Define BOOST_REGEX_UCHAR_IS_WCHAR_T, use it to disable overloads

Inheritance from std::iterator is deprecated in c++17. Therefore repl…

example/timer/regex_timer.cpp

@@ -67,18 +67,14 @@ ostream& operator << (ostream& os, const std::wstring& s)
 
 template <class S>
 class string_out_iterator 
-#ifndef BOOST_NO_STD_ITERATOR
-  : public std::iterator<std::output_iterator_tag, void, void, void, void>
-#endif // ndef BOOST_NO_STD_ITERATOR
 {
-#ifdef BOOST_NO_STD_ITERATOR
+public:
    typedef std::output_iterator_tag iterator_category;
    typedef void value_type;
    typedef void difference_type;
    typedef void pointer;
    typedef void reference;
-#endif // BOOST_NO_STD_ITERATOR
-
+private:
    S* out;
 public:
    string_out_iterator(S& s) : out(&s) {}

include/boost/regex/icu.hpp

@@ -26,6 +26,7 @@
 #include <boost/regex.hpp>
 #include <boost/regex/pending/unicode_iterator.hpp>
 #include <boost/mpl/int_fwd.hpp>
+#include <boost/static_assert.hpp>
 #include <bitset>
 
 #ifdef BOOST_MSVC
@@ -334,6 +335,34 @@ inline u32regex do_make_u32regex(InputIterator i,
 #endif
 }
 
+// BOOST_REGEX_UCHAR_IS_WCHAR_T
+//
+// Source inspection of unicode/umachine.h in ICU version 59 indicates that:
+//
+// On version 59, UChar is always char16_t in C++ mode (and uint16_t in C mode)
+//
+// On earlier versions, the logic is
+//
+// #if U_SIZEOF_WCHAR_T==2
+//   typedef wchar_t OldUChar;
+// #elif defined(__CHAR16_TYPE__)
+//   typedef __CHAR16_TYPE__ OldUChar;
+// #else
+//   typedef uint16_t OldUChar;
+// #endif
+//
+// That is, UChar is wchar_t only on versions below 59, when U_SIZEOF_WCHAR_T==2
+//
+// Hence,
+
+#define BOOST_REGEX_UCHAR_IS_WCHAR_T (U_ICU_VERSION_MAJOR_NUM < 59 && U_SIZEOF_WCHAR_T == 2)
+
+#if BOOST_REGEX_UCHAR_IS_WCHAR_T
+  BOOST_STATIC_ASSERT((boost::is_same<UChar, wchar_t>::value));
+#else
+  BOOST_STATIC_ASSERT(!(boost::is_same<UChar, wchar_t>::value));
+#endif
+
 //
 // Construction from an iterator pair:
 //
@@ -364,7 +393,7 @@ inline u32regex make_u32regex(const wchar_t* p, boost::regex_constants::syntax_o
    return BOOST_REGEX_DETAIL_NS::do_make_u32regex(p, p + std::wcslen(p), opt, static_cast<boost::mpl::int_<sizeof(wchar_t)> const*>(0));
 }
 #endif
-#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2)
+#if !BOOST_REGEX_UCHAR_IS_WCHAR_T
 inline u32regex make_u32regex(const UChar* p, boost::regex_constants::syntax_option_type opt = boost::regex_constants::perl)
 {
    return BOOST_REGEX_DETAIL_NS::do_make_u32regex(p, p + u_strlen(p), opt, static_cast<boost::mpl::int_<2> const*>(0));
@@ -481,7 +510,7 @@ inline bool u32regex_match(const UChar* p,
 {
    return BOOST_REGEX_DETAIL_NS::do_regex_match(p, p+u_strlen(p), m, e, flags, static_cast<mpl::int_<2> const*>(0));
 }
-#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX)
+#if !BOOST_REGEX_UCHAR_IS_WCHAR_T && !defined(BOOST_NO_WREGEX)
 inline bool u32regex_match(const wchar_t* p, 
                  match_results<const wchar_t*>& m, 
                  const u32regex& e, 
@@ -545,7 +574,7 @@ inline bool u32regex_match(const UChar* p,
    match_results<const UChar*> m;
    return BOOST_REGEX_DETAIL_NS::do_regex_match(p, p+u_strlen(p), m, e, flags, static_cast<mpl::int_<2> const*>(0));
 }
-#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX)
+#if !BOOST_REGEX_UCHAR_IS_WCHAR_T && !defined(BOOST_NO_WREGEX)
 inline bool u32regex_match(const wchar_t* p, 
                  const u32regex& e, 
                  match_flag_type flags = match_default)
@@ -666,7 +695,7 @@ inline bool u32regex_search(const UChar* p,
 {
    return BOOST_REGEX_DETAIL_NS::do_regex_search(p, p+u_strlen(p), m, e, flags, p, static_cast<mpl::int_<2> const*>(0));
 }
-#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX)
+#if !BOOST_REGEX_UCHAR_IS_WCHAR_T && !defined(BOOST_NO_WREGEX)
 inline bool u32regex_search(const wchar_t* p, 
                  match_results<const wchar_t*>& m, 
                  const u32regex& e, 
@@ -727,7 +756,7 @@ inline bool u32regex_search(const UChar* p,
    match_results<const UChar*> m;
    return BOOST_REGEX_DETAIL_NS::do_regex_search(p, p+u_strlen(p), m, e, flags, p, static_cast<mpl::int_<2> const*>(0));
 }
-#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX)
+#if !BOOST_REGEX_UCHAR_IS_WCHAR_T && !defined(BOOST_NO_WREGEX)
 inline bool u32regex_search(const wchar_t* p, 
                  const u32regex& e, 
                  match_flag_type flags = match_default)

include/boost/regex/v4/basic_regex_parser.hpp

@@ -183,7 +183,7 @@ void basic_regex_parser<charT, traits>::parse(const charT* p1, const charT* p2,
    // have had an unexpected ')' :
    if(!result)
    {
-      fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Found a closing ) with no corresponding openening parenthesis.");
+      fail(regex_constants::error_paren, ::boost::BOOST_REGEX_DETAIL_NS::distance(m_base, m_position), "Found a closing ) with no corresponding opening parenthesis.");
       return;
    }
    // if an error has been set then give up now:

include/boost/regex/v4/fileiter.hpp

@@ -195,9 +195,6 @@ public:
 };
 
 class BOOST_REGEX_DECL mapfile_iterator
-#if !defined(BOOST_NO_STD_ITERATOR) || defined(BOOST_MSVC_STD_ITERATOR)
-: public std::iterator<std::random_access_iterator_tag, char>
-#endif
 {
    typedef mapfile::pointer internal_pointer;
    internal_pointer* node;

include/boost/regex/v4/match_results.hpp

@@ -56,7 +56,9 @@ private:
 #endif
 public: 
    typedef          sub_match<BidiIterator>                         value_type;
-#if  !defined(BOOST_NO_STD_ALLOCATOR) && !(defined(BOOST_MSVC) && defined(_STLPORT_VERSION))
+#ifndef BOOST_NO_CXX11_ALLOCATOR
+   typedef typename std::allocator_traits<Allocator>::value_type const &    const_reference;
+#elif  !defined(BOOST_NO_STD_ALLOCATOR) && !(defined(BOOST_MSVC) && defined(_STLPORT_VERSION))
    typedef typename Allocator::const_reference                              const_reference;
 #else
    typedef          const value_type&                                       const_reference;
@@ -66,7 +68,11 @@ public:
    typedef          const_iterator                                          iterator;
    typedef typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<
                                     BidiIterator>::difference_type          difference_type;
+#ifdef BOOST_NO_CXX11_ALLOCATOR
    typedef typename Allocator::size_type                                    size_type;
+#else
+   typedef typename std::allocator_traits<Allocator>::size_type             size_type;
+#endif
    typedef          Allocator                                               allocator_type;
    typedef typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<
                                     BidiIterator>::value_type               char_type;

include/boost/regex/v4/regex_format.hpp

@@ -800,9 +800,6 @@ void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::put(co
 
 template <class S>
 class string_out_iterator
-#ifndef BOOST_NO_STD_ITERATOR
-   : public std::iterator<std::output_iterator_tag, typename S::value_type>
-#endif
 {
    S* out;
 public:
@@ -816,13 +813,11 @@ public:
       return *this; 
    }
 
-#ifdef BOOST_NO_STD_ITERATOR
    typedef std::ptrdiff_t difference_type;
    typedef typename S::value_type value_type;
    typedef value_type* pointer;
    typedef value_type& reference;
    typedef std::output_iterator_tag iterator_category;
-#endif
 };
 
 template <class OutputIterator, class Iterator, class Alloc, class ForwardIter, class traits>

include/boost/regex/v4/regex_iterator.hpp

@@ -85,14 +85,6 @@ template <class BidirectionalIterator,
           class charT = BOOST_DEDUCED_TYPENAME BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::value_type,
           class traits = regex_traits<charT> >
 class regex_iterator 
-#ifndef BOOST_NO_STD_ITERATOR
-   : public std::iterator<
-         std::forward_iterator_tag, 
-         match_results<BidirectionalIterator>,
-         typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type,
-         const match_results<BidirectionalIterator>*,
-         const match_results<BidirectionalIterator>& >         
-#endif
 {
 private:
    typedef regex_iterator_implementation<BidirectionalIterator, charT, traits> impl;

include/boost/regex/v4/regex_token_iterator.hpp

@@ -167,14 +167,6 @@ template <class BidirectionalIterator,
           class charT = BOOST_DEDUCED_TYPENAME BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::value_type,
           class traits = regex_traits<charT> >
 class regex_token_iterator 
-#ifndef BOOST_NO_STD_ITERATOR
-   : public std::iterator<
-         std::forward_iterator_tag, 
-         sub_match<BidirectionalIterator>,
-         typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type,
-         const sub_match<BidirectionalIterator>*,
-         const sub_match<BidirectionalIterator>& >         
-#endif
 {
 private:
    typedef regex_token_iterator_implementation<BidirectionalIterator, charT, traits> impl;

include/boost/regex/v4/u32regex_iterator.hpp

@@ -72,14 +72,6 @@ private:
 
 template <class BidirectionalIterator>
 class u32regex_iterator 
-#ifndef BOOST_NO_STD_ITERATOR
-   : public std::iterator<
-         std::forward_iterator_tag, 
-         match_results<BidirectionalIterator>,
-         typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type,
-         const match_results<BidirectionalIterator>*,
-         const match_results<BidirectionalIterator>& >         
-#endif
 {
 private:
    typedef u32regex_iterator_implementation<BidirectionalIterator> impl;

include/boost/regex/v4/u32regex_token_iterator.hpp

@@ -155,14 +155,6 @@ private:
 
 template <class BidirectionalIterator>
 class u32regex_token_iterator 
-#ifndef BOOST_NO_STD_ITERATOR
-   : public std::iterator<
-         std::forward_iterator_tag, 
-         sub_match<BidirectionalIterator>,
-         typename BOOST_REGEX_DETAIL_NS::regex_iterator_traits<BidirectionalIterator>::difference_type,
-         const sub_match<BidirectionalIterator>*,
-         const sub_match<BidirectionalIterator>& >         
-#endif
 {
 private:
    typedef u32regex_token_iterator_implementation<BidirectionalIterator> impl;

src/icu.cpp

@@ -481,7 +481,7 @@ icu_regex_traits::string_type icu_regex_traits::lookup_collatename(const char_ty
 bool icu_regex_traits::isctype(char_type c, char_class_type f) const
 {
    // check for standard catagories first:
-   char_class_type m = char_class_type(1u << u_charType(c));
+   char_class_type m = char_class_type(1uLL << u_charType(c));
    if((m & f) != 0) 
       return true;
    // now check for special cases:

test/Jamfile.v2

@@ -194,4 +194,5 @@ compile test_consolidated.cpp ;
 
 build-project ../example ;
 
-
+# `quick` target (for CI)
+run quick.cpp ../build//boost_regex ;

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																		

test/de_fuzz/corpus/0b7e65a2f97ea6135e0c46e75494a6dc0302192a

@@ -0,0 +1,73 @@
+'<27>  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*                          # optional leading comment
+(?:    (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?:                      # opening quote...
+[^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
+|                     #    or
+\\ [^\x80-\xff]           #   Escaped something (something != CR)
+)* "  # closing quote
+)                    # initial word
+(?:  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*  \.  (?: [\040\t] |  \(\x80-\xff]+    # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?:                      # opening quote...
+[^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
+|                     #    or
+\\ [^\x80-\xff]           #   Escaped something (something != CR)
+)* "  # closing quote
+)                    # initial word
+(?:  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*  \.  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*   (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|
+" (?:                      # opening quote...
+[^\\\x80-\xff\n\015"]                #   Anything except backslash and quote
+|                     #    or
+\\ [^\x80-\xff]           #   Escaped something (something != CR)
+)* "  # closing quote
+)  )* # further okay, if led by a period
+(?: (\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*  @  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*    (?:
+[^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|   \[                         # [
+(?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
+\]                        #           ]
+)                           # initial subdomain
+(?:                                  #
+(?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*  \.                        # if led by a period...
+(?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*   (?:
+[^(\04<30>)<>@,;:".\\\[\]\000-\037\x80-\xff]+    # some number of atom characters...
+(?![^(\040)<>@,;:".\\\[\]\000-\037\x80-\xff]) # ..not followed by something that could be part of an atom
+|   \[                         # [
+(?: [^\\\x80-\xff\n\015\[\]] |  \\ [^\x80-\xff]  )*    #    stuff
+\]                        #           ]
+)                     #   ...further okay
+)*
+#       address spec
+(?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  ^  | \\[\x80-\xff]  )* \)  )*
+\)  )*  > #                  trailing >
+# name and address
+)  (?: [\040\t] |  \(
+(?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  |  \( (?:  [^\\\x80-\xff\n\015()]  |  \\ [^\x80-\xff]  )* \)  )*
+\)  )*                       # optional trailing comment
+
+<user@dom.ain>