• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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
3 // found in the LICENSE file.
4 
5 #ifndef CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_
6 #define CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_
7 #pragma once
8 
9 #include <functional>
10 #include <map>
11 #include <set>
12 #include <string>
13 #include <vector>
14 
15 #include "app/sql/connection.h"
16 #include "base/basictypes.h"
17 #include "base/file_path.h"
18 #include "base/gtest_prod_util.h"
19 #include "base/memory/linked_ptr.h"
20 #include "base/memory/scoped_ptr.h"
21 #include "base/string16.h"
22 #include "chrome/browser/autocomplete/autocomplete_match.h"
23 #include "chrome/browser/autocomplete/history_provider_util.h"
24 #include "chrome/browser/history/history_types.h"
25 #include "chrome/browser/history/in_memory_url_index_cache.pb.h"
26 #include "testing/gtest/include/gtest/gtest_prod.h"
27 
28 class Profile;
29 
30 namespace base {
31 class Time;
32 }
33 
34 namespace in_memory_url_index {
35 class InMemoryURLIndexCacheItem;
36 }
37 
38 namespace history {
39 
40 namespace imui = in_memory_url_index;
41 
42 class URLDatabase;
43 
44 // Specifies where an omnibox term occurs within a string. Used for specifying
45 // highlights in AutocompleteMatches (ACMatchClassifications) and to assist in
46 // scoring a result.
47 struct TermMatch {
TermMatchTermMatch48   TermMatch(int term_num, size_t offset, size_t length)
49       : term_num(term_num),
50         offset(offset),
51         length(length) {}
52 
53   int term_num;  // The index of the term in the original search string.
54   size_t offset;  // The starting offset of the substring match.
55   size_t length;  // The length of the substring match.
56 };
57 typedef std::vector<TermMatch> TermMatches;
58 
59 // Used for intermediate history result operations.
60 struct ScoredHistoryMatch : public HistoryMatch {
61   ScoredHistoryMatch();  // Required by STL.
62   explicit ScoredHistoryMatch(const URLRow& url_info);
63   ~ScoredHistoryMatch();
64 
65   static bool MatchScoreGreater(const ScoredHistoryMatch& m1,
66                                 const ScoredHistoryMatch& m2);
67 
68   // An interim score taking into consideration location and completeness
69   // of the match.
70   int raw_score;
71   TermMatches url_matches;  // Term matches within the URL.
72   TermMatches title_matches;  // Term matches within the page title.
73   size_t prefix_adjust;  // The length of a prefix which should be ignored.
74 };
75 typedef std::vector<ScoredHistoryMatch> ScoredHistoryMatches;
76 
77 // The URL history source.
78 // Holds portions of the URL database in memory in an indexed form.  Used to
79 // quickly look up matching URLs for a given query string.  Used by
80 // the HistoryURLProvider for inline autocomplete and to provide URL
81 // matches to the omnibox.
82 //
83 // Note about multi-byte codepoints and the data structures in the
84 // InMemoryURLIndex class: One will quickly notice that no effort is made to
85 // insure that multi-byte character boundaries are detected when indexing the
86 // words and characters in the URL history database except when converting
87 // URL strings to lowercase. Multi-byte-edness makes no difference when
88 // indexing or when searching the index as the final filtering of results
89 // is dependent on the comparison of a string of bytes, not individual
90 // characters. While the lookup of those bytes during a search in the
91 // |char_word_map_| could serve up words in which the individual char16
92 // occurs as a portion of a composite character the next filtering step
93 // will eliminate such words except in the case where a single character
94 // is being searched on and which character occurs as the second char16 of a
95 // multi-char16 instance.
96 class InMemoryURLIndex {
97  public:
98   // |history_dir| is a path to the directory containing the history database
99   // within the profile wherein the cache and transaction journals will be
100   // stored.
101   explicit InMemoryURLIndex(const FilePath& history_dir);
102   ~InMemoryURLIndex();
103 
104   // Convenience types
105   typedef std::vector<string16> String16Vector;
106 
107   // Opens and indexes the URL history database.
108   // |languages| gives a list of language encodings with which the history
109   // URLs and omnibox searches are interpreted, i.e. when each is broken
110   // down into words and each word is broken down into characters.
111   bool Init(URLDatabase* history_db, const std::string& languages);
112 
113   // Reloads the history index. Attempts to reload from the cache unless
114   // |clear_cache| is true. If the cache is unavailable then reload the
115   // index from |history_db|.
116   bool ReloadFromHistory(URLDatabase* history_db, bool clear_cache);
117 
118   // Signals that any outstanding initialization should be canceled and
119   // flushes the cache to disk.
120   void ShutDown();
121 
122   // Restores the index's private data from the cache file stored in the
123   // profile directory and returns true if successful.
124   bool RestoreFromCacheFile();
125 
126   // Caches the index private data and writes the cache file to the profile
127   // directory.
128   bool SaveToCacheFile();
129 
130   // Given a vector containing one or more words as string16s, scans the
131   // history index and return a vector with all scored, matching history items.
132   // Each term must occur somewhere in the history item's URL or page title for
133   // the item to qualify; however, the terms do not necessarily have to be
134   // adjacent. Results are sorted with higher scoring items first. Each term
135   // from |terms| may contain punctuation but should not contain spaces.
136   // A search request which results in more than |kItemsToScoreLimit| total
137   // candidate items returns no matches (though the results set will be
138   // retained and used for subsequent calls to this function) as the scoring
139   // of such a large number of candidates may cause perceptible typing response
140   // delays in the omnibox. This is likely to occur for short omnibox terms
141   // such as 'h' and 'w' which will be found in nearly all history candidates.
142   ScoredHistoryMatches HistoryItemsForTerms(const String16Vector& terms);
143 
144   // Updates or adds an history item to the index if it meets the minimum
145   // 'quick' criteria.
146   void UpdateURL(URLID row_id, const URLRow& row);
147 
148   // Deletes indexing data for an history item. The item may not have actually
149   // been indexed (which is the case if it did not previously meet minimum
150   // 'quick' criteria).
151   void DeleteURL(URLID row_id);
152 
153   // Breaks the |uni_string| string down into individual words and return
154   // a vector with the individual words in their original order. If
155   // |break_on_space| is false then the resulting list will contain only words
156   // containing alpha-numeric characters. If |break_on_space| is true then the
157   // resulting list will contain strings broken at whitespace.
158   //
159   // Example:
160   //   Given: |uni_string|: "http://www.google.com/ harry the rabbit."
161   //   With |break_on_space| false the returned list will contain:
162   //    "http", "www", "google", "com", "harry", "the", "rabbit"
163   //   With |break_on_space| true the returned list will contain:
164   //    "http://", "www.google.com/", "harry", "the", "rabbit."
165   static String16Vector WordVectorFromString16(const string16& uni_string,
166                                                bool break_on_space);
167 
168   // Extract and return the offsets from |matches|.
169   static std::vector<size_t> OffsetsFromTermMatches(const TermMatches& matches);
170 
171   // Replace the offsets in |matches| with those given in |offsets|, deleting
172   // any which are npos, and return the updated list of matches.
173   static TermMatches ReplaceOffsetsInTermMatches(
174       const TermMatches& matches,
175       const std::vector<size_t>& offsets);
176 
177  private:
178   friend class AddHistoryMatch;
179   FRIEND_TEST_ALL_PREFIXES(LimitedInMemoryURLIndexTest, Initialization);
180   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheFilePath);
181   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, CacheSaveRestore);
182   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Char16Utilities);
183   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, Scoring);
184   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, StaticFunctions);
185   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, TitleSearch);
186   FRIEND_TEST_ALL_PREFIXES(InMemoryURLIndexTest, TypedCharacterCaching);
187 
188   // Signals that there are no previously cached results for the typed term.
189   static const size_t kNoCachedResultForTerm;
190 
191   // Creating one of me without a history path is not allowed (tests excepted).
192   InMemoryURLIndex();
193 
194   // Convenience types.
195   typedef std::set<string16> String16Set;
196   typedef std::set<char16> Char16Set;
197   typedef std::vector<char16> Char16Vector;
198 
199   // An index into list of all of the words we have indexed.
200   typedef int WordID;
201 
202   // A map allowing a WordID to be determined given a word.
203   typedef std::map<string16, WordID> WordMap;
204 
205   // A map from character to word_ids.
206   typedef std::set<WordID> WordIDSet;  // An index into the WordList.
207   typedef std::map<char16, WordIDSet> CharWordIDMap;
208 
209   // A map from word_id to history item.
210   // TODO(mrossetti): URLID is 64 bit: a memory bloat and performance hit.
211   // Consider using a smaller type.
212   typedef URLID HistoryID;
213   typedef std::set<HistoryID> HistoryIDSet;
214   typedef std::map<WordID, HistoryIDSet> WordIDHistoryMap;
215 
216   // Support caching of term character results so that we can optimize
217   // searches which build upon a previous search. Each entry in this vector
218   // represents a progressive reduction of the result set for each unique
219   // character found in the search term, with each character being taken as
220   // initially encountered. For example, once the results for the search
221   // term of "mextexarkana" have been fully determined, this vector will
222   // contain one entry for the characters: m, e, x, t, a, r, k, & n, in
223   // that order. The result sets will naturally shrink in size for each
224   // subsequent character as the sets intersections are taken in an
225   // incremental manner.
226   struct TermCharWordSet;
227   typedef std::vector<TermCharWordSet> TermCharWordSetVector;
228 
229   // TODO(rohitrao): Probably replace this with QueryResults.
230   typedef std::vector<URLRow> URLRowVector;
231 
232   // A map from history_id to the history's URL and title.
233   typedef std::map<HistoryID, URLRow> HistoryInfoMap;
234 
235   // A helper class which performs the final filter on each candidate
236   // history URL match, inserting accepted matches into |scored_matches_|
237   // and trimming the maximum number of matches to 10.
238   class AddHistoryMatch : public std::unary_function<HistoryID, void> {
239    public:
240     AddHistoryMatch(const InMemoryURLIndex& index,
241                     const String16Vector& lower_terms);
242     ~AddHistoryMatch();
243 
244     void operator()(const HistoryID history_id);
245 
ScoredMatches()246     ScoredHistoryMatches ScoredMatches() const { return scored_matches_; }
247 
248    private:
249     const InMemoryURLIndex& index_;
250     ScoredHistoryMatches scored_matches_;
251     const String16Vector& lower_terms_;
252   };
253 
254   // Initializes all index data members in preparation for restoring the index
255   // from the cache or a complete rebuild from the history database.
256   void ClearPrivateData();
257 
258   // Breaks a string down into individual words.
259   static String16Set WordSetFromString16(const string16& uni_string);
260 
261   // Given a vector of Char16s, representing the characters the user has typed
262   // into the omnibox, finds words containing those characters. If any
263   // existing, cached set is a proper subset then starts with that cached
264   // set. Updates the previously-typed-character cache.
265   WordIDSet WordIDSetForTermChars(const Char16Vector& uni_chars);
266 
267   // Given a vector of Char16s in |uni_chars|, compare those characters, in
268   // order, with the previously searched term, returning the index of the
269   // cached results in the term_char_word_set_cache_ of the set with best
270   // matching number of characters. Returns kNoCachedResultForTerm if there
271   // was no match at all (i.e. the first character of |uni-chars| is not the
272   // same as the character of the first entry in term_char_word_set_cache_).
273   size_t CachedResultsIndexForTerm(const Char16Vector& uni_chars);
274 
275   // Creates a TermMatches which has an entry for each occurrence of the string
276   // |term| found in the string |string|. Mark each match with |term_num| so
277   // that the resulting TermMatches can be merged with other TermMatches for
278   // other terms.
279   static TermMatches MatchTermInString(const string16& term,
280                                        const string16& string,
281                                        int term_num);
282 
283   // URL History indexing support functions.
284 
285   // Indexes one URL history item.
286   bool IndexRow(const URLRow& row);
287 
288   // Breaks a string down into unique, individual characters in the order
289   // in which the characters are first encountered in the |uni_word| string.
290   static Char16Vector Char16VectorFromString16(const string16& uni_word);
291 
292   // Breaks the |uni_word| string down into its individual characters.
293   // Note that this is temporarily intended to work on a single word, but
294   // _will_ work on a string of words, perhaps with unexpected results.
295   // TODO(mrossetti): Lots of optimizations possible here for not restarting
296   // a search if the user is just typing along. Also, change this to uniString
297   // and properly handle substring matches, scoring and sorting the results
298   // by score. Also, provide the metrics for where the matches occur so that
299   // the UI can highlight the matched sections.
300   static Char16Set Char16SetFromString16(const string16& uni_word);
301 
302   // Given a single word in |uni_word|, adds a reference for the containing
303   // history item identified by |history_id| to the index.
304   void AddWordToIndex(const string16& uni_word, HistoryID history_id);
305 
306   // Updates an existing entry in the word/history index by adding the
307   // |history_id| to set for |word_id| in the word_id_history_map_.
308   void UpdateWordHistory(WordID word_id, HistoryID history_id);
309 
310   // Creates a new entry in the word/history map for |word_id| and add
311   // |history_id| as the initial element of the word's set.
312   void AddWordHistory(const string16& uni_word, HistoryID history_id);
313 
314   // Clears the search term cache. This cache holds on to the intermediate
315   // word results for each previously typed character to eliminate the need
316   // to re-perform set operations for previously typed characters.
317   void ResetTermCharWordSetCache();
318 
319   // Composes a set of history item IDs by intersecting the set for each word
320   // in |uni_string|.
321   HistoryIDSet HistoryIDSetFromWords(const string16& uni_string);
322 
323   // Helper function to HistoryIDSetFromWords which composes a set of history
324   // ids for the given term given in |uni_word|.
325   HistoryIDSet HistoryIDsForTerm(const string16& uni_word);
326 
327   // Calculates a raw score for this history item by first determining
328   // if all of the terms in |terms_vector| occur in |row| and, if so,
329   // calculating a raw score based on 1) starting position of each term
330   // in the user input, 2) completeness of each term's match, 3) ordering
331   // of the occurrence of each term (i.e. they appear in order), 4) last
332   // visit time, and 5) number of visits.
333   // This raw score allows the results to be ordered and can be used
334   // to influence the final score calculated by the client of this
335   // index. Returns a ScoredHistoryMatch structure with the raw score and
336   // substring matching metrics.
337   static ScoredHistoryMatch ScoredMatchForURL(
338       const URLRow& row,
339       const String16Vector& terms_vector);
340 
341   // Calculates a component score based on position, ordering and total
342   // substring match size using metrics recorded in |matches|. |max_length|
343   // is the length of the string against which the terms are being searched.
344   static int ScoreComponentForMatches(const TermMatches& matches,
345                                       size_t max_length);
346 
347   // Sorts and removes overlapping substring matches from |matches| and
348   // returns the cleaned up matches.
349   static TermMatches SortAndDeoverlap(const TermMatches& matches);
350 
351   // Utility functions supporting RestoreFromCache and SaveToCache.
352 
353   // Construct a file path for the cache file within the same directory where
354   // the history database is kept and saves that path to |file_path|. Returns
355   // true if |file_path| can be successfully constructed. (This function
356   // provided as a hook for unit testing.)
357   bool GetCacheFilePath(FilePath* file_path);
358 
359   // Encode a data structure into the protobuf |cache|.
360   void SavePrivateData(imui::InMemoryURLIndexCacheItem* cache) const;
361   void SaveWordList(imui::InMemoryURLIndexCacheItem* cache) const;
362   void SaveWordMap(imui::InMemoryURLIndexCacheItem* cache) const;
363   void SaveCharWordMap(imui::InMemoryURLIndexCacheItem* cache) const;
364   void SaveWordIDHistoryMap(imui::InMemoryURLIndexCacheItem* cache) const;
365   void SaveHistoryInfoMap(imui::InMemoryURLIndexCacheItem* cache) const;
366 
367   // Decode a data structure from the protobuf |cache|. Return false if there
368   // is any kind of failure.
369   bool RestorePrivateData(const imui::InMemoryURLIndexCacheItem& cache);
370   bool RestoreWordList(const imui::InMemoryURLIndexCacheItem& cache);
371   bool RestoreWordMap(const imui::InMemoryURLIndexCacheItem& cache);
372   bool RestoreCharWordMap(const imui::InMemoryURLIndexCacheItem& cache);
373   bool RestoreWordIDHistoryMap(const imui::InMemoryURLIndexCacheItem& cache);
374   bool RestoreHistoryInfoMap(const imui::InMemoryURLIndexCacheItem& cache);
375 
376   // Directory where cache file resides. This is, except when unit testing,
377   // the same directory in which the profile's history database is found. It
378   // should never be empty.
379   FilePath history_dir_;
380 
381   // The timestamp of when the cache was last saved. This is used to validate
382   // the transaction journal's applicability to the cache. The timestamp is
383   // initialized to the NULL time, indicating that the cache was not used with
384   // the InMemoryURLIndex was last populated.
385   base::Time last_saved_;
386 
387   // A list of all of indexed words. The index of a word in this list is the
388   // ID of the word in the word_map_. It reduces the memory overhead by
389   // replacing a potentially long and repeated string with a simple index.
390   // NOTE: A word will _never_ be removed from this vector thus insuring
391   // the immutability of the word_id throughout the session, reducing
392   // maintenance complexity.
393   // TODO(mrossetti): Profile the vector allocation and determine if judicious
394   // 'reserve' calls are called for.
395   String16Vector word_list_;
396 
397   int history_item_count_;
398   WordMap word_map_;
399   CharWordIDMap char_word_map_;
400   WordIDHistoryMap word_id_history_map_;
401   TermCharWordSetVector term_char_word_set_cache_;
402   HistoryInfoMap history_info_map_;
403   std::string languages_;
404 
405   DISALLOW_COPY_AND_ASSIGN(InMemoryURLIndex);
406 };
407 
408 }  // namespace history
409 
410 #endif  // CHROME_BROWSER_HISTORY_IN_MEMORY_URL_INDEX_H_
411