Fix for bug in recursive expressions, see http://lists.boost.org/Archives/boost/2015/03/221018.php

Author: jzmaddock

include/boost/regex/v4/perl_matcher.hpp

@@ -253,23 +253,40 @@ class repeater_count
    int state_id;
    std::size_t count;        // the number of iterations so far
    BidiIterator start_pos;   // where the last repeat started
+
+   repeater_count* unwind_until(int n, repeater_count* p, int current_recursion_id)
+   { 
+      while(p && (p->state_id != n))
+      {
+         if(-2 - current_recursion_id == p->state_id)
+            return 0;
+         p = p->next;
+         if(p && (p->state_id < 0))
+         {
+            p = unwind_until(p->state_id, p, current_recursion_id);
+            if(!p)
+               return p;
+            p = p->next;
+         }
+      }
+      return p;
+   }
 public:
    repeater_count(repeater_count** s) : stack(s), next(0), state_id(-1), count(0), start_pos() {}
-
-   repeater_count(int i, repeater_count** s, BidiIterator start)
+   
+   repeater_count(int i, repeater_count** s, BidiIterator start, int current_recursion_id)
       : start_pos(start)
    {
       state_id = i;
       stack = s;
       next = *stack;
       *stack = this;
-      if(state_id > next->state_id)
+      if((state_id > next->state_id) && (next->state_id >= 0))
          count = 0;
       else
       {
          repeater_count* p = next;
-         while(p && (p->state_id != state_id))
-            p = p->next;
+         p = unwind_until(state_id, p, current_recursion_id);
          if(p)
          {
             count = p->count;

include/boost/regex/v4/perl_matcher_non_recursive.hpp

@@ -87,8 +87,8 @@ template <class BidiIterator>
 struct saved_repeater : public saved_state
 {
    repeater_count<BidiIterator> count;
-   saved_repeater(int i, repeater_count<BidiIterator>** s, BidiIterator start) 
-      : saved_state(saved_state_repeater_count), count(i,s,start){}
+   saved_repeater(int i, repeater_count<BidiIterator>** s, BidiIterator start, int current_recursion_id)
+      : saved_state(saved_state_repeater_count), count(i, s, start, current_recursion_id){}
 };
 
 struct saved_extra_block : public saved_state
@@ -309,7 +309,7 @@ inline void perl_matcher<BidiIterator, Allocator, traits>::push_repeater_count(i
       pmp = static_cast<saved_repeater<BidiIterator>*>(m_backup_state);
       --pmp;
    }
-   (void) new (pmp)saved_repeater<BidiIterator>(i, s, position);
+   (void) new (pmp)saved_repeater<BidiIterator>(i, s, position, this->recursion_stack.size() ? this->recursion_stack.back().idx : (INT_MIN + 3));
    m_backup_state = pmp;
 }
 
@@ -923,12 +923,12 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_recursion()
    recursion_stack.push_back(recursion_info<results_type>());
    recursion_stack.back().preturn_address = pstate->next.p;
    recursion_stack.back().results = *m_presult;
-   if(static_cast<const re_recurse*>(pstate)->state_id > 0)
-   {
-      push_repeater_count(static_cast<const re_recurse*>(pstate)->state_id, &next_count);
-   }
    pstate = static_cast<const re_jump*>(pstate)->alt.p;
    recursion_stack.back().idx = static_cast<const re_brace*>(pstate)->index;
+   //if(static_cast<const re_recurse*>(pstate)->state_id > 0)
+   {
+      push_repeater_count(-(2 + static_cast<const re_brace*>(pstate)->index), &next_count);
+   }
 
    return true;
 }
@@ -952,6 +952,7 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_endmark()
             *m_presult = recursion_stack.back().results;
             push_recursion(recursion_stack.back().idx, recursion_stack.back().preturn_address, &recursion_stack.back().results);
             recursion_stack.pop_back();
+            push_repeater_count(-(2 + index), &next_count);
          }
       }
    }

include/boost/regex/v4/perl_matcher_recursive.hpp

@@ -308,7 +308,7 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_rep()
    // Always copy the repeat count, so that the state is restored
    // when we exit this scope:
    //
-   repeater_count<BidiIterator> r(rep->state_id, &next_count, position);
+   repeater_count<BidiIterator> r(rep->state_id, &next_count, position, this->recursion_stack.size() ? this->recursion_stack.back().idx : INT_MIN + 3);
    //
    // If we've had at least one repeat already, and the last one 
    // matched the NULL string then set the repeat count to

test/regress/test_perl_ex.cpp

@@ -918,6 +918,7 @@ void test_recursion()
    TEST_REGEX_SEARCH("namespace\\s+(\\w+)\\s+(\\{(?:[^{}]*(?:(?2)[^{}]*)*)?\\})", perl, "namespace one { namespace two { int foo(); } }", match_default, make_array(0, 46, 10, 13, 14, 46, -2, -2));
    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));
 
    // 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));