| 1 | /* |
|---|
| 2 | * |
|---|
| 3 | * Copyright (c) 2002 |
|---|
| 4 | * John Maddock |
|---|
| 5 | * |
|---|
| 6 | * Use, modification and distribution are subject to the |
|---|
| 7 | * Boost Software License, Version 1.0. (See accompanying file |
|---|
| 8 | * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) |
|---|
| 9 | * |
|---|
| 10 | */ |
|---|
| 11 | |
|---|
| 12 | #include "./regex_comparison.hpp" |
|---|
| 13 | #include <map> |
|---|
| 14 | #include <cassert> |
|---|
| 15 | #include <boost/timer.hpp> |
|---|
| 16 | #include <boost/xpressive/xpressive.hpp> |
|---|
| 17 | |
|---|
| 18 | namespace sxpr |
|---|
| 19 | { |
|---|
| 20 | |
|---|
| 21 | using namespace boost::xpressive; |
|---|
| 22 | |
|---|
| 23 | // short matches |
|---|
| 24 | char const * sz1 = "^([0-9]+)(\\-| |$)(.*)$"; |
|---|
| 25 | sregex rx1 = bol >> (s1= +range('0','9')) >> (s2= as_xpr('-')|' '|eol) >> (s3= *_) >> eol; |
|---|
| 26 | |
|---|
| 27 | char const * sz2 = "([[:digit:]]{4}[- ]){3}[[:digit:]]{3,4}"; |
|---|
| 28 | sregex rx2 = bol >> repeat<3>(s1= repeat<4>(set[digit]) >> (set='-',' ')) >> repeat<3,4>(set[digit]); |
|---|
| 29 | |
|---|
| 30 | char const * sz3 = "^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$"; |
|---|
| 31 | sregex rx3 |
|---|
| 32 | = bol |
|---|
| 33 | >> (s1= +set[ range('a','z') | range('A','Z') | range('0','9') | '_' | '-' | '.' ]) |
|---|
| 34 | >> '@' |
|---|
| 35 | >> (s2= |
|---|
| 36 | (s3= '[' >> repeat<1,3>(range('0','9')) >> '.' >> repeat<1,3>(range('0','9')) |
|---|
| 37 | >> '.' >> repeat<1,3>(range('0','9')) >> '.' |
|---|
| 38 | ) |
|---|
| 39 | | |
|---|
| 40 | (s4= +(s5= +set[ range('a','z') | range('A','Z') | range('0','9') | '-' ] >> '.' ) ) |
|---|
| 41 | ) |
|---|
| 42 | >> (s6= repeat<2,4>(set[ range('a','z') | range('A','Z')]) | repeat<1,3>(range('0','9'))) |
|---|
| 43 | >> (s7= !as_xpr(']')) |
|---|
| 44 | >> eol |
|---|
| 45 | ; |
|---|
| 46 | |
|---|
| 47 | char const * sz4 = "^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$"; |
|---|
| 48 | sregex rx4 = rx3; |
|---|
| 49 | |
|---|
| 50 | char const * sz5 = "^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$"; |
|---|
| 51 | sregex rx5 = rx3; |
|---|
| 52 | |
|---|
| 53 | char const * sz6 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; |
|---|
| 54 | sregex rx6 = bol >> repeat<1,2>(set[ range('a','z') | range('A','Z') ]) |
|---|
| 55 | >> range('0','9') >> repeat<0,1>(set[range('0','9')|range('A','Z')|range('a','z')]) |
|---|
| 56 | >> repeat<0,1>(as_xpr(' ')) >> range('0','9') |
|---|
| 57 | >> repeat<2>(set[range('A','Z')|range('a','z')]) >> eol; |
|---|
| 58 | |
|---|
| 59 | char const * sz7 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; |
|---|
| 60 | sregex rx7 = rx6; |
|---|
| 61 | |
|---|
| 62 | char const * sz8 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; |
|---|
| 63 | sregex rx8 = rx6; |
|---|
| 64 | |
|---|
| 65 | char const * sz9 = "^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$"; |
|---|
| 66 | sregex rx9 = bol >> repeat<1,2>(set[digit]) >> '/' >> repeat<1,2>(set[digit]) >> '/' >> repeat<4>(set[digit]) >> eol; |
|---|
| 67 | |
|---|
| 68 | char const * sz10 = "^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$"; |
|---|
| 69 | sregex rx10 = rx9; |
|---|
| 70 | |
|---|
| 71 | char const * sz11 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; |
|---|
| 72 | sregex rx11 = bol >> !(set= '-','+') >> *set[digit] >> !as_xpr('.') >> *set[digit] >> eol ; |
|---|
| 73 | |
|---|
| 74 | char const * sz12 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; |
|---|
| 75 | sregex rx12 = rx11; |
|---|
| 76 | |
|---|
| 77 | char const * sz13 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; |
|---|
| 78 | sregex rx13 = rx11; |
|---|
| 79 | |
|---|
| 80 | // long matches |
|---|
| 81 | char const * sz14 = "Twain"; |
|---|
| 82 | boost::xpressive::sregex rx14 = as_xpr("Twain"); |
|---|
| 83 | |
|---|
| 84 | char const * sz15 = "Huck[[:alpha:]]+"; |
|---|
| 85 | boost::xpressive::sregex rx15 = "Huck" >> +set[alpha]; |
|---|
| 86 | |
|---|
| 87 | char const * sz16 = "[[:alpha:]]+ing"; |
|---|
| 88 | boost::xpressive::sregex rx16 = +set[alpha] >> "ing"; |
|---|
| 89 | |
|---|
| 90 | char const * sz17 = "^[^\n]*?Twain"; |
|---|
| 91 | boost::xpressive::sregex rx17 = bol >> -*~as_xpr('\n') >> "Twain"; |
|---|
| 92 | |
|---|
| 93 | char const * sz18 = "Tom|Sawyer|Huckleberry|Finn"; |
|---|
| 94 | boost::xpressive::sregex rx18 = ( as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ); |
|---|
| 95 | |
|---|
| 96 | //char const * sz18 = "Tom|Sawyer|.uckleberry|Finn"; |
|---|
| 97 | //boost::xpressive::sregex rx18 = ( as_xpr("Tom") | "Sawyer" | _ >> "uckleberry" | "Finn" ); |
|---|
| 98 | |
|---|
| 99 | char const * sz19 = "(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)"; |
|---|
| 100 | boost::xpressive::sregex rx19 = |
|---|
| 101 | (s1= as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ) |
|---|
| 102 | >> repeat<0,30>(_) |
|---|
| 103 | >> "river" |
|---|
| 104 | | |
|---|
| 105 | "river" |
|---|
| 106 | >> repeat<0,30>(_) |
|---|
| 107 | >> (s2= as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ); |
|---|
| 108 | |
|---|
| 109 | std::map< std::string, sregex > rxmap; |
|---|
| 110 | |
|---|
| 111 | struct map_init |
|---|
| 112 | { |
|---|
| 113 | map_init() |
|---|
| 114 | { |
|---|
| 115 | rxmap[ sz1 ] = rx1; |
|---|
| 116 | rxmap[ sz2 ] = rx2; |
|---|
| 117 | rxmap[ sz3 ] = rx3; |
|---|
| 118 | rxmap[ sz4 ] = rx4; |
|---|
| 119 | rxmap[ sz5 ] = rx5; |
|---|
| 120 | rxmap[ sz6 ] = rx6; |
|---|
| 121 | rxmap[ sz7 ] = rx7; |
|---|
| 122 | rxmap[ sz8 ] = rx8; |
|---|
| 123 | rxmap[ sz9 ] = rx9; |
|---|
| 124 | rxmap[ sz10 ] = rx10; |
|---|
| 125 | rxmap[ sz11 ] = rx11; |
|---|
| 126 | rxmap[ sz12 ] = rx12; |
|---|
| 127 | rxmap[ sz13 ] = rx13; |
|---|
| 128 | rxmap[ sz14 ] = rx14; |
|---|
| 129 | rxmap[ sz15 ] = rx15; |
|---|
| 130 | rxmap[ sz16 ] = rx16; |
|---|
| 131 | rxmap[ sz17 ] = rx17; |
|---|
| 132 | rxmap[ sz18 ] = rx18; |
|---|
| 133 | rxmap[ sz19 ] = rx19; |
|---|
| 134 | } |
|---|
| 135 | }; |
|---|
| 136 | |
|---|
| 137 | static map_init const i = map_init(); |
|---|
| 138 | |
|---|
| 139 | double time_match(const std::string& re, const std::string& text) |
|---|
| 140 | { |
|---|
| 141 | boost::xpressive::sregex const &e = rxmap[ re ]; |
|---|
| 142 | boost::xpressive::smatch what; |
|---|
| 143 | boost::timer tim; |
|---|
| 144 | int iter = 1; |
|---|
| 145 | int counter, repeats; |
|---|
| 146 | double result = 0; |
|---|
| 147 | double run; |
|---|
| 148 | assert(boost::xpressive::regex_match( text, what, e )); |
|---|
| 149 | do |
|---|
| 150 | { |
|---|
| 151 | tim.restart(); |
|---|
| 152 | for(counter = 0; counter < iter; ++counter) |
|---|
| 153 | { |
|---|
| 154 | boost::xpressive::regex_match( text, what, e ); |
|---|
| 155 | } |
|---|
| 156 | result = tim.elapsed(); |
|---|
| 157 | iter *= 2; |
|---|
| 158 | } while(result < 0.5); |
|---|
| 159 | iter /= 2; |
|---|
| 160 | |
|---|
| 161 | // repeat test and report least value for consistency: |
|---|
| 162 | for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) |
|---|
| 163 | { |
|---|
| 164 | tim.restart(); |
|---|
| 165 | for(counter = 0; counter < iter; ++counter) |
|---|
| 166 | { |
|---|
| 167 | boost::xpressive::regex_match( text, what, e ); |
|---|
| 168 | } |
|---|
| 169 | run = tim.elapsed(); |
|---|
| 170 | result = (std::min)(run, result); |
|---|
| 171 | } |
|---|
| 172 | return result / iter; |
|---|
| 173 | } |
|---|
| 174 | |
|---|
| 175 | struct noop |
|---|
| 176 | { |
|---|
| 177 | void operator()( boost::xpressive::smatch const & ) const |
|---|
| 178 | { |
|---|
| 179 | } |
|---|
| 180 | }; |
|---|
| 181 | |
|---|
| 182 | double time_find_all(const std::string& re, const std::string& text) |
|---|
| 183 | { |
|---|
| 184 | boost::xpressive::sregex const &e = rxmap[ re ]; |
|---|
| 185 | boost::xpressive::smatch what; |
|---|
| 186 | boost::timer tim; |
|---|
| 187 | int iter = 1; |
|---|
| 188 | int counter, repeats; |
|---|
| 189 | double result = 0; |
|---|
| 190 | double run; |
|---|
| 191 | do |
|---|
| 192 | { |
|---|
| 193 | tim.restart(); |
|---|
| 194 | for(counter = 0; counter < iter; ++counter) |
|---|
| 195 | { |
|---|
| 196 | boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; |
|---|
| 197 | std::for_each( begin, end, noop() ); |
|---|
| 198 | } |
|---|
| 199 | result = tim.elapsed(); |
|---|
| 200 | iter *= 2; |
|---|
| 201 | }while(result < 0.5); |
|---|
| 202 | iter /= 2; |
|---|
| 203 | |
|---|
| 204 | if(result >10) |
|---|
| 205 | return result / iter; |
|---|
| 206 | |
|---|
| 207 | // repeat test and report least value for consistency: |
|---|
| 208 | for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) |
|---|
| 209 | { |
|---|
| 210 | tim.restart(); |
|---|
| 211 | for(counter = 0; counter < iter; ++counter) |
|---|
| 212 | { |
|---|
| 213 | boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; |
|---|
| 214 | std::for_each( begin, end, noop() ); |
|---|
| 215 | } |
|---|
| 216 | run = tim.elapsed(); |
|---|
| 217 | result = (std::min)(run, result); |
|---|
| 218 | } |
|---|
| 219 | return result / iter; |
|---|
| 220 | } |
|---|
| 221 | |
|---|
| 222 | } |
|---|
| 223 | |
|---|
| 224 | |
|---|