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/v4/cpp_regex_traits.hpp | 1088 ++++++++++++++++++++++ 1 file changed, 1088 insertions(+) create mode 100644 cutl/details/boost/regex/v4/cpp_regex_traits.hpp (limited to 'cutl/details/boost/regex/v4/cpp_regex_traits.hpp') diff --git a/cutl/details/boost/regex/v4/cpp_regex_traits.hpp b/cutl/details/boost/regex/v4/cpp_regex_traits.hpp new file mode 100644 index 0000000..7281e84 --- /dev/null +++ b/cutl/details/boost/regex/v4/cpp_regex_traits.hpp @@ -0,0 +1,1088 @@ +/* + * + * 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 cpp_regex_traits.hpp + * VERSION see + * DESCRIPTION: Declares regular expression traits class cpp_regex_traits. + */ + +#ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED +#define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED + +#include + +#ifndef BOOST_NO_STD_LOCALE + +#ifndef BOOST_RE_PAT_EXCEPT_HPP +#include +#endif +#ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED +#include +#endif +#ifdef BOOST_HAS_THREADS +#include +#endif +#ifndef BOOST_REGEX_PRIMARY_TRANSFORM +#include +#endif +#ifndef BOOST_REGEX_OBJECT_CACHE_HPP +#include +#endif + +#include +#include +#include + +#ifdef BOOST_MSVC +#pragma warning(push) +#pragma warning(disable: 4103) +#endif +#ifdef BOOST_HAS_ABI_HEADERS +# include BOOST_ABI_PREFIX +#endif +#ifdef BOOST_MSVC +#pragma warning(pop) +#endif + +#ifdef BOOST_MSVC +#pragma warning(push) +#pragma warning(disable:4786 4251) +#endif + +namespace cutl_details_boost{ + +// +// forward declaration is needed by some compilers: +// +template +class cpp_regex_traits; + +namespace re_detail{ + +// +// class parser_buf: +// acts as a stream buffer which wraps around a pair of pointers: +// +template > +class parser_buf : public ::std::basic_streambuf +{ + typedef ::std::basic_streambuf base_type; + typedef typename base_type::int_type int_type; + typedef typename base_type::char_type char_type; + typedef typename base_type::pos_type pos_type; + typedef ::std::streamsize streamsize; + typedef typename base_type::off_type off_type; +public: + parser_buf() : base_type() { setbuf(0, 0); } + const charT* getnext() { return this->gptr(); } +protected: + std::basic_streambuf* setbuf(char_type* s, streamsize n); + typename parser_buf::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which); + typename parser_buf::pos_type seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which); +private: + parser_buf& operator=(const parser_buf&); + parser_buf(const parser_buf&); +}; + +template +std::basic_streambuf* +parser_buf::setbuf(char_type* s, streamsize n) +{ + this->setg(s, s, s + n); + return this; +} + +template +typename parser_buf::pos_type +parser_buf::seekoff(off_type off, ::std::ios_base::seekdir way, ::std::ios_base::openmode which) +{ + if(which & ::std::ios_base::out) + return pos_type(off_type(-1)); + std::ptrdiff_t size = this->egptr() - this->eback(); + std::ptrdiff_t pos = this->gptr() - this->eback(); + charT* g = this->eback(); + switch(way) + { + case ::std::ios_base::beg: + if((off < 0) || (off > size)) + return pos_type(off_type(-1)); + else + this->setg(g, g + off, g + size); + break; + case ::std::ios_base::end: + if((off < 0) || (off > size)) + return pos_type(off_type(-1)); + else + this->setg(g, g + size - off, g + size); + break; + case ::std::ios_base::cur: + { + std::ptrdiff_t newpos = static_cast(pos + off); + if((newpos < 0) || (newpos > size)) + return pos_type(off_type(-1)); + else + this->setg(g, g + newpos, g + size); + break; + } + default: ; + } +#ifdef BOOST_MSVC +#pragma warning(push) +#pragma warning(disable:4244) +#endif + return static_cast(this->gptr() - this->eback()); +#ifdef BOOST_MSVC +#pragma warning(pop) +#endif +} + +template +typename parser_buf::pos_type +parser_buf::seekpos(pos_type sp, ::std::ios_base::openmode which) +{ + if(which & ::std::ios_base::out) + return pos_type(off_type(-1)); + off_type size = static_cast(this->egptr() - this->eback()); + charT* g = this->eback(); + if(off_type(sp) <= size) + { + this->setg(g, g + off_type(sp), g + size); + } + return pos_type(off_type(-1)); +} + +// +// class cpp_regex_traits_base: +// acts as a container for locale and the facets we are using. +// +template +struct cpp_regex_traits_base +{ + cpp_regex_traits_base(const std::locale& l) + { imbue(l); } + std::locale imbue(const std::locale& l); + + std::locale m_locale; + std::ctype const* m_pctype; +#ifndef BOOST_NO_STD_MESSAGES + std::messages const* m_pmessages; +#endif + std::collate const* m_pcollate; + + bool operator<(const cpp_regex_traits_base& b)const + { + if(m_pctype == b.m_pctype) + { +#ifndef BOOST_NO_STD_MESSAGES + if(m_pmessages == b.m_pmessages) + { + return m_pcollate < b.m_pcollate; + } + return m_pmessages < b.m_pmessages; +#else + return m_pcollate < b.m_pcollate; +#endif + } + return m_pctype < b.m_pctype; + } + bool operator==(const cpp_regex_traits_base& b)const + { + return (m_pctype == b.m_pctype) +#ifndef BOOST_NO_STD_MESSAGES + && (m_pmessages == b.m_pmessages) +#endif + && (m_pcollate == b.m_pcollate); + } +}; + +template +std::locale cpp_regex_traits_base::imbue(const std::locale& l) +{ + std::locale result(m_locale); + m_locale = l; + m_pctype = &BOOST_USE_FACET(std::ctype, l); +#ifndef BOOST_NO_STD_MESSAGES + m_pmessages = BOOST_HAS_FACET(std::messages, l) ? &BOOST_USE_FACET(std::messages, l) : 0; +#endif + m_pcollate = &BOOST_USE_FACET(std::collate, l); + return result; +} + +// +// class cpp_regex_traits_char_layer: +// implements methods that require specialisation for narrow characters: +// +template +class cpp_regex_traits_char_layer : public cpp_regex_traits_base +{ + typedef std::basic_string string_type; + typedef std::map map_type; + typedef typename map_type::const_iterator map_iterator_type; +public: + cpp_regex_traits_char_layer(const std::locale& l) + : cpp_regex_traits_base(l) + { + init(); + } + cpp_regex_traits_char_layer(const cpp_regex_traits_base& b) + : cpp_regex_traits_base(b) + { + init(); + } + void init(); + + regex_constants::syntax_type syntax_type(charT c)const + { + map_iterator_type i = m_char_map.find(c); + return ((i == m_char_map.end()) ? 0 : i->second); + } + regex_constants::escape_syntax_type escape_syntax_type(charT c) const + { + map_iterator_type i = m_char_map.find(c); + if(i == m_char_map.end()) + { + if(this->m_pctype->is(std::ctype_base::lower, c)) return regex_constants::escape_type_class; + if(this->m_pctype->is(std::ctype_base::upper, c)) return regex_constants::escape_type_not_class; + return 0; + } + return i->second; + } + +private: + string_type get_default_message(regex_constants::syntax_type); + // TODO: use a hash table when available! + map_type m_char_map; +}; + +template +void cpp_regex_traits_char_layer::init() +{ + // we need to start by initialising our syntax map so we know which + // character is used for which purpose: +#ifndef BOOST_NO_STD_MESSAGES +#ifndef __IBMCPP__ + typename std::messages::catalog cat = static_cast::catalog>(-1); +#else + typename std::messages::catalog cat = reinterpret_cast::catalog>(-1); +#endif + std::string cat_name(cpp_regex_traits::get_catalog_name()); + if(cat_name.size() && (this->m_pmessages != 0)) + { + cat = this->m_pmessages->open( + cat_name, + this->m_locale); + if((int)cat < 0) + { + std::string m("Unable to open message catalog: "); + std::runtime_error err(m + cat_name); + cutl_details_boost::re_detail::raise_runtime_error(err); + } + } + // + // if we have a valid catalog then load our messages: + // + if((int)cat >= 0) + { +#ifndef BOOST_NO_EXCEPTIONS + try{ +#endif + for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i) + { + string_type mss = this->m_pmessages->get(cat, 0, i, get_default_message(i)); + for(typename string_type::size_type j = 0; j < mss.size(); ++j) + { + m_char_map[mss[j]] = i; + } + } + this->m_pmessages->close(cat); +#ifndef BOOST_NO_EXCEPTIONS + } + catch(...) + { + if(this->m_pmessages) + this->m_pmessages->close(cat); + throw; + } +#endif + } + else + { +#endif + for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i) + { + const char* ptr = get_default_syntax(i); + while(ptr && *ptr) + { + m_char_map[this->m_pctype->widen(*ptr)] = i; + ++ptr; + } + } +#ifndef BOOST_NO_STD_MESSAGES + } +#endif +} + +template +typename cpp_regex_traits_char_layer::string_type + cpp_regex_traits_char_layer::get_default_message(regex_constants::syntax_type i) +{ + const char* ptr = get_default_syntax(i); + string_type result; + while(ptr && *ptr) + { + result.append(1, this->m_pctype->widen(*ptr)); + ++ptr; + } + return result; +} + +// +// specialised version for narrow characters: +// +template <> +class BOOST_REGEX_DECL cpp_regex_traits_char_layer : public cpp_regex_traits_base +{ + typedef std::string string_type; +public: + cpp_regex_traits_char_layer(const std::locale& l) + : cpp_regex_traits_base(l) + { + init(); + } + cpp_regex_traits_char_layer(const cpp_regex_traits_base& l) + : cpp_regex_traits_base(l) + { + init(); + } + + regex_constants::syntax_type syntax_type(char c)const + { + return m_char_map[static_cast(c)]; + } + regex_constants::escape_syntax_type escape_syntax_type(char c) const + { + return m_char_map[static_cast(c)]; + } + +private: + regex_constants::syntax_type m_char_map[1u << CHAR_BIT]; + void init(); +}; + +#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET +enum +{ + char_class_space=1<<0, + char_class_print=1<<1, + char_class_cntrl=1<<2, + char_class_upper=1<<3, + char_class_lower=1<<4, + char_class_alpha=1<<5, + char_class_digit=1<<6, + char_class_punct=1<<7, + char_class_xdigit=1<<8, + char_class_alnum=char_class_alpha|char_class_digit, + char_class_graph=char_class_alnum|char_class_punct, + char_class_blank=1<<9, + char_class_word=1<<10, + char_class_unicode=1<<11, + char_class_horizontal_space=1<<12, + char_class_vertical_space=1<<13 +}; + +#endif + +// +// class cpp_regex_traits_implementation: +// provides pimpl implementation for cpp_regex_traits. +// +template +class cpp_regex_traits_implementation : public cpp_regex_traits_char_layer +{ +public: + typedef typename cpp_regex_traits::char_class_type char_class_type; + typedef typename std::ctype::mask native_mask_type; +#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET + BOOST_STATIC_CONSTANT(char_class_type, mask_blank = 1u << 24); + BOOST_STATIC_CONSTANT(char_class_type, mask_word = 1u << 25); + BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 1u << 26); + BOOST_STATIC_CONSTANT(char_class_type, mask_horizontal = 1u << 27); + BOOST_STATIC_CONSTANT(char_class_type, mask_vertical = 1u << 28); +#endif + + typedef std::basic_string string_type; + typedef charT char_type; + //cpp_regex_traits_implementation(); + cpp_regex_traits_implementation(const std::locale& l) + : cpp_regex_traits_char_layer(l) + { + init(); + } + cpp_regex_traits_implementation(const cpp_regex_traits_base& l) + : cpp_regex_traits_char_layer(l) + { + init(); + } + std::string error_string(regex_constants::error_type n) const + { + if(!m_error_strings.empty()) + { + std::map::const_iterator p = m_error_strings.find(n); + return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second; + } + return get_default_error_string(n); + } + char_class_type lookup_classname(const charT* p1, const charT* p2) const + { + char_class_type result = lookup_classname_imp(p1, p2); + if(result == 0) + { + string_type temp(p1, p2); + this->m_pctype->tolower(&*temp.begin(), &*temp.begin() + temp.size()); + result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size()); + } + return result; + } + string_type lookup_collatename(const charT* p1, const charT* p2) const; + string_type transform_primary(const charT* p1, const charT* p2) const; + string_type transform(const charT* p1, const charT* p2) const; +private: + std::map m_error_strings; // error messages indexed by numberic ID + std::map m_custom_class_names; // character class names + std::map m_custom_collate_names; // collating element names + unsigned m_collate_type; // the form of the collation string + charT m_collate_delim; // the collation group delimiter + // + // helpers: + // + char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const; + void init(); +#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET +public: + bool isctype(charT c, char_class_type m)const; +#endif +}; + +#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET +#if !defined(BOOST_NO_INCLASS_MEMBER_INITIALIZATION) + +template +typename cpp_regex_traits_implementation::char_class_type const cpp_regex_traits_implementation::mask_blank; +template +typename cpp_regex_traits_implementation::char_class_type const cpp_regex_traits_implementation::mask_word; +template +typename cpp_regex_traits_implementation::char_class_type const cpp_regex_traits_implementation::mask_unicode; +template +typename cpp_regex_traits_implementation::char_class_type const cpp_regex_traits_implementation::mask_vertical; +template +typename cpp_regex_traits_implementation::char_class_type const cpp_regex_traits_implementation::mask_horizontal; + +#endif +#endif + +template +typename cpp_regex_traits_implementation::string_type + cpp_regex_traits_implementation::transform_primary(const charT* p1, const charT* p2) const +{ + // + // PRECONDITIONS: + // + // A bug in gcc 3.2 (and maybe other versions as well) treats + // p1 as a null terminated string, for efficiency reasons + // we work around this elsewhere, but just assert here that + // we adhere to gcc's (buggy) preconditions... + // + BOOST_ASSERT(*p2 == 0); + + string_type result; + // + // swallowing all exceptions here is a bad idea + // however at least one std lib will always throw + // std::bad_alloc for certain arguments... + // + try{ + // + // What we do here depends upon the format of the sort key returned by + // sort key returned by this->transform: + // + switch(m_collate_type) + { + case sort_C: + case sort_unknown: + // the best we can do is translate to lower case, then get a regular sort key: + { + result.assign(p1, p2); + this->m_pctype->tolower(&*result.begin(), &*result.begin() + result.size()); + result = this->m_pcollate->transform(&*result.begin(), &*result.begin() + result.size()); + break; + } + case sort_fixed: + { + // get a regular sort key, and then truncate it: + result.assign(this->m_pcollate->transform(p1, p2)); + result.erase(this->m_collate_delim); + break; + } + case sort_delim: + // get a regular sort key, and then truncate everything after the delim: + result.assign(this->m_pcollate->transform(p1, p2)); + std::size_t i; + for(i = 0; i < result.size(); ++i) + { + if(result[i] == m_collate_delim) + break; + } + result.erase(i); + break; + } + }catch(...){} + while(result.size() && (charT(0) == *result.rbegin())) + result.erase(result.size() - 1); + if(result.empty()) + { + // character is ignorable at the primary level: + result = string_type(1, charT(0)); + } + return result; +} + +template +typename cpp_regex_traits_implementation::string_type + cpp_regex_traits_implementation::transform(const charT* p1, const charT* p2) const +{ + // + // PRECONDITIONS: + // + // A bug in gcc 3.2 (and maybe other versions as well) treats + // p1 as a null terminated string, for efficiency reasons + // we work around this elsewhere, but just assert here that + // we adhere to gcc's (buggy) preconditions... + // + BOOST_ASSERT(*p2 == 0); + // + // swallowing all exceptions here is a bad idea + // however at least one std lib will always throw + // std::bad_alloc for certain arguments... + // + string_type result; + try{ + result = this->m_pcollate->transform(p1, p2); + // + // Borland's STLPort version returns a NULL-terminated + // string that has garbage at the end - each call to + // std::collate::transform returns a different string! + // So as a workaround, we'll truncate the string at the first NULL + // which _seems_ to work.... +#if BOOST_WORKAROUND(__BORLANDC__, < 0x580) + result.erase(result.find(charT(0))); +#else + // + // some implementations (Dinkumware) append unnecessary trailing \0's: + while(result.size() && (charT(0) == *result.rbegin())) + result.erase(result.size() - 1); +#endif + BOOST_ASSERT(std::find(result.begin(), result.end(), charT(0)) == result.end()); + } + catch(...) + { + } + return result; +} + + +template +typename cpp_regex_traits_implementation::string_type + cpp_regex_traits_implementation::lookup_collatename(const charT* p1, const charT* p2) const +{ + typedef typename std::map::const_iterator iter_type; + if(m_custom_collate_names.size()) + { + iter_type pos = m_custom_collate_names.find(string_type(p1, p2)); + if(pos != m_custom_collate_names.end()) + return pos->second; + } +#if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\ + && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\ + && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551) + std::string name(p1, p2); +#else + std::string name; + const charT* p0 = p1; + while(p0 != p2) + name.append(1, char(*p0++)); +#endif + name = lookup_default_collate_name(name); +#if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\ + && !BOOST_WORKAROUND(BOOST_MSVC, < 1300)\ + && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551) + if(name.size()) + return string_type(name.begin(), name.end()); +#else + if(name.size()) + { + string_type result; + typedef std::string::const_iterator iter; + iter b = name.begin(); + iter e = name.end(); + while(b != e) + result.append(1, charT(*b++)); + return result; + } +#endif + if(p2 - p1 == 1) + return string_type(1, *p1); + return string_type(); +} + +template +void cpp_regex_traits_implementation::init() +{ +#ifndef BOOST_NO_STD_MESSAGES +#ifndef __IBMCPP__ + typename std::messages::catalog cat = static_cast::catalog>(-1); +#else + typename std::messages::catalog cat = reinterpret_cast::catalog>(-1); +#endif + std::string cat_name(cpp_regex_traits::get_catalog_name()); + if(cat_name.size() && (this->m_pmessages != 0)) + { + cat = this->m_pmessages->open( + cat_name, + this->m_locale); + if((int)cat < 0) + { + std::string m("Unable to open message catalog: "); + std::runtime_error err(m + cat_name); + cutl_details_boost::re_detail::raise_runtime_error(err); + } + } + // + // if we have a valid catalog then load our messages: + // + if((int)cat >= 0) + { + // + // Error messages: + // + for(cutl_details_boost::regex_constants::error_type i = static_cast(0); + i <= cutl_details_boost::regex_constants::error_unknown; + i = static_cast(i + 1)) + { + const char* p = get_default_error_string(i); + string_type default_message; + while(*p) + { + default_message.append(1, this->m_pctype->widen(*p)); + ++p; + } + string_type s = this->m_pmessages->get(cat, 0, i+200, default_message); + std::string result; + for(std::string::size_type j = 0; j < s.size(); ++j) + { + result.append(1, this->m_pctype->narrow(s[j], 0)); + } + m_error_strings[i] = result; + } + // + // Custom class names: + // +#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET + static const char_class_type masks[16] = + { + std::ctype::alnum, + std::ctype::alpha, + std::ctype::cntrl, + std::ctype::digit, + std::ctype::graph, + cpp_regex_traits_implementation::mask_horizontal, + std::ctype::lower, + std::ctype::print, + std::ctype::punct, + std::ctype::space, + std::ctype::upper, + cpp_regex_traits_implementation::mask_vertical, + std::ctype::xdigit, + cpp_regex_traits_implementation::mask_blank, + cpp_regex_traits_implementation::mask_word, + cpp_regex_traits_implementation::mask_unicode, + }; +#else + static const char_class_type masks[16] = + { + ::cutl_details_boost::re_detail::char_class_alnum, + ::cutl_details_boost::re_detail::char_class_alpha, + ::cutl_details_boost::re_detail::char_class_cntrl, + ::cutl_details_boost::re_detail::char_class_digit, + ::cutl_details_boost::re_detail::char_class_graph, + ::cutl_details_boost::re_detail::char_class_horizontal_space, + ::cutl_details_boost::re_detail::char_class_lower, + ::cutl_details_boost::re_detail::char_class_print, + ::cutl_details_boost::re_detail::char_class_punct, + ::cutl_details_boost::re_detail::char_class_space, + ::cutl_details_boost::re_detail::char_class_upper, + ::cutl_details_boost::re_detail::char_class_vertical_space, + ::cutl_details_boost::re_detail::char_class_xdigit, + ::cutl_details_boost::re_detail::char_class_blank, + ::cutl_details_boost::re_detail::char_class_word, + ::cutl_details_boost::re_detail::char_class_unicode, + }; +#endif + static const string_type null_string; + for(unsigned int j = 0; j <= 13; ++j) + { + string_type s(this->m_pmessages->get(cat, 0, j+300, null_string)); + if(s.size()) + this->m_custom_class_names[s] = masks[j]; + } + } +#endif + // + // get the collation format used by m_pcollate: + // + m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim); +} + +template +typename cpp_regex_traits_implementation::char_class_type + cpp_regex_traits_implementation::lookup_classname_imp(const charT* p1, const charT* p2) const +{ +#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET + static const char_class_type masks[22] = + { + 0, + std::ctype::alnum, + std::ctype::alpha, + cpp_regex_traits_implementation::mask_blank, + std::ctype::cntrl, + std::ctype::digit, + std::ctype::digit, + std::ctype::graph, + cpp_regex_traits_implementation::mask_horizontal, + std::ctype::lower, + std::ctype::lower, + std::ctype::print, + std::ctype::punct, + std::ctype::space, + std::ctype::space, + std::ctype::upper, + cpp_regex_traits_implementation::mask_unicode, + std::ctype::upper, + cpp_regex_traits_implementation::mask_vertical, + std::ctype::alnum | cpp_regex_traits_implementation::mask_word, + std::ctype::alnum | cpp_regex_traits_implementation::mask_word, + std::ctype::xdigit, + }; +#else + static const char_class_type masks[22] = + { + 0, + ::cutl_details_boost::re_detail::char_class_alnum, + ::cutl_details_boost::re_detail::char_class_alpha, + ::cutl_details_boost::re_detail::char_class_blank, + ::cutl_details_boost::re_detail::char_class_cntrl, + ::cutl_details_boost::re_detail::char_class_digit, + ::cutl_details_boost::re_detail::char_class_digit, + ::cutl_details_boost::re_detail::char_class_graph, + ::cutl_details_boost::re_detail::char_class_horizontal_space, + ::cutl_details_boost::re_detail::char_class_lower, + ::cutl_details_boost::re_detail::char_class_lower, + ::cutl_details_boost::re_detail::char_class_print, + ::cutl_details_boost::re_detail::char_class_punct, + ::cutl_details_boost::re_detail::char_class_space, + ::cutl_details_boost::re_detail::char_class_space, + ::cutl_details_boost::re_detail::char_class_upper, + ::cutl_details_boost::re_detail::char_class_unicode, + ::cutl_details_boost::re_detail::char_class_upper, + ::cutl_details_boost::re_detail::char_class_vertical_space, + ::cutl_details_boost::re_detail::char_class_alnum | ::cutl_details_boost::re_detail::char_class_word, + ::cutl_details_boost::re_detail::char_class_alnum | ::cutl_details_boost::re_detail::char_class_word, + ::cutl_details_boost::re_detail::char_class_xdigit, + }; +#endif + if(m_custom_class_names.size()) + { + typedef typename std::map, char_class_type>::const_iterator map_iter; + map_iter pos = m_custom_class_names.find(string_type(p1, p2)); + if(pos != m_custom_class_names.end()) + return pos->second; + } + std::size_t state_id = 1 + re_detail::get_default_class_id(p1, p2); + BOOST_ASSERT(state_id < sizeof(masks) / sizeof(masks[0])); + return masks[state_id]; +} + +#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET +template +bool cpp_regex_traits_implementation::isctype(const charT c, char_class_type mask) const +{ + return + ((mask & ::cutl_details_boost::re_detail::char_class_space) && (this->m_pctype->is(std::ctype::space, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_print) && (this->m_pctype->is(std::ctype::print, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_cntrl) && (this->m_pctype->is(std::ctype::cntrl, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_upper) && (this->m_pctype->is(std::ctype::upper, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_lower) && (this->m_pctype->is(std::ctype::lower, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_alpha) && (this->m_pctype->is(std::ctype::alpha, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_digit) && (this->m_pctype->is(std::ctype::digit, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_punct) && (this->m_pctype->is(std::ctype::punct, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_xdigit) && (this->m_pctype->is(std::ctype::xdigit, c))) + || ((mask & ::cutl_details_boost::re_detail::char_class_blank) && (this->m_pctype->is(std::ctype::space, c)) && !::cutl_details_boost::re_detail::is_separator(c)) + || ((mask & ::cutl_details_boost::re_detail::char_class_word) && (c == '_')) + || ((mask & ::cutl_details_boost::re_detail::char_class_unicode) && ::cutl_details_boost::re_detail::is_extended(c)) + || ((mask & ::cutl_details_boost::re_detail::char_class_vertical_space) && (is_separator(c) || (c == '\v'))) + || ((mask & ::cutl_details_boost::re_detail::char_class_horizontal_space) && this->m_pctype->is(std::ctype::space, c) && !(is_separator(c) || (c == '\v'))); +} +#endif + + +template +inline cutl_details_boost::shared_ptr > create_cpp_regex_traits(const std::locale& l BOOST_APPEND_EXPLICIT_TEMPLATE_TYPE(charT)) +{ + cpp_regex_traits_base key(l); + return ::cutl_details_boost::object_cache, cpp_regex_traits_implementation >::get(key, 5); +} + +} // re_detail + +template +class cpp_regex_traits +{ +private: + typedef std::ctype ctype_type; +public: + typedef charT char_type; + typedef std::size_t size_type; + typedef std::basic_string string_type; + typedef std::locale locale_type; + typedef cutl_details_boost::uint_least32_t char_class_type; + + struct boost_extensions_tag{}; + + cpp_regex_traits() + : m_pimpl(re_detail::create_cpp_regex_traits(std::locale())) + { } + static size_type length(const char_type* p) + { + return std::char_traits::length(p); + } + regex_constants::syntax_type syntax_type(charT c)const + { + return m_pimpl->syntax_type(c); + } + regex_constants::escape_syntax_type escape_syntax_type(charT c) const + { + return m_pimpl->escape_syntax_type(c); + } + charT translate(charT c) const + { + return c; + } + charT translate_nocase(charT c) const + { + return m_pimpl->m_pctype->tolower(c); + } + charT translate(charT c, bool icase) const + { + return icase ? m_pimpl->m_pctype->tolower(c) : c; + } + charT tolower(charT c) const + { + return m_pimpl->m_pctype->tolower(c); + } + charT toupper(charT c) const + { + return m_pimpl->m_pctype->toupper(c); + } + string_type transform(const charT* p1, const charT* p2) const + { + return m_pimpl->transform(p1, p2); + } + string_type transform_primary(const charT* p1, const charT* p2) const + { + return m_pimpl->transform_primary(p1, p2); + } + char_class_type lookup_classname(const charT* p1, const charT* p2) const + { + return m_pimpl->lookup_classname(p1, p2); + } + string_type lookup_collatename(const charT* p1, const charT* p2) const + { + return m_pimpl->lookup_collatename(p1, p2); + } + bool isctype(charT c, char_class_type f) const + { +#ifndef BOOST_REGEX_BUGGY_CTYPE_FACET + typedef typename std::ctype::mask ctype_mask; + + static const ctype_mask mask_base = + static_cast( + std::ctype::alnum + | std::ctype::alpha + | std::ctype::cntrl + | std::ctype::digit + | std::ctype::graph + | std::ctype::lower + | std::ctype::print + | std::ctype::punct + | std::ctype::space + | std::ctype::upper + | std::ctype::xdigit); + + if((f & mask_base) + && (m_pimpl->m_pctype->is( + static_cast(f & mask_base), c))) + return true; + else if((f & re_detail::cpp_regex_traits_implementation::mask_unicode) && re_detail::is_extended(c)) + return true; + else if((f & re_detail::cpp_regex_traits_implementation::mask_word) && (c == '_')) + return true; + else if((f & re_detail::cpp_regex_traits_implementation::mask_blank) + && m_pimpl->m_pctype->is(std::ctype::space, c) + && !re_detail::is_separator(c)) + return true; + else if((f & re_detail::cpp_regex_traits_implementation::mask_vertical) + && (::cutl_details_boost::re_detail::is_separator(c) || (c == '\v'))) + return true; + else if((f & re_detail::cpp_regex_traits_implementation::mask_horizontal) + && this->isctype(c, std::ctype::space) && !this->isctype(c, re_detail::cpp_regex_traits_implementation::mask_vertical)) + return true; + return false; +#else + return m_pimpl->isctype(c, f); +#endif + } + int toi(const charT*& p1, const charT* p2, int radix)const; + int value(charT c, int radix)const + { + const charT* pc = &c; + return toi(pc, pc + 1, radix); + } + locale_type imbue(locale_type l) + { + std::locale result(getloc()); + m_pimpl = re_detail::create_cpp_regex_traits(l); + return result; + } + locale_type getloc()const + { + return m_pimpl->m_locale; + } + std::string error_string(regex_constants::error_type n) const + { + return m_pimpl->error_string(n); + } + + // + // extension: + // set the name of the message catalog in use (defaults to "boost_regex"). + // + static std::string catalog_name(const std::string& name); + static std::string get_catalog_name(); + +private: + cutl_details_boost::shared_ptr > m_pimpl; + // + // catalog name handler: + // + static std::string& get_catalog_name_inst(); + +#ifdef BOOST_HAS_THREADS + static static_mutex& get_mutex_inst(); +#endif +}; + + +template +int cpp_regex_traits::toi(const charT*& first, const charT* last, int radix)const +{ + re_detail::parser_buf sbuf; // buffer for parsing numbers. + std::basic_istream is(&sbuf); // stream for parsing numbers. + + // we do NOT want to parse any thousands separators inside the stream: + last = std::find(first, last, BOOST_USE_FACET(std::numpunct, is.getloc()).thousands_sep()); + + sbuf.pubsetbuf(const_cast(static_cast(first)), static_cast(last-first)); + is.clear(); + if(std::abs(radix) == 16) is >> std::hex; + else if(std::abs(radix) == 8) is >> std::oct; + else is >> std::dec; + int val; + if(is >> val) + { + first = first + ((last - first) - sbuf.in_avail()); + return val; + } + else + return -1; +} + +template +std::string cpp_regex_traits::catalog_name(const std::string& name) +{ +#ifdef BOOST_HAS_THREADS + static_mutex::scoped_lock lk(get_mutex_inst()); +#endif + std::string result(get_catalog_name_inst()); + get_catalog_name_inst() = name; + return result; +} + +template +std::string& cpp_regex_traits::get_catalog_name_inst() +{ + static std::string s_name; + return s_name; +} + +template +std::string cpp_regex_traits::get_catalog_name() +{ +#ifdef BOOST_HAS_THREADS + static_mutex::scoped_lock lk(get_mutex_inst()); +#endif + std::string result(get_catalog_name_inst()); + return result; +} + +#ifdef BOOST_HAS_THREADS +template +static_mutex& cpp_regex_traits::get_mutex_inst() +{ + static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT; + return s_mutex; +} +#endif + + +} // boost + +#ifdef BOOST_MSVC +#pragma warning(pop) +#endif + +#ifdef BOOST_MSVC +#pragma warning(push) +#pragma warning(disable: 4103) +#endif +#ifdef BOOST_HAS_ABI_HEADERS +# include BOOST_ABI_SUFFIX +#endif +#ifdef BOOST_MSVC +#pragma warning(pop) +#endif + +#endif + +#endif + + -- cgit v1.1