Merge branch 'develop'

Fix infinite loop in parser.

build/Jamfile.v2

@@ -57,7 +57,7 @@ if ! $(disable-icu)
 
    if $(ICU_ICUUC_NAME)
    {
-      lib icuuc : : <name>$(ICU_ICUUC_NAME) ;
+      lib icuuc : : <name>$(ICU_ICUUC_NAME) <conditional>@path_options ;
    }
    else
    {
@@ -71,7 +71,7 @@ if ! $(disable-icu)
    }
    if $(ICU_ICUDT_NAME)
    {
-      lib icudt : : <name>$(ICU_ICUDT_NAME) ;
+      lib icudt : : <name>$(ICU_ICUDT_NAME) <conditional>@path_options ;
    }
    else
    {
@@ -85,7 +85,7 @@ if ! $(disable-icu)
    }
    if $(ICU_ICUIN_NAME)
    {
-      lib icuin : : <name>$(ICU_ICUIN_NAME) ;
+      lib icuin : : <name>$(ICU_ICUIN_NAME) <conditional>@path_options ;
    }
    else
    {
@@ -110,10 +110,33 @@ if ! $(disable-icu)
       <runtime-link>static:<library>icuuc
       <runtime-link>static:<library>icudt
       <runtime-link>static:<library>icuin
+      <target-os>windows,<toolset>clang:<linkflags>"advapi32.lib"
       <define>BOOST_HAS_ICU=1 
       <runtime-link>static:<define>U_STATIC_IMPLEMENTATION=1
       ;
 
+   if [ modules.peek : ICU_DATA_DIR ]
+   {
+       rule data-dir-options ( properties * )
+       {
+          local result ;
+          local data_dir = [ modules.peek : ICU_DATA_DIR ] ;
+          if <toolset>emscripten in $(properties)
+          {
+             result = <define>ICU_DATA_DIR=\\\"/$(data_dir:BS)\\\"
+                      "<linkflags>--embed-file $(data_dir)@/$(data_dir:BS)"
+                      ;
+          }
+          else
+          {
+             result = <define>ICU_DATA_DIR=\\\"$(data_dir)\\\" ;
+          }
+          return $(result) ;
+       }
+
+       ICU_OPTS += <conditional>@data-dir-options ;
+   }
+
 }
 
 exe has_icu : has_icu_test.cpp : $(ICU_OPTS) ;

build/has_icu_test.cpp

@@ -28,6 +28,10 @@ void print_error(UErrorCode err, const char* func)
 
 int main()
 {
+#ifdef ICU_DATA_DIR
+   ::u_setDataDirectory(ICU_DATA_DIR);
+#endif
+
    // To detect possible binary mismatches between the installed ICU build, and whatever
    // C++ std lib's we're using, we need to:
    // * Make sure we call ICU C++ API's

include/boost/regex/concepts.hpp

@@ -29,7 +29,7 @@
 #endif
 #include <bitset>
 #include <vector>
-#include <iostream>
+#include <ostream>
 
 #ifdef BOOST_REGEX_CXX03
 #define RW_NS boost

include/boost/regex/v4/basic_regex_parser.hpp

@@ -1124,6 +1124,9 @@ bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_
                   }
                   else
                      contin = false;
+                  break;
+               default:
+                  contin = false;
                }
             }
             else

include/boost/regex/v4/unicode_iterator.hpp

@@ -62,7 +62,6 @@ Accepts UTF-32 code points and forwards them on as UTF-16 code points.
 #define BOOST_REGEX_V4_UNICODE_ITERATOR_HPP
 #include <boost/cstdint.hpp>
 #include <boost/regex/config.hpp>
-#include <boost/iterator/iterator_facade.hpp>
 #include <boost/static_assert.hpp>
 #include <boost/throw_exception.hpp>
 #include <stdexcept>
@@ -145,10 +144,7 @@ inline void invalid_utf32_code_point(::boost::uint32_t val)
 
 template <class BaseIterator, class U16Type = ::boost::uint16_t>
 class u32_to_u16_iterator
-   : public boost::iterator_facade<u32_to_u16_iterator<BaseIterator, U16Type>, U16Type, std::bidirectional_iterator_tag, const U16Type>
 {
-   typedef boost::iterator_facade<u32_to_u16_iterator<BaseIterator, U16Type>, U16Type, std::bidirectional_iterator_tag, const U16Type> base_type;
-
 #if !defined(BOOST_NO_STD_ITERATOR_TRAITS)
    typedef typename std::iterator_traits<BaseIterator>::value_type base_value_type;
 
@@ -157,14 +153,19 @@ class u32_to_u16_iterator
 #endif
 
 public:
-   typename base_type::reference
-      dereference()const
+   typedef std::ptrdiff_t     difference_type;
+   typedef U16Type            value_type;
+   typedef value_type const*  pointer;
+   typedef value_type const   reference;
+   typedef std::bidirectional_iterator_tag iterator_category;
+
+   reference operator*()const
    {
       if(m_current == 2)
          extract_current();
       return m_values[m_current];
    }
-   bool equal(const u32_to_u16_iterator& that)const
+   bool operator==(const u32_to_u16_iterator& that)const
    {
       if(m_position == that.m_position)
       {
@@ -174,7 +175,11 @@ public:
       }
       return false;
    }
-   void increment()
+   bool operator!=(const u32_to_u16_iterator& that)const
+   {
+      return !(*this == that);
+   }
+   u32_to_u16_iterator& operator++()
    {
       // if we have a pending read then read now, so that we know whether
       // to skip a position, or move to a low-surrogate:
@@ -191,8 +196,15 @@ public:
          m_current = 2;
          ++m_position;
       }
+      return *this;
    }
-   void decrement()
+   u32_to_u16_iterator operator++(int)
+   {
+      u32_to_u16_iterator r(*this);
+      ++(*this);
+      return r;
+   }
+   u32_to_u16_iterator& operator--()
    {
       if(m_current != 1)
       {
@@ -205,6 +217,13 @@ public:
       {
          m_current = 0;
       }
+      return *this;
+   }
+   u32_to_u16_iterator operator--(int)
+   {
+      u32_to_u16_iterator r(*this);
+      --(*this);
+      return r;
    }
    BaseIterator base()const
    {
@@ -258,9 +277,7 @@ private:
 
 template <class BaseIterator, class U32Type = ::boost::uint32_t>
 class u16_to_u32_iterator
-   : public boost::iterator_facade<u16_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type>
 {
-   typedef boost::iterator_facade<u16_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type> base_type;
    // special values for pending iterator reads:
    BOOST_STATIC_CONSTANT(U32Type, pending_read = 0xffffffffu);
 
@@ -272,31 +289,54 @@ class u16_to_u32_iterator
 #endif
 
 public:
-   typename base_type::reference
-      dereference()const
+   typedef std::ptrdiff_t     difference_type;
+   typedef U32Type            value_type;
+   typedef value_type const*  pointer;
+   typedef value_type const   reference;
+   typedef std::bidirectional_iterator_tag iterator_category;
+
+   reference operator*()const
    {
       if(m_value == pending_read)
          extract_current();
       return m_value;
    }
-   bool equal(const u16_to_u32_iterator& that)const
+   bool operator==(const u16_to_u32_iterator& that)const
    {
       return m_position == that.m_position;
    }
-   void increment()
+   bool operator!=(const u16_to_u32_iterator& that)const
+   {
+      return !(*this == that);
+   }
+   u16_to_u32_iterator& operator++()
    {
       // skip high surrogate first if there is one:
       if(detail::is_high_surrogate(*m_position)) ++m_position;
       ++m_position;
       m_value = pending_read;
+      return *this;
    }
-   void decrement()
+   u16_to_u32_iterator operator++(int)
+   {
+      u16_to_u32_iterator r(*this);
+      ++(*this);
+      return r;
+   }
+   u16_to_u32_iterator& operator--()
    {
       --m_position;
       // if we have a low surrogate then go back one more:
       if(detail::is_low_surrogate(*m_position)) 
          --m_position;
       m_value = pending_read;
+      return *this;
+   }
+   u16_to_u32_iterator operator--(int)
+   {
+      u16_to_u32_iterator r(*this);
+      --(*this);
+      return r;
    }
    BaseIterator base()const
    {
@@ -375,10 +415,7 @@ private:
 
 template <class BaseIterator, class U8Type = ::boost::uint8_t>
 class u32_to_u8_iterator
-   : public boost::iterator_facade<u32_to_u8_iterator<BaseIterator, U8Type>, U8Type, std::bidirectional_iterator_tag, const U8Type>
 {
-   typedef boost::iterator_facade<u32_to_u8_iterator<BaseIterator, U8Type>, U8Type, std::bidirectional_iterator_tag, const U8Type> base_type;
-   
 #if !defined(BOOST_NO_STD_ITERATOR_TRAITS)
    typedef typename std::iterator_traits<BaseIterator>::value_type base_value_type;
 
@@ -387,14 +424,19 @@ class u32_to_u8_iterator
 #endif
 
 public:
-   typename base_type::reference
-      dereference()const
+   typedef std::ptrdiff_t     difference_type;
+   typedef U8Type             value_type;
+   typedef value_type const*  pointer;
+   typedef value_type const   reference;
+   typedef std::bidirectional_iterator_tag iterator_category;
+
+   reference operator*()const
    {
       if(m_current == 4)
          extract_current();
       return m_values[m_current];
    }
-   bool equal(const u32_to_u8_iterator& that)const
+   bool operator==(const u32_to_u8_iterator& that)const
    {
       if(m_position == that.m_position)
       {
@@ -405,7 +447,11 @@ public:
       }
       return false;
    }
-   void increment()
+   bool operator!=(const u32_to_u8_iterator& that)const
+   {
+      return !(*this == that);
+   }
+   u32_to_u8_iterator& operator++()
    {
       // if we have a pending read then read now, so that we know whether
       // to skip a position, or move to a low-surrogate:
@@ -422,8 +468,15 @@ public:
          m_current = 4;
          ++m_position;
       }
+      return *this;
    }
-   void decrement()
+   u32_to_u8_iterator operator++(int)
+   {
+      u32_to_u8_iterator r(*this);
+      ++(*this);
+      return r;
+   }
+   u32_to_u8_iterator& operator--()
    {
       if((m_current & 3) == 0)
       {
@@ -435,6 +488,13 @@ public:
       }
       else
          --m_current;
+      return *this;
+   }
+   u32_to_u8_iterator operator--(int)
+   {
+      u32_to_u8_iterator r(*this);
+      --(*this);
+      return r;
    }
    BaseIterator base()const
    {
@@ -501,9 +561,7 @@ private:
 
 template <class BaseIterator, class U32Type = ::boost::uint32_t>
 class u8_to_u32_iterator
-   : public boost::iterator_facade<u8_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type>
 {
-   typedef boost::iterator_facade<u8_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type> base_type;
    // special values for pending iterator reads:
    BOOST_STATIC_CONSTANT(U32Type, pending_read = 0xffffffffu);
 
@@ -515,18 +573,27 @@ class u8_to_u32_iterator
 #endif
 
 public:
-   typename base_type::reference
-      dereference()const
+   typedef std::ptrdiff_t     difference_type;
+   typedef U32Type            value_type;
+   typedef value_type const*  pointer;
+   typedef value_type const   reference;
+   typedef std::bidirectional_iterator_tag iterator_category;
+
+   reference operator*()const
    {
       if(m_value == pending_read)
          extract_current();
       return m_value;
    }
-   bool equal(const u8_to_u32_iterator& that)const
+   bool operator==(const u8_to_u32_iterator& that)const
    {
       return m_position == that.m_position;
    }
-   void increment()
+   bool operator!=(const u8_to_u32_iterator& that)const
+   {
+      return !(*this == that);
+   }
+   u8_to_u32_iterator& operator++()
    {
       // We must not start with a continuation character:
       if((static_cast<boost::uint8_t>(*m_position) & 0xC0) == 0x80)
@@ -549,8 +616,15 @@ public:
          std::advance(m_position, c);
       }
       m_value = pending_read;
+      return *this;
    }
-   void decrement()
+   u8_to_u32_iterator operator++(int)
+   {
+      u8_to_u32_iterator r(*this);
+      ++(*this);
+      return r;
+   }
+   u8_to_u32_iterator& operator--()
    {
       // Keep backtracking until we don't have a trailing character:
       unsigned count = 0;
@@ -559,6 +633,13 @@ public:
       if(count != detail::utf8_trailing_byte_count(*m_position))
          invalid_sequence();
       m_value = pending_read;
+      return *this;
+   }
+   u8_to_u32_iterator operator--(int)
+   {
+      u8_to_u32_iterator r(*this);
+      --(*this);
+      return r;
    }
    BaseIterator base()const
    {

include/boost/regex/v5/basic_regex_parser.hpp

@@ -1098,6 +1098,9 @@ bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_
                   }
                   else
                      contin = false;
+                  break;
+               default:
+                  contin = false;
                }
             }
             else

include/boost/regex/v5/mem_block_cache.hpp

@@ -85,10 +85,10 @@ struct mem_block_node
 struct mem_block_cache
 {
    // this member has to be statically initialsed:
-   mem_block_node* next;
-   unsigned cached_blocks;
+   mem_block_node* next { nullptr };
+   unsigned cached_blocks { 0 };
 #ifdef BOOST_HAS_THREADS
-   boost::static_mutex mut;
+   std::mutex mut;
 #endif
 
    ~mem_block_cache()
@@ -133,11 +133,7 @@ struct mem_block_cache
    }
    static mem_block_cache& instance()
    {
-#ifdef BOOST_HAS_THREADS
-      static mem_block_cache block_cache = { 0, 0, BOOST_STATIC_MUTEX_INIT, };
-#else
-      static mem_block_cache block_cache = { 0, 0, };
-#endif
+      static mem_block_cache block_cache;
       return block_cache;
    }
 };

include/boost/regex/v5/regex_traits_defaults.hpp

@@ -29,6 +29,7 @@
 #include <cctype>
 #include <locale>
 #include <cwctype>
+#include <limits>
 
 namespace boost{ namespace BOOST_REGEX_DETAIL_NS{
 

include/boost/regex/v5/unicode_iterator.hpp

@@ -67,8 +67,6 @@ Accepts UTF-32 code points and forwards them on as UTF-16 code points.
 #include <ios>
 #include <limits.h> // CHAR_BIT
 
-#include <iostream>
-
 #ifndef BOOST_REGEX_STANDALONE
 #include <boost/throw_exception.hpp>
 #endif

include/boost/regex/v5/w32_regex_traits.hpp

@@ -29,8 +29,12 @@
 #include <boost/regex/v5/primary_transform.hpp>
 #include <boost/regex/v5/object_cache.hpp>
 
-#define VC_EXTRALEAN
-#define WIN32_LEAN_AND_MEAN
+#ifndef VC_EXTRALEAN
+#  define VC_EXTRALEAN
+#endif
+#ifndef WIN32_LEAN_AND_MEAN
+#  define WIN32_LEAN_AND_MEAN
+#endif
 #include <windows.h>
 
 #if defined(_MSC_VER) && !defined(_WIN32_WCE) && !defined(UNDER_CE)

test/regress/test_perl_ex.cpp

@@ -58,6 +58,8 @@ void test_independent_subs()
    TEST_REGEX_SEARCH("word (?>[a-zA-Z0-9]+ ){0,30}otherword", perl, "word cat dog elephant mussel cow horse canary baboon snake shark the quick brown fox and the lazy dog and several other words getting close to thirty by now I really really hope otherword", match_default, make_array(-2, -2));
    TEST_REGEX_SEARCH("((?>Z)+|A)+", perl, "ZABCDEFG", match_default, make_array(0, 2, 1, 2, -2, -2));
    TEST_INVALID_REGEX("((?>)+|A)+", perl);
+   // Bug https://github.com/boostorg/regex/issues/140:
+   TEST_REGEX_SEARCH("1?+(?#)1", perl, "22113", match_default, make_array(2, 4, -2, -2));
 }
 
 void test_conditionals()