base/gtest_prod_util.h"
 #include "chrome/browser/history/in_memory_url_index_types.h"
 #include "chrome/browser/history/in_memory_url_index_cache.pb.h"
 
+class InMemoryURLIndexTest;
+
 namespace in_memory_url_index {
 class InMemoryURLIndexCacheItem;
 }
 
 namespace history {
 
 namespace imui = in_memory_url_index;
 
 // A structure describing the InMemoryURLIndex's internal data and providing for
 // restoring, rebuilding and updating that internal data.
 class URLIndexPrivateData {
  public:
   URLIndexPrivateData();
   ~URLIndexPrivateData();
 
  private:
+  friend class AddHistoryMatch;
+  friend class HistoryQuickProviderTest;
   friend class InMemoryURLIndex;
-  friend class AddHistoryMatch;
   friend class InMemoryURLIndexTest;
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheSaveRestore);
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, HugeResultSet);
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Scoring);
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, TitleSearch);
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, TypedCharacterCaching);
   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, WhitelistedURLs);
   FRIEND_TEST_ALL_PREFIXES(LimitedInMemoryURLIndexTest, Initialization);
 
+#if 1
+void DUMP() const;
+#endif

[Peter Kasting, 2012/01/14 00:12:49]

This seems like it should have disappeared before uploading...

[mrossetti, 2012/03/03 05:05:56]

Done.

+
   // Support caching of term results so that we can optimize searches which
   // build upon a previous search. Each entry in this map represents one
   // search term from the most recent search. For example, if the user had
   // typed "google blog trans" and then typed an additional 'l' (at the end,
   // of course) then there would be four items in the cache: 'blog', 'google',
   // 'trans', and 'transl'. All would be marked as being in use except for the
   // 'trans' item; its cached data would have been used when optimizing the
   // construction of the search results candidates for 'transl' but then would
   // no longer needed.
   //
@@ skipping 66 matching lines @@
   // descending score. The full results set (i.e. beyond the
   // |kItemsToScoreLimit| limit) will be retained and used for subsequent calls
   // to this function.
   ScoredHistoryMatches HistoryItemsForTerms(const string16& term_string);
 
   // Sets the |languages| to a list of language encodings with which the history
   // URLs and omnibox searches are interpreted, i.e. how each is broken
   // down into words and each word is broken down into characters.
   void set_languages(const std::string& languages) { languages_ = languages; }
 
-  // Restores the index's private data from the cache file stored in the
-  // profile directory and returns true if successful.
-  bool RestoreFromFile(const FilePath& file_path);
+  // Constructs a new object by restoring its contents from the file at |path|.
+  // Returns the new URLIndexPrivateData which on success will contain the
+  // restored data but upon failure will be empty.
+  static URLIndexPrivateData* RestoreFromFile(const FilePath& path);
+
+  // Constructs a new object by rebuilding its contents from the history
+  // database in |db|. Returns the new URLIndexPrivateData which on success
+  // will contain the rebuilt data but upon failure will be empty.
+  static URLIndexPrivateData* RebuildFromHistory(URLDatabase* history_db);
 
   // Caches the index private data and writes the cache file to the profile
   // directory.
   bool SaveToFile(const FilePath& file_path);
 
-  // Reloads the history index from |history_db|.
-  bool ReloadFromHistory(URLDatabase* history_db);
-
   // Initializes all index data members in preparation for restoring the index
   // from the cache or a complete rebuild from the history database.
   void Clear();
 
   // Adds |word_id| to |history_id|'s entry in the history/word map,
   // creating a new entry if one does not already exist.
   void AddToHistoryIDWordMap(HistoryID history_id, WordID word_id);
 
   // Given a set of Char16s, finds words containing those characters.
   WordIDSet WordIDSetForTermChars(const Char16Set& term_chars);
 
   // Initializes the whitelist of URL schemes.
   static void InitializeSchemeWhitelist(std::set<std::string>* whitelist);
 
   // URL History indexing support functions.
 
   // Indexes one URL history item.
   void IndexRow(const URLRow& row);
 
-  // Updates or adds an history item to the index if it meets the minimum
-  // 'quick' criteria.
-  void UpdateURL(URLID row_id, const URLRow& row);
+  // Updates or adds the history item in |row| to the index if it meets the
+  // minimum 'quick' criteria.
+  void UpdateURL(const URLRow& row);
 
-  // Deletes indexing data for an history item. The item may not have actually
-  // been indexed (which is the case if it did not previously meet minimum
-  // 'quick' criteria).
-  void DeleteURL(URLID row_id);
+  // Deletes indexing data for the history item with the URL given in |url|.
+  // The item may not have actually been indexed, which is the case if it did
+  // not previously meet minimum 'quick' criteria.
+  void DeleteURL(const GURL& url);
 
   // Parses and indexes the words in the URL and page title of |row|.
   void AddRowWordsToIndex(const URLRow& row);
 
   // Removes |row| and all associated words and characters from the index.
   void RemoveRowFromIndex(const URLRow& row);
 
   // Removes all words and characters associated with |row| from the index.
   void RemoveRowWordsFromIndex(const URLRow& row);
 
@@ skipping 33 matching lines @@
   static ScoredHistoryMatch ScoredMatchForURL(
       const URLRow& row,
       const String16Vector& terms_vector);
 
   // Calculates a component score based on position, ordering and total
   // substring match size using metrics recorded in |matches|. |max_length|
   // is the length of the string against which the terms are being searched.
   static int ScoreComponentForMatches(const TermMatches& matches,
                                       size_t max_length);
 
-  // Determines if |gurl| has a whitelisted scheme and returns true if so.
-  bool URLSchemeIsWhitelisted(const GURL& gurl) const;
-
   // Encode a data structure into the protobuf |cache|.
   void SavePrivateData(imui::InMemoryURLIndexCacheItem* cache) const;
   void SaveWordList(imui::InMemoryURLIndexCacheItem* cache) const;
   void SaveWordMap(imui::InMemoryURLIndexCacheItem* cache) const;
   void SaveCharWordMap(imui::InMemoryURLIndexCacheItem* cache) const;
   void SaveWordIDHistoryMap(imui::InMemoryURLIndexCacheItem* cache) const;
   void SaveHistoryInfoMap(imui::InMemoryURLIndexCacheItem* cache) const;
 
   // Decode a data structure from the protobuf |cache|. Return false if there
   // is any kind of failure.
   bool RestorePrivateData(const imui::InMemoryURLIndexCacheItem& cache);
   bool RestoreWordList(const imui::InMemoryURLIndexCacheItem& cache);
   bool RestoreWordMap(const imui::InMemoryURLIndexCacheItem& cache);
   bool RestoreCharWordMap(const imui::InMemoryURLIndexCacheItem& cache);
   bool RestoreWordIDHistoryMap(const imui::InMemoryURLIndexCacheItem& cache);
   bool RestoreHistoryInfoMap(const imui::InMemoryURLIndexCacheItem& cache);
 
+  // Determines if |gurl| has a whitelisted scheme and returns true if so.
+  bool URLSchemeIsWhitelisted(const GURL& gurl) const;
+
   // Cache of search terms.
   SearchTermCacheMap search_term_cache_;
 
   // Languages used during the word-breaking process during indexing.
   std::string languages_;
 
   // Only URLs with a whitelisted scheme are indexed.
   std::set<std::string> scheme_whitelist_;
 
   // Start of data members that are cached -------------------------------------
@@ skipping 39 matching lines @@
   // Used for unit testing only. Records the number of candidate history items
   // at three stages in the index searching process.
   size_t pre_filter_item_count_;    // After word index is queried.
   size_t post_filter_item_count_;   // After trimming large result set.
   size_t post_scoring_item_count_;  // After performing final filter/scoring.
 };
 
 }  // namespace history
 
 #endif  // CHROME_BROWSER_HISTORY_URL_INDEX_PRIVATE_DATA_H_