blob: 6bb0787cb6a5481410cff11191cc3d1412782590 [file] [log] [blame]
[email protected]13f698d2011-05-12 21:55:451// Copyright (c) 2011 The Chromium Authors. All rights reserved.
license.botbf09a502008-08-24 00:55:552// Use of this source code is governed by a BSD-style license that can be
3// found in the LICENSE file.
initial.commit09911bf2008-07-26 23:55:294
[email protected]e53668962010-06-23 15:35:255#ifndef CHROME_BROWSER_HISTORY_QUERY_PARSER_H_
6#define CHROME_BROWSER_HISTORY_QUERY_PARSER_H_
initial.commit09911bf2008-07-26 23:55:297
initial.commit09911bf2008-07-26 23:55:298#include <vector>
9
[email protected]13f698d2011-05-12 21:55:4510#include "base/basictypes.h"
[email protected]d8830562013-06-10 22:01:5411#include "base/strings/string16.h"
[email protected]6956cd62008-08-29 19:48:5812#include "chrome/browser/history/snippet.h"
13
initial.commit09911bf2008-07-26 23:55:2914class QueryNodeList;
15
[email protected]6956cd62008-08-29 19:48:5816// Used by HasMatchIn.
17struct QueryWord {
18 // The work to match against.
[email protected]e53668962010-06-23 15:35:2519 string16 word;
[email protected]6956cd62008-08-29 19:48:5820
21 // The starting position of the word in the original text.
[email protected]c29962f22008-12-03 00:47:5822 size_t position;
[email protected]6956cd62008-08-29 19:48:5823};
24
[email protected]13f698d2011-05-12 21:55:4525// QueryNode is used by QueryParser to represent the elements that constitute a
26// query. While QueryNode is exposed by way of ParseQuery, it really isn't meant
27// for external usage.
initial.commit09911bf2008-07-26 23:55:2928class QueryNode {
29 public:
30 virtual ~QueryNode() {}
31
32 // Serialize ourselves out to a string that can be passed to SQLite. Returns
33 // the number of words in this node.
[email protected]e53668962010-06-23 15:35:2534 virtual int AppendToSQLiteQuery(string16* query) const = 0;
initial.commit09911bf2008-07-26 23:55:2935
[email protected]13f698d2011-05-12 21:55:4536 // Return true if this is a QueryNodeWord, false if it's a QueryNodeList.
initial.commit09911bf2008-07-26 23:55:2937 virtual bool IsWord() const = 0;
38
[email protected]13f698d2011-05-12 21:55:4539 // Returns true if this node matches |word|. If |exact| is true, the string
40 // must exactly match. Otherwise, this uses a starts with comparison.
[email protected]e53668962010-06-23 15:35:2541 virtual bool Matches(const string16& word, bool exact) const = 0;
initial.commit09911bf2008-07-26 23:55:2942
[email protected]25320602012-10-18 22:05:5643 // Returns true if this node matches at least one of the words in |words|. An
44 // entry is added to |match_positions| for all matching words giving the
45 // matching regions.
[email protected]6956cd62008-08-29 19:48:5846 virtual bool HasMatchIn(const std::vector<QueryWord>& words,
47 Snippet::MatchPositions* match_positions) const = 0;
[email protected]7de99592008-12-09 19:16:0248
49 // Appends the words that make up this node in |words|.
[email protected]e53668962010-06-23 15:35:2550 virtual void AppendWords(std::vector<string16>* words) const = 0;
initial.commit09911bf2008-07-26 23:55:2951};
52
[email protected]13f698d2011-05-12 21:55:4553// This class is used to parse queries entered into the history search into more
54// normalized queries that can be passed to the SQLite backend.
initial.commit09911bf2008-07-26 23:55:2955class QueryParser {
56 public:
57 QueryParser();
58
[email protected]85d911c2009-05-19 03:59:4259 // For CJK ideographs and Korean Hangul, even a single character
60 // can be useful in prefix matching, but that may give us too many
61 // false positives. Moreover, the current ICU word breaker gives us
62 // back every single Chinese character as a word so that there's no
63 // point doing anything for them and we only adjust the minimum length
64 // to 2 for Korean Hangul while using 3 for others. This is a temporary
65 // hack until we have a segmentation support.
[email protected]e53668962010-06-23 15:35:2566 static bool IsWordLongEnoughForPrefixSearch(const string16& word);
[email protected]85d911c2009-05-19 03:59:4267
initial.commit09911bf2008-07-26 23:55:2968 // Parse a query into a SQLite query. The resulting query is placed in
[email protected]13f698d2011-05-12 21:55:4569 // |sqlite_query| and the number of words is returned.
70 int ParseQuery(const string16& query, string16* sqlite_query);
initial.commit09911bf2008-07-26 23:55:2971
[email protected]8c793c82011-05-19 00:41:3372 // Parses |query|, returning the words that make up it. Any words in quotes
73 // are put in |words| without the quotes. For example, the query text
74 // "foo bar" results in two entries being added to words, one for foo and one
75 // for bar.
76 void ParseQueryWords(const string16& query, std::vector<string16>* words);
initial.commit09911bf2008-07-26 23:55:2977
[email protected]8c793c82011-05-19 00:41:3378 // Parses |query|, returning the nodes that constitute the valid words in the
79 // query. This is intended for later usage with DoesQueryMatch. Ownership of
80 // the nodes passes to the caller.
81 void ParseQueryNodes(const string16& query, std::vector<QueryNode*>* nodes);
[email protected]7de99592008-12-09 19:16:0282
initial.commit09911bf2008-07-26 23:55:2983 // Returns true if the string text matches the query nodes created by a call
[email protected]13f698d2011-05-12 21:55:4584 // to ParseQuery. If the query does match, each of the matching positions in
[email protected]6956cd62008-08-29 19:48:5885 // the text is added to |match_positions|.
[email protected]e53668962010-06-23 15:35:2586 bool DoesQueryMatch(const string16& text,
[email protected]6956cd62008-08-29 19:48:5887 const std::vector<QueryNode*>& nodes,
88 Snippet::MatchPositions* match_positions);
initial.commit09911bf2008-07-26 23:55:2989
90 private:
[email protected]13f698d2011-05-12 21:55:4591 // Does the work of parsing |query|; creates nodes in |root| as appropriate.
92 // This is invoked from both of the ParseQuery methods.
93 bool ParseQueryImpl(const string16& query, QueryNodeList* root);
initial.commit09911bf2008-07-26 23:55:2994
[email protected]13f698d2011-05-12 21:55:4595 // Extracts the words from |text|, placing each word into |words|.
96 void ExtractQueryWords(const string16& text, std::vector<QueryWord>* words);
97
98 DISALLOW_COPY_AND_ASSIGN(QueryParser);
initial.commit09911bf2008-07-26 23:55:2999};
100
[email protected]e53668962010-06-23 15:35:25101#endif // CHROME_BROWSER_HISTORY_QUERY_PARSER_H_