test_simple_repeats.cpp

C++的一个好库。。。现在很流行

/*
 *
 * Copyright (c) 2004
 * 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 "test.hpp"

#ifdef BOOST_MSVC
#pragma warning(disable:4127)
#endif

void test_simple_repeats2();

void test_simple_repeats()
{
   using namespace boost::regex_constants;
   // simple repeats:
   TEST_REGEX_SEARCH("a*", perl, "b", match_default, make_array(0, 0, -2, 1, 1, -2, -2));
   TEST_REGEX_SEARCH("ab*", perl, "ab", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab*", basic, "ab", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab*", perl, "sssabbbbbbsss", match_default, make_array(3, 10, -2, -2));
   TEST_REGEX_SEARCH("ab*c*", perl, "a", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("ab*c*", perl, "abbb", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("ab*c*", perl, "accc", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("ab*c*", perl, "abbcc", match_default, make_array(0, 5, -2, -2));
   TEST_REGEX_SEARCH("*a", basic, "*a", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("^*a", basic, "*a", match_default, make_array(0, 2, -2, -2));
   TEST_INVALID_REGEX("*a", perl);
   TEST_INVALID_REGEX("\\<*", perl);
   TEST_INVALID_REGEX("\\>*", perl);
   TEST_REGEX_SEARCH("\n*", perl, "\n\n", match_default, make_array(0, 2, -2, 2, 2, -2, -2));
   TEST_REGEX_SEARCH("\\**", perl, "**", match_default, make_array(0, 2, -2, 2, 2, -2, -2));
   TEST_REGEX_SEARCH("\\*", perl, "*", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("(ab)*", perl, "abab", match_default, make_array(0, 4, 2, 4, -2, 4, 4, -2, -2));

   TEST_INVALID_REGEX("(*)", perl);
   TEST_INVALID_REGEX("(*)", extended);
   TEST_INVALID_REGEX("\\(*\\)", basic);
   TEST_INVALID_REGEX("^*", perl);
   TEST_INVALID_REGEX("^*", extended);
   TEST_INVALID_REGEX("$*", perl);
   TEST_INVALID_REGEX("$*", extended);
   TEST_INVALID_REGEX("$*", basic);
   TEST_INVALID_REGEX("\\b*", perl);
   TEST_INVALID_REGEX("\\B*", perl);
   TEST_INVALID_REGEX("\\A*", perl);
   TEST_INVALID_REGEX("\\z*", perl);
   TEST_INVALID_REGEX("\\Z*", perl);
   TEST_INVALID_REGEX("\\A*", perl);
   TEST_INVALID_REGEX("a|*", perl);
   TEST_INVALID_REGEX("a|*", extended);
   TEST_INVALID_REGEX("(+)", perl);
   TEST_INVALID_REGEX("(+)", extended);
   TEST_INVALID_REGEX("^+", perl);
   TEST_INVALID_REGEX("^+", extended);
   TEST_INVALID_REGEX("$+", perl);
   TEST_INVALID_REGEX("$+", extended);
   TEST_INVALID_REGEX("\\b+", perl);
   TEST_INVALID_REGEX("\\B+", perl);
   TEST_INVALID_REGEX("\\A+", perl);
   TEST_INVALID_REGEX("\\z+", perl);
   TEST_INVALID_REGEX("\\Z+", perl);
   TEST_INVALID_REGEX("\\A+", perl);
   TEST_INVALID_REGEX("a|+", perl);
   TEST_INVALID_REGEX("a|+", extended);
   TEST_INVALID_REGEX("(?)", perl);
   TEST_INVALID_REGEX("(?)", extended);
   TEST_INVALID_REGEX("^?", perl);
   TEST_INVALID_REGEX("^?", extended);
   TEST_INVALID_REGEX("$?", perl);
   TEST_INVALID_REGEX("$?", extended);
   TEST_INVALID_REGEX("\\b?", perl);
   TEST_INVALID_REGEX("\\B?", perl);
   TEST_INVALID_REGEX("\\A?", perl);
   TEST_INVALID_REGEX("\\z?", perl);
   TEST_INVALID_REGEX("\\Z?", perl);
   TEST_INVALID_REGEX("\\A?", perl);
   TEST_INVALID_REGEX("a|?", perl);
   TEST_INVALID_REGEX("a|?", extended);
   TEST_INVALID_REGEX("({1,2})", perl);
   TEST_INVALID_REGEX("({1,2})", extended);
   TEST_INVALID_REGEX("^{1,2}", perl);
   TEST_INVALID_REGEX("^{1,2}", extended);
   TEST_INVALID_REGEX("${1,2}", perl);
   TEST_INVALID_REGEX("${1,2}", extended);
   TEST_INVALID_REGEX("\\b{1,2}", perl);
   TEST_INVALID_REGEX("\\B{1,2}", perl);
   TEST_INVALID_REGEX("\\A{1,2}", perl);
   TEST_INVALID_REGEX("\\z{1,2}", perl);
   TEST_INVALID_REGEX("\\Z{1,2}", perl);
   TEST_INVALID_REGEX("\\A{1,2}", perl);
   TEST_INVALID_REGEX("a|{1,2}", perl);
   TEST_INVALID_REGEX("a|{1,2}", extended);

   // now try operator + :
   TEST_REGEX_SEARCH("ab+", perl, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("ab+", perl, "ab", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab+", perl, "sssabbbbbbsss", match_default, make_array(3, 10, -2, -2));
   TEST_REGEX_SEARCH("ab+c+", perl, "abbb", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("ab+c+", perl, "accc", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("ab+c+", perl, "abbcc", match_default, make_array(0, 5, -2, -2));
   TEST_INVALID_REGEX("+a", perl);
   TEST_INVALID_REGEX("\\<+", perl);
   TEST_INVALID_REGEX("\\>+", perl);
   TEST_REGEX_SEARCH("\n+", perl, "\n\n", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("\\+", perl, "+", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("\\+", perl, "++", match_default, make_array(0, 1, -2, 1, 2, -2, -2));
   TEST_REGEX_SEARCH("\\++", perl, "++", match_default, make_array(0, 2, -2, -2));

   TEST_REGEX_SEARCH("+", basic|bk_plus_qm, "+", match_default, make_array(0, 1, -2, -2));
   TEST_INVALID_REGEX("\\+", basic|bk_plus_qm);
   TEST_REGEX_SEARCH("a\\+", basic|bk_plus_qm, "aa", match_default, make_array(0, 2, -2, -2));

   // now try operator ?
   TEST_REGEX_SEARCH("a?", perl, "b", match_default, make_array(0, 0, -2, 1, 1, -2, -2));
   TEST_REGEX_SEARCH("ab?", perl, "a", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("ab?", perl, "ab", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab?", perl, "sssabbbbbbsss", match_default, make_array(3, 5, -2, -2));
   TEST_REGEX_SEARCH("ab?c?", perl, "a", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("ab?c?", perl, "abbb", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab?c?", perl, "accc", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("ab?c?", perl, "abcc", match_default, make_array(0, 3, -2, -2));
   TEST_INVALID_REGEX("?a", perl);
   TEST_INVALID_REGEX("\\<?", perl);
   TEST_INVALID_REGEX("\\>?", perl);
   TEST_REGEX_SEARCH("\n?", perl, "\n\n", match_default, make_array(0, 1, -2, 1, 2, -2, 2, 2, -2, -2));
   TEST_REGEX_SEARCH("\\?", perl, "?", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("\\?", perl, "?", match_default, make_array(0, 1, -2, -2));
   TEST_REGEX_SEARCH("\\??", perl, "??", match_default, make_array(0, 1, -2, 1, 2, -2, 2, 2, -2, -2));
   TEST_REGEX_SEARCH("?", basic|bk_plus_qm, "?", match_default, make_array(0, 1, -2, -2));
   TEST_INVALID_REGEX("\\?", basic|bk_plus_qm);
   TEST_REGEX_SEARCH("a\\?", basic|bk_plus_qm, "aa", match_default, make_array(0, 1, -2, 1, 2, -2, 2, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\?", basic|bk_plus_qm, "b", match_default, make_array(0, 0, -2, 1, 1, -2, -2));

   TEST_REGEX_SEARCH("a?", basic, "a?", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a+", basic, "a+", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\?", basic, "a?", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\+", basic, "a+", match_default, make_array(0, 2, -2, -2));

   // now try operator {}
   TEST_REGEX_SEARCH("a{2}", perl, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a{2}", perl, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2}", perl, "aaa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,}", perl, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a{2,}", perl, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,}", perl, "aaaaa", match_default, make_array(0, 5, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}", perl, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a{2,4}", perl, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}", perl, "aaa", match_default, make_array(0, 3, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}", perl, "aaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}", perl, "aaaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a{ 2 , 4 }", perl, "aaaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a{ 2 , }", perl, "aaaaa", match_default, make_array(0, 5, -2, -2));
   TEST_REGEX_SEARCH("a{ 2 }", perl, "aaa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\{\\}", perl, "a{}", match_default, make_array(0, 3, -2, -2));
   
   TEST_REGEX_SEARCH("a{2,4}?", perl, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a{2,4}?", perl, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?", perl, "aaa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?", perl, "aaaa", match_default, make_array(0, 2, -2, 2, 4, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?", perl, "aaaaa", match_default, make_array(0, 2, -2, 2, 4, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?$", perl, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?$", perl, "aaa", match_default, make_array(0, 3, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?$", perl, "aaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a{2,4}?$", perl, "aaaaa", match_default, make_array(1, 5, -2, -2));
   TEST_REGEX_SEARCH("^a{0,1}?$", perl, "aaaaa", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("^(?:a){0,1}?$", perl, "aaaaa", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("^a(?:bc)?", perl, "abcbc", match_any|match_all, make_array(-2, -2));
}

void test_simple_repeats2()
{
   using namespace boost::regex_constants;

   TEST_INVALID_REGEX("a{}", perl);
   TEST_INVALID_REGEX("a{", perl);
   TEST_INVALID_REGEX("a{1", perl);
   TEST_INVALID_REGEX("a{1,", perl);
   TEST_INVALID_REGEX("a{1,2", perl);
   TEST_INVALID_REGEX("a{ 1 , 2 ", perl);
   TEST_INVALID_REGEX("a{ }", perl);
   TEST_INVALID_REGEX("a}", perl);
   TEST_INVALID_REGEX("{1}", perl);
   TEST_INVALID_REGEX("a{b}", perl);
   TEST_INVALID_REGEX("a{1b}", perl);
   TEST_INVALID_REGEX("a{1,b}", perl);
   TEST_INVALID_REGEX("a{1,2v}", perl);
   TEST_INVALID_REGEX("a{2,1}", perl);

   // now try operator \\{\\} for POSIX basic regexes
   TEST_REGEX_SEARCH("a\\{2\\}", basic, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a\\{2\\}", basic|no_intervals, "a{2}", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a\\{2\\}", basic, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\{2\\}", basic, "aaa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,\\}", basic, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a\\{2,\\}", basic, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,\\}", basic, "aaaaa", match_default, make_array(0, 5, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,4\\}", basic, "a", match_default, make_array(-2, -2));
   TEST_REGEX_SEARCH("a\\{2,4\\}", basic, "aa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,4\\}", basic, "aaa", match_default, make_array(0, 3, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,4\\}", basic, "aaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a\\{2,4\\}", basic, "aaaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a\\{ 2 , 4 \\}", basic, "aaaaa", match_default, make_array(0, 4, -2, -2));
   TEST_REGEX_SEARCH("a\\{ 2 , \\}", basic, "aaaaa", match_default, make_array(0, 5, -2, -2));
   TEST_REGEX_SEARCH("a\\{ 2 \\}", basic, "aaa", match_default, make_array(0, 2, -2, -2));
   TEST_REGEX_SEARCH("a{}", basic, "a{}", match_default, make_array(0, 3, -2, -2));
   TEST_INVALID_REGEX("a\\{\\}", basic);