6 * Use, modification and distribution are subject to the
7 * Boost Software License, Version 1.0. (See accompanying file
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
13 * LOCATION: see http://www.boost.org for most recent version.
14 * FILE w32_regex_traits.hpp
15 * VERSION see <boost/version.hpp>
16 * DESCRIPTION: Declares regular expression traits class w32_regex_traits.
19 #ifndef BOOST_W32_REGEX_TRAITS_HPP_INCLUDED
20 #define BOOST_W32_REGEX_TRAITS_HPP_INCLUDED
22 #ifndef BOOST_REGEX_NO_WIN32_LOCALE
24 #ifndef BOOST_RE_PAT_EXCEPT_HPP
25 #include <boost/regex/pattern_except.hpp>
27 #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
28 #include <boost/regex/v4/regex_traits_defaults.hpp>
30 #ifdef BOOST_HAS_THREADS
31 #include <boost/regex/pending/static_mutex.hpp>
33 #ifndef BOOST_REGEX_PRIMARY_TRANSFORM
34 #include <boost/regex/v4/primary_transform.hpp>
36 #ifndef BOOST_REGEX_OBJECT_CACHE_HPP
37 #include <boost/regex/pending/object_cache.hpp>
42 #pragma warning(disable: 4103)
44 #ifdef BOOST_HAS_ABI_HEADERS
45 # include BOOST_ABI_PREFIX
53 #pragma warning(disable:4786)
54 #pragma warning(disable:4800)
60 // forward declaration is needed by some compilers:
62 template <class charT>
63 class w32_regex_traits;
65 namespace BOOST_REGEX_DETAIL_NS{
68 // start by typedeffing the types we'll need:
70 typedef ::boost::uint32_t lcid_type; // placeholder for LCID.
71 typedef ::boost::shared_ptr<void> cat_type; // placeholder for dll HANDLE.
74 // then add wrappers around the actual Win32 API's (ie implementation hiding):
76 BOOST_REGEX_DECL lcid_type BOOST_REGEX_CALL w32_get_default_locale();
77 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(char, lcid_type);
78 #ifndef BOOST_NO_WREGEX
79 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(wchar_t, lcid_type);
80 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
81 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(unsigned short ca, lcid_type state_id);
84 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(char, lcid_type);
85 #ifndef BOOST_NO_WREGEX
86 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(wchar_t, lcid_type);
87 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
88 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(unsigned short ca, lcid_type state_id);
91 BOOST_REGEX_DECL cat_type BOOST_REGEX_CALL w32_cat_open(const std::string& name);
92 BOOST_REGEX_DECL std::string BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type state_id, int i, const std::string& def);
93 #ifndef BOOST_NO_WREGEX
94 BOOST_REGEX_DECL std::wstring BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type state_id, int i, const std::wstring& def);
95 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
96 BOOST_REGEX_DECL std::basic_string<unsigned short> BOOST_REGEX_CALL w32_cat_get(const cat_type& cat, lcid_type, int i, const std::basic_string<unsigned short>& def);
99 BOOST_REGEX_DECL std::string BOOST_REGEX_CALL w32_transform(lcid_type state_id, const char* p1, const char* p2);
100 #ifndef BOOST_NO_WREGEX
101 BOOST_REGEX_DECL std::wstring BOOST_REGEX_CALL w32_transform(lcid_type state_id, const wchar_t* p1, const wchar_t* p2);
102 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
103 BOOST_REGEX_DECL std::basic_string<unsigned short> BOOST_REGEX_CALL w32_transform(lcid_type state_id, const unsigned short* p1, const unsigned short* p2);
106 BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_tolower(char c, lcid_type);
107 #ifndef BOOST_NO_WREGEX
108 BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_tolower(wchar_t c, lcid_type);
109 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
110 BOOST_REGEX_DECL unsigned short BOOST_REGEX_CALL w32_tolower(unsigned short c, lcid_type state_id);
113 BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_toupper(char c, lcid_type);
114 #ifndef BOOST_NO_WREGEX
115 BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_toupper(wchar_t c, lcid_type);
117 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, char c);
118 #ifndef BOOST_NO_WREGEX
119 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, wchar_t c);
120 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
121 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type state_id, boost::uint32_t m, unsigned short c);
125 // class w32_regex_traits_base:
126 // acts as a container for locale and the facets we are using.
128 template <class charT>
129 struct w32_regex_traits_base
131 w32_regex_traits_base(lcid_type l)
133 lcid_type imbue(lcid_type l);
138 template <class charT>
139 inline lcid_type w32_regex_traits_base<charT>::imbue(lcid_type l)
141 lcid_type result(m_locale);
147 // class w32_regex_traits_char_layer:
148 // implements methods that require specialisation for narrow characters:
150 template <class charT>
151 class w32_regex_traits_char_layer : public w32_regex_traits_base<charT>
153 typedef std::basic_string<charT> string_type;
154 typedef std::map<charT, regex_constants::syntax_type> map_type;
155 typedef typename map_type::const_iterator map_iterator_type;
157 w32_regex_traits_char_layer(const lcid_type l);
159 regex_constants::syntax_type syntax_type(charT c)const
161 map_iterator_type i = m_char_map.find(c);
162 return ((i == m_char_map.end()) ? 0 : i->second);
164 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
166 map_iterator_type i = m_char_map.find(c);
167 if(i == m_char_map.end())
169 if(::boost::BOOST_REGEX_DETAIL_NS::w32_is_lower(c, this->m_locale)) return regex_constants::escape_type_class;
170 if(::boost::BOOST_REGEX_DETAIL_NS::w32_is_upper(c, this->m_locale)) return regex_constants::escape_type_not_class;
175 charT tolower(charT c)const
177 return ::boost::BOOST_REGEX_DETAIL_NS::w32_tolower(c, this->m_locale);
179 bool isctype(boost::uint32_t mask, charT c)const
181 return ::boost::BOOST_REGEX_DETAIL_NS::w32_is(this->m_locale, mask, c);
185 string_type get_default_message(regex_constants::syntax_type);
186 // TODO: use a hash table when available!
190 template <class charT>
191 w32_regex_traits_char_layer<charT>::w32_regex_traits_char_layer(::boost::BOOST_REGEX_DETAIL_NS::lcid_type l)
192 : w32_regex_traits_base<charT>(l)
194 // we need to start by initialising our syntax map so we know which
195 // character is used for which purpose:
197 std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
200 cat = ::boost::BOOST_REGEX_DETAIL_NS::w32_cat_open(cat_name);
203 std::string m("Unable to open message catalog: ");
204 std::runtime_error err(m + cat_name);
205 boost::BOOST_REGEX_DETAIL_NS::raise_runtime_error(err);
209 // if we have a valid catalog then load our messages:
213 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
215 string_type mss = ::boost::BOOST_REGEX_DETAIL_NS::w32_cat_get(cat, this->m_locale, i, get_default_message(i));
216 for(typename string_type::size_type j = 0; j < mss.size(); ++j)
218 this->m_char_map[mss[j]] = i;
224 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
226 const char* ptr = get_default_syntax(i);
229 this->m_char_map[static_cast<charT>(*ptr)] = i;
236 template <class charT>
237 typename w32_regex_traits_char_layer<charT>::string_type
238 w32_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
240 const char* ptr = get_default_syntax(i);
244 result.append(1, static_cast<charT>(*ptr));
251 // specialised version for narrow characters:
254 class BOOST_REGEX_DECL w32_regex_traits_char_layer<char> : public w32_regex_traits_base<char>
256 typedef std::string string_type;
258 w32_regex_traits_char_layer(::boost::BOOST_REGEX_DETAIL_NS::lcid_type l)
259 : w32_regex_traits_base<char>(l)
264 regex_constants::syntax_type syntax_type(char c)const
266 return m_char_map[static_cast<unsigned char>(c)];
268 regex_constants::escape_syntax_type escape_syntax_type(char c) const
270 return m_char_map[static_cast<unsigned char>(c)];
272 char tolower(char c)const
274 return m_lower_map[static_cast<unsigned char>(c)];
276 bool isctype(boost::uint32_t mask, char c)const
278 return m_type_map[static_cast<unsigned char>(c)] & mask;
282 regex_constants::syntax_type m_char_map[1u << CHAR_BIT];
283 char m_lower_map[1u << CHAR_BIT];
284 boost::uint16_t m_type_map[1u << CHAR_BIT];
289 // class w32_regex_traits_implementation:
290 // provides pimpl implementation for w32_regex_traits.
292 template <class charT>
293 class w32_regex_traits_implementation : public w32_regex_traits_char_layer<charT>
296 typedef typename w32_regex_traits<charT>::char_class_type char_class_type;
297 BOOST_STATIC_CONSTANT(char_class_type, mask_word = 0x0400); // must be C1_DEFINED << 1
298 BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 0x0800); // must be C1_DEFINED << 2
299 BOOST_STATIC_CONSTANT(char_class_type, mask_horizontal = 0x1000); // must be C1_DEFINED << 3
300 BOOST_STATIC_CONSTANT(char_class_type, mask_vertical = 0x2000); // must be C1_DEFINED << 4
301 BOOST_STATIC_CONSTANT(char_class_type, mask_base = 0x3ff); // all the masks used by the CT_CTYPE1 group
303 typedef std::basic_string<charT> string_type;
304 typedef charT char_type;
305 w32_regex_traits_implementation(::boost::BOOST_REGEX_DETAIL_NS::lcid_type l);
306 std::string error_string(regex_constants::error_type n) const
308 if(!m_error_strings.empty())
310 std::map<int, std::string>::const_iterator p = m_error_strings.find(n);
311 return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;
313 return get_default_error_string(n);
315 char_class_type lookup_classname(const charT* p1, const charT* p2) const
317 char_class_type result = lookup_classname_imp(p1, p2);
320 typedef typename string_type::size_type size_type;
321 string_type temp(p1, p2);
322 for(size_type i = 0; i < temp.size(); ++i)
323 temp[i] = this->tolower(temp[i]);
324 result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());
328 string_type lookup_collatename(const charT* p1, const charT* p2) const;
329 string_type transform_primary(const charT* p1, const charT* p2) const;
330 string_type transform(const charT* p1, const charT* p2) const
332 return ::boost::BOOST_REGEX_DETAIL_NS::w32_transform(this->m_locale, p1, p2);
335 std::map<int, std::string> m_error_strings; // error messages indexed by numberic ID
336 std::map<string_type, char_class_type> m_custom_class_names; // character class names
337 std::map<string_type, string_type> m_custom_collate_names; // collating element names
338 unsigned m_collate_type; // the form of the collation string
339 charT m_collate_delim; // the collation group delimiter
343 char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;
346 template <class charT>
347 typename w32_regex_traits_implementation<charT>::string_type
348 w32_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const
352 // What we do here depends upon the format of the sort key returned by
353 // sort key returned by this->transform:
355 switch(m_collate_type)
359 // the best we can do is translate to lower case, then get a regular sort key:
361 result.assign(p1, p2);
362 typedef typename string_type::size_type size_type;
363 for(size_type i = 0; i < result.size(); ++i)
364 result[i] = this->tolower(result[i]);
365 result = this->transform(&*result.begin(), &*result.begin() + result.size());
370 // get a regular sort key, and then truncate it:
371 result.assign(this->transform(p1, p2));
372 result.erase(this->m_collate_delim);
376 // get a regular sort key, and then truncate everything after the delim:
377 result.assign(this->transform(p1, p2));
379 for(i = 0; i < result.size(); ++i)
381 if(result[i] == m_collate_delim)
388 result = string_type(1, charT(0));
392 template <class charT>
393 typename w32_regex_traits_implementation<charT>::string_type
394 w32_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const
396 typedef typename std::map<string_type, string_type>::const_iterator iter_type;
397 if(m_custom_collate_names.size())
399 iter_type pos = m_custom_collate_names.find(string_type(p1, p2));
400 if(pos != m_custom_collate_names.end())
403 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
404 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
405 std::string name(p1, p2);
408 const charT* p0 = p1;
410 name.append(1, char(*p0++));
412 name = lookup_default_collate_name(name);
413 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
414 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
416 return string_type(name.begin(), name.end());
421 typedef std::string::const_iterator iter;
422 iter b = name.begin();
425 result.append(1, charT(*b++));
430 return string_type(1, *p1);
431 return string_type();
434 template <class charT>
435 w32_regex_traits_implementation<charT>::w32_regex_traits_implementation(::boost::BOOST_REGEX_DETAIL_NS::lcid_type l)
436 : w32_regex_traits_char_layer<charT>(l)
439 std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
442 cat = ::boost::BOOST_REGEX_DETAIL_NS::w32_cat_open(cat_name);
445 std::string m("Unable to open message catalog: ");
446 std::runtime_error err(m + cat_name);
447 boost::BOOST_REGEX_DETAIL_NS::raise_runtime_error(err);
451 // if we have a valid catalog then load our messages:
458 for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0);
459 i <= boost::regex_constants::error_unknown;
460 i = static_cast<boost::regex_constants::error_type>(i + 1))
462 const char* p = get_default_error_string(i);
463 string_type default_message;
466 default_message.append(1, static_cast<charT>(*p));
469 string_type s = ::boost::BOOST_REGEX_DETAIL_NS::w32_cat_get(cat, this->m_locale, i+200, default_message);
471 for(std::string::size_type j = 0; j < s.size(); ++j)
473 result.append(1, static_cast<char>(s[j]));
475 m_error_strings[i] = result;
478 // Custom class names:
480 static const char_class_type masks[14] =
482 0x0104u, // C1_ALPHA | C1_DIGIT
486 (~(0x0020u|0x0008u) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE
488 (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
492 0x0080u, // C1_XDIGIT
494 w32_regex_traits_implementation<charT>::mask_word,
495 w32_regex_traits_implementation<charT>::mask_unicode,
497 static const string_type null_string;
498 for(unsigned int j = 0; j <= 13; ++j)
500 string_type s(::boost::BOOST_REGEX_DETAIL_NS::w32_cat_get(cat, this->m_locale, j+300, null_string));
502 this->m_custom_class_names[s] = masks[j];
506 // get the collation format used by m_pcollate:
508 m_collate_type = BOOST_REGEX_DETAIL_NS::find_sort_syntax(this, &m_collate_delim);
511 template <class charT>
512 typename w32_regex_traits_implementation<charT>::char_class_type
513 w32_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
515 static const char_class_type masks[22] =
518 0x0104u, // C1_ALPHA | C1_DIGIT
524 (~(0x0020u|0x0008u|0x0040) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE or C1_BLANK
525 w32_regex_traits_implementation<charT>::mask_horizontal,
528 (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
533 w32_regex_traits_implementation<charT>::mask_unicode,
535 w32_regex_traits_implementation<charT>::mask_vertical,
536 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
537 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
538 0x0080u, // C1_XDIGIT
540 if(m_custom_class_names.size())
542 typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
543 map_iter pos = m_custom_class_names.find(string_type(p1, p2));
544 if(pos != m_custom_class_names.end())
547 std::size_t state_id = 1 + BOOST_REGEX_DETAIL_NS::get_default_class_id(p1, p2);
548 if(state_id < sizeof(masks) / sizeof(masks[0]))
549 return masks[state_id];
554 template <class charT>
555 boost::shared_ptr<const w32_regex_traits_implementation<charT> > create_w32_regex_traits(::boost::BOOST_REGEX_DETAIL_NS::lcid_type l)
557 // TODO: create a cache for previously constructed objects.
558 return boost::object_cache< ::boost::BOOST_REGEX_DETAIL_NS::lcid_type, w32_regex_traits_implementation<charT> >::get(l, 5);
561 } // BOOST_REGEX_DETAIL_NS
563 template <class charT>
564 class w32_regex_traits
567 typedef charT char_type;
568 typedef std::size_t size_type;
569 typedef std::basic_string<char_type> string_type;
570 typedef ::boost::BOOST_REGEX_DETAIL_NS::lcid_type locale_type;
571 typedef boost::uint_least32_t char_class_type;
573 struct boost_extensions_tag{};
576 : m_pimpl(BOOST_REGEX_DETAIL_NS::create_w32_regex_traits<charT>(::boost::BOOST_REGEX_DETAIL_NS::w32_get_default_locale()))
578 static size_type length(const char_type* p)
580 return std::char_traits<charT>::length(p);
582 regex_constants::syntax_type syntax_type(charT c)const
584 return m_pimpl->syntax_type(c);
586 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
588 return m_pimpl->escape_syntax_type(c);
590 charT translate(charT c) const
594 charT translate_nocase(charT c) const
596 return this->m_pimpl->tolower(c);
598 charT translate(charT c, bool icase) const
600 return icase ? this->m_pimpl->tolower(c) : c;
602 charT tolower(charT c) const
604 return this->m_pimpl->tolower(c);
606 charT toupper(charT c) const
608 return ::boost::BOOST_REGEX_DETAIL_NS::w32_toupper(c, this->m_pimpl->m_locale);
610 string_type transform(const charT* p1, const charT* p2) const
612 return ::boost::BOOST_REGEX_DETAIL_NS::w32_transform(this->m_pimpl->m_locale, p1, p2);
614 string_type transform_primary(const charT* p1, const charT* p2) const
616 return m_pimpl->transform_primary(p1, p2);
618 char_class_type lookup_classname(const charT* p1, const charT* p2) const
620 return m_pimpl->lookup_classname(p1, p2);
622 string_type lookup_collatename(const charT* p1, const charT* p2) const
624 return m_pimpl->lookup_collatename(p1, p2);
626 bool isctype(charT c, char_class_type f) const
628 if((f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_base)
629 && (this->m_pimpl->isctype(f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_base, c)))
631 else if((f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_unicode) && BOOST_REGEX_DETAIL_NS::is_extended(c))
633 else if((f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_word) && (c == '_'))
635 else if((f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_vertical)
636 && (::boost::BOOST_REGEX_DETAIL_NS::is_separator(c) || (c == '\v')))
638 else if((f & BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_horizontal)
639 && this->isctype(c, 0x0008u) && !this->isctype(c, BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT>::mask_vertical))
643 boost::intmax_t toi(const charT*& p1, const charT* p2, int radix)const
645 return ::boost::BOOST_REGEX_DETAIL_NS::global_toi(p1, p2, radix, *this);
647 int value(charT c, int radix)const
649 int result = (int)::boost::BOOST_REGEX_DETAIL_NS::global_value(c);
650 return result < radix ? result : -1;
652 locale_type imbue(locale_type l)
654 ::boost::BOOST_REGEX_DETAIL_NS::lcid_type result(getloc());
655 m_pimpl = BOOST_REGEX_DETAIL_NS::create_w32_regex_traits<charT>(l);
658 locale_type getloc()const
660 return m_pimpl->m_locale;
662 std::string error_string(regex_constants::error_type n) const
664 return m_pimpl->error_string(n);
669 // set the name of the message catalog in use (defaults to "boost_regex").
671 static std::string catalog_name(const std::string& name);
672 static std::string get_catalog_name();
675 boost::shared_ptr<const BOOST_REGEX_DETAIL_NS::w32_regex_traits_implementation<charT> > m_pimpl;
677 // catalog name handler:
679 static std::string& get_catalog_name_inst();
681 #ifdef BOOST_HAS_THREADS
682 static static_mutex& get_mutex_inst();
686 template <class charT>
687 std::string w32_regex_traits<charT>::catalog_name(const std::string& name)
689 #ifdef BOOST_HAS_THREADS
690 static_mutex::scoped_lock lk(get_mutex_inst());
692 std::string result(get_catalog_name_inst());
693 get_catalog_name_inst() = name;
697 template <class charT>
698 std::string& w32_regex_traits<charT>::get_catalog_name_inst()
700 static std::string s_name;
704 template <class charT>
705 std::string w32_regex_traits<charT>::get_catalog_name()
707 #ifdef BOOST_HAS_THREADS
708 static_mutex::scoped_lock lk(get_mutex_inst());
710 std::string result(get_catalog_name_inst());
714 #ifdef BOOST_HAS_THREADS
715 template <class charT>
716 static_mutex& w32_regex_traits<charT>::get_mutex_inst()
718 static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT;
731 #pragma warning(push)
732 #pragma warning(disable: 4103)
734 #ifdef BOOST_HAS_ABI_HEADERS
735 # include BOOST_ABI_SUFFIX
741 #endif // BOOST_REGEX_NO_WIN32_LOCALE