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 | #include "re2/filtered_re2.h" |
6 | |
7 | #include <stddef.h> |
8 | #include <string> |
9 | |
10 | #include "util/util.h" |
11 | #include "util/logging.h" |
12 | #include "re2/prefilter.h" |
13 | #include "re2/prefilter_tree.h" |
14 | |
15 | namespace re2 { |
16 | |
17 | FilteredRE2::FilteredRE2() |
18 | : compiled_(false), |
19 | prefilter_tree_(new PrefilterTree()) { |
20 | } |
21 | |
22 | FilteredRE2::FilteredRE2(int min_atom_len) |
23 | : compiled_(false), |
24 | prefilter_tree_(new PrefilterTree(min_atom_len)) { |
25 | } |
26 | |
27 | FilteredRE2::~FilteredRE2() { |
28 | for (size_t i = 0; i < re2_vec_.size(); i++) |
29 | delete re2_vec_[i]; |
30 | delete prefilter_tree_; |
31 | } |
32 | |
33 | RE2::ErrorCode FilteredRE2::Add(const StringPiece& pattern, |
34 | const RE2::Options& options, int* id) { |
35 | RE2* re = new RE2(pattern, options); |
36 | RE2::ErrorCode code = re->error_code(); |
37 | |
38 | if (!re->ok()) { |
39 | if (options.log_errors()) { |
40 | LOG(ERROR) << "Couldn't compile regular expression, skipping: " |
41 | << re << " due to error " << re->error(); |
42 | } |
43 | delete re; |
44 | } else { |
45 | *id = static_cast<int>(re2_vec_.size()); |
46 | re2_vec_.push_back(re); |
47 | } |
48 | |
49 | return code; |
50 | } |
51 | |
52 | void FilteredRE2::Compile(std::vector<std::string>* atoms) { |
53 | if (compiled_) { |
54 | LOG(ERROR) << "Compile called already." ; |
55 | return; |
56 | } |
57 | |
58 | if (re2_vec_.empty()) { |
59 | LOG(ERROR) << "Compile called before Add." ; |
60 | return; |
61 | } |
62 | |
63 | for (size_t i = 0; i < re2_vec_.size(); i++) { |
64 | Prefilter* prefilter = Prefilter::FromRE2(re2_vec_[i]); |
65 | prefilter_tree_->Add(prefilter); |
66 | } |
67 | atoms->clear(); |
68 | prefilter_tree_->Compile(atoms); |
69 | compiled_ = true; |
70 | } |
71 | |
72 | int FilteredRE2::SlowFirstMatch(const StringPiece& text) const { |
73 | for (size_t i = 0; i < re2_vec_.size(); i++) |
74 | if (RE2::PartialMatch(text, *re2_vec_[i])) |
75 | return static_cast<int>(i); |
76 | return -1; |
77 | } |
78 | |
79 | int FilteredRE2::FirstMatch(const StringPiece& text, |
80 | const std::vector<int>& atoms) const { |
81 | if (!compiled_) { |
82 | LOG(DFATAL) << "FirstMatch called before Compile." ; |
83 | return -1; |
84 | } |
85 | std::vector<int> regexps; |
86 | prefilter_tree_->RegexpsGivenStrings(atoms, ®exps); |
87 | for (size_t i = 0; i < regexps.size(); i++) |
88 | if (RE2::PartialMatch(text, *re2_vec_[regexps[i]])) |
89 | return regexps[i]; |
90 | return -1; |
91 | } |
92 | |
93 | bool FilteredRE2::AllMatches( |
94 | const StringPiece& text, |
95 | const std::vector<int>& atoms, |
96 | std::vector<int>* matching_regexps) const { |
97 | matching_regexps->clear(); |
98 | std::vector<int> regexps; |
99 | prefilter_tree_->RegexpsGivenStrings(atoms, ®exps); |
100 | for (size_t i = 0; i < regexps.size(); i++) |
101 | if (RE2::PartialMatch(text, *re2_vec_[regexps[i]])) |
102 | matching_regexps->push_back(regexps[i]); |
103 | return !matching_regexps->empty(); |
104 | } |
105 | |
106 | void FilteredRE2::AllPotentials( |
107 | const std::vector<int>& atoms, |
108 | std::vector<int>* potential_regexps) const { |
109 | prefilter_tree_->RegexpsGivenStrings(atoms, potential_regexps); |
110 | } |
111 | |
112 | void FilteredRE2::RegexpsGivenStrings(const std::vector<int>& matched_atoms, |
113 | std::vector<int>* passed_regexps) { |
114 | prefilter_tree_->RegexpsGivenStrings(matched_atoms, passed_regexps); |
115 | } |
116 | |
117 | void FilteredRE2::PrintPrefilter(int regexpid) { |
118 | prefilter_tree_->PrintPrefilter(regexpid); |
119 | } |
120 | |
121 | } // namespace re2 |
122 | |