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_RE_PAT_EXCEPT_HPP
23 #include <boost/regex/pattern_except.hpp>
25 #ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED
26 #include <boost/regex/v4/regex_traits_defaults.hpp>
28 #ifdef BOOST_HAS_THREADS
29 #include <boost/regex/pending/static_mutex.hpp>
31 #ifndef BOOST_REGEX_PRIMARY_TRANSFORM
32 #include <boost/regex/v4/primary_transform.hpp>
34 #ifndef BOOST_REGEX_OBJECT_CACHE_HPP
35 #include <boost/regex/pending/object_cache.hpp>
40 #pragma warning(disable: 4103)
42 #ifdef BOOST_HAS_ABI_HEADERS
43 # include BOOST_ABI_PREFIX
51 #pragma warning(disable:4786)
52 #pragma warning(disable:4800)
58 // forward declaration is needed by some compilers:
60 template <class charT>
61 class w32_regex_traits;
66 // start by typedeffing the types we'll need:
68 typedef ::boost::uint32_t lcid_type; // placeholder for LCID.
69 typedef ::boost::shared_ptr<void> cat_type; // placeholder for dll HANDLE.
72 // then add wrappers around the actual Win32 API's (ie implementation hiding):
74 BOOST_REGEX_DECL lcid_type BOOST_REGEX_CALL w32_get_default_locale();
75 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(char, lcid_type);
76 #ifndef BOOST_NO_WREGEX
77 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(wchar_t, lcid_type);
78 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
79 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_lower(unsigned short ca, lcid_type state_id);
82 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(char, lcid_type);
83 #ifndef BOOST_NO_WREGEX
84 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(wchar_t, lcid_type);
85 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
86 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is_upper(unsigned short ca, lcid_type state_id);
89 BOOST_REGEX_DECL cat_type BOOST_REGEX_CALL w32_cat_open(const std::string& name);
90 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);
91 #ifndef BOOST_NO_WREGEX
92 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);
93 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
94 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);
97 BOOST_REGEX_DECL std::string BOOST_REGEX_CALL w32_transform(lcid_type state_id, const char* p1, const char* p2);
98 #ifndef BOOST_NO_WREGEX
99 BOOST_REGEX_DECL std::wstring BOOST_REGEX_CALL w32_transform(lcid_type state_id, const wchar_t* p1, const wchar_t* p2);
100 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
101 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);
104 BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_tolower(char c, lcid_type);
105 #ifndef BOOST_NO_WREGEX
106 BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_tolower(wchar_t c, lcid_type);
107 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
108 BOOST_REGEX_DECL unsigned short BOOST_REGEX_CALL w32_tolower(unsigned short c, lcid_type state_id);
111 BOOST_REGEX_DECL char BOOST_REGEX_CALL w32_toupper(char c, lcid_type);
112 #ifndef BOOST_NO_WREGEX
113 BOOST_REGEX_DECL wchar_t BOOST_REGEX_CALL w32_toupper(wchar_t c, lcid_type);
115 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, char c);
116 #ifndef BOOST_NO_WREGEX
117 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type, boost::uint32_t mask, wchar_t c);
118 #ifdef BOOST_REGEX_HAS_OTHER_WCHAR_T
119 BOOST_REGEX_DECL bool BOOST_REGEX_CALL w32_is(lcid_type state_id, boost::uint32_t m, unsigned short c);
123 // class w32_regex_traits_base:
124 // acts as a container for locale and the facets we are using.
126 template <class charT>
127 struct w32_regex_traits_base
129 w32_regex_traits_base(lcid_type l)
131 lcid_type imbue(lcid_type l);
136 template <class charT>
137 inline lcid_type w32_regex_traits_base<charT>::imbue(lcid_type l)
139 lcid_type result(m_locale);
145 // class w32_regex_traits_char_layer:
146 // implements methods that require specialisation for narrow characters:
148 template <class charT>
149 class w32_regex_traits_char_layer : public w32_regex_traits_base<charT>
151 typedef std::basic_string<charT> string_type;
152 typedef std::map<charT, regex_constants::syntax_type> map_type;
153 typedef typename map_type::const_iterator map_iterator_type;
155 w32_regex_traits_char_layer(const lcid_type l);
157 regex_constants::syntax_type syntax_type(charT c)const
159 map_iterator_type i = m_char_map.find(c);
160 return ((i == m_char_map.end()) ? 0 : i->second);
162 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
164 map_iterator_type i = m_char_map.find(c);
165 if(i == m_char_map.end())
167 if(::boost::re_detail::w32_is_lower(c, this->m_locale)) return regex_constants::escape_type_class;
168 if(::boost::re_detail::w32_is_upper(c, this->m_locale)) return regex_constants::escape_type_not_class;
173 charT tolower(charT c)const
175 return ::boost::re_detail::w32_tolower(c, this->m_locale);
177 bool isctype(boost::uint32_t mask, charT c)const
179 return ::boost::re_detail::w32_is(this->m_locale, mask, c);
183 string_type get_default_message(regex_constants::syntax_type);
184 // TODO: use a hash table when available!
188 template <class charT>
189 w32_regex_traits_char_layer<charT>::w32_regex_traits_char_layer(::boost::re_detail::lcid_type l)
190 : w32_regex_traits_base<charT>(l)
192 // we need to start by initialising our syntax map so we know which
193 // character is used for which purpose:
195 std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
198 cat = ::boost::re_detail::w32_cat_open(cat_name);
201 std::string m("Unable to open message catalog: ");
202 std::runtime_error err(m + cat_name);
203 boost::re_detail::raise_runtime_error(err);
207 // if we have a valid catalog then load our messages:
211 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
213 string_type mss = ::boost::re_detail::w32_cat_get(cat, this->m_locale, i, get_default_message(i));
214 for(typename string_type::size_type j = 0; j < mss.size(); ++j)
216 this->m_char_map[mss[j]] = i;
222 for(regex_constants::syntax_type i = 1; i < regex_constants::syntax_max; ++i)
224 const char* ptr = get_default_syntax(i);
227 this->m_char_map[static_cast<charT>(*ptr)] = i;
234 template <class charT>
235 typename w32_regex_traits_char_layer<charT>::string_type
236 w32_regex_traits_char_layer<charT>::get_default_message(regex_constants::syntax_type i)
238 const char* ptr = get_default_syntax(i);
242 result.append(1, static_cast<charT>(*ptr));
249 // specialised version for narrow characters:
252 class BOOST_REGEX_DECL w32_regex_traits_char_layer<char> : public w32_regex_traits_base<char>
254 typedef std::string string_type;
256 w32_regex_traits_char_layer(::boost::re_detail::lcid_type l)
257 : w32_regex_traits_base<char>(l)
262 regex_constants::syntax_type syntax_type(char c)const
264 return m_char_map[static_cast<unsigned char>(c)];
266 regex_constants::escape_syntax_type escape_syntax_type(char c) const
268 return m_char_map[static_cast<unsigned char>(c)];
270 char tolower(char c)const
272 return m_lower_map[static_cast<unsigned char>(c)];
274 bool isctype(boost::uint32_t mask, char c)const
276 return m_type_map[static_cast<unsigned char>(c)] & mask;
280 regex_constants::syntax_type m_char_map[1u << CHAR_BIT];
281 char m_lower_map[1u << CHAR_BIT];
282 boost::uint16_t m_type_map[1u << CHAR_BIT];
287 // class w32_regex_traits_implementation:
288 // provides pimpl implementation for w32_regex_traits.
290 template <class charT>
291 class w32_regex_traits_implementation : public w32_regex_traits_char_layer<charT>
294 typedef typename w32_regex_traits<charT>::char_class_type char_class_type;
295 BOOST_STATIC_CONSTANT(char_class_type, mask_word = 0x0400); // must be C1_DEFINED << 1
296 BOOST_STATIC_CONSTANT(char_class_type, mask_unicode = 0x0800); // must be C1_DEFINED << 2
297 BOOST_STATIC_CONSTANT(char_class_type, mask_horizontal = 0x1000); // must be C1_DEFINED << 3
298 BOOST_STATIC_CONSTANT(char_class_type, mask_vertical = 0x2000); // must be C1_DEFINED << 4
299 BOOST_STATIC_CONSTANT(char_class_type, mask_base = 0x3ff); // all the masks used by the CT_CTYPE1 group
301 typedef std::basic_string<charT> string_type;
302 typedef charT char_type;
303 w32_regex_traits_implementation(::boost::re_detail::lcid_type l);
304 std::string error_string(regex_constants::error_type n) const
306 if(!m_error_strings.empty())
308 std::map<int, std::string>::const_iterator p = m_error_strings.find(n);
309 return (p == m_error_strings.end()) ? std::string(get_default_error_string(n)) : p->second;
311 return get_default_error_string(n);
313 char_class_type lookup_classname(const charT* p1, const charT* p2) const
315 char_class_type result = lookup_classname_imp(p1, p2);
318 typedef typename string_type::size_type size_type;
319 string_type temp(p1, p2);
320 for(size_type i = 0; i < temp.size(); ++i)
321 temp[i] = this->tolower(temp[i]);
322 result = lookup_classname_imp(&*temp.begin(), &*temp.begin() + temp.size());
326 string_type lookup_collatename(const charT* p1, const charT* p2) const;
327 string_type transform_primary(const charT* p1, const charT* p2) const;
328 string_type transform(const charT* p1, const charT* p2) const
330 return ::boost::re_detail::w32_transform(this->m_locale, p1, p2);
333 std::map<int, std::string> m_error_strings; // error messages indexed by numberic ID
334 std::map<string_type, char_class_type> m_custom_class_names; // character class names
335 std::map<string_type, string_type> m_custom_collate_names; // collating element names
336 unsigned m_collate_type; // the form of the collation string
337 charT m_collate_delim; // the collation group delimiter
341 char_class_type lookup_classname_imp(const charT* p1, const charT* p2) const;
344 template <class charT>
345 typename w32_regex_traits_implementation<charT>::string_type
346 w32_regex_traits_implementation<charT>::transform_primary(const charT* p1, const charT* p2) const
350 // What we do here depends upon the format of the sort key returned by
351 // sort key returned by this->transform:
353 switch(m_collate_type)
357 // the best we can do is translate to lower case, then get a regular sort key:
359 result.assign(p1, p2);
360 typedef typename string_type::size_type size_type;
361 for(size_type i = 0; i < result.size(); ++i)
362 result[i] = this->tolower(result[i]);
363 result = this->transform(&*result.begin(), &*result.begin() + result.size());
368 // get a regular sort key, and then truncate it:
369 result.assign(this->transform(p1, p2));
370 result.erase(this->m_collate_delim);
374 // get a regular sort key, and then truncate everything after the delim:
375 result.assign(this->transform(p1, p2));
377 for(i = 0; i < result.size(); ++i)
379 if(result[i] == m_collate_delim)
386 result = string_type(1, charT(0));
390 template <class charT>
391 typename w32_regex_traits_implementation<charT>::string_type
392 w32_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const
394 typedef typename std::map<string_type, string_type>::const_iterator iter_type;
395 if(m_custom_collate_names.size())
397 iter_type pos = m_custom_collate_names.find(string_type(p1, p2));
398 if(pos != m_custom_collate_names.end())
401 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
402 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
403 std::string name(p1, p2);
406 const charT* p0 = p1;
408 name.append(1, char(*p0++));
410 name = lookup_default_collate_name(name);
411 #if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\
412 && !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)
414 return string_type(name.begin(), name.end());
419 typedef std::string::const_iterator iter;
420 iter b = name.begin();
423 result.append(1, charT(*b++));
428 return string_type(1, *p1);
429 return string_type();
432 template <class charT>
433 w32_regex_traits_implementation<charT>::w32_regex_traits_implementation(::boost::re_detail::lcid_type l)
434 : w32_regex_traits_char_layer<charT>(l)
437 std::string cat_name(w32_regex_traits<charT>::get_catalog_name());
440 cat = ::boost::re_detail::w32_cat_open(cat_name);
443 std::string m("Unable to open message catalog: ");
444 std::runtime_error err(m + cat_name);
445 boost::re_detail::raise_runtime_error(err);
449 // if we have a valid catalog then load our messages:
456 for(boost::regex_constants::error_type i = static_cast<boost::regex_constants::error_type>(0);
457 i <= boost::regex_constants::error_unknown;
458 i = static_cast<boost::regex_constants::error_type>(i + 1))
460 const char* p = get_default_error_string(i);
461 string_type default_message;
464 default_message.append(1, static_cast<charT>(*p));
467 string_type s = ::boost::re_detail::w32_cat_get(cat, this->m_locale, i+200, default_message);
469 for(std::string::size_type j = 0; j < s.size(); ++j)
471 result.append(1, static_cast<char>(s[j]));
473 m_error_strings[i] = result;
476 // Custom class names:
478 static const char_class_type masks[14] =
480 0x0104u, // C1_ALPHA | C1_DIGIT
484 (~(0x0020u|0x0008u) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE
486 (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
490 0x0080u, // C1_XDIGIT
492 w32_regex_traits_implementation<charT>::mask_word,
493 w32_regex_traits_implementation<charT>::mask_unicode,
495 static const string_type null_string;
496 for(unsigned int j = 0; j <= 13; ++j)
498 string_type s(::boost::re_detail::w32_cat_get(cat, this->m_locale, j+300, null_string));
500 this->m_custom_class_names[s] = masks[j];
504 // get the collation format used by m_pcollate:
506 m_collate_type = re_detail::find_sort_syntax(this, &m_collate_delim);
509 template <class charT>
510 typename w32_regex_traits_implementation<charT>::char_class_type
511 w32_regex_traits_implementation<charT>::lookup_classname_imp(const charT* p1, const charT* p2) const
513 static const char_class_type masks[22] =
516 0x0104u, // C1_ALPHA | C1_DIGIT
522 (~(0x0020u|0x0008u|0x0040) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE or C1_BLANK
523 w32_regex_traits_implementation<charT>::mask_horizontal,
526 (~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL
531 w32_regex_traits_implementation<charT>::mask_unicode,
533 w32_regex_traits_implementation<charT>::mask_vertical,
534 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
535 0x0104u | w32_regex_traits_implementation<charT>::mask_word,
536 0x0080u, // C1_XDIGIT
538 if(m_custom_class_names.size())
540 typedef typename std::map<std::basic_string<charT>, char_class_type>::const_iterator map_iter;
541 map_iter pos = m_custom_class_names.find(string_type(p1, p2));
542 if(pos != m_custom_class_names.end())
545 std::size_t state_id = 1 + re_detail::get_default_class_id(p1, p2);
546 if(state_id < sizeof(masks) / sizeof(masks[0]))
547 return masks[state_id];
552 template <class charT>
553 boost::shared_ptr<const w32_regex_traits_implementation<charT> > create_w32_regex_traits(::boost::re_detail::lcid_type l)
555 // TODO: create a cache for previously constructed objects.
556 return boost::object_cache< ::boost::re_detail::lcid_type, w32_regex_traits_implementation<charT> >::get(l, 5);
561 template <class charT>
562 class w32_regex_traits
565 typedef charT char_type;
566 typedef std::size_t size_type;
567 typedef std::basic_string<char_type> string_type;
568 typedef ::boost::re_detail::lcid_type locale_type;
569 typedef boost::uint_least32_t char_class_type;
571 struct boost_extensions_tag{};
574 : m_pimpl(re_detail::create_w32_regex_traits<charT>(::boost::re_detail::w32_get_default_locale()))
576 static size_type length(const char_type* p)
578 return std::char_traits<charT>::length(p);
580 regex_constants::syntax_type syntax_type(charT c)const
582 return m_pimpl->syntax_type(c);
584 regex_constants::escape_syntax_type escape_syntax_type(charT c) const
586 return m_pimpl->escape_syntax_type(c);
588 charT translate(charT c) const
592 charT translate_nocase(charT c) const
594 return this->m_pimpl->tolower(c);
596 charT translate(charT c, bool icase) const
598 return icase ? this->m_pimpl->tolower(c) : c;
600 charT tolower(charT c) const
602 return this->m_pimpl->tolower(c);
604 charT toupper(charT c) const
606 return ::boost::re_detail::w32_toupper(c, this->m_pimpl->m_locale);
608 string_type transform(const charT* p1, const charT* p2) const
610 return ::boost::re_detail::w32_transform(this->m_pimpl->m_locale, p1, p2);
612 string_type transform_primary(const charT* p1, const charT* p2) const
614 return m_pimpl->transform_primary(p1, p2);
616 char_class_type lookup_classname(const charT* p1, const charT* p2) const
618 return m_pimpl->lookup_classname(p1, p2);
620 string_type lookup_collatename(const charT* p1, const charT* p2) const
622 return m_pimpl->lookup_collatename(p1, p2);
624 bool isctype(charT c, char_class_type f) const
626 if((f & re_detail::w32_regex_traits_implementation<charT>::mask_base)
627 && (this->m_pimpl->isctype(f & re_detail::w32_regex_traits_implementation<charT>::mask_base, c)))
629 else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_unicode) && re_detail::is_extended(c))
631 else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_word) && (c == '_'))
633 else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_vertical)
634 && (::boost::re_detail::is_separator(c) || (c == '\v')))
636 else if((f & re_detail::w32_regex_traits_implementation<charT>::mask_horizontal)
637 && this->isctype(c, 0x0008u) && !this->isctype(c, re_detail::w32_regex_traits_implementation<charT>::mask_vertical))
641 int toi(const charT*& p1, const charT* p2, int radix)const
643 return ::boost::re_detail::global_toi(p1, p2, radix, *this);
645 int value(charT c, int radix)const
647 int result = ::boost::re_detail::global_value(c);
648 return result < radix ? result : -1;
650 locale_type imbue(locale_type l)
652 ::boost::re_detail::lcid_type result(getloc());
653 m_pimpl = re_detail::create_w32_regex_traits<charT>(l);
656 locale_type getloc()const
658 return m_pimpl->m_locale;
660 std::string error_string(regex_constants::error_type n) const
662 return m_pimpl->error_string(n);
667 // set the name of the message catalog in use (defaults to "boost_regex").
669 static std::string catalog_name(const std::string& name);
670 static std::string get_catalog_name();
673 boost::shared_ptr<const re_detail::w32_regex_traits_implementation<charT> > m_pimpl;
675 // catalog name handler:
677 static std::string& get_catalog_name_inst();
679 #ifdef BOOST_HAS_THREADS
680 static static_mutex& get_mutex_inst();
684 template <class charT>
685 std::string w32_regex_traits<charT>::catalog_name(const std::string& name)
687 #ifdef BOOST_HAS_THREADS
688 static_mutex::scoped_lock lk(get_mutex_inst());
690 std::string result(get_catalog_name_inst());
691 get_catalog_name_inst() = name;
695 template <class charT>
696 std::string& w32_regex_traits<charT>::get_catalog_name_inst()
698 static std::string s_name;
702 template <class charT>
703 std::string w32_regex_traits<charT>::get_catalog_name()
705 #ifdef BOOST_HAS_THREADS
706 static_mutex::scoped_lock lk(get_mutex_inst());
708 std::string result(get_catalog_name_inst());
712 #ifdef BOOST_HAS_THREADS
713 template <class charT>
714 static_mutex& w32_regex_traits<charT>::get_mutex_inst()
716 static static_mutex s_mutex = BOOST_STATIC_MUTEX_INIT;
729 #pragma warning(push)
730 #pragma warning(disable: 4103)
732 #ifdef BOOST_HAS_ABI_HEADERS
733 # include BOOST_ABI_SUFFIX