Alexander Gutkin | 0d4c523 | 2013-02-28 13:47:27 +0000 | [diff] [blame] | 1 | // Copyright 2008 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 | // Comparative tester for regular expression matching. |
| 6 | // Checks all implementations against each other. |
| 7 | |
| 8 | #ifndef RE2_TESTING_TESTER_H__ |
| 9 | #define RE2_TESTING_TESTER_H__ |
| 10 | |
| 11 | #include "re2/stringpiece.h" |
| 12 | #include "re2/prog.h" |
| 13 | #include "re2/regexp.h" |
| 14 | #include "re2/re2.h" |
| 15 | #include "util/pcre.h" |
| 16 | |
| 17 | namespace re2 { |
| 18 | |
| 19 | class Regexp; |
| 20 | |
| 21 | // All the supported regexp engines. |
| 22 | enum Engine { |
| 23 | kEngineBacktrack = 0, // Prog::BadSearchBacktrack |
| 24 | kEngineNFA, // Prog::SearchNFA |
| 25 | kEngineDFA, // Prog::SearchDFA, only ask whether it matched |
| 26 | kEngineDFA1, // Prog::SearchDFA, ask for match[0] |
| 27 | kEngineOnePass, // Prog::SearchOnePass, if applicable |
| 28 | kEngineBitState, // Prog::SearchBitState |
| 29 | kEngineRE2, // RE2, all submatches |
| 30 | kEngineRE2a, // RE2, only ask for match[0] |
| 31 | kEngineRE2b, // RE2, only ask whether it matched |
| 32 | kEnginePCRE, // PCRE (util/pcre.h) |
| 33 | |
| 34 | kEngineMax, |
| 35 | }; |
| 36 | |
| 37 | // Make normal math on the enum preserve the type. |
| 38 | // By default, C++ doesn't define ++ on enum, and e+1 has type int. |
| 39 | static inline void operator++(Engine& e, int unused) { |
| 40 | e = static_cast<Engine>(e+1); |
| 41 | } |
| 42 | |
| 43 | static inline Engine operator+(Engine e, int i) { |
| 44 | return static_cast<Engine>(static_cast<int>(e)+i); |
| 45 | } |
| 46 | |
| 47 | // A TestInstance caches per-regexp state for a given |
| 48 | // regular expression in a given configuration |
| 49 | // (UTF-8 vs Latin1, longest vs first match, etc.). |
| 50 | class TestInstance { |
| 51 | public: |
| 52 | struct Result; |
| 53 | |
| 54 | TestInstance(const StringPiece& regexp, Prog::MatchKind kind, |
| 55 | Regexp::ParseFlags flags); |
| 56 | ~TestInstance(); |
| 57 | Regexp::ParseFlags flags() { return flags_; } |
| 58 | bool error() { return error_; } |
| 59 | |
| 60 | // Runs a single test case: search in text, which is in context, |
| 61 | // using the given anchoring. |
| 62 | bool RunCase(const StringPiece& text, const StringPiece& context, |
| 63 | Prog::Anchor anchor); |
| 64 | |
| 65 | private: |
| 66 | // Runs a single search using the named engine type. |
| 67 | void RunSearch(Engine type, |
| 68 | const StringPiece& text, const StringPiece& context, |
| 69 | Prog::Anchor anchor, |
| 70 | Result *result); |
| 71 | |
| 72 | void LogMatch(const char* prefix, Engine e, const StringPiece& text, |
| 73 | const StringPiece& context, Prog::Anchor anchor); |
| 74 | |
| 75 | const StringPiece& regexp_str_; // regexp being tested |
| 76 | Prog::MatchKind kind_; // kind of match |
| 77 | Regexp::ParseFlags flags_; // flags for parsing regexp_str_ |
| 78 | bool error_; // error during constructor? |
| 79 | |
| 80 | Regexp* regexp_; // parsed regexp |
| 81 | int num_captures_; // regexp_->NumCaptures() cached |
| 82 | Prog* prog_; // compiled program |
| 83 | Prog* rprog_; // compiled reverse program |
| 84 | PCRE* re_; // PCRE implementation |
| 85 | RE2* re2_; // RE2 implementation |
| 86 | |
| 87 | DISALLOW_EVIL_CONSTRUCTORS(TestInstance); |
| 88 | }; |
| 89 | |
| 90 | // A group of TestInstances for all possible configurations. |
| 91 | class Tester { |
| 92 | public: |
| 93 | explicit Tester(const StringPiece& regexp); |
| 94 | ~Tester(); |
| 95 | |
| 96 | bool error() { return error_; } |
| 97 | |
| 98 | // Runs a single test case: search in text, which is in context, |
| 99 | // using the given anchoring. |
| 100 | bool TestCase(const StringPiece& text, const StringPiece& context, |
| 101 | Prog::Anchor anchor); |
| 102 | |
| 103 | // Run TestCase(text, text, anchor) for all anchoring modes. |
| 104 | bool TestInput(const StringPiece& text); |
| 105 | |
| 106 | // Run TestCase(text, context, anchor) for all anchoring modes. |
| 107 | bool TestInputInContext(const StringPiece& text, const StringPiece& context); |
| 108 | |
| 109 | private: |
| 110 | bool error_; |
| 111 | vector<TestInstance*> v_; |
| 112 | |
| 113 | DISALLOW_EVIL_CONSTRUCTORS(Tester); |
| 114 | }; |
| 115 | |
| 116 | // Run all possible tests using regexp and text. |
| 117 | bool TestRegexpOnText(const StringPiece& regexp, const StringPiece& text); |
| 118 | |
| 119 | } // namespace re2 |
| 120 | |
| 121 | #endif // RE2_TESTING_TESTER_H__ |