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 "base/memory/ref_counted.h"
 #include "chrome/browser/history/in_memory_url_index_types.h"
 #include "chrome/browser/history/in_memory_url_index_cache.pb.h"
+#include "content/public/browser/notification_details.h"
 
 class HistoryQuickProviderTest;
 
 namespace in_memory_url_index {
 class InMemoryURLIndexCacheItem;
 }
 
 namespace history {
 
 namespace imui = in_memory_url_index;
 
 class HistoryDatabase;
+class InMemoryURLIndex;
+class RefCountedBool;
 
 // Current version of the cache file.
 static const int kCurrentCacheFileVersion = 1;
 
 // A structure describing the InMemoryURLIndex's internal data and providing for
 // restoring, rebuilding and updating that internal data.
-class URLIndexPrivateData {
+class URLIndexPrivateData
+    : public base::RefCountedThreadSafe<URLIndexPrivateData> {
  public:
   URLIndexPrivateData();
+
+ private:
+  friend class base::RefCountedThreadSafe<URLIndexPrivateData>;
   ~URLIndexPrivateData();
 
- private:
   friend class AddHistoryMatch;
   friend class ::HistoryQuickProviderTest;
   friend class InMemoryURLIndex;
   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);
@@ skipping 75 matching lines @@
   // or page title. Scores are then calculated on no more than
   // |kItemsToScoreLimit| candidates, as the scoring of such a large number of
   // candidates may cause perceptible typing response delays in the omnibox.
   // This is likely to occur for short omnibox terms such as 'h' and 'w' which
   // will be found in nearly all history candidates. Results are sorted by
   // 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; }
+  // Creates a new URLIndexPrivateData object, populates it from the contents
+  // of the cache file stored in |file_path|, and assigns it to |private_data|.
+  // |languages| will be used to break URLs and page titles into words and is
+  // deliberately passed by value.
+  static void RestoreFromFileTask(
+      const FilePath& file_path,
+      scoped_refptr<URLIndexPrivateData> private_data,
+      std::string 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.  |languages| will be used to
+  // break URLs and page titles into words
+  static scoped_refptr<URLIndexPrivateData> RestoreFromFile(
+      const FilePath& path,
+      const std::string& languages);
 
   // Constructs a new object by rebuilding its contents from the history
   // database in |history_db|. Returns the new URLIndexPrivateData which on
   // success will contain the rebuilt data but upon failure will be empty.
-  static URLIndexPrivateData* RebuildFromHistory(HistoryDatabase* history_db);
+  // |languages| gives a list of language encodings by which the URLs and page
+  // titles are broken down into words and characters.
+  static scoped_refptr<URLIndexPrivateData> RebuildFromHistory(
+      HistoryDatabase* history_db,
+      const std::string& languages,
+      const std::set<std::string>& scheme_whitelist);
+
+  // Writes |private_data| as a cache file to |file_path| and returns success
+  // via |succeeded|.
+  static void WritePrivateDataToCacheFileTask(
+      scoped_refptr<URLIndexPrivateData> private_data,
+      const FilePath& file_path,
+      scoped_refptr<RefCountedBool> succeeded);
 
   // Caches the index private data and writes the cache file to the profile
-  // directory.
+  // directory.  Called by WritePrivateDataToCacheFileTask.
   bool SaveToFile(const FilePath& file_path);
 
   // Initializes all index data members in preparation for restoring the index
   // from the cache or a complete rebuild from the history database.
   void Clear();
 
+  // Returns true if there is no data in the index.
+  bool Empty() const;
+
+  // Creates a copy of ourself.
+  scoped_refptr<URLIndexPrivateData> Duplicate() const;
+
   // 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 as described by |row|. Returns true if the
-  // row was actually indexed.
-  bool IndexRow(const URLRow& row);
+  // row was actually indexed. |languages| gives a list of language encodings by
+  // which the URLs and page titles are broken down into words and characters.
+  // |scheme_whitelist| is used to filter non-qualifying schemes.
+  bool IndexRow(const URLRow& row,
+                const std::string& languages,
+                const std::set<std::string>& scheme_whitelist);
 
   // Adds the history item in |row| to the index if it does not already already
   // exist and it meets the minimum 'quick' criteria. If the row already exists
   // in the index then the index will be updated if the row still meets the
   // criteria, otherwise the row will be removed from the index. Returns true
-  // if the index was actually updated.
-  bool UpdateURL(const URLRow& row);
+  // if the index was actually updated. |languages| gives a list of language
+  // encodings by which the URLs and page titles are broken down into words and
+  // characters. |scheme_whitelist| is used to filter non-qualifying schemes.
+  bool UpdateURL(const URLRow& row,
+                 const std::string& languages,
+                 const std::set<std::string>& scheme_whitelist);
 
   // 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. Returns true if the index
   // was actually updated.
   bool DeleteURL(const GURL& url);
 
   // Parses and indexes the words in the URL and page title of |row| and
   // calculate the word starts in each, saving the starts in |word_starts|.
-  void AddRowWordsToIndex(const URLRow& row, RowWordStarts* word_starts);
+  // |languages| gives a list of language encodings by which the URLs and page
+  // titles are broken down into words and characters.
+  void AddRowWordsToIndex(const URLRow& row,
+                          RowWordStarts* word_starts,
+                          const std::string& languages);
 
   // 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);
 
   // Given a single word in |uni_word|, adds a reference for the containing
   // history item identified by |history_id| to the index.
   void AddWordToIndex(const string16& uni_word, HistoryID history_id);
@@ skipping 32 matching lines @@
       const string16& lower_string,
       const String16Vector& terms_vector,
       const RowWordStarts& word_starts);
 
   // 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;
-
-  // Sets the version of the cache file that will be saved when calling
-  // SavePrivateData(). For unit testing only.
-  void set_saved_cache_version(int version) { saved_cache_version_ = version; }
-
   // 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;
   void SaveWordStartsMap(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);
+  // is any kind of failure. |languages| will be used to break URLs and page
+  // titles into words
+  bool RestorePrivateData(const imui::InMemoryURLIndexCacheItem& cache,
+                          const std::string& languages);
   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);
-  bool RestoreWordStartsMap(const imui::InMemoryURLIndexCacheItem& cache);
+  bool RestoreWordStartsMap(const imui::InMemoryURLIndexCacheItem& cache,
+                            const std::string& languages);
+
+  // Determines if |gurl| has a whitelisted scheme and returns true if so.
+  static bool URLSchemeIsWhitelisted(const GURL& gurl,
+                                     const std::set<std::string>& whitelist);
 
   // 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 -------------------------------------
 
   // The version of the cache file most recently used to restore this instance
   // of the private data. If the private data was rebuilt from the history
   // database this will be 0.
   int restored_cache_version_;
 
   // A list of all of indexed words. The index of a word in this list is the
   // ID of the word in the word_map_. It reduces the memory overhead by
   // replacing a potentially long and repeated string with a simple index.
@@ skipping 44 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_