Merge branch 'de_fuzz' into develop

doc/syntax_perl.qbk

@@ -28,6 +28,8 @@ following special characters:
 
 [pre .\[{}()\\\*+?|^$]
 
+Other characters are special only in certain situations - for example `]` is special only after an opening `[`.
+
 [h4 Wildcard]
 
 The single character '.' when used outside of a character set will match 

include/boost/regex/config.hpp

@@ -152,7 +152,7 @@
 #  if defined(BOOST_REGEX_NO_W32) || BOOST_PLAT_WINDOWS_RUNTIME
 #    define BOOST_REGEX_NO_FILEITER
 #  endif
-#else // defined(_WIN32)
+#else /* defined(_WIN32) */
 #  if !defined(BOOST_HAS_DIRENT_H)
 #    define BOOST_REGEX_NO_FILEITER
 #  endif

include/boost/regex/icu.hpp

@@ -410,7 +410,7 @@ void copy_results(MR1& out, MR2 const& in)
       if(in[i].captures().size())
       {
          out[i].get_captures().assign(in[i].captures().size(), typename MR1::value_type());
-         for(int j = 0; j < out[i].captures().size(); ++j)
+         for(int j = 0; j < (int)out[i].captures().size(); ++j)
          {
             out[i].get_captures()[j].first = in[i].captures()[j].first.base();
             out[i].get_captures()[j].second = in[i].captures()[j].second.base();

include/boost/regex/pending/unicode_iterator.hpp

@@ -117,7 +117,10 @@ inline unsigned utf8_trailing_byte_count(boost::uint8_t c)
 #pragma warning(push)
 #pragma warning(disable:4100)
 #endif
-BOOST_NORETURN inline void invalid_utf32_code_point(::boost::uint32_t val)
+#ifndef BOOST_NO_EXCEPTIONS
+BOOST_NORETURN
+#endif
+inline void invalid_utf32_code_point(::boost::uint32_t val)
 {
 #ifndef BOOST_NO_STD_LOCALE
    std::stringstream ss;

include/boost/regex/v4/basic_regex_creator.hpp

@@ -77,15 +77,15 @@ public:
 
    void add_single(const digraph_type& s)
    {
-      m_singles.insert(m_singles.end(), s);
+      m_singles.insert(s);
       if(s.second)
          m_has_digraphs = true;
       m_empty = false;
    }
    void add_range(const digraph_type& first, const digraph_type& end)
    {
-      m_ranges.insert(m_ranges.end(), first);
-      m_ranges.insert(m_ranges.end(), end);
+      m_ranges.push_back(first);
+      m_ranges.push_back(end);
       if(first.second)
       {
          m_has_digraphs = true;
@@ -110,7 +110,7 @@ public:
    }
    void add_equivalent(const digraph_type& s)
    {
-      m_equivalents.insert(m_equivalents.end(), s);
+      m_equivalents.insert(s);
       if(s.second)
       {
          m_has_digraphs = true;
@@ -136,11 +136,12 @@ public:
       return m_negate;
    }
    typedef typename std::vector<digraph_type>::const_iterator  list_iterator;
-   list_iterator singles_begin()const
+   typedef typename std::set<digraph_type>::const_iterator     set_iterator;
+   set_iterator singles_begin()const
    {
       return m_singles.begin();
    }
-   list_iterator singles_end()const
+   set_iterator singles_end()const
    {
       return m_singles.end();
    }
@@ -152,11 +153,11 @@ public:
    {
       return m_ranges.end();
    }
-   list_iterator equivalents_begin()const
+   set_iterator equivalents_begin()const
    {
       return m_equivalents.begin();
    }
-   list_iterator equivalents_end()const
+   set_iterator equivalents_end()const
    {
       return m_equivalents.end();
    }
@@ -173,14 +174,14 @@ public:
       return m_empty;
    }
 private:
-   std::vector<digraph_type> m_singles;         // a list of single characters to match
+   std::set<digraph_type>    m_singles;         // a list of single characters to match
    std::vector<digraph_type> m_ranges;          // a list of end points of our ranges
    bool                      m_negate;          // true if the set is to be negated
    bool                      m_has_digraphs;    // true if we have digraphs present
    m_type                    m_classes;         // character classes to match
    m_type                    m_negated_classes; // negated character classes to match
    bool                      m_empty;           // whether we've added anything yet
-   std::vector<digraph_type> m_equivalents;     // a list of equivalence classes
+   std::set<digraph_type>    m_equivalents;     // a list of equivalence classes
 };
    
 template <class charT, class traits>
@@ -239,7 +240,7 @@ protected:
    unsigned                      m_backrefs;           // bitmask of permitted backrefs
    boost::uintmax_t              m_bad_repeats;        // bitmask of repeats we can't deduce a startmap for;
    bool                          m_has_recursions;     // set when we have recursive expresisons to fixup
-   std::vector<bool>             m_recursion_checks;   // notes which recursions we've followed while analysing this expression
+   std::vector<unsigned char>    m_recursion_checks;   // notes which recursions we've followed while analysing this expression
    typename traits::char_class_type m_word_mask;       // mask used to determine if a character is a word character
    typename traits::char_class_type m_mask_space;      // mask used to determine if a character is a word character
    typename traits::char_class_type m_lower_mask;       // mask used to determine if a character is a lowercase character
@@ -365,6 +366,7 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
 {
    typedef typename traits::string_type string_type;
    typedef typename basic_char_set<charT, traits>::list_iterator item_iterator;
+   typedef typename basic_char_set<charT, traits>::set_iterator  set_iterator;
    typedef typename traits::char_class_type m_type;
    
    re_set_long<m_type>* result = static_cast<re_set_long<m_type>*>(append_state(syntax_element_long_set, sizeof(re_set_long<m_type>)));
@@ -395,20 +397,25 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
    // now extend with all the singles:
    //
    item_iterator first, last;
-   first = char_set.singles_begin();
-   last = char_set.singles_end();
-   while(first != last)
+   set_iterator sfirst, slast;
+   sfirst = char_set.singles_begin();
+   slast = char_set.singles_end();
+   while(sfirst != slast)
    {
-      charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (first->second ? 3 : 2)));
-      p[0] = m_traits.translate(first->first, m_icase);
-      if(first->second)
+      charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (sfirst->first == static_cast<charT>(0) ? 1 : sfirst->second ? 3 : 2)));
+      p[0] = m_traits.translate(sfirst->first, m_icase);
+      if(sfirst->first == static_cast<charT>(0))
       {
-         p[1] = m_traits.translate(first->second, m_icase);
+         p[0] = 0;
+      }
+      else if(sfirst->second)
+      {
+         p[1] = m_traits.translate(sfirst->second, m_icase);
          p[2] = 0;
       }
       else
          p[1] = 0;
-      ++first;
+      ++sfirst;
    }
    //
    // now extend with all the ranges:
@@ -472,24 +479,24 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
    //
    // now process the equivalence classes:
    //
-   first = char_set.equivalents_begin();
-   last = char_set.equivalents_end();
-   while(first != last)
+   sfirst = char_set.equivalents_begin();
+   slast = char_set.equivalents_end();
+   while(sfirst != slast)
    {
       string_type s;
-      if(first->second)
+      if(sfirst->second)
       {
-         charT cs[3] = { first->first, first->second, charT(0), };
+         charT cs[3] = { sfirst->first, sfirst->second, charT(0), };
          s = m_traits.transform_primary(cs, cs+2);
       }
       else
-         s = m_traits.transform_primary(&first->first, &first->first+1);
+         s = m_traits.transform_primary(&sfirst->first, &sfirst->first+1);
       if(s.empty())
          return 0;  // invalid or unsupported equivalence class
       charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (s.size()+1) ) );
       BOOST_REGEX_DETAIL_NS::copy(s.begin(), s.end(), p);
       p[s.size()] = charT(0);
-      ++first;
+      ++sfirst;
    }
    //
    // finally reset the address of our last state:
@@ -518,7 +525,8 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
 {
    typedef typename traits::string_type string_type;
    typedef typename basic_char_set<charT, traits>::list_iterator item_iterator;
-   
+   typedef typename basic_char_set<charT, traits>::set_iterator set_iterator;
+
    re_set* result = static_cast<re_set*>(append_state(syntax_element_set, sizeof(re_set)));
    bool negate = char_set.is_negated();
    std::memset(result->_map, 0, sizeof(result->_map));
@@ -526,17 +534,18 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
    // handle singles first:
    //
    item_iterator first, last;
-   first = char_set.singles_begin();
-   last = char_set.singles_end();
-   while(first != last)
+   set_iterator sfirst, slast;
+   sfirst = char_set.singles_begin();
+   slast = char_set.singles_end();
+   while(sfirst != slast)
    {
       for(unsigned int i = 0; i < (1 << CHAR_BIT); ++i)
       {
          if(this->m_traits.translate(static_cast<charT>(i), this->m_icase)
-            == this->m_traits.translate(first->first, this->m_icase))
+            == this->m_traits.translate(sfirst->first, this->m_icase))
             result->_map[i] = true;
       }
-      ++first;
+      ++sfirst;
    }
    //
    // OK now handle ranges:
@@ -623,13 +632,13 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
    //
    // now process the equivalence classes:
    //
-   first = char_set.equivalents_begin();
-   last = char_set.equivalents_end();
-   while(first != last)
+   sfirst = char_set.equivalents_begin();
+   slast = char_set.equivalents_end();
+   while(sfirst != slast)
    {
       string_type s;
-      BOOST_ASSERT(static_cast<charT>(0) == first->second);
-      s = m_traits.transform_primary(&first->first, &first->first+1);
+      BOOST_ASSERT(static_cast<charT>(0) == sfirst->second);
+      s = m_traits.transform_primary(&sfirst->first, &sfirst->first+1);
       if(s.empty())
          return 0;  // invalid or unsupported equivalence class
       for(unsigned i = 0; i < (1u << CHAR_BIT); ++i)
@@ -639,7 +648,7 @@ re_syntax_base* basic_regex_creator<charT, traits>::append_set(
          if(s == s2)
             result->_map[i] = true;
       }
-      ++first;
+      ++sfirst;
    }
    if(negate)
    {
@@ -690,7 +699,7 @@ void basic_regex_creator<charT, traits>::finalize(const charT* p1, const charT*
 
    m_bad_repeats = 0;
    if(m_has_recursions)
-      m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);
+      m_recursion_checks.assign(1 + m_pdata->m_mark_count, 0u);
    create_startmap(m_pdata->m_first_state, m_pdata->m_startmap, &(m_pdata->m_can_be_null), mask_all);
    // get the restart type:
    m_pdata->m_restart_type = get_restart_type(m_pdata->m_first_state);
@@ -792,50 +801,57 @@ void basic_regex_creator<charT, traits>::fixup_recursions(re_syntax_base* state)
                //
                idx = m_pdata->get_id(static_cast<int>(idx));
             }
-            while(p)
+            if(idx < 0)
             {
-               if((p->type == syntax_element_startmark) && (static_cast<re_brace*>(p)->index == idx))
+               ok = false;
+            }
+            else
+            {
+               while(p)
                {
-                  //
-                  // We've found the target of the recursion, set the jump target:
-                  //
-                  static_cast<re_jump*>(state)->alt.p = p;
-                  ok = true;
-                  // 
-                  // Now scan the target for nested repeats:
-                  //
-                  p = p->next.p;
-                  int next_rep_id = 0;
-                  while(p)
+                  if((p->type == syntax_element_startmark) && (static_cast<re_brace*>(p)->index == idx))
                   {
-                     switch(p->type)
-                     {
-                     case syntax_element_rep:
-                     case syntax_element_dot_rep:
-                     case syntax_element_char_rep:
-                     case syntax_element_short_set_rep:
-                     case syntax_element_long_set_rep:
-                        next_rep_id = static_cast<re_repeat*>(p)->state_id;
-                        break;
-                     case syntax_element_endmark:
-                        if(static_cast<const re_brace*>(p)->index == idx)
-                           next_rep_id = -1;
-                        break;
-                     default: 
-                        break;
-                     }
-                     if(next_rep_id)
-                        break;
+                     //
+                     // We've found the target of the recursion, set the jump target:
+                     //
+                     static_cast<re_jump*>(state)->alt.p = p;
+                     ok = true;
+                     // 
+                     // Now scan the target for nested repeats:
+                     //
                      p = p->next.p;
-                  }
-                  if(next_rep_id > 0)
-                  {
-                     static_cast<re_recurse*>(state)->state_id = next_rep_id - 1;
-                  }
+                     int next_rep_id = 0;
+                     while(p)
+                     {
+                        switch(p->type)
+                        {
+                        case syntax_element_rep:
+                        case syntax_element_dot_rep:
+                        case syntax_element_char_rep:
+                        case syntax_element_short_set_rep:
+                        case syntax_element_long_set_rep:
+                           next_rep_id = static_cast<re_repeat*>(p)->state_id;
+                           break;
+                        case syntax_element_endmark:
+                           if(static_cast<const re_brace*>(p)->index == idx)
+                              next_rep_id = -1;
+                           break;
+                        default:
+                           break;
+                        }
+                        if(next_rep_id)
+                           break;
+                        p = p->next.p;
+                     }
+                     if(next_rep_id > 0)
+                     {
+                        static_cast<re_recurse*>(state)->state_id = next_rep_id - 1;
+                     }
 
-                  break;
+                     break;
+                  }
+                  p = p->next.p;
                }
-               p = p->next.p;
             }
             if(!ok)
             {
@@ -934,7 +950,7 @@ void basic_regex_creator<charT, traits>::create_startmaps(re_syntax_base* state)
    {
       // Initialize m_recursion_checks if we need it:
       if(m_has_recursions)
-         m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);
+         m_recursion_checks.assign(1 + m_pdata->m_mark_count, 0u);
 
       const std::pair<bool, re_syntax_base*>& p = v.back();
       m_icase = p.first;
@@ -947,7 +963,7 @@ void basic_regex_creator<charT, traits>::create_startmaps(re_syntax_base* state)
       m_bad_repeats = 0;
 
       if(m_has_recursions)
-         m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);
+         m_recursion_checks.assign(1 + m_pdata->m_mark_count, 0u);
       create_startmap(static_cast<re_alt*>(state)->alt.p, static_cast<re_alt*>(state)->_map, &static_cast<re_alt*>(state)->can_be_null, mask_skip);
       // adjust the type of the state to allow for faster matching:
       state->type = this->get_repeat_type(state);
@@ -1102,11 +1118,9 @@ void basic_regex_creator<charT, traits>::create_startmap(re_syntax_base* state,
       }
       case syntax_element_recurse:
          {
-            if(state->type == syntax_element_startmark)
-               recursion_sub = static_cast<re_brace*>(state)->index;
-            else
-               recursion_sub = 0;
-            if(m_recursion_checks[recursion_sub])
+            BOOST_ASSERT(static_cast<const re_jump*>(state)->alt.p->type == syntax_element_startmark);
+            recursion_sub = static_cast<re_brace*>(static_cast<const re_jump*>(state)->alt.p)->index;
+            if(m_recursion_checks[recursion_sub] & 1u)
             {
                // Infinite recursion!!
                if(0 == this->m_pdata->m_status) // update the error code if not already set
@@ -1131,10 +1145,10 @@ void basic_regex_creator<charT, traits>::create_startmap(re_syntax_base* state,
                recursion_start = state;
                recursion_restart = state->next.p;
                state = static_cast<re_jump*>(state)->alt.p;
-               m_recursion_checks[recursion_sub] = true;
+               m_recursion_checks[recursion_sub] |= 1u;
                break;
             }
-            m_recursion_checks[recursion_sub] = true;
+            m_recursion_checks[recursion_sub] |= 1u;
             // can't handle nested recursion here...
             BOOST_FALLTHROUGH;
          }
@@ -1328,8 +1342,9 @@ void basic_regex_creator<charT, traits>::create_startmap(re_syntax_base* state,
                }
                p = p->next.p;
             }
-            if(ok)
+            if(ok && ((m_recursion_checks[static_cast<re_brace*>(state)->index] & 2u) == 0))
             {
+               m_recursion_checks[static_cast<re_brace*>(state)->index] |= 2u;
                create_startmap(p->next.p, l_map, pnull, mask);
             }
          }
@@ -1419,7 +1434,7 @@ bool basic_regex_creator<charT, traits>::is_bad_repeat(re_syntax_base* pt)
    case syntax_element_long_set_rep:
       {
          unsigned state_id = static_cast<re_repeat*>(pt)->state_id;
-         if(state_id > sizeof(m_bad_repeats) * CHAR_BIT)
+         if(state_id >= sizeof(m_bad_repeats) * CHAR_BIT)
             return true;  // run out of bits, assume we can't traverse this one.
          static const boost::uintmax_t one = 1uL;
          return m_bad_repeats & (one << state_id);

include/boost/regex/v4/basic_regex_parser.hpp

@@ -511,7 +511,8 @@ bool basic_regex_parser<charT, traits>::parse_open_paren()
       this->fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_end));
       return false;
    }
-   BOOST_ASSERT(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_mark);
+   if(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark)
+      return false;
 #ifndef BOOST_NO_STD_DISTANCE
    if(markid && (this->flags() & regbase::save_subexpression_location))
       this->m_pdata->m_subs.at(markid - 1).second = std::distance(m_base, m_position);
@@ -901,7 +902,7 @@ escape_type_class_jump:
          }
          if(negative)
             i = 1 + m_mark_count - i;
-         if(((i > 0) && (this->m_backrefs & (1u << (i-1)))) || ((i > 10000) && (this->m_pdata->get_id(i) > 0) && (this->m_backrefs & (1u << (this->m_pdata->get_id(i)-1)))))
+         if(((i > 0) && (i < std::numeric_limits<unsigned>::digits) && (i - 1 < static_cast<boost::intmax_t>(sizeof(unsigned) * CHAR_BIT)) && (this->m_backrefs & (1u << (i-1)))) || ((i > 10000) && (this->m_pdata->get_id(i) > 0) && (this->m_pdata->get_id(i)-1 < static_cast<boost::intmax_t>(sizeof(unsigned) * CHAR_BIT)) && (this->m_backrefs & (1u << (this->m_pdata->get_id(i)-1)))))
          {
             m_position = pc;
             re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace)));
@@ -2596,7 +2597,7 @@ option_group_jump:
          re_alt* alt = static_cast<re_alt*>(this->insert_state(expected_alt_point, syntax_element_alt, sizeof(re_alt)));
          alt->alt.i = this->m_pdata->m_data.size() - this->getoffset(alt);
       }
-      else if(this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt)
+      else if(((std::ptrdiff_t)this->m_pdata->m_data.size() > (static_cast<re_alt*>(b)->alt.i + this->getoffset(b))) && (static_cast<re_alt*>(b)->alt.i > 0) && this->getaddress(static_cast<re_alt*>(b)->alt.i, b)->type == syntax_element_alt)
       {
          // Can't have seen more than one alternative:
          // Rewind to start of (? sequence:
@@ -2860,6 +2861,10 @@ bool basic_regex_parser<charT, traits>::parse_perl_verb()
       }
       break;
    }
+   // Rewind to start of (* sequence:
+   --m_position;
+   while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;
+   fail(regex_constants::error_perl_extension, m_position - m_base);
    return false;
 }
 

include/boost/regex/v4/match_flags.hpp

@@ -68,7 +68,18 @@ typedef enum _match_flags
    format_no_copy = format_all << 1,                 /* don't copy non-matching segments. */
    format_first_only = format_no_copy << 1,          /* Only replace first occurance. */
    format_is_if = format_first_only << 1,            /* internal use only. */
-   format_literal = format_is_if << 1                /* treat string as a literal */
+   format_literal = format_is_if << 1,               /* treat string as a literal */
+
+   match_not_any = match_not_bol | match_not_eol | match_not_bob 
+      | match_not_eob | match_not_bow | match_not_eow | match_not_dot_newline 
+      | match_not_dot_null | match_prev_avail | match_init | match_not_null
+      | match_continuous | match_partial | match_stop | match_not_initial_null 
+      | match_stop | match_all | match_perl | match_posix | match_nosubs
+      | match_extra | match_single_line | match_unused1 | match_unused2 
+      | match_unused3 | match_max | format_perl | format_default | format_sed
+      | format_all | format_no_copy | format_first_only | format_is_if
+      | format_literal
+
 
 } match_flags;
 

include/boost/regex/v4/perl_matcher.hpp

@@ -161,9 +161,9 @@ iterator BOOST_REGEX_CALL re_is_set_member(iterator next,
       if(*p == static_cast<charT>(0))
       {
          // treat null string as special case:
-         if(traits_inst.translate(*ptr, icase) != *p)
+         if(traits_inst.translate(*ptr, icase))
          {
-            while(*p == static_cast<charT>(0))++p;
+            ++p;
             continue;
          }
          return set_->isnot ? next : (ptr == next) ? ++next : ptr;
@@ -348,6 +348,7 @@ struct recursion_info
    const re_syntax_base* preturn_address;
    Results results;
    repeater_count<iterator>* repeater_stack;
+   iterator location_of_start;
 };
 
 #ifdef BOOST_MSVC

include/boost/regex/v4/perl_matcher_common.hpp

@@ -90,7 +90,7 @@ void perl_matcher<BidiIterator, Allocator, traits>::construct_init(const basic_r
    match_any_mask = static_cast<unsigned char>((f & match_not_dot_newline) ? BOOST_REGEX_DETAIL_NS::test_not_newline : BOOST_REGEX_DETAIL_NS::test_newline);
    // Disable match_any if requested in the state machine:
    if(e.get_data().m_disable_match_any)
-      m_match_flags &= ~regex_constants::match_any;
+      m_match_flags &= regex_constants::match_not_any;
 }
 
 template <class BidiIterator, class Allocator, class traits>

include/boost/regex/v4/perl_matcher_non_recursive.hpp

@@ -131,8 +131,7 @@ template <class Results>
 struct saved_recursion : public saved_state
 {
    saved_recursion(int idx, const re_syntax_base* p, Results* pr) 
-      : saved_state(14), recursion_id(idx), preturn_address(p), results(*pr)
-   {}
+      : saved_state(14), recursion_id(idx), preturn_address(p), results(*pr) {}
    int recursion_id;
    const re_syntax_base* preturn_address;
    Results results;
@@ -405,7 +404,11 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_startmark()
          m_independent = true;
          const re_syntax_base* next_pstate = static_cast<const re_jump*>(pstate->next.p)->alt.p->next.p;
          pstate = pstate->next.p->next.p;
-         bool r = match_all_states();
+         bool r = false;
+#if !defined(BOOST_NO_EXCEPTIONS)
+      try{
+#endif
+         r = match_all_states();
          if(!r && !m_independent)
          {
             // Must be unwinding from a COMMIT/SKIP/PRUNE and the independent 
@@ -413,8 +416,20 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_startmark()
             while(unwind(false));
             return false;
          }
+#if !defined(BOOST_NO_EXCEPTIONS)
+      }
+      catch(...)
+      {
          pstate = next_pstate;
-         m_independent = old_independent;
+         // unwind all pushed states, apart from anything else this
+         // ensures that all the states are correctly destructed
+         // not just the memory freed.
+         while(unwind(true)) {}
+         throw;
+      }
+#endif
+      pstate = next_pstate;
+      m_independent = old_independent;
 #ifdef BOOST_REGEX_MATCH_EXTRA
          if(r && (m_match_flags & match_extra))
          {
@@ -428,8 +443,23 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_startmark()
             for(i = 0; i < temp_match.size(); ++i)
                (*m_presult)[i].get_captures().clear();
             // match everything else:
-            r = match_all_states();
-            // now place the stored captures back:
+#if !defined(BOOST_NO_EXCEPTIONS)
+            try{
+#endif
+               r = match_all_states();
+#if !defined(BOOST_NO_EXCEPTIONS)
+            }
+            catch(...)
+            {
+               pstate = next_pstate;
+               // unwind all pushed states, apart from anything else this
+               // ensures that all the states are correctly destructed
+               // not just the memory freed.
+               while(unwind(true)) {}
+               throw;
+            }
+#endif
+         // now place the stored captures back:
             for(i = 0; i < temp_match.size(); ++i)
             {
                typedef typename sub_match<BidiIterator>::capture_sequence_type seq;
@@ -464,14 +494,29 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_startmark()
          BidiIterator saved_position = position;
          const re_syntax_base* next_pstate = static_cast<const re_jump*>(pstate->next.p)->alt.p->next.p;
          pstate = pstate->next.p->next.p;
-         bool r = match_all_states();
-         position = saved_position;
-         if(negated)
-            r = !r;
-         if(r)
+#if !defined(BOOST_NO_EXCEPTIONS)
+         try{
+#endif
+            bool r = match_all_states();
+            position = saved_position;
+            if(negated)
+               r = !r;
+            if(r)
+               pstate = next_pstate;
+            else
+               pstate = alt->alt.p;
+#if !defined(BOOST_NO_EXCEPTIONS)
+         }
+         catch(...)
+         {
             pstate = next_pstate;
-         else
-            pstate = alt->alt.p;
+            // unwind all pushed states, apart from anything else this
+            // ensures that all the states are correctly destructed
+            // not just the memory freed.
+            while(unwind(true)){}
+            throw;
+         }
+#endif
          break;
       }
       }
@@ -953,6 +998,19 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_recursion()
 {
    BOOST_ASSERT(pstate->type == syntax_element_recurse);
    //
+   // See if we've seen this recursion before at this location, if we have then
+   // we need to prevent infinite recursion:
+   //
+   for(typename std::vector<recursion_info<results_type> >::const_reverse_iterator i = recursion_stack.rbegin(); i != recursion_stack.rend(); ++i)
+   {
+      if(i->idx == static_cast<const re_brace*>(static_cast<const re_jump*>(pstate)->alt.p)->index)
+      {
+         if(i->location_of_start == position)
+            return false;
+         break;
+      }
+   }
+   //
    // Backup call stack:
    //
    push_recursion_pop();
@@ -968,6 +1026,7 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_recursion()
    recursion_stack.back().results = *m_presult;
    pstate = static_cast<const re_jump*>(pstate)->alt.p;
    recursion_stack.back().idx = static_cast<const re_brace*>(pstate)->index;
+   recursion_stack.back().location_of_start = position;
    //if(static_cast<const re_recurse*>(pstate)->state_id > 0)
    {
       push_repeater_count(-(2 + static_cast<const re_brace*>(pstate)->index), &next_count);
@@ -1705,6 +1764,7 @@ bool perl_matcher<BidiIterator, Allocator, traits>::unwind_recursion(bool r)
       recursion_stack.back().idx = pmp->recursion_id;
       recursion_stack.back().preturn_address = pmp->preturn_address;
       recursion_stack.back().results = pmp->results;
+      recursion_stack.back().location_of_start = position;
    }
    boost::BOOST_REGEX_DETAIL_NS::inplace_destroy(pmp++);
    m_backup_state = pmp;

include/boost/regex/v4/perl_matcher_recursive.hpp

@@ -900,10 +900,27 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_recursion()
    {
       recursion_stack.reserve(50);
    }
+   //
+   // See if we've seen this recursion before at this location, if we have then
+   // we need to prevent infinite recursion:
+   //
+   for(typename std::vector<recursion_info<results_type> >::const_reverse_iterator i = recursion_stack.rbegin(); i != recursion_stack.rend(); ++i)
+   {
+      if(i->idx == static_cast<const re_brace*>(static_cast<const re_jump*>(pstate)->alt.p)->index)
+      {
+         if(i->location_of_start == position)
+            return false;
+         break;
+      }
+   }
+   //
+   // Now get on with it:
+   //
    recursion_stack.push_back(recursion_info<results_type>());
    recursion_stack.back().preturn_address = pstate->next.p;
    recursion_stack.back().results = *m_presult;
    recursion_stack.back().repeater_stack = next_count;
+   recursion_stack.back().location_of_start = position;
    pstate = static_cast<const re_jump*>(pstate)->alt.p;
    recursion_stack.back().idx = static_cast<const re_brace*>(pstate)->index;
 
@@ -979,6 +996,7 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_match()
          recursion_stack.push_back(recursion_info<results_type>());
          recursion_stack.back().preturn_address = saved_state;
          recursion_stack.back().results = *m_presult;
+         recursion_stack.back().location_of_start = position;
          return false;
       }
       return true;

include/boost/regex/v4/regex_workaround.hpp

@@ -33,6 +33,7 @@
 #include <algorithm>
 #include <iosfwd>
 #include <vector>
+#include <set>
 #include <map>
 #include <boost/limits.hpp>
 #include <boost/assert.hpp>

test/captures/Jamfile.v2

@@ -26,10 +26,10 @@ EX_SOURCES =
    winstances.cpp 
    usinstances.cpp ;
        
-lib boost_regex_extra : $(EX_SOURCES)
+lib boost_regex_extra : $(EX_SOURCES) ../../build//icu_options
     : 
         <define>BOOST_REGEX_MATCH_EXTRA=1
-	     <link>shared:<define>BOOST_REGEX_DYN_LINK=1
+	<link>shared:<define>BOOST_REGEX_DYN_LINK=1
     : 
     ;
 

test/de_fuzz/Jamfile.v2

@@ -0,0 +1,37 @@
+# copyright John Maddock 2003
+# 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.
+
+import testing ;
+
+lib Fuzzer : : <search>. ;
+
+run narrow.cpp [ glob ../../src/*.cpp ] Fuzzer
+         :  # additional args
+         -dict=dictionary.txt -workers=3 corpus -runs=5000
+         :  # test-files
+         :  # requirements
+            <toolset>clang <cxxflags>-fsanitize-coverage=trace-pc-guard 
+            <cxxflags>-fsanitize=address <cxxflags>-fsanitize=undefined 
+            <cxxflags>-fno-sanitize-recover=undefined <cxxflags>-fno-optimize-sibling-calls 
+            <cxxflags>-fno-omit-frame-pointer
+            <include>../../../..
+            <linkflags>-fsanitize=address <linkflags>-fsanitize=undefined 
+            debug
+;
+
+run wide.cpp [ glob ../../src/*.cpp ] Fuzzer
+         :  # additional args
+         -dict=dictionary.txt -workers=3 corpus -runs=5000
+         :  # test-files
+         :  # requirements
+            <toolset>clang <cxxflags>-fsanitize-coverage=trace-pc-guard 
+            <cxxflags>-fsanitize=address <cxxflags>-fsanitize=undefined 
+            <cxxflags>-fno-sanitize-recover=undefined <cxxflags>-fno-optimize-sibling-calls 
+            <cxxflags>-fno-omit-frame-pointer
+            <include>../../../..
+            <linkflags>-fsanitize=address <linkflags>-fsanitize=undefined 
+            debug
+;
+

test/de_fuzz/corpus/0334845542326f79e3fafbe94969c6b369f004bc

@@ -0,0 +1 @@
+
Z(((((((a+)+)+)+)+)+)+)+|Y(((((((a+)+)+)+)+)+)+)+|X(((((((a+)+)+)+)+)+)+)+|W(((((((a+)+)+)+)+)+)+)+|V(((((((a+)+)+)+)+)+)+)+|CZ(((((((a+)+)+)+)+)+)+)+|CY(((((((a+)+)+)+)+)+)+;+|CX(((((((a+)+)+)+)+)+)+)+|CW(((((((a+)+)+)+)+)+)+)+|CV(((((((a+)+)+)+)+)+)+)+|(a+)+b)

test/de_fuzz/corpus/07284125555baeddc03d1a3637df0ccbcab9d5fb

@@ -0,0 +1 @@
+
Z(((((((a+)+)+)+)+<2B><><EFBFBD><EFBFBD>)+|Y(((((((a+)+)+)+)+)++)+)|X(((((((a+)+)+)+)+)+)+)+|W((<28><>(((a+)+)+)+)+)+)+)+;|V((()++)+)+|CW(((((((a+)+)+)+((((a+)+)+)+)+)+)+)+|CZ(((((((a+)+)+)+)+)+)+)+|CY(((((((a+)+)+)+)+)+)+)+|CX(((((((a+)+)+)+)+)+)+)+|CW(((((((a+)+)+)+)+)+)+)+|CV,((((((a+)+)+)+)+)+)+)+|(a+)+bc

test/de_fuzz/corpus/09ddbffe27633f26351167c38ef10a942862462f

@@ -0,0 +1 @@
+8^\l*(?:((.<2E>*?)\W*(?1)\<5C>*\2[abc]|	|((.)\)W*(?1)\W*\4|\ZZZZZZZZZZZZZZZZZZZZZZ<1B><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\pd\)W*(?1)\W*\4|\ZZZZZZZZZZZZZZZZZZZZZZ<1B><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><1B><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\pd<1B><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><1B><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\pdquick brown )ox