Merge pull request #101 from boostorg/git_issue_87

Fix recursive expressions where the recursion appears more than once.

Author: jzmaddock

include/boost/regex/v4/basic_regex_creator.hpp

@@ -594,7 +594,7 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
             return 0;
          }
          // everything in range matches:
-         std::memset(result->_map + static_cast<unsigned char>(c1), true, 1 + static_cast<unsigned char>(c2) - static_cast<unsigned char>(c1));
+         std::memset(result->_map + static_cast<unsigned char>(c1), true, static_cast<unsigned char>(1u) + static_cast<unsigned char>(static_cast<unsigned char>(c2) - static_cast<unsigned char>(c1)));
       }
    }
    //
@@ -1070,9 +1070,21 @@ int basic_regex_creator<charT, traits>::calculate_backstep(re_syntax_base* state
    return -1;
 }
 
+struct recursion_saver
+{
+   std::vector<unsigned char> saved_state;
+   std::vector<unsigned char>* state;
+   recursion_saver(std::vector<unsigned char>* p) : saved_state(*p), state(p) {}
+   ~recursion_saver()
+   {
+      state->swap(saved_state);
+   }
+};
+
 template <class charT, class traits>
 void basic_regex_creator<charT, traits>::create_startmap(re_syntax_base* state, unsigned char* l_map, unsigned int* pnull, unsigned char mask)
 {
+   recursion_saver saved_recursions(&m_recursion_checks);
    int not_last_jump = 1;
    re_syntax_base* recursion_start = 0;
    int recursion_sub = 0;

include/boost/regex/v4/basic_regex_parser.hpp

@@ -197,7 +197,7 @@ void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2,
    if(this->m_pdata->m_status)
       return;
    // fill in our sub-expression count:
-   this->m_pdata->m_mark_count = 1 + m_mark_count;
+   this->m_pdata->m_mark_count = 1u + (std::size_t)m_mark_count;
    this->finalize(p1, p2);
 }
 
@@ -2723,7 +2723,7 @@ bool basic_regex_parser<charT, traits>::parse_perl_extension()
    {
 #ifndef BOOST_NO_STD_DISTANCE
       if(this->flags() & regbase::save_subexpression_location)
-         this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position) - 1;
+         this->m_pdata->m_subs.at((std::size_t)markid - 1).second = std::distance(m_base, m_position) - 1;
 #else
       if(this->flags() & regbase::save_subexpression_location)
          this->m_pdata->m_subs.at(markid - 1).second = (m_position - m_base) - 1;

include/boost/regex/v4/w32_regex_traits.hpp

@@ -546,7 +546,7 @@ typename w32_regex_traits_implementation<charT>::char_class_type
       if(pos != m_custom_class_names.end())
          return pos->second;
    }
-   std::size_t state_id = 1 + BOOST_REGEX_DETAIL_NS::get_default_class_id(p1, p2);
+   std::size_t state_id = 1u + (std::size_t)BOOST_REGEX_DETAIL_NS::get_default_class_id(p1, p2);
    if(state_id < sizeof(masks) / sizeof(masks[0]))
       return masks[state_id];
    return masks[0];

src/c_regex_traits.cpp

@@ -157,7 +157,7 @@ c_regex_traits<char>::char_class_type BOOST_REGEX_CALL c_regex_traits<char>::loo
          s[i] = static_cast<char>((std::tolower)(static_cast<unsigned char>(s[i])));
       idx = ::boost::BOOST_REGEX_DETAIL_NS::get_default_class_id(&*s.begin(), &*s.begin() + s.size());
    }
-   BOOST_ASSERT(std::size_t(idx+1) < sizeof(masks) / sizeof(masks[0]));
+   BOOST_ASSERT(std::size_t(idx) + 1u < sizeof(masks) / sizeof(masks[0]));
    return masks[idx+1];
 }
 

test/regress/test_perl_ex.cpp

@@ -935,6 +935,7 @@ void test_recursion()
    TEST_REGEX_SEARCH("namespace\\s+(\\w+)\\s+(\\{(?:[^{}]*(?:(?2)[^{}]*)*)?\\})", perl, "namespace one { namespace two { int foo(){} } { {{{ }  } } } {}}", match_default, make_array(0, 64, 10, 13, 14, 64, -2, -2));
    TEST_INVALID_REGEX("((?1)|a)", perl);
    TEST_REGEX_SEARCH("a(?0)?", perl, "aaaaa", match_default, make_array(0, 5, -2, -2));
+   TEST_REGEX_SEARCH("((?(DEFINE)(?'a'A)(?'b'(?&a)?(?&a)))(?&b)?)", perl, "AA", match_default, make_array(0, 2, 0, 2, -1, -1, -2, 2, 2, 2, 2, -1, -1, -2, -2));
 
    // Recursion to a named sub with a name that is used multiple times:
    TEST_REGEX_SEARCH("(?:(?<A>a+)|(?<A>b+))\\.(?&A)", perl, "aaaa.aa", match_default, make_array(0, 7, 0, 4, -1, -1, -2, -2));