diff options
Diffstat (limited to 'src/boost/libs/xpressive/perf')
| -rw-r--r-- | src/boost/libs/xpressive/perf/Jamfile.v2 | 44 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/command_line.cpp | 261 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/gcc/long_twain_search.xml | 37 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/gcc/short_matches.xml | 57 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/gcc/short_twain_search.xml | 7 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/main.cpp | 193 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/msvc/long_twain_search.xml | 37 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/msvc/short_matches.xml | 57 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/msvc/short_twain_search.xml | 7 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/regex_comparison.hpp | 139 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/time_boost.cpp | 105 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/time_dynamic_xpressive.cpp | 105 | ||||
| -rw-r--r-- | src/boost/libs/xpressive/perf/time_static_xpressive.cpp | 224 |
13 files changed, 1273 insertions, 0 deletions
diff --git a/src/boost/libs/xpressive/perf/Jamfile.v2 b/src/boost/libs/xpressive/perf/Jamfile.v2 new file mode 100644 index 000000000..7769420e4 --- /dev/null +++ b/src/boost/libs/xpressive/perf/Jamfile.v2 @@ -0,0 +1,44 @@ +# Copyright Eric Niebler 2006 + +# Use, modification, and distribution is subject to the Boost Software +# License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at +# http://www.boost.org/LICENSE_1_0.txt) + +#project +# : requirements <library>../../../../boost/libs/regex/build//boost_regex +# ; + +BOOST_REGEX_SOURCES = + c_regex_traits + cpp_regex_traits + cregex + fileiter + icu + instances + posix_api + regex + regex_debug + regex_raw_buffer + regex_traits_defaults + static_mutex + w32_regex_traits + wc_regex_traits + wide_posix_api + winstances + usinstances ; + +exe xprperf + : + command_line.cpp + main.cpp + time_boost.cpp + time_dynamic_xpressive.cpp + time_static_xpressive.cpp + $(BOOST_ROOT)/libs/regex/src/$(BOOST_REGEX_SOURCES).cpp + : + <include>$(BOOST_ROOT) + <define>BOOST_REGEX_NO_LIB=1 + <define>BOOST_REGEX_RECURSIVE + <define>BOOST_REGEX_USE_CPP_LOCALE + <define>BOOST_XPRESSIVE_USE_CPP_TRAITS + ; diff --git a/src/boost/libs/xpressive/perf/command_line.cpp b/src/boost/libs/xpressive/perf/command_line.cpp new file mode 100644 index 000000000..1c77805d3 --- /dev/null +++ b/src/boost/libs/xpressive/perf/command_line.cpp @@ -0,0 +1,261 @@ +// (C) Copyright Eric Niebler 2006. +// Distributed under the Boost +// Software License, Version 1.0. (See accompanying file +// LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + +#include <iostream> +#include <iomanip> +#include <fstream> +#include <deque> +#include <sstream> +#include <stdexcept> +#include <iterator> +#include "./regex_comparison.hpp" + +// +// globals: +// +bool time_boost = false; +bool time_greta = false; +bool time_safe_greta = false; +bool time_dynamic_xpressive = false; +bool time_static_xpressive = false; +//bool time_posix = false; +//bool time_pcre = false; + +bool test_matches = false; +bool test_short_twain = false; +bool test_long_twain = false; + +std::string xml_out_file; +std::string xml_contents; +std::list<results> result_list; + +int handle_argument(const std::string& what) +{ + if(what == "-b") + time_boost = true; + else if(what == "-g") + time_greta = true; + else if(what == "-gs") + time_safe_greta = true; + else if(what == "-dx") + time_dynamic_xpressive = true; + else if(what == "-sx") + time_static_xpressive = true; + //else if(what == "-posix") + // time_posix = true; + //else if(what == "-pcre") + // time_pcre = true; + else if(what == "-all") + { + time_boost = true; + time_greta = true; + time_safe_greta = true; + time_dynamic_xpressive = true; + time_static_xpressive = true; + //time_posix = true; + //time_pcre = true; + } + else if(what == "-test-matches") + test_matches = true; + else if(what == "-test-short-twain") + test_short_twain = true; + else if(what == "-test-long-twain") + test_long_twain = true; + else if(what == "-test-all") + { + test_matches = true; + test_short_twain = true; + test_long_twain = true; + } + else if((what == "-h") || (what == "--help")) + return show_usage(); + else if((what[0] == '-') || (what[0] == '/')) + { + std::cerr << "Unknown argument: \"" << what << "\"" << std::endl; + return 1; + } + else if(xml_out_file.size() == 0) + { + xml_out_file = what; + } + else + { + std::cerr << "Unexpected argument: \"" << what << "\"" << std::endl; + return 1; + } + return 0; +} + +int show_usage() +{ + std::cout << + "Usage\n" + "xprperf [-h] [library options] [test options] [xml_output_file]\n" + " -h Show help\n\n" + " library options:\n" + " -b Apply tests to boost library\n" + //" -ba Apply tests to boost library with a custom allocator\n" + //" -be Apply tests to experimental boost library\n" + //" -g Apply tests to GRETA library\n" + //" -gs Apply tests to GRETA library (in non-recursive mode)\n" + " -dx Apply tests to dynamic xpressive library\n" + " -sx Apply tests to static xpressive library\n" + //" -posix Apply tests to POSIX library\n" + //" -pcre Apply tests to PCRE library\n" + " -all Apply tests to all libraries\n\n" + " test options:\n" + " -test-matches Test short matches\n" + " -test-short-twain Test short searches\n" + " -test-long-twain Test long searches\n" + " -test-all Test everthing\n"; + return 1; +} + +void load_file(std::string& text, const char* file) +{ + std::deque<char> temp_copy; + std::ifstream is(file); + if(!is.good()) + { + std::string msg("Unable to open file: \""); + msg.append(file); + msg.append("\""); + throw std::runtime_error(msg); + } + std::istreambuf_iterator<char> it(is); + std::copy(it, std::istreambuf_iterator<char>(), std::back_inserter(temp_copy)); + text.erase(); + text.reserve(temp_copy.size()); + text.append(temp_copy.begin(), temp_copy.end()); +} + +struct xml_double +{ + double d_; + xml_double( double d ) : d_(d) {} + friend std::ostream & operator<<( std::ostream & out, xml_double const & xd ) + { + std::ostringstream tmp; + tmp << std::setprecision(out.precision()) << xd.d_; + std::string str = tmp.str(); + std::string::size_type i = str.find( '-' ); + if( i != std::string::npos ) + str.replace( i, 1, "‑" ); + return out << str; + } +}; + +void print_result(std::ostream& os, double time, double best) +{ + static const char* suffixes[] = {"s", "ms", "us", "ns", "ps", }; + + if(time < 0) + { + os << "<entry>NA</entry>"; + return; + } + double rel = time / best; + bool highlight = ((rel > 0) && (rel < 1.1)); + unsigned suffix = 0; + while(time < 0) + { + time *= 1000; + ++suffix; + } + os << "<entry>"; + if(highlight) + os << "<phrase role=\"highlight\">"; + if(rel <= 1000) + os << std::setprecision(3) << xml_double(rel); + else + os << (int)rel; + os << "<para/>("; + if(time <= 1000) + os << std::setprecision(3) << xml_double(time); + else + os << (int)time; + os << suffixes[suffix] << ")"; + if(highlight) + os << "</phrase>"; + os << "</entry>"; +} + +void output_xml_results(bool show_description, const std::string& title, const std::string& filename) +{ + std::stringstream os; + // Generate the copyright and license on the output file + os << "<!--\n" + " Copyright 2004 Eric Niebler.\n" + "\n" + " Distributed under the Boost Software License, Version 1.0.\n" + " (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)\n" + "-->\n"; + + if(result_list.size()) + { + // calculate the number of columns in this table + int num_cols = 1 + show_description + time_greta + time_safe_greta + + time_dynamic_xpressive + time_static_xpressive + time_boost; + + // + // start by outputting the table header: + // + os << "<informaltable frame=\"all\">\n"; + os << "<bridgehead renderas=\"sect4\">" + "<phrase role=\"table-title\">" << title << "</phrase>" + "</bridgehead>\n"; + os << "<tgroup cols=\"" << num_cols << "\">\n"; + os << "<thead>\n"; + os << "<row>\n"; + + if(time_static_xpressive) os << "<entry>static xpressive</entry>"; + if(time_dynamic_xpressive) os << "<entry>dynamic xpressive</entry>"; + if(time_greta) os << "<entry>GRETA</entry>"; + if(time_safe_greta) os << "<entry>GRETA<para/>(non-recursive mode)</entry>"; + if(time_boost) os << "<entry>Boost</entry>"; + //if(time_posix) os << "<entry>POSIX</entry>"; + //if(time_pcre) os << "<entry>PCRE</entry>"; + if(show_description) os << "<entry>Text</entry>"; + os << "<entry>Expression</entry>"; + os << "\n</row>\n"; + os << "</thead>\n"; + os << "<tbody>\n"; + + // + // Now enumerate through all the test results: + // + std::list<results>::const_iterator first, last; + first = result_list.begin(); + last = result_list.end(); + while(first != last) + { + os << "<row>\n"; + if(time_static_xpressive) print_result(os, first->static_xpressive_time, first->factor); + if(time_dynamic_xpressive) print_result(os, first->dynamic_xpressive_time, first->factor); + if(time_greta) print_result(os, first->greta_time, first->factor); + if(time_safe_greta) print_result(os, first->safe_greta_time, first->factor); + if(time_boost) print_result(os, first->boost_time, first->factor); + //if(time_posix) print_result(os, first->posix_time, first->factor); + //if(time_pcre) print_result(os, first->pcre_time, first->factor); + if(show_description) os << "<entry>" << first->description << "</entry>"; + os << "<entry><literal>" << first->expression << "</literal></entry>"; + os << "\n</row>\n"; + ++first; + } + + os << "</tbody>\n" + "</tgroup>\n" + "</informaltable>\n"; + + result_list.clear(); + } + else + { + os << "<para><emphasis>Results not available...</emphasis></para>\n"; + } + + std::ofstream file(filename.c_str()); + file << os.str(); +} diff --git a/src/boost/libs/xpressive/perf/gcc/long_twain_search.xml b/src/boost/libs/xpressive/perf/gcc/long_twain_search.xml new file mode 100644 index 000000000..71efa3da9 --- /dev/null +++ b/src/boost/libs/xpressive/perf/gcc/long_twain_search.xml @@ -0,0 +1,37 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<informaltable frame="all"> +<bridgehead renderas="sect4"><phrase role="table-title">Long Searches</phrase></bridgehead> +<tgroup cols="4"> +<thead> +<row> +<entry>static xpressive</entry><entry>dynamic xpressive</entry><entry>Boost</entry><entry>Expression</entry> +</row> +</thead> +<tbody> +<row> +<entry><phrase role="highlight">1<para/>(0.0263s)</phrase></entry><entry><phrase role="highlight">1<para/>(0.0263s)</phrase></entry><entry>1.78<para/>(0.0469s)</entry><entry><literal>Twain</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(0.0234s)</phrase></entry><entry><phrase role="highlight">1<para/>(0.0234s)</phrase></entry><entry>1.79<para/>(0.042s)</entry><entry><literal>Huck[[:alpha:]]+</literal></entry> +</row> +<row> +<entry>1.84<para/>(1.26s)</entry><entry>2.21<para/>(1.51s)</entry><entry><phrase role="highlight">1<para/>(0.687s)</phrase></entry><entry><literal>[[:alpha:]]+ing</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1.09<para/>(0.192s)</phrase></entry><entry>2<para/>(0.351s)</entry><entry><phrase role="highlight">1<para/>(0.176s)</phrase></entry><entry><literal>^[^ +]*?Twain</literal></entry> +</row> +<row> +<entry>1.41<para/>(0.08s)</entry><entry>1.21<para/>(0.0684s)</entry><entry><phrase role="highlight">1<para/>(0.0566s)</phrase></entry><entry><literal>Tom|Sawyer|Huckleberry|Finn</literal></entry> +</row> +<row> +<entry>1.56<para/>(0.195s)</entry><entry>1.12<para/>(0.141s)</entry><entry><phrase role="highlight">1<para/>(0.125s)</phrase></entry><entry><literal>(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)</literal></entry> +</row> +</tbody> +</tgroup> +</informaltable> diff --git a/src/boost/libs/xpressive/perf/gcc/short_matches.xml b/src/boost/libs/xpressive/perf/gcc/short_matches.xml new file mode 100644 index 000000000..ee84e61a2 --- /dev/null +++ b/src/boost/libs/xpressive/perf/gcc/short_matches.xml @@ -0,0 +1,57 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<informaltable frame="all"> +<bridgehead renderas="sect4"><phrase role="table-title">Short Matches</phrase></bridgehead> +<tgroup cols="5"> +<thead> +<row> +<entry>static xpressive</entry><entry>dynamic xpressive</entry><entry>Boost</entry><entry>Text</entry><entry>Expression</entry> +</row> +</thead> +<tbody> +<row> +<entry><phrase role="highlight">1<para/>(8.79e‑07s)</phrase></entry><entry><phrase role="highlight">1.08<para/>(9.54e‑07s)</phrase></entry><entry>2.51<para/>(2.2e‑06s)</entry><entry>100- this is a line of ftp response which contains a message string</entry><entry><literal>^([0-9]+)(\-| |$)(.*)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1.06<para/>(1.07e‑06s)</phrase></entry><entry><phrase role="highlight">1<para/>(1.01e‑06s)</phrase></entry><entry>4.01<para/>(4.06e‑06s)</entry><entry>1234-5678-1234-456</entry><entry><literal>([[:digit:]]{4}[- ]){3}[[:digit:]]{3,4}</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(1.4e‑06s)</phrase></entry><entry>1.13<para/>(1.58e‑06s)</entry><entry>2.89<para/>(4.05e‑06s)</entry><entry>john_maddock@compuserve.com</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(1.28e‑06s)</phrase></entry><entry>1.16<para/>(1.49e‑06s)</entry><entry>3.07<para/>(3.94e‑06s)</entry><entry>foo12@foo.edu</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(1.22e‑06s)</phrase></entry><entry>1.2<para/>(1.46e‑06s)</entry><entry>3.22<para/>(3.93e‑06s)</entry><entry>bob.smith@foo.tv</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1.04<para/>(8.64e‑07s)</phrase></entry><entry><phrase role="highlight">1<para/>(8.34e‑07s)</phrase></entry><entry>2.5<para/>(2.09e‑06s)</entry><entry>EH10 2QQ</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry>1.11<para/>(9.09e‑07s)</entry><entry><phrase role="highlight">1<para/>(8.19e‑07s)</phrase></entry><entry>2.47<para/>(2.03e‑06s)</entry><entry>G1 1AA</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry>1.12<para/>(9.38e‑07s)</entry><entry><phrase role="highlight">1<para/>(8.34e‑07s)</phrase></entry><entry>2.5<para/>(2.08e‑06s)</entry><entry>SW1 1ZZ</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(7.9e‑07s)</phrase></entry><entry><phrase role="highlight">1.06<para/>(8.34e‑07s)</phrase></entry><entry>2.49<para/>(1.96e‑06s)</entry><entry>4/1/2001</entry><entry><literal>^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(8.19e‑07s)</phrase></entry><entry><phrase role="highlight">1.04<para/>(8.49e‑07s)</phrase></entry><entry>2.4<para/>(1.97e‑06s)</entry><entry>12/12/2001</entry><entry><literal>^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1.09<para/>(8.95e‑07s)</phrase></entry><entry><phrase role="highlight">1<para/>(8.19e‑07s)</phrase></entry><entry>2.4<para/>(1.96e‑06s)</entry><entry>123</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +<row> +<entry>1.11<para/>(8.79e‑07s)</entry><entry><phrase role="highlight">1<para/>(7.9e‑07s)</phrase></entry><entry>2.57<para/>(2.03e‑06s)</entry><entry>+3.14159</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1.09<para/>(8.94e‑07s)</phrase></entry><entry><phrase role="highlight">1<para/>(8.19e‑07s)</phrase></entry><entry>2.47<para/>(2.03e‑06s)</entry><entry>-3.14159</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +</tbody> +</tgroup> +</informaltable> diff --git a/src/boost/libs/xpressive/perf/gcc/short_twain_search.xml b/src/boost/libs/xpressive/perf/gcc/short_twain_search.xml new file mode 100644 index 000000000..52bc47898 --- /dev/null +++ b/src/boost/libs/xpressive/perf/gcc/short_twain_search.xml @@ -0,0 +1,7 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<para><emphasis>Results not available...</emphasis></para> diff --git a/src/boost/libs/xpressive/perf/main.cpp b/src/boost/libs/xpressive/perf/main.cpp new file mode 100644 index 000000000..19ea02e65 --- /dev/null +++ b/src/boost/libs/xpressive/perf/main.cpp @@ -0,0 +1,193 @@ +/* + * + * Copyright (c) 2002 + * John Maddock + * + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0. (See accompanying file + * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + * + */ + +#include <iostream> +#include <fstream> +#include <iterator> +#include <cassert> +#include <boost/test/execution_monitor.hpp> +#include "./regex_comparison.hpp" + +void test_match(const std::string& re, const std::string& text, const std::string& description) +{ + double time; + results r(re, description); + + std::cout << "Testing: \"" << re << "\" against \"" << description << "\"" << std::endl; + if(time_greta == true) + { + // time = g::time_match(re, text); + // r.greta_time = time; + // std::cout << "\tGRETA regex: " << time << "s\n"; + } + if(time_safe_greta == true) + { + // time = gs::time_match(re, text); + // r.safe_greta_time = time; + // std::cout << "\tSafe GRETA regex: " << time << "s\n"; + } + if(time_dynamic_xpressive == true) + { + time = dxpr::time_match(re, text); + r.dynamic_xpressive_time = time; + std::cout << "\tdynamic xpressive regex: " << time << "s\n"; + } + if(time_static_xpressive == true) + { + time = sxpr::time_match(re, text); + r.static_xpressive_time = time; + std::cout << "\tstatic xpressive regex: " << time << "s\n"; + } + if(time_boost == true) + { + time = b::time_match(re, text); + r.boost_time = time; + std::cout << "\tBoost regex: " << time << "s\n"; + } + //if(time_posix == true) + //{ + // time = posix::time_match(re, text); + // r.posix_time = time; + // std::cout << "\tPOSIX regex: " << time << "s\n"; + //} + //if(time_pcre == true) + //{ + // time = pcr::time_match(re, text); + // r.pcre_time = time; + // std::cout << "\tPCRE regex: " << time << "s\n"; + //} + r.finalise(); + result_list.push_back(r); +} + +void test_find_all(const std::string& re, const std::string& text, const std::string& description) +{ + std::cout << "Testing: " << re << std::endl; + + double time; + results r(re, description); + +#if defined(_MSC_VER) && (_MSC_VER >= 1300) + if(time_greta == true) + { + // time = g::time_find_all(re, text); + // r.greta_time = time; + // std::cout << "\tGRETA regex: " << time << "s\n"; + } + if(time_safe_greta == true) + { + // time = gs::time_find_all(re, text); + // r.safe_greta_time = time; + // std::cout << "\tSafe GRETA regex: " << time << "s\n"; + } +#endif + if(time_dynamic_xpressive == true) + { + time = dxpr::time_find_all(re, text); + r.dynamic_xpressive_time = time; + std::cout << "\tdynamic xpressive regex: " << time << "s\n"; + } + if(time_static_xpressive == true) + { + time = sxpr::time_find_all(re, text); + r.static_xpressive_time = time; + std::cout << "\tstatic xpressive regex: " << time << "s\n"; + } + if(time_boost == true) + { + time = b::time_find_all(re, text); + r.boost_time = time; + std::cout << "\tBoost regex: " << time << "s\n"; + } + //if(time_posix == true) + //{ + // time = posix::time_find_all(re, text); + // r.posix_time = time; + // std::cout << "\tPOSIX regex: " << time << "s\n"; + //} + //if(time_pcre == true) + //{ + // time = pcr::time_find_all(re, text); + // r.pcre_time = time; + // std::cout << "\tPCRE regex: " << time << "s\n"; + //} + r.finalise(); + result_list.push_back(r); +} + +//int cpp_main(int argc, char**const argv) +int main(int argc, char**const argv) +{ + // start by processing the command line args: + if(argc < 2) + return show_usage(); + int result = 0; + for(int c = 1; c < argc; ++c) + { + result += handle_argument(argv[c]); + } + if(result) + return result; + + if(test_matches) + { + // these are from the regex docs: + test_match("^([0-9]+)(\\-| |$)(.*)$", "100- this is a line of ftp response which contains a message string"); + test_match("([[:digit:]]{4}[- ]){3}[[:digit:]]{3,4}", "1234-5678-1234-456"); + // these are from http://www.regxlib.com/ + test_match("^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$", "john_maddock@compuserve.com"); + test_match("^([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})(\\]?)$", "foo12@foo.edu"); + test_match("^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$", "bob.smith@foo.tv"); + test_match("^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$", "EH10 2QQ"); + test_match("^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$", "G1 1AA"); + test_match("^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$", "SW1 1ZZ"); + test_match("^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$", "4/1/2001"); + test_match("^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$", "12/12/2001"); + test_match("^[-+]?[[:digit:]]*\\.?[[:digit:]]*$", "123"); + test_match("^[-+]?[[:digit:]]*\\.?[[:digit:]]*$", "+3.14159"); + test_match("^[-+]?[[:digit:]]*\\.?[[:digit:]]*$", "-3.14159"); + + output_xml_results(true, "Short Matches", "short_matches.xml"); + } + std::string twain; + + if(test_short_twain) + { + load_file(twain, "short_twain.txt"); + + test_find_all("Twain", twain); + test_find_all("Huck[[:alpha:]]+", twain); + test_find_all("[[:alpha:]]+ing", twain); + test_find_all("^[^\n]*?Twain", twain); + test_find_all("Tom|Sawyer|Huckleberry|Finn", twain); + test_find_all("(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)", twain); + + output_xml_results(false, "Moderate Searches", "short_twain_search.xml"); + } + + if(test_long_twain) + { + load_file(twain, "3200.txt"); + + test_find_all("Twain", twain); + test_find_all("Huck[[:alpha:]]+", twain); + test_find_all("[[:alpha:]]+ing", twain); + test_find_all("^[^\n]*?Twain", twain); + test_find_all("Tom|Sawyer|Huckleberry|Finn", twain); + //time_posix = false; + test_find_all("(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)", twain); + //time_posix = true; + + output_xml_results(false, "Long Searches", "long_twain_search.xml"); + } + + return 0; +} diff --git a/src/boost/libs/xpressive/perf/msvc/long_twain_search.xml b/src/boost/libs/xpressive/perf/msvc/long_twain_search.xml new file mode 100644 index 000000000..bf8d5f1f2 --- /dev/null +++ b/src/boost/libs/xpressive/perf/msvc/long_twain_search.xml @@ -0,0 +1,37 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<informaltable frame="all"> +<bridgehead renderas="sect4"><phrase role="table-title">Long Searches</phrase></bridgehead> +<tgroup cols="4"> +<thead> +<row> +<entry>static xpressive</entry><entry>dynamic xpressive</entry><entry>Boost</entry><entry>Expression</entry> +</row> +</thead> +<tbody> +<row> +<entry><phrase role="highlight">1<para/>(0.019s)</phrase></entry><entry><phrase role="highlight">1<para/>(0.019s)</phrase></entry><entry>2.98<para/>(0.0566s)</entry><entry><literal>Twain</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(0.0176s)</phrase></entry><entry><phrase role="highlight">1<para/>(0.0176s)</phrase></entry><entry>3.17<para/>(0.0556s)</entry><entry><literal>Huck[[:alpha:]]+</literal></entry> +</row> +<row> +<entry>3.62<para/>(1.78s)</entry><entry>3.97<para/>(1.95s)</entry><entry><phrase role="highlight">1<para/>(0.492s)</phrase></entry><entry><literal>[[:alpha:]]+ing</literal></entry> +</row> +<row> +<entry>2.32<para/>(0.344s)</entry><entry>3.06<para/>(0.453s)</entry><entry><phrase role="highlight">1<para/>(0.148s)</phrase></entry><entry><literal>^[^ +]*?Twain</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(0.0576s)</phrase></entry><entry><phrase role="highlight">1.05<para/>(0.0606s)</phrase></entry><entry>1.15<para/>(0.0664s)</entry><entry><literal>Tom|Sawyer|Huckleberry|Finn</literal></entry> +</row> +<row> +<entry>1.24<para/>(0.164s)</entry><entry>1.44<para/>(0.191s)</entry><entry><phrase role="highlight">1<para/>(0.133s)</phrase></entry><entry><literal>(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)</literal></entry> +</row> +</tbody> +</tgroup> +</informaltable> diff --git a/src/boost/libs/xpressive/perf/msvc/short_matches.xml b/src/boost/libs/xpressive/perf/msvc/short_matches.xml new file mode 100644 index 000000000..0adcdb3d7 --- /dev/null +++ b/src/boost/libs/xpressive/perf/msvc/short_matches.xml @@ -0,0 +1,57 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<informaltable frame="all"> +<bridgehead renderas="sect4"><phrase role="table-title">Short Matches</phrase></bridgehead> +<tgroup cols="5"> +<thead> +<row> +<entry>static xpressive</entry><entry>dynamic xpressive</entry><entry>Boost</entry><entry>Text</entry><entry>Expression</entry> +</row> +</thead> +<tbody> +<row> +<entry><phrase role="highlight">1<para/>(3.2e‑007s)</phrase></entry><entry>1.37<para/>(4.4e‑007s)</entry><entry>2.38<para/>(7.6e‑007s)</entry><entry>100- this is a line of ftp response which contains a message string</entry><entry><literal>^([0-9]+)(\-| |$)(.*)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(6.4e‑007s)</phrase></entry><entry>1.12<para/>(7.15e‑007s)</entry><entry>1.72<para/>(1.1e‑006s)</entry><entry>1234-5678-1234-456</entry><entry><literal>([[:digit:]]{4}[- ]){3}[[:digit:]]{3,4}</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(9.82e‑007s)</phrase></entry><entry>1.3<para/>(1.28e‑006s)</entry><entry>1.61<para/>(1.58e‑006s)</entry><entry>john_maddock@compuserve.com</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(8.94e‑007s)</phrase></entry><entry>1.3<para/>(1.16e‑006s)</entry><entry>1.7<para/>(1.52e‑006s)</entry><entry>foo12@foo.edu</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(9.09e‑007s)</phrase></entry><entry>1.28<para/>(1.16e‑006s)</entry><entry>1.67<para/>(1.52e‑006s)</entry><entry>bob.smith@foo.tv</entry><entry><literal>^([a-zA-Z0-9_\-\.]+)@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.)|(([a-zA-Z0-9\-]+\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\]?)$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(3.06e‑007s)</phrase></entry><entry><phrase role="highlight">1.07<para/>(3.28e‑007s)</phrase></entry><entry>1.95<para/>(5.96e‑007s)</entry><entry>EH10 2QQ</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(3.13e‑007s)</phrase></entry><entry><phrase role="highlight">1.09<para/>(3.42e‑007s)</phrase></entry><entry>1.86<para/>(5.81e‑007s)</entry><entry>G1 1AA</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(3.2e‑007s)</phrase></entry><entry><phrase role="highlight">1.09<para/>(3.5e‑007s)</phrase></entry><entry>1.86<para/>(5.96e‑007s)</entry><entry>SW1 1ZZ</entry><entry><literal>^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(2.68e‑007s)</phrase></entry><entry>1.22<para/>(3.28e‑007s)</entry><entry>2<para/>(5.36e‑007s)</entry><entry>4/1/2001</entry><entry><literal>^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(2.76e‑007s)</phrase></entry><entry>1.16<para/>(3.2e‑007s)</entry><entry>1.94<para/>(5.36e‑007s)</entry><entry>12/12/2001</entry><entry><literal>^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(2.98e‑007s)</phrase></entry><entry><phrase role="highlight">1.03<para/>(3.06e‑007s)</phrase></entry><entry>1.85<para/>(5.51e‑007s)</entry><entry>123</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(3.2e‑007s)</phrase></entry><entry>1.12<para/>(3.58e‑007s)</entry><entry>1.81<para/>(5.81e‑007s)</entry><entry>+3.14159</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +<row> +<entry><phrase role="highlight">1<para/>(3.28e‑007s)</phrase></entry><entry>1.11<para/>(3.65e‑007s)</entry><entry>1.77<para/>(5.81e‑007s)</entry><entry>-3.14159</entry><entry><literal>^[-+]?[[:digit:]]*\.?[[:digit:]]*$</literal></entry> +</row> +</tbody> +</tgroup> +</informaltable> diff --git a/src/boost/libs/xpressive/perf/msvc/short_twain_search.xml b/src/boost/libs/xpressive/perf/msvc/short_twain_search.xml new file mode 100644 index 000000000..52bc47898 --- /dev/null +++ b/src/boost/libs/xpressive/perf/msvc/short_twain_search.xml @@ -0,0 +1,7 @@ +<!-- + Copyright 2004 Eric Niebler. + + Distributed under the Boost Software License, Version 1.0. + (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) +--> +<para><emphasis>Results not available...</emphasis></para> diff --git a/src/boost/libs/xpressive/perf/regex_comparison.hpp b/src/boost/libs/xpressive/perf/regex_comparison.hpp new file mode 100644 index 000000000..d293049d6 --- /dev/null +++ b/src/boost/libs/xpressive/perf/regex_comparison.hpp @@ -0,0 +1,139 @@ +/* + * + * Copyright (c) 2002 + * John Maddock + * + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0. (See accompanying file + * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + * + */ + + +#ifndef REGEX_COMPARISON_HPP +#define REGEX_COMPARISON_HPP + +#include <string> +#include <list> +#include <boost/limits.hpp> + +// +// globals: +// +extern bool time_boost; +extern bool time_greta; +extern bool time_safe_greta; +extern bool time_dynamic_xpressive; +extern bool time_static_xpressive; +//extern bool time_posix; +//extern bool time_pcre; + +extern bool test_matches; +extern bool test_short_twain; +extern bool test_long_twain; + +extern std::string xml_out_file; +extern std::string xml_contents; + + +int handle_argument(const std::string& what); +int show_usage(); +void load_file(std::string& text, const char* file); +void output_xml_results(bool show_description, const std::string& title, const std::string& filename); + +struct results +{ + double boost_time; + double greta_time; + double safe_greta_time; + double dynamic_xpressive_time; + double static_xpressive_time; + //double posix_time; + //double pcre_time; + double factor; + std::string expression; + std::string description; + results(const std::string& ex, const std::string& desc) + : boost_time(-1), + greta_time(-1), + safe_greta_time(-1), + dynamic_xpressive_time(-1), + static_xpressive_time(-1), + //posix_time(-1), + //pcre_time(-1), + factor((std::numeric_limits<double>::max)()), + expression(ex), + description(desc) + {} + void finalise() + { + if((boost_time >= 0) && (boost_time < factor)) + factor = boost_time; + if((greta_time >= 0) && (greta_time < factor)) + factor = greta_time; + if((safe_greta_time >= 0) && (safe_greta_time < factor)) + factor = safe_greta_time; + if((dynamic_xpressive_time >= 0) && (dynamic_xpressive_time < factor)) + factor = dynamic_xpressive_time; + if((static_xpressive_time >= 0) && (static_xpressive_time < factor)) + factor = static_xpressive_time; + //if((posix_time >= 0) && (posix_time < factor)) + // factor = posix_time; + //if((pcre_time >= 0) && (pcre_time < factor)) + // factor = pcre_time; + if((factor >= 0) && (factor < factor)) + factor = factor; + } +}; + +extern std::list<results> result_list; + + +namespace b { +// boost tests: +double time_match(const std::string& re, const std::string& text); +double time_find_all(const std::string& re, const std::string& text); +} +//namespace posix { +//// posix tests: +//double time_match(const std::string& re, const std::string& text); +//double time_find_all(const std::string& re, const std::string& text); +// +//} +//namespace pcr { +//// pcre tests: +//double time_match(const std::string& re, const std::string& text); +//double time_find_all(const std::string& re, const std::string& text); +// +//} +namespace g { +// greta tests: +double time_match(const std::string& re, const std::string& text); +double time_find_all(const std::string& re, const std::string& text); +} +namespace gs { +// safe greta tests: +double time_match(const std::string& re, const std::string& text); +double time_find_all(const std::string& re, const std::string& text); +} +namespace dxpr { +// dynamic xpressive tests: +double time_match(const std::string& re, const std::string& text); +double time_find_all(const std::string& re, const std::string& text); +} +namespace sxpr { +// static xpressive tests: +double time_match(const std::string& re, const std::string& text); +double time_find_all(const std::string& re, const std::string& text); +} +void test_match(const std::string& re, const std::string& text, const std::string& description); +void test_find_all(const std::string& re, const std::string& text, const std::string& description); +inline void test_match(const std::string& re, const std::string& text) +{ test_match(re, text, text); } +inline void test_find_all(const std::string& re, const std::string& text) +{ test_find_all(re, text, ""); } + + +#define REPEAT_COUNT 10 + +#endif diff --git a/src/boost/libs/xpressive/perf/time_boost.cpp b/src/boost/libs/xpressive/perf/time_boost.cpp new file mode 100644 index 000000000..420f141c2 --- /dev/null +++ b/src/boost/libs/xpressive/perf/time_boost.cpp @@ -0,0 +1,105 @@ +/* + * + * Copyright (c) 2002 + * John Maddock + * + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0. (See accompanying file + * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + * + */ + +#include "./regex_comparison.hpp" +#include <boost/timer.hpp> +#include <boost/regex.hpp> + +namespace b{ + +double time_match(const std::string& re, const std::string& text) +{ + boost::regex e(re, boost::regex_constants::ECMAScript | boost::regex_constants::optimize); + boost::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::regex_match(text, what, e); + } + result = tim.elapsed(); + iter *= 2; + }while(result < 0.5); + iter /= 2; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::regex_match(text, what, e); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +//bool dummy_grep_proc(const boost::smatch&) +//{ return true; } + +struct noop +{ + void operator()( boost::smatch const & ) const + { + } +}; + +double time_find_all(const std::string& re, const std::string& text) +{ + boost::regex e(re, boost::regex_constants::ECMAScript | boost::regex_constants::optimize); + boost::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + //boost::regex_grep(&dummy_grep_proc, text, e); + } + result = tim.elapsed(); + iter *= 2; + }while(result < 0.5); + iter /= 2; + + if(result >10) + return result / iter; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + //boost::regex_grep(&dummy_grep_proc, text, e); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +} diff --git a/src/boost/libs/xpressive/perf/time_dynamic_xpressive.cpp b/src/boost/libs/xpressive/perf/time_dynamic_xpressive.cpp new file mode 100644 index 000000000..d6f0a98f0 --- /dev/null +++ b/src/boost/libs/xpressive/perf/time_dynamic_xpressive.cpp @@ -0,0 +1,105 @@ +/* + * + * Copyright (c) 2002 + * John Maddock + * + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0. (See accompanying file + * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + * + */ + +#include "./regex_comparison.hpp" +#include <cassert> +#include <boost/timer.hpp> +#include <boost/xpressive/xpressive.hpp> + +namespace dxpr +{ + +double time_match(const std::string& re, const std::string& text) +{ + boost::xpressive::sregex e(boost::xpressive::sregex::compile(re, boost::xpressive::regex_constants::optimize)); + boost::xpressive::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + assert(boost::xpressive::regex_match( text, what, e )); + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::regex_match( text, what, e ); + } + result = tim.elapsed(); + iter *= 2; + } while(result < 0.5); + iter /= 2; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::regex_match( text, what, e ); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +struct noop +{ + void operator()( boost::xpressive::smatch const & ) const + { + } +}; + +double time_find_all(const std::string& re, const std::string& text) +{ + boost::xpressive::sregex e(boost::xpressive::sregex::compile(re, boost::xpressive::regex_constants::optimize)); + boost::xpressive::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + } + result = tim.elapsed(); + iter *= 2; + }while(result < 0.5); + iter /= 2; + + if(result >10) + return result / iter; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +} + + diff --git a/src/boost/libs/xpressive/perf/time_static_xpressive.cpp b/src/boost/libs/xpressive/perf/time_static_xpressive.cpp new file mode 100644 index 000000000..b90d6b32f --- /dev/null +++ b/src/boost/libs/xpressive/perf/time_static_xpressive.cpp @@ -0,0 +1,224 @@ +/* + * + * Copyright (c) 2002 + * John Maddock + * + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0. (See accompanying file + * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + * + */ + +#include "./regex_comparison.hpp" +#include <map> +#include <cassert> +#include <boost/timer.hpp> +#include <boost/xpressive/xpressive.hpp> + +namespace sxpr +{ + +using namespace boost::xpressive; + +// short matches +char const * sz1 = "^([0-9]+)(\\-| |$)(.*)$"; +sregex rx1 = bol >> (s1= +range('0','9')) >> (s2= as_xpr('-')|' '|eol) >> (s3= *_) >> eol; + +char const * sz2 = "([[:digit:]]{4}[- ]){3}[[:digit:]]{3,4}"; +sregex rx2 = bol >> repeat<3>(s1= repeat<4>(set[digit]) >> (set='-',' ')) >> repeat<3,4>(set[digit]); + +char const * sz3 = "^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$"; +sregex rx3 + = bol + >> (s1= +set[ range('a','z') | range('A','Z') | range('0','9') | '_' | '-' | '.' ]) + >> '@' + >> (s2= + (s3= '[' >> repeat<1,3>(range('0','9')) >> '.' >> repeat<1,3>(range('0','9')) + >> '.' >> repeat<1,3>(range('0','9')) >> '.' + ) + | + (s4= +(s5= +set[ range('a','z') | range('A','Z') | range('0','9') | '-' ] >> '.' ) ) + ) + >> (s6= repeat<2,4>(set[ range('a','z') | range('A','Z')]) | repeat<1,3>(range('0','9'))) + >> (s7= !as_xpr(']')) + >> eol + ; + +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})(\\]?)$"; +sregex rx4 = rx3; + +char const * sz5 = "^([a-zA-Z0-9_\\-\\.]+)@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.)|(([a-zA-Z0-9\\-]+\\.)+))([a-zA-Z]{2,4}|[0-9]{1,3})(\\]?)$"; +sregex rx5 = rx3; + +char const * sz6 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; +sregex rx6 = bol >> repeat<1,2>(set[ range('a','z') | range('A','Z') ]) + >> range('0','9') >> repeat<0,1>(set[range('0','9')|range('A','Z')|range('a','z')]) + >> repeat<0,1>(as_xpr(' ')) >> range('0','9') + >> repeat<2>(set[range('A','Z')|range('a','z')]) >> eol; + +char const * sz7 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; +sregex rx7 = rx6; + +char const * sz8 = "^[a-zA-Z]{1,2}[0-9][0-9A-Za-z]{0,1} {0,1}[0-9][A-Za-z]{2}$"; +sregex rx8 = rx6; + +char const * sz9 = "^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$"; +sregex rx9 = bol >> repeat<1,2>(set[digit]) >> '/' >> repeat<1,2>(set[digit]) >> '/' >> repeat<4>(set[digit]) >> eol; + +char const * sz10 = "^[[:digit:]]{1,2}/[[:digit:]]{1,2}/[[:digit:]]{4}$"; +sregex rx10 = rx9; + +char const * sz11 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; +sregex rx11 = bol >> !(set= '-','+') >> *set[digit] >> !as_xpr('.') >> *set[digit] >> eol ; + +char const * sz12 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; +sregex rx12 = rx11; + +char const * sz13 = "^[-+]?[[:digit:]]*\\.?[[:digit:]]*$"; +sregex rx13 = rx11; + +// long matches +char const * sz14 = "Twain"; +boost::xpressive::sregex rx14 = as_xpr("Twain"); + +char const * sz15 = "Huck[[:alpha:]]+"; +boost::xpressive::sregex rx15 = "Huck" >> +set[alpha]; + +char const * sz16 = "[[:alpha:]]+ing"; +boost::xpressive::sregex rx16 = +set[alpha] >> "ing"; + +char const * sz17 = "^[^\n]*?Twain"; +boost::xpressive::sregex rx17 = bol >> -*~as_xpr('\n') >> "Twain"; + +char const * sz18 = "Tom|Sawyer|Huckleberry|Finn"; +boost::xpressive::sregex rx18 = ( as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ); + +//char const * sz18 = "Tom|Sawyer|.uckleberry|Finn"; +//boost::xpressive::sregex rx18 = ( as_xpr("Tom") | "Sawyer" | _ >> "uckleberry" | "Finn" ); + +char const * sz19 = "(Tom|Sawyer|Huckleberry|Finn).{0,30}river|river.{0,30}(Tom|Sawyer|Huckleberry|Finn)"; +boost::xpressive::sregex rx19 = + (s1= as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ) + >> repeat<0,30>(_) + >> "river" + | + "river" + >> repeat<0,30>(_) + >> (s2= as_xpr("Tom") | "Sawyer" | "Huckleberry" | "Finn" ); + +std::map< std::string, sregex > rxmap; + +struct map_init +{ + map_init() + { + rxmap[ sz1 ] = rx1; + rxmap[ sz2 ] = rx2; + rxmap[ sz3 ] = rx3; + rxmap[ sz4 ] = rx4; + rxmap[ sz5 ] = rx5; + rxmap[ sz6 ] = rx6; + rxmap[ sz7 ] = rx7; + rxmap[ sz8 ] = rx8; + rxmap[ sz9 ] = rx9; + rxmap[ sz10 ] = rx10; + rxmap[ sz11 ] = rx11; + rxmap[ sz12 ] = rx12; + rxmap[ sz13 ] = rx13; + rxmap[ sz14 ] = rx14; + rxmap[ sz15 ] = rx15; + rxmap[ sz16 ] = rx16; + rxmap[ sz17 ] = rx17; + rxmap[ sz18 ] = rx18; + rxmap[ sz19 ] = rx19; + } +}; + +static map_init const i = map_init(); + +double time_match(const std::string& re, const std::string& text) +{ + boost::xpressive::sregex const &e = rxmap[ re ]; + boost::xpressive::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + assert(boost::xpressive::regex_match( text, what, e )); + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::regex_match( text, what, e ); + } + result = tim.elapsed(); + iter *= 2; + } while(result < 0.5); + iter /= 2; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::regex_match( text, what, e ); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +struct noop +{ + void operator()( boost::xpressive::smatch const & ) const + { + } +}; + +double time_find_all(const std::string& re, const std::string& text) +{ + boost::xpressive::sregex const &e = rxmap[ re ]; + boost::xpressive::smatch what; + boost::timer tim; + int iter = 1; + int counter, repeats; + double result = 0; + double run; + do + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + } + result = tim.elapsed(); + iter *= 2; + }while(result < 0.5); + iter /= 2; + + if(result >10) + return result / iter; + + // repeat test and report least value for consistency: + for(repeats = 0; repeats < REPEAT_COUNT; ++repeats) + { + tim.restart(); + for(counter = 0; counter < iter; ++counter) + { + boost::xpressive::sregex_iterator begin( text.begin(), text.end(), e ), end; + std::for_each( begin, end, noop() ); + } + run = tim.elapsed(); + result = (std::min)(run, result); + } + return result / iter; +} + +} + + |