1 // Copyright 2009 The RE2 Authors.  All Rights Reserved.
2 // Use of this source code is governed by a BSD-style
3 // license that can be found in the LICENSE file.
4 
5 // The class FilteredRE2 is used as a wrapper to multiple RE2 regexps.
6 // It provides a prefilter mechanism that helps in cutting down the
7 // number of regexps that need to be actually searched.
8 //
9 // By design, it does not include a string matching engine. This is to
10 // allow the user of the class to use their favorite string match
11 // engine. The overall flow is: Add all the regexps using Add, then
12 // Compile the FilteredRE2. The compile returns strings that need to
13 // be matched. Note that all returned strings are lowercase. For
14 // applying regexps to a search text, the caller does the string
15 // matching using the strings returned. When doing the string match,
16 // note that the caller has to do that on lower cased version of the
17 // search text. Then call FirstMatch or AllMatches with a vector of
18 // indices of strings that were found in the text to get the actual
19 // regexp matches.
20 
21 #ifndef RE2_FILTERED_RE2_H_
22 #define RE2_FILTERED_RE2_H_
23 
24 #include <vector>
25 #include "re2/re2.h"
26 
27 namespace re2 {
28 using std::vector;
29 
30 class PrefilterTree;
31 
32 class FilteredRE2 {
33  public:
34   FilteredRE2();
35   ~FilteredRE2();
36 
37   // Uses RE2 constructor to create a RE2 object (re). Returns
38   // re->error_code(). If error_code is other than NoError, then re is
39   // deleted and not added to re2_vec_.
40   RE2::ErrorCode Add(const StringPiece& pattern,
41                      const RE2::Options& options,
42                      int *id);
43 
44   // Prepares the regexps added by Add for filtering.  Returns a set
45   // of strings that the caller should check for in candidate texts.
46   // The returned strings are lowercased. When doing string matching,
47   // the search text should be lowercased first to find matching
48   // strings from the set of strings returned by Compile.  Call after
49   // all Add calls are done.
50   void Compile(vector<string>* strings_to_match);
51 
52   // Returns the index of the first matching regexp.
53   // Returns -1 on no match. Can be called prior to Compile.
54   // Does not do any filtering: simply tries to Match the
55   // regexps in a loop.
56   int SlowFirstMatch(const StringPiece& text) const;
57 
58   // Returns the index of the first matching regexp.
59   // Returns -1 on no match. Compile has to be called before
60   // calling this.
61   int FirstMatch(const StringPiece& text,
62                  const vector<int>& atoms) const;
63 
64   // Returns the indices of all matching regexps, after first clearing
65   // matched_regexps.
66   bool AllMatches(const StringPiece& text,
67                   const vector<int>& atoms,
68                   vector<int>* matching_regexps) const;
69 
70   // The number of regexps added.
NumRegexps()71   int NumRegexps() const { return re2_vec_.size(); }
72 
73  private:
74 
75   // Get the individual RE2 objects. Useful for testing.
GetRE2(int regexpid)76   RE2* GetRE2(int regexpid) const { return re2_vec_[regexpid]; }
77 
78   // Print prefilter.
79   void PrintPrefilter(int regexpid);
80 
81   // Useful for testing and debugging.
82   void RegexpsGivenStrings(const vector<int>& matched_atoms,
83                            vector<int>* passed_regexps);
84 
85   // All the regexps in the FilteredRE2.
86   vector<RE2*> re2_vec_;
87 
88   // Has the FilteredRE2 been compiled using Compile()
89   bool compiled_;
90 
91   // An AND-OR tree of string atoms used for filtering regexps.
92   PrefilterTree* prefilter_tree_;
93 
94   //DISALLOW_EVIL_CONSTRUCTORS(FilteredRE2);
95   FilteredRE2(const FilteredRE2&);
96   void operator=(const FilteredRE2&);
97 };
98 
99 }  // namespace re2
100 
101 #endif  // RE2_FILTERED_RE2_H_
102