From b51965dddbed68f23c5e8c169c23c794313ce5f6 Mon Sep 17 00:00:00 2001 From: Boris Kolpackov Date: Tue, 28 Jun 2011 17:17:23 +0200 Subject: Add boost subset as an implementation detail --- cutl/details/boost/regex/icu.hpp | 1029 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 1029 insertions(+) create mode 100644 cutl/details/boost/regex/icu.hpp (limited to 'cutl/details/boost/regex/icu.hpp') diff --git a/cutl/details/boost/regex/icu.hpp b/cutl/details/boost/regex/icu.hpp new file mode 100644 index 0000000..02b6085 --- /dev/null +++ b/cutl/details/boost/regex/icu.hpp @@ -0,0 +1,1029 @@ +/* + * + * Copyright (c) 2004 + * John Maddock + * + * Use, modification and distribution are subject to 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) + * + */ + + /* + * LOCATION: see http://www.boost.org for most recent version. + * FILE icu.hpp + * VERSION see + * DESCRIPTION: Unicode regular expressions on top of the ICU Library. + */ + +#ifndef BOOST_REGEX_ICU_HPP +#define BOOST_REGEX_ICU_HPP + +#include +#include +#include +#include +#include +#include +#include + +#ifdef BOOST_MSVC +#pragma warning (push) +#pragma warning (disable: 4251) +#endif + +namespace cutl_details_boost{ + +namespace re_detail{ + +// +// Implementation details: +// +class BOOST_REGEX_DECL icu_regex_traits_implementation +{ + typedef UChar32 char_type; + typedef std::size_t size_type; + typedef std::vector string_type; + typedef U_NAMESPACE_QUALIFIER Locale locale_type; + typedef cutl_details_boost::uint_least32_t char_class_type; +public: + icu_regex_traits_implementation(const U_NAMESPACE_QUALIFIER Locale& l) + : m_locale(l) + { + UErrorCode success = U_ZERO_ERROR; + m_collator.reset(U_NAMESPACE_QUALIFIER Collator::createInstance(l, success)); + if(U_SUCCESS(success) == 0) + init_error(); + m_collator->setStrength(U_NAMESPACE_QUALIFIER Collator::IDENTICAL); + success = U_ZERO_ERROR; + m_primary_collator.reset(U_NAMESPACE_QUALIFIER Collator::createInstance(l, success)); + if(U_SUCCESS(success) == 0) + init_error(); + m_primary_collator->setStrength(U_NAMESPACE_QUALIFIER Collator::PRIMARY); + } + U_NAMESPACE_QUALIFIER Locale getloc()const + { + return m_locale; + } + string_type do_transform(const char_type* p1, const char_type* p2, const U_NAMESPACE_QUALIFIER Collator* pcoll) const; + string_type transform(const char_type* p1, const char_type* p2) const + { + return do_transform(p1, p2, m_collator.get()); + } + string_type transform_primary(const char_type* p1, const char_type* p2) const + { + return do_transform(p1, p2, m_primary_collator.get()); + } +private: + void init_error() + { + std::runtime_error e("Could not initialize ICU resources"); + cutl_details_boost::throw_exception(e); + } + U_NAMESPACE_QUALIFIER Locale m_locale; // The ICU locale that we're using + cutl_details_boost::scoped_ptr< U_NAMESPACE_QUALIFIER Collator> m_collator; // The full collation object + cutl_details_boost::scoped_ptr< U_NAMESPACE_QUALIFIER Collator> m_primary_collator; // The primary collation object +}; + +inline cutl_details_boost::shared_ptr get_icu_regex_traits_implementation(const U_NAMESPACE_QUALIFIER Locale& loc) +{ + return cutl_details_boost::shared_ptr(new icu_regex_traits_implementation(loc)); +} + +} + +class BOOST_REGEX_DECL icu_regex_traits +{ +public: + typedef UChar32 char_type; + typedef std::size_t size_type; + typedef std::vector string_type; + typedef U_NAMESPACE_QUALIFIER Locale locale_type; +#ifdef BOOST_NO_INT64_T + typedef std::bitset<64> char_class_type; +#else + typedef cutl_details_boost::uint64_t char_class_type; +#endif + + struct boost_extensions_tag{}; + + icu_regex_traits() + : m_pimpl(re_detail::get_icu_regex_traits_implementation(U_NAMESPACE_QUALIFIER Locale())) + { + } + static size_type length(const char_type* p); + + ::cutl_details_boost::regex_constants::syntax_type syntax_type(char_type c)const + { + return ((c < 0x7f) && (c > 0)) ? re_detail::get_default_syntax_type(static_cast(c)) : regex_constants::syntax_char; + } + ::cutl_details_boost::regex_constants::escape_syntax_type escape_syntax_type(char_type c) const + { + return ((c < 0x7f) && (c > 0)) ? re_detail::get_default_escape_syntax_type(static_cast(c)) : regex_constants::syntax_char; + } + char_type translate(char_type c) const + { + return c; + } + char_type translate_nocase(char_type c) const + { + return ::u_tolower(c); + } + char_type translate(char_type c, bool icase) const + { + return icase ? translate_nocase(c) : translate(c); + } + char_type tolower(char_type c) const + { + return ::u_tolower(c); + } + char_type toupper(char_type c) const + { + return ::u_toupper(c); + } + string_type transform(const char_type* p1, const char_type* p2) const + { + return m_pimpl->transform(p1, p2); + } + string_type transform_primary(const char_type* p1, const char_type* p2) const + { + return m_pimpl->transform_primary(p1, p2); + } + char_class_type lookup_classname(const char_type* p1, const char_type* p2) const; + string_type lookup_collatename(const char_type* p1, const char_type* p2) const; + bool isctype(char_type c, char_class_type f) const; + int toi(const char_type*& p1, const char_type* p2, int radix)const + { + return re_detail::global_toi(p1, p2, radix, *this); + } + int value(char_type c, int radix)const + { + return u_digit(c, static_cast< ::int8_t>(radix)); + } + locale_type imbue(locale_type l) + { + locale_type result(m_pimpl->getloc()); + m_pimpl = re_detail::get_icu_regex_traits_implementation(l); + return result; + } + locale_type getloc()const + { + return locale_type(); + } + std::string error_string(::cutl_details_boost::regex_constants::error_type n) const + { + return re_detail::get_default_error_string(n); + } +private: + icu_regex_traits(const icu_regex_traits&); + icu_regex_traits& operator=(const icu_regex_traits&); + + // + // define the bitmasks offsets we need for additional character properties: + // + enum{ + offset_blank = U_CHAR_CATEGORY_COUNT, + offset_space = U_CHAR_CATEGORY_COUNT+1, + offset_xdigit = U_CHAR_CATEGORY_COUNT+2, + offset_underscore = U_CHAR_CATEGORY_COUNT+3, + offset_unicode = U_CHAR_CATEGORY_COUNT+4, + offset_any = U_CHAR_CATEGORY_COUNT+5, + offset_ascii = U_CHAR_CATEGORY_COUNT+6, + offset_horizontal = U_CHAR_CATEGORY_COUNT+7, + offset_vertical = U_CHAR_CATEGORY_COUNT+8 + }; + + // + // and now the masks: + // + static const char_class_type mask_blank; + static const char_class_type mask_space; + static const char_class_type mask_xdigit; + static const char_class_type mask_underscore; + static const char_class_type mask_unicode; + static const char_class_type mask_any; + static const char_class_type mask_ascii; + static const char_class_type mask_horizontal; + static const char_class_type mask_vertical; + + static char_class_type lookup_icu_mask(const ::UChar32* p1, const ::UChar32* p2); + + cutl_details_boost::shared_ptr< ::cutl_details_boost::re_detail::icu_regex_traits_implementation> m_pimpl; +}; + +} // namespace cutl_details_boost + +// +// template instances: +// +#define BOOST_REGEX_CHAR_T UChar32 +#undef BOOST_REGEX_TRAITS_T +#define BOOST_REGEX_TRAITS_T , icu_regex_traits +#define BOOST_REGEX_ICU_INSTANCES +#ifdef BOOST_REGEX_ICU_INSTANTIATE +# define BOOST_REGEX_INSTANTIATE +#endif +#include +#undef BOOST_REGEX_CHAR_T +#undef BOOST_REGEX_TRAITS_T +#undef BOOST_REGEX_ICU_INSTANCES +#ifdef BOOST_REGEX_INSTANTIATE +# undef BOOST_REGEX_INSTANTIATE +#endif + +namespace cutl_details_boost{ + +// types: +typedef basic_regex< ::UChar32, icu_regex_traits> u32regex; +typedef match_results u32match; +typedef match_results u16match; + +// +// Construction of 32-bit regex types from UTF-8 and UTF-16 primitives: +// +namespace re_detail{ + +#if !defined(BOOST_NO_MEMBER_TEMPLATES) && !defined(__IBMCPP__) +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<1>*) +{ + typedef cutl_details_boost::u8_to_u32_iterator conv_type; + return u32regex(conv_type(i), conv_type(j), opt); +} + +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<2>*) +{ + typedef cutl_details_boost::u16_to_u32_iterator conv_type; + return u32regex(conv_type(i), conv_type(j), opt); +} + +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<4>*) +{ + return u32regex(i, j, opt); +} +#else +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<1>*) +{ + typedef cutl_details_boost::u8_to_u32_iterator conv_type; + typedef std::vector vector_type; + vector_type v; + conv_type a(i), b(j); + while(a != b) + { + v.push_back(*a); + ++a; + } + if(v.size()) + return u32regex(&*v.begin(), v.size(), opt); + return u32regex(static_cast(0), static_cast(0), opt); +} + +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<2>*) +{ + typedef cutl_details_boost::u16_to_u32_iterator conv_type; + typedef std::vector vector_type; + vector_type v; + conv_type a(i), b(j); + while(a != b) + { + v.push_back(*a); + ++a; + } + if(v.size()) + return u32regex(&*v.begin(), v.size(), opt); + return u32regex(static_cast(0), static_cast(0), opt); +} + +template +inline u32regex do_make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt, + const cutl_details_boost::mpl::int_<4>*) +{ + typedef std::vector vector_type; + vector_type v; + while(i != j) + { + v.push_back((UChar32)(*i)); + ++i; + } + if(v.size()) + return u32regex(&*v.begin(), v.size(), opt); + return u32regex(static_cast(0), static_cast(0), opt); +} +#endif +} + +// +// Construction from an iterator pair: +// +template +inline u32regex make_u32regex(InputIterator i, + InputIterator j, + cutl_details_boost::regex_constants::syntax_option_type opt) +{ + return re_detail::do_make_u32regex(i, j, opt, static_cast const*>(0)); +} +// +// construction from UTF-8 nul-terminated strings: +// +inline u32regex make_u32regex(const char* p, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(p, p + std::strlen(p), opt, static_cast const*>(0)); +} +inline u32regex make_u32regex(const unsigned char* p, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(p, p + std::strlen(reinterpret_cast(p)), opt, static_cast const*>(0)); +} +// +// construction from UTF-16 nul-terminated strings: +// +#ifndef BOOST_NO_WREGEX +inline u32regex make_u32regex(const wchar_t* p, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(p, p + std::wcslen(p), opt, static_cast const*>(0)); +} +#endif +#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) +inline u32regex make_u32regex(const UChar* p, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(p, p + u_strlen(p), opt, static_cast const*>(0)); +} +#endif +// +// construction from basic_string class-template: +// +template +inline u32regex make_u32regex(const std::basic_string& s, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(s.begin(), s.end(), opt, static_cast const*>(0)); +} +// +// Construction from ICU string type: +// +inline u32regex make_u32regex(const U_NAMESPACE_QUALIFIER UnicodeString& s, cutl_details_boost::regex_constants::syntax_option_type opt = cutl_details_boost::regex_constants::perl) +{ + return re_detail::do_make_u32regex(s.getBuffer(), s.getBuffer() + s.length(), opt, static_cast const*>(0)); +} + +// +// regex_match overloads that widen the character type as appropriate: +// +namespace re_detail{ +template +void copy_results(MR1& out, MR2 const& in) +{ + // copy results from an adapted MR2 match_results: + out.set_size(in.size(), in.prefix().first.base(), in.suffix().second.base()); + out.set_base(in.base().base()); + for(int i = 0; i < (int)in.size(); ++i) + { + if(in[i].matched) + { + out.set_first(in[i].first.base(), i); + out.set_second(in[i].second.base(), i); + } + } +} + +template +inline bool do_regex_match(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + cutl_details_boost::mpl::int_<4> const*) +{ + return ::cutl_details_boost::regex_match(first, last, m, e, flags); +} +template +bool do_regex_match(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + cutl_details_boost::mpl::int_<2> const*) +{ + typedef u16_to_u32_iterator conv_type; + typedef match_results match_type; + typedef typename match_type::allocator_type alloc_type; + match_type what; + bool result = ::cutl_details_boost::regex_match(conv_type(first), conv_type(last), what, e, flags); + // copy results across to m: + if(result) copy_results(m, what); + return result; +} +template +bool do_regex_match(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + cutl_details_boost::mpl::int_<1> const*) +{ + typedef u8_to_u32_iterator conv_type; + typedef match_results match_type; + typedef typename match_type::allocator_type alloc_type; + match_type what; + bool result = ::cutl_details_boost::regex_match(conv_type(first), conv_type(last), what, e, flags); + // copy results across to m: + if(result) copy_results(m, what); + return result; +} +} // namespace re_detail + +template +inline bool u32regex_match(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(first, last, m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const UChar* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(p, p+u_strlen(p), m, e, flags, static_cast const*>(0)); +} +#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX) +inline bool u32regex_match(const wchar_t* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(p, p+std::wcslen(p), m, e, flags, static_cast const*>(0)); +} +#endif +inline bool u32regex_match(const char* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(p, p+std::strlen(p), m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const unsigned char* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(p, p+std::strlen((const char*)p), m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const std::string& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(s.begin(), s.end(), m, e, flags, static_cast const*>(0)); +} +#ifndef BOOST_NO_STD_WSTRING +inline bool u32regex_match(const std::wstring& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(s.begin(), s.end(), m, e, flags, static_cast const*>(0)); +} +#endif +inline bool u32regex_match(const U_NAMESPACE_QUALIFIER UnicodeString& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_match(s.getBuffer(), s.getBuffer() + s.length(), m, e, flags, static_cast const*>(0)); +} +// +// regex_match overloads that do not return what matched: +// +template +inline bool u32regex_match(BidiIterator first, BidiIterator last, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(first, last, m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const UChar* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(p, p+u_strlen(p), m, e, flags, static_cast const*>(0)); +} +#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX) +inline bool u32regex_match(const wchar_t* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(p, p+std::wcslen(p), m, e, flags, static_cast const*>(0)); +} +#endif +inline bool u32regex_match(const char* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(p, p+std::strlen(p), m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const unsigned char* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(p, p+std::strlen((const char*)p), m, e, flags, static_cast const*>(0)); +} +inline bool u32regex_match(const std::string& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(s.begin(), s.end(), m, e, flags, static_cast const*>(0)); +} +#ifndef BOOST_NO_STD_WSTRING +inline bool u32regex_match(const std::wstring& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(s.begin(), s.end(), m, e, flags, static_cast const*>(0)); +} +#endif +inline bool u32regex_match(const U_NAMESPACE_QUALIFIER UnicodeString& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_match(s.getBuffer(), s.getBuffer() + s.length(), m, e, flags, static_cast const*>(0)); +} + +// +// regex_search overloads that widen the character type as appropriate: +// +namespace re_detail{ +template +inline bool do_regex_search(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + BidiIterator base, + cutl_details_boost::mpl::int_<4> const*) +{ + return ::cutl_details_boost::regex_search(first, last, m, e, flags, base); +} +template +bool do_regex_search(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + BidiIterator base, + cutl_details_boost::mpl::int_<2> const*) +{ + typedef u16_to_u32_iterator conv_type; + typedef match_results match_type; + typedef typename match_type::allocator_type alloc_type; + match_type what; + bool result = ::cutl_details_boost::regex_search(conv_type(first), conv_type(last), what, e, flags, conv_type(base)); + // copy results across to m: + if(result) copy_results(m, what); + return result; +} +template +bool do_regex_search(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + BidiIterator base, + cutl_details_boost::mpl::int_<1> const*) +{ + typedef u8_to_u32_iterator conv_type; + typedef match_results match_type; + typedef typename match_type::allocator_type alloc_type; + match_type what; + bool result = ::cutl_details_boost::regex_search(conv_type(first), conv_type(last), what, e, flags, conv_type(base)); + // copy results across to m: + if(result) copy_results(m, what); + return result; +} +} + +template +inline bool u32regex_search(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(first, last, m, e, flags, first, static_cast const*>(0)); +} +template +inline bool u32regex_search(BidiIterator first, BidiIterator last, + match_results& m, + const u32regex& e, + match_flag_type flags, + BidiIterator base) +{ + return re_detail::do_regex_search(first, last, m, e, flags, base, static_cast const*>(0)); +} +inline bool u32regex_search(const UChar* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(p, p+u_strlen(p), m, e, flags, p, static_cast const*>(0)); +} +#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX) +inline bool u32regex_search(const wchar_t* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(p, p+std::wcslen(p), m, e, flags, p, static_cast const*>(0)); +} +#endif +inline bool u32regex_search(const char* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(p, p+std::strlen(p), m, e, flags, p, static_cast const*>(0)); +} +inline bool u32regex_search(const unsigned char* p, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(p, p+std::strlen((const char*)p), m, e, flags, p, static_cast const*>(0)); +} +inline bool u32regex_search(const std::string& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(s.begin(), s.end(), m, e, flags, s.begin(), static_cast const*>(0)); +} +#ifndef BOOST_NO_STD_WSTRING +inline bool u32regex_search(const std::wstring& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(s.begin(), s.end(), m, e, flags, s.begin(), static_cast const*>(0)); +} +#endif +inline bool u32regex_search(const U_NAMESPACE_QUALIFIER UnicodeString& s, + match_results& m, + const u32regex& e, + match_flag_type flags = match_default) +{ + return re_detail::do_regex_search(s.getBuffer(), s.getBuffer() + s.length(), m, e, flags, s.getBuffer(), static_cast const*>(0)); +} +template +inline bool u32regex_search(BidiIterator first, BidiIterator last, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(first, last, m, e, flags, first, static_cast const*>(0)); +} +inline bool u32regex_search(const UChar* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(p, p+u_strlen(p), m, e, flags, p, static_cast const*>(0)); +} +#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2) && !defined(BOOST_NO_WREGEX) +inline bool u32regex_search(const wchar_t* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(p, p+std::wcslen(p), m, e, flags, p, static_cast const*>(0)); +} +#endif +inline bool u32regex_search(const char* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(p, p+std::strlen(p), m, e, flags, p, static_cast const*>(0)); +} +inline bool u32regex_search(const unsigned char* p, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(p, p+std::strlen((const char*)p), m, e, flags, p, static_cast const*>(0)); +} +inline bool u32regex_search(const std::string& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(s.begin(), s.end(), m, e, flags, s.begin(), static_cast const*>(0)); +} +#ifndef BOOST_NO_STD_WSTRING +inline bool u32regex_search(const std::wstring& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(s.begin(), s.end(), m, e, flags, s.begin(), static_cast const*>(0)); +} +#endif +inline bool u32regex_search(const U_NAMESPACE_QUALIFIER UnicodeString& s, + const u32regex& e, + match_flag_type flags = match_default) +{ + match_results m; + return re_detail::do_regex_search(s.getBuffer(), s.getBuffer() + s.length(), m, e, flags, s.getBuffer(), static_cast const*>(0)); +} + +// +// overloads for regex_replace with utf-8 and utf-16 data types: +// +namespace re_detail{ +template +inline std::pair< cutl_details_boost::u8_to_u32_iterator, cutl_details_boost::u8_to_u32_iterator > + make_utf32_seq(I i, I j, mpl::int_<1> const*) +{ + return std::pair< cutl_details_boost::u8_to_u32_iterator, cutl_details_boost::u8_to_u32_iterator >(cutl_details_boost::u8_to_u32_iterator(i), cutl_details_boost::u8_to_u32_iterator(j)); +} +template +inline std::pair< cutl_details_boost::u16_to_u32_iterator, cutl_details_boost::u16_to_u32_iterator > + make_utf32_seq(I i, I j, mpl::int_<2> const*) +{ + return std::pair< cutl_details_boost::u16_to_u32_iterator, cutl_details_boost::u16_to_u32_iterator >(cutl_details_boost::u16_to_u32_iterator(i), cutl_details_boost::u16_to_u32_iterator(j)); +} +template +inline std::pair< I, I > + make_utf32_seq(I i, I j, mpl::int_<4> const*) +{ + return std::pair< I, I >(i, j); +} +template +inline std::pair< cutl_details_boost::u8_to_u32_iterator, cutl_details_boost::u8_to_u32_iterator > + make_utf32_seq(const charT* p, mpl::int_<1> const*) +{ + return std::pair< cutl_details_boost::u8_to_u32_iterator, cutl_details_boost::u8_to_u32_iterator >(cutl_details_boost::u8_to_u32_iterator(p), cutl_details_boost::u8_to_u32_iterator(p+std::strlen((const char*)p))); +} +template +inline std::pair< cutl_details_boost::u16_to_u32_iterator, cutl_details_boost::u16_to_u32_iterator > + make_utf32_seq(const charT* p, mpl::int_<2> const*) +{ + return std::pair< cutl_details_boost::u16_to_u32_iterator, cutl_details_boost::u16_to_u32_iterator >(cutl_details_boost::u16_to_u32_iterator(p), cutl_details_boost::u16_to_u32_iterator(p+u_strlen((const UChar*)p))); +} +template +inline std::pair< const charT*, const charT* > + make_utf32_seq(const charT* p, mpl::int_<4> const*) +{ + return std::pair< const charT*, const charT* >(p, p+icu_regex_traits::length((UChar32 const*)p)); +} +template +inline OutputIterator make_utf32_out(OutputIterator o, mpl::int_<4> const*) +{ + return o; +} +template +inline utf16_output_iterator make_utf32_out(OutputIterator o, mpl::int_<2> const*) +{ + return o; +} +template +inline utf8_output_iterator make_utf32_out(OutputIterator o, mpl::int_<1> const*) +{ + return o; +} + +template +OutputIterator do_regex_replace(OutputIterator out, + std::pair const& in, + const u32regex& e, + const std::pair& fmt, + match_flag_type flags + ) +{ + // unfortunately we have to copy the format string in order to pass in onward: + std::vector f; +#ifndef BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS + f.assign(fmt.first, fmt.second); +#else + f.clear(); + I2 pos = fmt.first; + while(pos != fmt.second) + f.push_back(*pos++); +#endif + + regex_iterator i(in.first, in.second, e, flags); + regex_iterator j; + if(i == j) + { + if(!(flags & regex_constants::format_no_copy)) + out = re_detail::copy(in.first, in.second, out); + } + else + { + I1 last_m = in.first; + while(i != j) + { + if(!(flags & regex_constants::format_no_copy)) + out = re_detail::copy(i->prefix().first, i->prefix().second, out); + if(f.size()) + out = ::cutl_details_boost::re_detail::regex_format_imp(out, *i, &*f.begin(), &*f.begin() + f.size(), flags, e.get_traits()); + else + out = ::cutl_details_boost::re_detail::regex_format_imp(out, *i, static_cast(0), static_cast(0), flags, e.get_traits()); + last_m = (*i)[0].second; + if(flags & regex_constants::format_first_only) + break; + ++i; + } + if(!(flags & regex_constants::format_no_copy)) + out = re_detail::copy(last_m, in.second, out); + } + return out; +} +template +inline const BaseIterator& extract_output_base(const BaseIterator& b) +{ + return b; +} +template +inline BaseIterator extract_output_base(const utf8_output_iterator& b) +{ + return b.base(); +} +template +inline BaseIterator extract_output_base(const utf16_output_iterator& b) +{ + return b.base(); +} +} // re_detail + +template +inline OutputIterator u32regex_replace(OutputIterator out, + BidirectionalIterator first, + BidirectionalIterator last, + const u32regex& e, + const charT* fmt, + match_flag_type flags = match_default) +{ + return re_detail::extract_output_base +#if BOOST_WORKAROUND(BOOST_MSVC, <= 1300) + +#endif + ( + re_detail::do_regex_replace( + re_detail::make_utf32_out(out, static_cast const*>(0)), + re_detail::make_utf32_seq(first, last, static_cast const*>(0)), + e, + re_detail::make_utf32_seq(fmt, static_cast const*>(0)), + flags) + ); +} + +template +inline OutputIterator u32regex_replace(OutputIterator out, + Iterator first, + Iterator last, + const u32regex& e, + const std::basic_string& fmt, + match_flag_type flags = match_default) +{ + return re_detail::extract_output_base +#if BOOST_WORKAROUND(BOOST_MSVC, <= 1300) + +#endif + ( + re_detail::do_regex_replace( + re_detail::make_utf32_out(out, static_cast const*>(0)), + re_detail::make_utf32_seq(first, last, static_cast const*>(0)), + e, + re_detail::make_utf32_seq(fmt.begin(), fmt.end(), static_cast const*>(0)), + flags) + ); +} + +template +inline OutputIterator u32regex_replace(OutputIterator out, + Iterator first, + Iterator last, + const u32regex& e, + const U_NAMESPACE_QUALIFIER UnicodeString& fmt, + match_flag_type flags = match_default) +{ + return re_detail::extract_output_base +#if BOOST_WORKAROUND(BOOST_MSVC, <= 1300) + +#endif + ( + re_detail::do_regex_replace( + re_detail::make_utf32_out(out, static_cast const*>(0)), + re_detail::make_utf32_seq(first, last, static_cast const*>(0)), + e, + re_detail::make_utf32_seq(fmt.getBuffer(), fmt.getBuffer() + fmt.length(), static_cast const*>(0)), + flags) + ); +} + +template +std::basic_string u32regex_replace(const std::basic_string& s, + const u32regex& e, + const charT* fmt, + match_flag_type flags = match_default) +{ + std::basic_string result; + re_detail::string_out_iterator > i(result); + u32regex_replace(i, s.begin(), s.end(), e, fmt, flags); + return result; +} + +template +std::basic_string u32regex_replace(const std::basic_string& s, + const u32regex& e, + const std::basic_string& fmt, + match_flag_type flags = match_default) +{ + std::basic_string result; + re_detail::string_out_iterator > i(result); + u32regex_replace(i, s.begin(), s.end(), e, fmt.c_str(), flags); + return result; +} + +namespace re_detail{ + +class unicode_string_out_iterator +{ + U_NAMESPACE_QUALIFIER UnicodeString* out; +public: + unicode_string_out_iterator(U_NAMESPACE_QUALIFIER UnicodeString& s) : out(&s) {} + unicode_string_out_iterator& operator++() { return *this; } + unicode_string_out_iterator& operator++(int) { return *this; } + unicode_string_out_iterator& operator*() { return *this; } + unicode_string_out_iterator& operator=(UChar v) + { + *out += v; + return *this; + } + typedef std::ptrdiff_t difference_type; + typedef UChar value_type; + typedef value_type* pointer; + typedef value_type& reference; + typedef std::output_iterator_tag iterator_category; +}; + +} + +inline U_NAMESPACE_QUALIFIER UnicodeString u32regex_replace(const U_NAMESPACE_QUALIFIER UnicodeString& s, + const u32regex& e, + const UChar* fmt, + match_flag_type flags = match_default) +{ + U_NAMESPACE_QUALIFIER UnicodeString result; + re_detail::unicode_string_out_iterator i(result); + u32regex_replace(i, s.getBuffer(), s.getBuffer()+s.length(), e, fmt, flags); + return result; +} + +inline U_NAMESPACE_QUALIFIER UnicodeString u32regex_replace(const U_NAMESPACE_QUALIFIER UnicodeString& s, + const u32regex& e, + const U_NAMESPACE_QUALIFIER UnicodeString& fmt, + match_flag_type flags = match_default) +{ + U_NAMESPACE_QUALIFIER UnicodeString result; + re_detail::unicode_string_out_iterator i(result); + re_detail::do_regex_replace( + re_detail::make_utf32_out(i, static_cast const*>(0)), + re_detail::make_utf32_seq(s.getBuffer(), s.getBuffer()+s.length(), static_cast const*>(0)), + e, + re_detail::make_utf32_seq(fmt.getBuffer(), fmt.getBuffer() + fmt.length(), static_cast const*>(0)), + flags); + return result; +} + +} // namespace cutl_details_boost. + +#ifdef BOOST_MSVC +#pragma warning (pop) +#endif + +#include +#include + +#endif -- cgit v1.1