components/query_parser/snippet.h

   1 // Copyright 2014 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 // This module computes snippets of queries based on hits in the documents
   6 // for display in history search results.
   7
   8 #ifndef COMPONENTS_QUERY_PARSER_SNIPPET_H__
   9 #define COMPONENTS_QUERY_PARSER_SNIPPET_H__
  10
  11 #include <vector>
  12
  13 #include "base/strings/string16.h"
  14
  15 namespace query_parser {
  16
  17 class Snippet {
  18  public:
  19   // Each MatchPosition is the [begin, end) positions of a match within a
  20   // string.
  21   typedef std::pair<size_t, size_t> MatchPosition;
  22   typedef std::vector<MatchPosition> MatchPositions;
  23
  24   // Parses an offsets string as returned from a sqlite full text index. An
  25   // offsets string encodes information about why a row matched a text query.
  26   // The information is encoded in the string as a set of matches, where each
  27   // match consists of the column, term-number, location, and length of the
  28   // match. Each element of the match is separated by a space, as is each match
  29   // from other matches.
  30   //
  31   // This method adds the start and end of each match whose column is
  32   // column_num to match_positions. The pairs are ordered based on first,
  33   // with no overlapping elements.
  34   //
  35   // NOTE: the positions returned are in terms of UTF8 encoding. To convert the
  36   // offsets to UTF-16, use ConvertMatchPositionsToWide
  37   static void ExtractMatchPositions(const std::string& offsets_str,
  38                                     const std::string& column_num,
  39                                     MatchPositions* match_positions);
  40
  41   // Converts match positions as returned from ExtractMatchPositions to be in
  42   // terms of a UTF-16 2-byte code unit.
  43   static void ConvertMatchPositionsToWide(
  44       const std::string& utf8_string,
  45       Snippet::MatchPositions* match_positions);
  46
  47   Snippet();
  48   ~Snippet();
  49
  50   // Given |matches|, the match positions within |document|, compute the snippet
  51   // for the document.
  52   // Note that |document| is UTF-8 and the offsets in |matches| are byte
  53   // offsets.
  54   void ComputeSnippet(const MatchPositions& matches,
  55                       const std::string& document);
  56
  57   const base::string16& text() const { return text_; }
  58   const MatchPositions& matches() const { return matches_; }
  59
  60   // Efficiently swaps the contents of this snippet with the other.
  61   void Swap(Snippet* other);
  62
  63  private:
  64   // The text of the snippet.
  65   base::string16 text_;
  66
  67   // The matches within text_.
  68   MatchPositions matches_;
  69 };
  70
  71 }  // namespace query_parser
  72
  73 #endif  // COMPONENTS_QUERY_PARSER_SNIPPET_H__