OLD | NEW |
| (Empty) |
1 // Copyright (c) 2012 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_AUTOCOMPLETE_SCORED_HISTORY_MATCH_H_ | |
6 #define CHROME_BROWSER_AUTOCOMPLETE_SCORED_HISTORY_MATCH_H_ | |
7 | |
8 #include <string> | |
9 #include <vector> | |
10 | |
11 #include "base/strings/string16.h" | |
12 #include "base/time/time.h" | |
13 #include "components/history/core/browser/history_match.h" | |
14 #include "components/history/core/browser/history_types.h" | |
15 #include "components/omnibox/in_memory_url_index_types.h" | |
16 #include "testing/gtest/include/gtest/gtest_prod.h" | |
17 | |
18 class ScoredHistoryMatchTest; | |
19 | |
20 // An HistoryMatch that has a score as well as metrics defining where in the | |
21 // history item's URL and/or page title matches have occurred. | |
22 struct ScoredHistoryMatch : public history::HistoryMatch { | |
23 // ScoreMaxRelevance maps from an intermediate-score to the maximum | |
24 // final-relevance score given to a URL for this intermediate score. | |
25 // This is used to store the score ranges of HQP relevance buckets. | |
26 // Please see GetFinalRelevancyScore() for details. | |
27 typedef std::pair<double, int> ScoreMaxRelevance; | |
28 | |
29 // Required for STL, we don't use this directly. | |
30 ScoredHistoryMatch(); | |
31 | |
32 // Initializes the ScoredHistoryMatch with a raw score calculated for the | |
33 // history item given in |row| with recent visits as indicated in |visits|. It | |
34 // first determines if the row qualifies by seeing if all of the terms in | |
35 // |terms_vector| occur in |row|. If so, calculates a raw score. This raw | |
36 // score is in part determined by whether the matches occur at word | |
37 // boundaries, the locations of which are stored in |word_starts|. For some | |
38 // terms, it's appropriate to look for the word boundary within the term. For | |
39 // instance, the term ".net" should look for a word boundary at the "n". These | |
40 // offsets (".net" should have an offset of 1) come from | |
41 // |terms_to_word_starts_offsets|. |is_url_bookmarked| indicates whether the | |
42 // match's URL is referenced by any bookmarks, which can also affect the raw | |
43 // score. The raw score allows the matches to be ordered and can be used to | |
44 // influence the final score calculated by the client of this index. If the | |
45 // row does not qualify the raw score will be 0. |languages| is used to help | |
46 // parse/format the URL before looking for the terms. | |
47 ScoredHistoryMatch(const history::URLRow& row, | |
48 const VisitInfoVector& visits, | |
49 const std::string& languages, | |
50 const base::string16& lower_string, | |
51 const String16Vector& terms_vector, | |
52 const WordStarts& terms_to_word_starts_offsets, | |
53 const RowWordStarts& word_starts, | |
54 bool is_url_bookmarked, | |
55 base::Time now); | |
56 | |
57 ~ScoredHistoryMatch(); | |
58 | |
59 // Compares two matches by score. Functor supporting URLIndexPrivateData's | |
60 // HistoryItemsForTerms function. Looks at particular fields within | |
61 // with url_info to make tie-breaking a bit smarter. | |
62 static bool MatchScoreGreater(const ScoredHistoryMatch& m1, | |
63 const ScoredHistoryMatch& m2); | |
64 | |
65 // Returns |term_matches| after removing all matches that are not at a | |
66 // word break that are in the range [|start_pos|, |end_pos|). | |
67 // start_pos == string::npos is treated as start_pos = length of string. | |
68 // (In other words, no matches will be filtered.) | |
69 // end_pos == string::npos is treated as end_pos = length of string. | |
70 static TermMatches FilterTermMatchesByWordStarts( | |
71 const TermMatches& term_matches, | |
72 const WordStarts& terms_to_word_starts_offsets, | |
73 const WordStarts& word_starts, | |
74 size_t start_pos, | |
75 size_t end_pos); | |
76 | |
77 // The maximum number of recent visits to examine in GetFrequency(). | |
78 // Public so url_index_private_data.cc knows how many visits it is | |
79 // expected to deliver (at minimum) to this class. | |
80 static const size_t kMaxVisitsToScore; | |
81 | |
82 // An interim score taking into consideration location and completeness | |
83 // of the match. | |
84 int raw_score; | |
85 | |
86 // Both these TermMatches contain the set of matches that are considered | |
87 // important. At this time, that means they exclude mid-word matches | |
88 // except in the hostname of the URL. (Technically, during early | |
89 // construction of ScoredHistoryMatch, they may contain all matches, but | |
90 // unimportant matches are eliminated by GetTopicalityScore(), called | |
91 // during construction.) | |
92 | |
93 // Term matches within the URL. | |
94 TermMatches url_matches; | |
95 // Term matches within the page title. | |
96 TermMatches title_matches; | |
97 | |
98 // True if this is a candidate for in-line autocompletion. | |
99 bool can_inline; | |
100 | |
101 private: | |
102 friend class ScoredHistoryMatchTest; | |
103 FRIEND_TEST_ALL_PREFIXES(ScoredHistoryMatchTest, GetFinalRelevancyScore); | |
104 FRIEND_TEST_ALL_PREFIXES(ScoredHistoryMatchTest, GetHQPBucketsFromString); | |
105 FRIEND_TEST_ALL_PREFIXES(ScoredHistoryMatchTest, ScoringBookmarks); | |
106 FRIEND_TEST_ALL_PREFIXES(ScoredHistoryMatchTest, ScoringScheme); | |
107 FRIEND_TEST_ALL_PREFIXES(ScoredHistoryMatchTest, ScoringTLD); | |
108 | |
109 // Initialize ScoredHistoryMatch statics. Must be called before any other | |
110 // method of ScoredHistoryMatch and before creating any instances. | |
111 static void Init(); | |
112 | |
113 // Return a topicality score based on how many matches appear in the url and | |
114 // the page's title and where they are (e.g., at word boundaries). Revises | |
115 // url_matches and title_matches in the process so they only reflect matches | |
116 // used for scoring. (For instance, some mid-word matches are not given | |
117 // credit in scoring.) | |
118 float GetTopicalityScore(const int num_terms, | |
119 const base::string16& cleaned_up_url, | |
120 const WordStarts& terms_to_word_starts_offsets, | |
121 const RowWordStarts& word_starts); | |
122 | |
123 // Returns a recency score based on |last_visit_days_ago|, which is | |
124 // how many days ago the page was last visited. | |
125 float GetRecencyScore(int last_visit_days_ago) const; | |
126 | |
127 // Examines the first kMaxVisitsToScore and return a score (higher is | |
128 // better) based the rate of visits, whether the page is bookmarked, and | |
129 // how often those visits are typed navigations (i.e., explicitly | |
130 // invoked by the user). |now| is passed in to avoid unnecessarily | |
131 // recomputing it frequently. | |
132 float GetFrequency(const base::Time& now, | |
133 const bool bookmarked, | |
134 const VisitInfoVector& visits) const; | |
135 | |
136 // Combines the two component scores into a final score that's | |
137 // an appropriate value to use as a relevancy score. Scoring buckets are | |
138 // specified through |hqp_relevance_buckets|. Please see the function | |
139 // implementation for more details. | |
140 static float GetFinalRelevancyScore( | |
141 float topicality_score, | |
142 float frequency_score, | |
143 const std::vector<ScoreMaxRelevance>& hqp_relevance_buckets); | |
144 | |
145 // Initializes the HQP experimental params: |hqp_relevance_buckets_| | |
146 // to default buckets. If hqp experimental scoring is enabled, it | |
147 // fetches the |hqp_experimental_scoring_enabled_|, |topicality_threshold_| | |
148 // and |hqp_relevance_buckets_| from omnibox field trials. | |
149 static void InitHQPExperimentalParams(); | |
150 | |
151 // Helper function to parse the string containing the scoring buckets. | |
152 // For example, | |
153 // String: "0.0:400,1.5:600,12.0:1300,20.0:1399" | |
154 // Buckets: vector[(0.0, 400),(1.5,600),(12.0,1300),(20.0,1399)] | |
155 // Returns false, in case if it fail to parse the string. | |
156 static bool GetHQPBucketsFromString( | |
157 const std::string& buckets_str, | |
158 std::vector<ScoreMaxRelevance>* hqp_buckets); | |
159 | |
160 // If true, assign raw scores to be max(whatever it normally would be, a | |
161 // score that's similar to the score HistoryURL provider would assign). | |
162 static bool also_do_hup_like_scoring_; | |
163 | |
164 // Untyped visits to bookmarked pages score this, compared to 1 for | |
165 // untyped visits to non-bookmarked pages and 20 for typed visits. | |
166 static int bookmark_value_; | |
167 | |
168 // True if we should fix certain bugs in frequency scoring. | |
169 static bool fix_frequency_bugs_; | |
170 | |
171 // If true, we allow input terms to match in the TLD (e.g., ".com"). | |
172 static bool allow_tld_matches_; | |
173 | |
174 // If true, we allow input terms to match in the scheme (e.g., "http://"). | |
175 static bool allow_scheme_matches_; | |
176 | |
177 // The number of title words examined when computing topicality scores. | |
178 // Words beyond this number are ignored. | |
179 static size_t num_title_words_to_allow_; | |
180 | |
181 // True, if hqp experimental scoring is enabled. | |
182 static bool hqp_experimental_scoring_enabled_; | |
183 | |
184 // |topicality_threshold_| is used to control the topicality scoring. | |
185 // If |topicality_threshold_| > 0, then URLs with topicality-score < threshold | |
186 // are given topicality score of 0. By default it is initalized to -1. | |
187 static float topicality_threshold_; | |
188 | |
189 // |hqp_relevance_buckets_| gives mapping from (topicality*frequency) | |
190 // to the final relevance scoring. Please see GetFinalRelevancyScore() | |
191 // for more details and scoring method. | |
192 static std::vector<ScoreMaxRelevance>* hqp_relevance_buckets_; | |
193 }; | |
194 typedef std::vector<ScoredHistoryMatch> ScoredHistoryMatches; | |
195 | |
196 #endif // CHROME_BROWSER_AUTOCOMPLETE_SCORED_HISTORY_MATCH_H_ | |
OLD | NEW |