basic_regex_implementation(){}

basic_regex_implementation(const ::boost::shared_ptr<

::boost::regex_traits_wrapper<traits> >& t)

: regex_data<charT, traits>(t) {}

void assign(const charT* arg_first,

const charT* arg_last,

flag_type f)

{

regex_data<charT, traits>* pdat = this;

basic_regex_parser<charT, traits> parser(pdat);

parser.parse(arg_first, arg_last, f);

}

locale_type BOOST_REGEX_CALL imbue(locale_type l)

{

return this->m_ptraits->imbue(l);

}

locale_type BOOST_REGEX_CALL getloc()const

{

return this->m_ptraits->getloc();

}

std::basic_string<charT> BOOST_REGEX_CALL str()const

{

std::basic_string<charT> result;

if(this->m_status == 0)

result = std::basic_string<charT>(this->m_expression, this->m_expression_len);

return result;

}

const_iterator BOOST_REGEX_CALL expression()const

{

return this->m_expression;

}

std::pair<const_iterator, const_iterator> BOOST_REGEX_CALL subexpression(std::size_t n)const

{

std::pair<const_iterator, const_iterator> p(expression() + pi.first, expression() + pi.second);

return p;

}

//

// begin, end:

const_iterator BOOST_REGEX_CALL begin()const

{

return (this->m_status ? 0 : this->m_expression);

}

const_iterator BOOST_REGEX_CALL end()const

{

return (this->m_status ? 0 : this->m_expression + this->m_expression_len);

}

flag_type BOOST_REGEX_CALL flags()const

{

return this->m_flags;

}

size_type BOOST_REGEX_CALL size()const

{

return this->m_expression_len;

}

int BOOST_REGEX_CALL status()const

{

return this->m_status;

}

size_type BOOST_REGEX_CALL mark_count()const

{

}

const re_detail::re_syntax_base* get_first_state()const

{

return this->m_first_state;

}

unsigned get_restart_type()const

{

return this->m_restart_type;

}

const unsigned char* get_map()const

{

return this->m_startmap;

}

const ::boost::regex_traits_wrapper<traits>& get_traits()const

{

return *(this->m_ptraits);

}

bool can_be_null()const

{

return this->m_can_be_null;

}

const regex_data<charT, traits>& get_data()const

{

basic_regex_implementation<charT, traits> const* p = this;

return *static_cast<const regex_data<charT, traits>*>(p);

}

};

} // namespace re_detail

//

// class basic_regex:

// represents the compiled

// regular expression:

//

* VERSION see <boost/version.hpp>

* DESCRIPTION: Declares template class basic_regex_creator which fills in

* the data members of a regex_data object.

*/

#ifndef BOOST_REGEX_V4_BASIC_REGEX_CREATOR_HPP

#define BOOST_REGEX_V4_BASIC_REGEX_CREATOR_HPP

#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: 4800)

#endif

namespace boost{

namespace re_detail{

template <class charT>

struct digraph : public std::pair<charT, charT>

{

digraph() : std::pair<charT, charT>(0, 0){}

digraph(charT c1) : std::pair<charT, charT>(c1, 0){}

digraph(charT c1, charT c2) : std::pair<charT, charT>(c1, c2)

{}

digraph(const digraph<charT>& d) : std::pair<charT, charT>(d.first, d.second){}

template <class Seq>

digraph(const Seq& s) : std::pair<charT, charT>()

{

BOOST_ASSERT(s.size() <= 2);

BOOST_ASSERT(s.size());

this->first = s[0];

this->second = (s.size() > 1) ? s[1] : 0;

}

};

template <class charT, class traits>

class basic_char_set

{

public:

typedef digraph<charT> digraph_type;

typedef typename traits::string_type string_type;

typedef typename traits::char_class_type m_type;

basic_char_set()

{

m_negate = false;

m_has_digraphs = false;

m_classes = 0;

m_negated_classes = 0;

m_empty = true;

}

void add_single(const digraph_type& s)

{

m_singles.insert(m_singles.end(), s);

if(s.second)

m_has_digraphs = true;

m_empty = false;

}

void add_range(const digraph_type& first, const digraph_type& end)

m_pdata->m_data.align();

// set the offset to the next state in our last one:

if(m_last_state)

m_last_state->next.i = m_pdata->m_data.size() - getoffset(m_last_state);

// remember the last state position:

std::ptrdiff_t off = getoffset(m_last_state) + s;

// now actually insert our data:

re_syntax_base* new_state = static_cast<re_syntax_base*>(m_pdata->m_data.insert(pos, s));

// fill in boilerplate options in the new state:

new_state->next.i = s;

new_state->type = t;

m_last_state = getaddress(off);

return new_state;

}

template <class charT, class traits>

re_literal* basic_regex_creator<charT, traits>::append_literal(charT c)

{

re_literal* result;

// start by seeing if we have an existing re_literal we can extend:

if((0 == m_last_state) || (m_last_state->type != syntax_element_literal))

{

// no existing re_literal, create a new one:

result = static_cast<re_literal*>(append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT)));

result->length = 1;

*static_cast<charT*>(static_cast<void*>(result+1)) = m_traits.translate(c, m_icase);

}

else

{

// we have an existing re_literal, extend it:

std::ptrdiff_t off = getoffset(m_last_state);

m_pdata->m_data.extend(sizeof(charT));

m_last_state = result = static_cast<re_literal*>(getaddress(off));

charT* characters = static_cast<charT*>(static_cast<void*>(result+1));

characters[result->length] = m_traits.translate(c, m_icase);

}

return result;

}

template <class charT, class traits>

inline re_syntax_base* basic_regex_creator<charT, traits>::append_set(

const basic_char_set<charT, traits>& char_set)

{

typedef mpl::bool_< (sizeof(charT) == 1) > truth_type;

return char_set.has_digraphs()

? append_set(char_set, static_cast<mpl::false_*>(0))

: append_set(char_set, static_cast<truth_type*>(0));

}

template <class charT, class traits>

re_syntax_base* basic_regex_creator<charT, traits>::append_set(

const basic_char_set<charT, traits>& char_set, mpl::false_*)

{

typedef typename traits::string_type string_type;

typedef typename basic_char_set<charT, traits>::list_iterator item_iterator;

typedef typename traits::char_class_type m_type;

re_set_long<m_type>* result = static_cast<re_set_long<m_type>*>(append_state(syntax_element_long_set, sizeof(re_set_long<m_type>)));

//

// fill in the basics:

//

result->csingles = static_cast<unsigned int>(::boost::re_detail::distance(char_set.singles_begin(), char_set.singles_end()));

result->cranges = static_cast<unsigned int>(::boost::re_detail::distance(char_set.ranges_begin(), char_set.ranges_end())) / 2;

result->cequivalents = static_cast<unsigned int>(::boost::re_detail::distance(char_set.equivalents_begin(), char_set.equivalents_end()));

result->cclasses = char_set.classes();

result->cnclasses = char_set.negated_classes();

if(flags() & regbase::icase)

{

// adjust classes as needed:

if(((result->cclasses & m_lower_mask) == m_lower_mask) || ((result->cclasses & m_upper_mask) == m_upper_mask))

last = char_set.singles_end();

while(first != last)

{

charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (first->second ? 3 : 2)));

p[0] = m_traits.translate(first->first, m_icase);

if(first->second)

{

p[1] = m_traits.translate(first->second, m_icase);

p[2] = 0;

}

else

p[1] = 0;

++first;

}

//

// now extend with all the ranges:

//

first = char_set.ranges_begin();

last = char_set.ranges_end();

while(first != last)

{

// first grab the endpoints of the range:

digraph<charT> c1 = *first;

c1.first = this->m_traits.translate(c1.first, this->m_icase);

c1.second = this->m_traits.translate(c1.second, this->m_icase);

++first;

digraph<charT> c2 = *first;

c2.first = this->m_traits.translate(c2.first, this->m_icase);

c2.second = this->m_traits.translate(c2.second, this->m_icase);

++first;

string_type s1, s2;

// different actions now depending upon whether collation is turned on:

if(flags() & regex_constants::collate)

{

// we need to transform our range into sort keys:

charT a1[3] = { c1.first, c1.second, charT(0), };

charT a2[3] = { c2.first, c2.second, charT(0), };

s1 = this->m_traits.transform(a1, (a1[1] ? a1+2 : a1+1));

s2 = this->m_traits.transform(a2, (a2[1] ? a2+2 : a2+1));

if(s1.size() == 0)

s1 = string_type(1, charT(0));

if(s2.size() == 0)

s2 = string_type(1, charT(0));

}

else

{

if(c1.second)

{

s1.insert(s1.end(), c1.first);

s1.insert(s1.end(), c1.second);

}

else

s1 = string_type(1, c1.first);

if(c2.second)

{

s2.insert(s2.end(), c2.first);

s2.insert(s2.end(), c2.second);

}

else

s2.insert(s2.end(), c2.first);

}

if(s1 > s2)

{

// Oops error:

return 0;

}

charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (s1.size() + s2.size() + 2) ) );

re_detail::copy(s1.begin(), s1.end(), p);

p[s1.size()] = charT(0);

p += s1.size() + 1;

re_detail::copy(s2.begin(), s2.end(), p);

p[s2.size()] = charT(0);

}

//

// now process the equivalence classes:

//

first = char_set.equivalents_begin();

last = char_set.equivalents_end();

while(first != last)

{

string_type s;

if(first->second)

{

charT cs[3] = { first->first, first->second, charT(0), };

s = m_traits.transform_primary(cs, cs+2);

}

else

s = m_traits.transform_primary(&first->first, &first->first+1);

if(s.empty())

return 0; // invalid or unsupported equivalence class

charT* p = static_cast<charT*>(this->m_pdata->m_data.extend(sizeof(charT) * (s.size()+1) ) );

re_detail::copy(s.begin(), s.end(), p);

p[s.size()] = charT(0);

++first;

}

//

// finally reset the address of our last state:

//

m_last_state = result = static_cast<re_set_long<m_type>*>(getaddress(offset));

return result;

}

template<class T>

inline bool char_less(T t1, T t2)

{

return t1 < t2;

}

inline bool char_less(char t1, char t2)

{

return static_cast<unsigned char>(t1) < static_cast<unsigned char>(t2);

}

inline bool char_less(signed char t1, signed char t2)

{

return static_cast<unsigned char>(t1) < static_cast<unsigned char>(t2);

}

template <class charT, class traits>

re_syntax_base* basic_regex_creator<charT, traits>::append_set(

const basic_char_set<charT, traits>& char_set, mpl::true_*)

{

// create main startmap:

std::memset(m_pdata->m_startmap, 0, sizeof(m_pdata->m_startmap));

m_pdata->m_can_be_null = 0;

m_bad_repeats = 0;

if(m_has_recursions)

m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);

create_startmap(m_pdata->m_first_state, m_pdata->m_startmap, &(m_pdata->m_can_be_null), mask_all);

// get the restart type:

m_pdata->m_restart_type = get_restart_type(m_pdata->m_first_state);

// optimise a leading repeat if there is one:

probe_leading_repeat(m_pdata->m_first_state);

}

template <class charT, class traits>

void basic_regex_creator<charT, traits>::fixup_pointers(re_syntax_base* state)

{

while(state)

{

switch(state->type)

{

case syntax_element_recurse:

m_has_recursions = true;

if(state->next.i)

state->next.p = getaddress(state->next.i, state);

else

state->next.p = 0;

break;

case syntax_element_rep:

case syntax_element_dot_rep:

case syntax_element_char_rep:

case syntax_element_short_set_rep:

case syntax_element_long_set_rep:

// set the state_id of this repeat:

static_cast<re_repeat*>(state)->state_id = m_repeater_id++;

case syntax_element_alt:

std::memset(static_cast<re_alt*>(state)->_map, 0, sizeof(static_cast<re_alt*>(state)->_map));

static_cast<re_alt*>(state)->can_be_null = 0;

case syntax_element_jump:

static_cast<re_jump*>(state)->alt.p = getaddress(static_cast<re_jump*>(state)->alt.i, state);

default:

if(state->next.i)

state->next.p = getaddress(state->next.i, state);

else

state->next.p = 0;

}

state = state->next.p;

}

}

template <class charT, class traits>

void basic_regex_creator<charT, traits>::fixup_recursions(re_syntax_base* state)

{

re_syntax_base* base = state;

while(state)

{

switch(state->type)

{

case syntax_element_assert_backref:

{

// just check that the index is valid:

int idx = static_cast<const re_brace*>(state)->index;

if(idx < 0)

{

idx = -idx-1;

if(idx >= 10000)

{

idx = m_pdata->get_id(idx);

if(idx <= 0)

{

// check of sub-expression that doesn't exist:

if(0 == this->m_pdata->m_status) // update the error code if not already set

this->m_pdata->m_status = boost::regex_constants::error_bad_pattern;

//

// clear the expression, we should be empty:

}

if(next_rep_id)

break;

p = p->next.p;

}

if(next_rep_id > 0)

{

static_cast<re_recurse*>(state)->state_id = next_rep_id - 1;

}

break;

}

p = p->next.p;

}

if(!ok)

{

// recursion to sub-expression that doesn't exist:

if(0 == this->m_pdata->m_status) // update the error code if not already set

this->m_pdata->m_status = boost::regex_constants::error_bad_pattern;

//

// clear the expression, we should be empty:

//

this->m_pdata->m_expression = 0;

this->m_pdata->m_expression_len = 0;

//

// and throw if required:

//

if(0 == (this->flags() & regex_constants::no_except))

{

std::string message = "Encountered a forward reference to a recursive sub-expression that does not exist.";

boost::regex_error e(message, boost::regex_constants::error_bad_pattern, 0);

e.raise();

}

}

}

default:

break;

}

state = state->next.p;

}

}

template <class charT, class traits>

void basic_regex_creator<charT, traits>::create_startmaps(re_syntax_base* state)

{

// non-recursive implementation:

// create the last map in the machine first, so that earlier maps

// can make use of the result...

//

// This was originally a recursive implementation, but that caused stack

// overflows with complex expressions on small stacks (think COM+).

// start by saving the case setting:

bool l_icase = m_icase;

std::vector<std::pair<bool, re_syntax_base*> > v;

while(state)

{

switch(state->type)

{

case syntax_element_toggle_case:

// we need to track case changes here:

m_icase = static_cast<re_case*>(state)->icase;

state = state->next.p;

continue;

case syntax_element_alt:

case syntax_element_rep:

case syntax_element_dot_rep:

case syntax_element_char_rep:

case syntax_element_short_set_rep:

case syntax_element_long_set_rep:

// just push the state onto our stack for now:

v.push_back(std::pair<bool, re_syntax_base*>(m_icase, state));

state = state->next.p;

break;

case syntax_element_backstep:

// we need to calculate how big the backstep is:

static_cast<re_brace*>(state)->index

= this->calculate_backstep(state->next.p);

if(static_cast<re_brace*>(state)->index < 0)

{

// Oops error:

if(0 == this->m_pdata->m_status) // update the error code if not already set

this->m_pdata->m_status = boost::regex_constants::error_bad_pattern;

//

// clear the expression, we should be empty:

//

this->m_pdata->m_expression = 0;

this->m_pdata->m_expression_len = 0;

//

// and throw if required:

//

if(0 == (this->flags() & regex_constants::no_except))

{

std::string message = "Invalid lookbehind assertion encountered in the regular expression.";

boost::regex_error e(message, boost::regex_constants::error_bad_pattern, 0);

e.raise();

}

}

default:

state = state->next.p;

}

}

// now work through our list, building all the maps as we go:

while(v.size())

{

// Initialize m_recursion_checks if we need it:

if(m_has_recursions)

m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);

const std::pair<bool, re_syntax_base*>& p = v.back();

m_icase = p.first;

state = p.second;

v.pop_back();

// Build maps:

m_bad_repeats = 0;

create_startmap(state->next.p, static_cast<re_alt*>(state)->_map, &static_cast<re_alt*>(state)->can_be_null, mask_take);

m_bad_repeats = 0;

if(m_has_recursions)

m_recursion_checks.assign(1 + m_pdata->m_mark_count, false);

create_startmap(static_cast<re_alt*>(state)->alt.p, static_cast<re_alt*>(state)->_map, &static_cast<re_alt*>(state)->can_be_null, mask_skip);

// adjust the type of the state to allow for faster matching:

state->type = this->get_repeat_type(state);

}

// restore case sensitivity:

m_icase = l_icase;

}

template <class charT, class traits>

int basic_regex_creator<charT, traits>::calculate_backstep(re_syntax_base* state)

{

case syntax_element_recurse:

{

if(state->type == syntax_element_startmark)

recursion_sub = static_cast<re_brace*>(state)->index;

else

recursion_sub = 0;

if(m_recursion_checks[recursion_sub])

{

// Infinite recursion!!

if(0 == this->m_pdata->m_status) // update the error code if not already set

this->m_pdata->m_status = boost::regex_constants::error_bad_pattern;

//

// clear the expression, we should be empty:

//

this->m_pdata->m_expression = 0;

this->m_pdata->m_expression_len = 0;

//

// and throw if required:

//

if(0 == (this->flags() & regex_constants::no_except))

{

std::string message = "Encountered an infinite recursion.";

boost::regex_error e(message, boost::regex_constants::error_bad_pattern, 0);

e.raise();

}

}

else if(recursion_start == 0)

{

recursion_start = state;

recursion_restart = state->next.p;

state = static_cast<re_jump*>(state)->alt.p;

m_recursion_checks[recursion_sub] = true;

break;

}

m_recursion_checks[recursion_sub] = true;

}

case syntax_element_backref:

// can be null, and any character can match:

if(pnull)

*pnull |= mask;

case syntax_element_wild:

{

// can't be null, any character can match:

set_all_masks(l_map, mask);

return;

}

case syntax_element_match:

{

// must be null, any character can match:

set_all_masks(l_map, mask);

if(pnull)

*pnull |= mask;

return;

}

case syntax_element_word_start:

{

// recurse, then AND with all the word characters:

create_startmap(state->next.p, l_map, pnull, mask);

if(l_map)

{

l_map[0] |= mask_init;

for(unsigned int i = 0; i < (1u << CHAR_BIT); ++i)

{

if(!m_traits.isctype(static_cast<charT>(i), m_word_mask))

l_map[i] &= static_cast<unsigned char>(~mask);

}

}

return;

}

case syntax_element_word_end:

{

// recurse, then AND with all the word characters:

create_startmap(state->next.p, l_map, pnull, mask);

if(l_map)

{

// case we should check whatever follows that recursion, as well as whatever

// follows this state:

//

if(m_pdata->m_has_recursions && static_cast<re_brace*>(state)->index)

{

bool ok = false;

re_syntax_base* p = m_pdata->m_first_state;

while(p)

{

if(p->type == syntax_element_recurse)

{

re_brace* p2 = static_cast<re_brace*>(static_cast<re_jump*>(p)->alt.p);

if((p2->type == syntax_element_startmark) && (p2->index == static_cast<re_brace*>(state)->index))

{

ok = true;

break;

}

}

p = p->next.p;

}

if(ok)

{

create_startmap(p->next.p, l_map, pnull, mask);

}

}

state = state->next.p;

break;

case syntax_element_startmark:

// need to handle independent subs as a special case:

if(static_cast<re_brace*>(state)->index == -3)

{

state = state->next.p->next.p;

break;

}

default:

state = state->next.p;

}

++not_last_jump;

}

}

template <class charT, class traits>

unsigned basic_regex_creator<charT, traits>::get_restart_type(re_syntax_base* state)

{

//

// find out how the machine starts, so we can optimise the search:

//

while(state)

{

switch(state->type)

{

case syntax_element_startmark:

case syntax_element_endmark:

state = state->next.p;

continue;

case syntax_element_start_line:

return regbase::restart_line;

case syntax_element_word_start:

return regbase::restart_word;

case syntax_element_buffer_start:

return regbase::restart_buf;

case syntax_element_restart_continue:

return regbase::restart_continue;

default:

state = 0;

continue;

}

}

return regbase::restart_any;

switch(pt->type)

{

case syntax_element_rep:

case syntax_element_dot_rep:

case syntax_element_char_rep:

case syntax_element_short_set_rep:

case syntax_element_long_set_rep:

{

unsigned state_id = static_cast<re_repeat*>(pt)->state_id;

if(state_id > sizeof(m_bad_repeats) * CHAR_BIT)

return true; // run out of bits, assume we can't traverse this one.

static const boost::uintmax_t one = 1uL;

return m_bad_repeats & (one << state_id);

}

default:

return false;

}

}

template <class charT, class traits>

void basic_regex_creator<charT, traits>::set_bad_repeat(re_syntax_base* pt)

{

switch(pt->type)

{

case syntax_element_rep:

case syntax_element_dot_rep:

case syntax_element_char_rep:

case syntax_element_short_set_rep:

case syntax_element_long_set_rep:

{

unsigned state_id = static_cast<re_repeat*>(pt)->state_id;

static const boost::uintmax_t one = 1uL;

if(state_id <= sizeof(m_bad_repeats) * CHAR_BIT)

m_bad_repeats |= (one << state_id);

}

default:

break;

}

}

template <class charT, class traits>

syntax_element_type basic_regex_creator<charT, traits>::get_repeat_type(re_syntax_base* state)

{

typedef typename traits::char_class_type m_type;

if(state->type == syntax_element_rep)

{

// check to see if we are repeating a single state:

if(state->next.p->next.p->next.p == static_cast<re_alt*>(state)->alt.p)

{

switch(state->next.p->type)

{

case re_detail::syntax_element_wild:

return re_detail::syntax_element_dot_rep;

case re_detail::syntax_element_literal:

return re_detail::syntax_element_char_rep;

case re_detail::syntax_element_set:

return re_detail::syntax_element_short_set_rep;

case re_detail::syntax_element_long_set:

if(static_cast<re_detail::re_set_long<m_type>*>(state->next.p)->singleton)

return re_detail::syntax_element_long_set_rep;

break;

default:

break;

}

}

}

return state->type;

}

template <class charT, class traits>

state = state->next.p;

continue;

}

if((static_cast<re_brace*>(state)->index == -1)

|| (static_cast<re_brace*>(state)->index == -2))

{

// skip past the zero width assertion:

state = static_cast<const re_jump*>(state->next.p)->alt.p->next.p;

continue;

}

if(static_cast<re_brace*>(state)->index == -3)

{

// Have to skip the leading jump state:

state = state->next.p->next.p;

continue;

}

return;

case syntax_element_endmark:

case syntax_element_start_line:

case syntax_element_end_line:

case syntax_element_word_boundary:

case syntax_element_within_word:

case syntax_element_word_start:

case syntax_element_word_end:

case syntax_element_buffer_start:

case syntax_element_buffer_end:

case syntax_element_restart_continue:

state = state->next.p;

break;

case syntax_element_dot_rep:

case syntax_element_char_rep:

case syntax_element_short_set_rep:

case syntax_element_long_set_rep:

if(this->m_has_backrefs == 0)

static_cast<re_repeat*>(state)->leading = true;

default:

return;

}

}while(state);

}

} // namespace re_detail

} // namespace 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

case regex_constants::syntax_plus:

if(m_position == this->m_base)

{

fail(regex_constants::error_badrepeat, 0, "The repeat operator \"+\" cannot start a regular expression.");

return false;

}

++m_position;

return parse_repeat(1);

case regex_constants::syntax_open_brace:

++m_position;

return parse_repeat_range(false);

case regex_constants::syntax_close_brace:

fail(regex_constants::error_brace, this->m_position - this->m_base, "Found a closing repetition operator } with no corresponding {.");

return false;

case regex_constants::syntax_or:

return parse_alt();

case regex_constants::syntax_open_set:

return parse_set();

case regex_constants::syntax_newline:

if(this->flags() & regbase::newline_alt)

return parse_alt();

else

return parse_literal();

case regex_constants::syntax_hash:

//

// If we have a mod_x flag set, then skip until

// we get to a newline character:

//

if((this->flags()

& (regbase::no_perl_ex|regbase::mod_x))

== regbase::mod_x)

{

while((m_position != m_end) && !is_separator(*m_position++)){}

return true;

}

default:

result = parse_literal();

break;

}

return result;

}

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_literal()

{

// append this as a literal provided it's not a space character

// or the perl option regbase::mod_x is not set:

if(

((this->flags()

& (regbase::main_option_type|regbase::mod_x|regbase::no_perl_ex))

!= regbase::mod_x)

|| !this->m_traits.isctype(*m_position, this->m_mask_space))

this->append_literal(*m_position);

++m_position;

return true;

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_open_paren()

{

//

// skip the '(' and error check:

//

if(++m_position == m_end)

{

fail(regex_constants::error_paren, m_position - m_base);

return false;

case regex_constants::escape_type_not_word_assert:

if(this->flags() & regbase::emacs_ex)

{

++m_position;

this->append_state(syntax_element_within_word);

}

else

result = parse_literal();

break;

case regex_constants::escape_type_left_word:

if(this->flags() & regbase::emacs_ex)

{

++m_position;

this->append_state(syntax_element_word_start);

}

else

result = parse_literal();

break;

case regex_constants::escape_type_right_word:

if(this->flags() & regbase::emacs_ex)

{

++m_position;

this->append_state(syntax_element_word_end);

}

else

result = parse_literal();

break;

default:

if(this->flags() & regbase::emacs_ex)

{

bool negate = true;

switch(*m_position)

{

case 'w':

negate = false;

case 'W':

{

basic_char_set<charT, traits> char_set;

if(negate)

char_set.negate();

char_set.add_class(this->m_word_mask);

if(0 == this->append_set(char_set))

{

fail(regex_constants::error_ctype, m_position - m_base);

return false;

}

++m_position;

return true;

}

case 's':

negate = false;

case 'S':

return add_emacs_code(negate);

case 'c':

case 'C':

// not supported yet:

fail(regex_constants::error_escape, m_position - m_base, "The \\c and \\C escape sequences are not supported by POSIX basic regular expressions: try the Perl syntax instead.");

return false;

default:

break;

}

}

result = parse_literal();

break;

}

return result;

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_extended_escape()

{

++m_position;

if(m_position == m_end)

{

fail(regex_constants::error_escape, m_position - m_base, "Incomplete escape sequence found.");

return false;

}

bool negate = false; // in case this is a character class escape: \w \d etc

switch(this->m_traits.escape_syntax_type(*m_position))

{

case regex_constants::escape_type_not_class:

negate = true;

case regex_constants::escape_type_class:

{

escape_type_class_jump:

typedef typename traits::char_class_type m_type;

m_type m = this->m_traits.lookup_classname(m_position, m_position+1);

if(m != 0)

{

basic_char_set<charT, traits> char_set;

if(negate)

char_set.negate();

char_set.add_class(m);

if(0 == this->append_set(char_set))

{

fail(regex_constants::error_ctype, m_position - m_base);

return false;

}

++m_position;

return true;

}

//

// not a class, just a regular unknown escape:

//

this->append_literal(unescape_character());

break;

}

case regex_constants::syntax_digit:

return parse_backref();

case regex_constants::escape_type_left_word:

++m_position;

this->append_state(syntax_element_word_start);

break;

case regex_constants::escape_type_right_word:

++m_position;

this->append_state(syntax_element_word_end);

break;

case regex_constants::escape_type_start_buffer:

++m_position;

this->append_state(syntax_element_buffer_start);

break;

case regex_constants::escape_type_end_buffer:

++m_position;

this->append_state(syntax_element_buffer_end);

break;

case regex_constants::escape_type_word_assert:

++m_position;

this->append_state(syntax_element_word_boundary);

break;

case regex_constants::escape_type_not_word_assert:

++m_position;

this->append_state(syntax_element_within_word);

break;

case regex_constants::escape_type_Z:

++m_position;

this->append_state(syntax_element_soft_buffer_end);

break;

case regex_constants::escape_type_Q:

return parse_QE();

case regex_constants::escape_type_C:

return parse_match_any();

case regex_constants::escape_type_X:

++m_position;

this->append_state(syntax_element_combining);

break;

case regex_constants::escape_type_G:

++m_position;

this->append_state(syntax_element_restart_continue);

break;

case regex_constants::escape_type_not_property:

negate = true;

case regex_constants::escape_type_property:

{

++m_position;

char_class_type m;

if(m_position == m_end)

{

fail(regex_constants::error_escape, m_position - m_base, "Incomplete property escape found.");

return false;

}

// maybe have \p{ddd}

if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace)

{

const charT* base = m_position;

// skip forward until we find enclosing brace:

while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))

++m_position;

if(m_position == m_end)

{

fail(regex_constants::error_escape, m_position - m_base, "Closing } missing from property escape sequence.");

return false;

}

m = this->m_traits.lookup_classname(++base, m_position++);

}

else

{

m = this->m_traits.lookup_classname(m_position, m_position+1);

++m_position;

}

if(m != 0)

{

basic_char_set<charT, traits> char_set;

if(negate)

char_set.negate();

char_set.add_class(m);

if(0 == this->append_set(char_set))

++m_position;

}

i = hash_value_from_capture_name(base, m_position);

pc = m_position;

}

if(negative)

i = 1 + m_mark_count - i;

if(((i > 0) && (this->m_backrefs & (1u << (i-1)))) || ((i > 10000) && (this->m_pdata->get_id(i) > 0) && (this->m_backrefs & (1u << (this->m_pdata->get_id(i)-1)))))

{

m_position = pc;

re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_backref, sizeof(re_brace)));

pb->index = i;

pb->icase = this->flags() & regbase::icase;

}

else

{

fail(regex_constants::error_backref, m_position - m_base);

return false;

}

m_position = pc;

if(have_brace)

{

if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end))

{

fail(regex_constants::error_escape, m_position - m_base, incomplete_message);

return false;

}

++m_position;

}

return true;

}

goto escape_type_class_jump;

case regex_constants::escape_type_control_v:

if(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))

goto escape_type_class_jump;

default:

this->append_literal(unescape_character());

break;

}

return true;

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_match_any()

{

//

// we have a '.' that can match any character:

//

++m_position;

static_cast<re_dot*>(

this->append_state(syntax_element_wild, sizeof(re_dot))

)->mask = static_cast<unsigned char>(this->flags() & regbase::no_mod_s

? re_detail::force_not_newline

: this->flags() & regbase::mod_s ?

re_detail::force_newline : re_detail::dont_care);

return true;

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_repeat(std::size_t low, std::size_t high)

{

bool greedy = true;

bool pocessive = false;

std::size_t insert_point;

//

// when we get to here we may have a non-greedy ? mark still to come:

//

if((m_position != m_end)

&& (

(0 == (this->flags() & (regbase::main_option_type | regbase::no_perl_ex)))

|| ((regbase::basic_syntax_group|regbase::emacs_ex) == (this->flags() & (regbase::main_option_type | regbase::emacs_ex)))

)

)

{

// OK we have a perl or emacs regex, check for a '?':

if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_question)

{

greedy = false;

++m_position;

}

// for perl regexes only check for pocessive ++ repeats.

if((m_position != m_end)

&& (0 == (this->flags() & regbase::main_option_type))

&& (this->m_traits.syntax_type(*m_position) == regex_constants::syntax_plus))

{

pocessive = true;

++m_position;

}

}

if(0 == this->m_last_state)

{

fail(regex_constants::error_badrepeat, ::boost::re_detail::distance(m_base, m_position), "Nothing to repeat.");

return false;

}

if(this->m_last_state->type == syntax_element_endmark)

{

// insert a repeat before the '(' matching the last ')':

insert_point = this->m_paren_start;

}

else if((this->m_last_state->type == syntax_element_literal) && (static_cast<re_literal*>(this->m_last_state)->length > 1))

{

// the last state was a literal with more than one character, split it in two:

re_literal* lit = static_cast<re_literal*>(this->m_last_state);

charT c = (static_cast<charT*>(static_cast<void*>(lit+1)))[lit->length - 1];

// now append new state:

lit = static_cast<re_literal*>(this->append_state(syntax_element_literal, sizeof(re_literal) + sizeof(charT)));

lit->length = 1;

(static_cast<charT*>(static_cast<void*>(lit+1)))[0] = c;

insert_point = this->getoffset(this->m_last_state);

}

else

{

// repeat the last state whatever it was, need to add some error checking here:

switch(this->m_last_state->type)

{

case syntax_element_start_line:

case syntax_element_end_line:

case syntax_element_word_boundary:

case syntax_element_within_word:

case syntax_element_word_start:

case syntax_element_word_end:

case syntax_element_buffer_start:

case syntax_element_buffer_end:

case syntax_element_alt:

case syntax_element_soft_buffer_end:

case syntax_element_restart_continue:

case syntax_element_jump:

case syntax_element_startmark:

case syntax_element_backstep:

// can't legally repeat any of the above:

fail(regex_constants::error_badrepeat, m_position - m_base);

return false;

default:

// do nothing...

break;

}

insert_point = this->getoffset(this->m_last_state);

}

//

switch(this->m_traits.syntax_type(*m_position))

{

case regex_constants::syntax_star:

case regex_constants::syntax_plus:

case regex_constants::syntax_question:

case regex_constants::syntax_open_brace:

fail(regex_constants::error_badrepeat, m_position - m_base);

return false;

}

}

re_brace* pb = static_cast<re_brace*>(this->insert_state(insert_point, syntax_element_startmark, sizeof(re_brace)));

pb->index = -3;

pb->icase = this->flags() & regbase::icase;

jmp = static_cast<re_jump*>(this->insert_state(insert_point + sizeof(re_brace), syntax_element_jump, sizeof(re_jump)));

this->m_pdata->m_data.align();

jmp->alt.i = this->m_pdata->m_data.size() - this->getoffset(jmp);

pb = static_cast<re_brace*>(this->append_state(syntax_element_endmark, sizeof(re_brace)));

pb->index = -3;

pb->icase = this->flags() & regbase::icase;

}

return true;

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_repeat_range(bool isbasic)

{

static const char* incomplete_message = "Missing } in quantified repetition.";

//

// parse a repeat-range:

//

std::size_t min, max;

int v;

// skip whitespace:

while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))

++m_position;

if(this->m_position == this->m_end)

{

}

// get min:

v = this->m_traits.toi(m_position, m_end, 10);

// skip whitespace:

if(v < 0)

{

}

{

}

min = v;

// see if we have a comma:

if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_comma)

{

// move on and error check:

++m_position;

// skip whitespace:

while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))

++m_position;

if(this->m_position == this->m_end)

{

}

// get the value if any:

v = this->m_traits.toi(m_position, m_end, 10);

}

else

{

// no comma, max = min:

max = min;

}

// skip whitespace:

while((m_position != m_end) && this->m_traits.isctype(*m_position, this->m_mask_space))

++m_position;

// OK now check trailing }:

if(this->m_position == this->m_end)

{

}

if(isbasic)

{

if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_escape)

{

++m_position;

if(this->m_position == this->m_end)

{

fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);

return false;

}

}

else

{

fail(regex_constants::error_brace, this->m_position - this->m_base, incomplete_message);

return false;

}

}

if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_close_brace)

++m_position;

else

{

}

//

// finally go and add the repeat, unless error:

//

if(min > max)

{

// Backtrack to error location:

m_position -= 2;

while(this->m_traits.isctype(*m_position, this->m_word_mask)) --m_position;

++m_position;

fail(regex_constants::error_badbrace, m_position - m_base);

return false;

}

return parse_repeat(min, max);

}

template <class charT, class traits>

bool basic_regex_parser<charT, traits>::parse_alt()

{

//

// error check: if there have been no previous states,

// or if the last state was a '(' then error:

//

if(

((this->m_last_state == 0) || (this->m_last_state->type == syntax_element_startmark))

&&

!(

((this->flags() & regbase::main_option_type) == regbase::perl_syntax_group)

&&

((this->flags() & regbase::no_empty_expressions) == 0)

)

)

{

fail(regex_constants::error_empty, this->m_position - this->m_base, "A regular expression can start with the alternation operator |.");

return false;

while((m_position != m_end)

&& (this->m_traits.syntax_type(*m_position++) != regex_constants::syntax_close_mark))

{}

return true;

}

//

// backup some state, and prepare the way:

//

int markid = 0;

std::ptrdiff_t jump_offset = 0;

re_brace* pb = static_cast<re_brace*>(this->append_state(syntax_element_startmark, sizeof(re_brace)));

pb->icase = this->flags() & regbase::icase;

std::ptrdiff_t last_paren_start = this->getoffset(pb);

// back up insertion point for alternations, and set new point:

std::ptrdiff_t last_alt_point = m_alt_insert_point;

this->m_pdata->m_data.align();

m_alt_insert_point = this->m_pdata->m_data.size();

std::ptrdiff_t expected_alt_point = m_alt_insert_point;

bool restore_flags = true;

regex_constants::syntax_option_type old_flags = this->flags();

bool old_case_change = m_has_case_change;

m_has_case_change = false;

charT name_delim;

int mark_reset = m_mark_reset;

int max_mark = m_max_mark;

m_mark_reset = -1;

m_max_mark = m_mark_count;

int v;

//

// select the actual extension used:

//

switch(this->m_traits.syntax_type(*m_position))

{

case regex_constants::syntax_or:

m_mark_reset = m_mark_count;

case regex_constants::syntax_colon:

//

// a non-capturing mark:

//

pb->index = markid = 0;

++m_position;

break;

case regex_constants::syntax_digit:

{

//

// a recursive subexpression:

//

v = this->m_traits.toi(m_position, m_end, 10);

if((v < 0) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_mark))

{

// Rewind to start of (? sequence:

--m_position;

while(this->m_traits.syntax_type(*m_position) != regex_constants::syntax_open_mark) --m_position;

fail(regex_constants::error_perl_extension, m_position - m_base, "The recursive sub-expression refers to an invalid marking group, or is unterminated.");

return false;

}

insert_recursion:

pb->index = markid = 0;

re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse)));

pr->alt.i = v;

pr->state_id = 0;

static_cast<re_case*>(

this->append_state(syntax_element_toggle_case, sizeof(re_case))

)->icase = this->flags() & regbase::icase;

break;

}

case regex_constants::syntax_plus:

//

// A forward-relative recursive subexpression:

//

/*

*

* Copyright (c) 2004 John Maddock

* Copyright 2011 Garmin Ltd. or its subsidiaries

*

* 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 <boost/version.hpp>

* DESCRIPTION: Declares regular expression traits class cpp_regex_traits.

*/

#ifndef BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED

#define BOOST_CPP_REGEX_TRAITS_HPP_INCLUDED

#include <boost/config.hpp>

#ifndef BOOST_NO_STD_LOCALE

#ifndef BOOST_RE_PAT_EXCEPT_HPP

#include <boost/regex/pattern_except.hpp>

#endif

#ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED

#include <boost/regex/v4/regex_traits_defaults.hpp>

#endif

#ifdef BOOST_HAS_THREADS

#include <boost/regex/pending/static_mutex.hpp>

#endif

#ifndef BOOST_REGEX_PRIMARY_TRANSFORM

#include <boost/regex/v4/primary_transform.hpp>

#endif

#ifndef BOOST_REGEX_OBJECT_CACHE_HPP

#include <boost/regex/pending/object_cache.hpp>

#endif

#include <istream>

#include <ios>

#include <climits>

#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

//

template <class charT,

class traits = ::std::char_traits<charT> >

class parser_buf : public ::std::basic_streambuf<charT, traits>

{

typedef ::std::basic_streambuf<charT, traits> 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<charT, traits>* setbuf(char_type* s, streamsize n);

typename parser_buf<charT, traits>::pos_type seekpos(pos_type sp, ::std::ios_base::openmode which);

typename parser_buf<charT, traits>::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<class charT, class traits>

std::basic_streambuf<charT, traits>*

parser_buf<charT, traits>::setbuf(char_type* s, streamsize n)

{

this->setg(s, s, s + n);

return this;

}

template<class charT, class traits>

typename parser_buf<charT, traits>::pos_type

parser_buf<charT, traits>::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();

{

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<std::ptrdiff_t>(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<pos_type>(this->gptr() - this->eback());

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

}

template<class charT, class traits>

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 <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_blank;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_word;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_unicode;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_vertical;

template <class charT>

typename cpp_regex_traits_implementation<charT>::char_class_type const cpp_regex_traits_implementation<charT>::mask_horizontal;

#endif

#endif

template <class charT>

typename cpp_regex_traits_implementation<charT>::string_type

cpp_regex_traits_implementation<charT>::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...

//

#ifndef BOOST_NO_EXCEPTIONS

try{

#endif

//

// 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)

}

result.erase(i);

break;

}

#ifndef BOOST_NO_EXCEPTIONS

}catch(...){}

#endif

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 <class charT>

typename cpp_regex_traits_implementation<charT>::string_type

cpp_regex_traits_implementation<charT>::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...

//

#ifndef BOOST_NO_EXCEPTIONS

try{

#endif

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<wchar_t>::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

#ifndef BOOST_NO_EXCEPTIONS

}

catch(...)

{

}

#endif

}

template <class charT>

typename cpp_regex_traits_implementation<charT>::string_type

cpp_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const

{

typedef typename std::map<string_type, string_type>::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(__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(__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 <class charT>

void cpp_regex_traits_implementation<charT>::init()

{

#ifndef BOOST_NO_STD_MESSAGES

#ifndef __IBMCPP__

typename std::messages<charT>::catalog cat = static_cast<std::messages<char>::catalog>(-1);

#else

typename std::messages<charT>::catalog cat = reinterpret_cast<std::messages<char>::catalog>(-1);

#endif

std::string cat_name(cpp_regex_traits<charT>::get_catalog_name());

if(cat_name.size() && (this->m_pmessages != 0))

{

cat = this->m_pmessages->open(

cat_name,

this->m_locale);

{

typedef typename std::map<std::basic_string<charT>, 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 <class charT>

bool cpp_regex_traits_implementation<charT>::isctype(const charT c, char_class_type mask) const

{

return

((mask & ::boost::re_detail::char_class_space) && (this->m_pctype->is(std::ctype<charT>::space, c)))

|| ((mask & ::boost::re_detail::char_class_print) && (this->m_pctype->is(std::ctype<charT>::print, c)))

|| ((mask & ::boost::re_detail::char_class_cntrl) && (this->m_pctype->is(std::ctype<charT>::cntrl, c)))

|| ((mask & ::boost::re_detail::char_class_upper) && (this->m_pctype->is(std::ctype<charT>::upper, c)))

|| ((mask & ::boost::re_detail::char_class_lower) && (this->m_pctype->is(std::ctype<charT>::lower, c)))

|| ((mask & ::boost::re_detail::char_class_alpha) && (this->m_pctype->is(std::ctype<charT>::alpha, c)))

|| ((mask & ::boost::re_detail::char_class_digit) && (this->m_pctype->is(std::ctype<charT>::digit, c)))

|| ((mask & ::boost::re_detail::char_class_punct) && (this->m_pctype->is(std::ctype<charT>::punct, c)))

|| ((mask & ::boost::re_detail::char_class_xdigit) && (this->m_pctype->is(std::ctype<charT>::xdigit, c)))

|| ((mask & ::boost::re_detail::char_class_blank) && (this->m_pctype->is(std::ctype<charT>::space, c)) && !::boost::re_detail::is_separator(c))

|| ((mask & ::boost::re_detail::char_class_word) && (c == '_'))

|| ((mask & ::boost::re_detail::char_class_unicode) && ::boost::re_detail::is_extended(c))

|| ((mask & ::boost::re_detail::char_class_vertical_space) && (is_separator(c) || (c == '\v')))

|| ((mask & ::boost::re_detail::char_class_horizontal_space) && this->m_pctype->is(std::ctype<charT>::space, c) && !(is_separator(c) || (c == '\v')));

}

#endif

template <class charT>

{

cpp_regex_traits_base<charT> key(l);

return ::boost::object_cache<cpp_regex_traits_base<charT>, cpp_regex_traits_implementation<charT> >::get(key, 5);

}

} // re_detail

template <class charT>

class cpp_regex_traits

{

private:

typedef std::ctype<charT> ctype_type;

public:

typedef charT char_type;

typedef std::size_t size_type;

typedef std::basic_string<char_type> string_type;

typedef std::locale locale_type;

typedef boost::uint_least32_t char_class_type;

struct boost_extensions_tag{};

cpp_regex_traits()

: m_pimpl(re_detail::create_cpp_regex_traits<charT>(std::locale()))

{ }

static size_type length(const char_type* p)

{

return std::char_traits<charT>::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);

# ifndef BOOST_REGEX_INSTANTIATE

# pragma option pop

# endif

#ifdef BOOST_HAS_ABI_HEADERS

# include BOOST_ABI_SUFFIX

#endif

#elif defined(BOOST_MSVC) || defined(__ICL)

# ifndef BOOST_REGEX_INSTANTIATE

# ifdef __GNUC__

# define template __extension__ extern template

# else

# if BOOST_MSVC > 1310

# define BOOST_REGEX_TEMPLATE_DECL

# endif

# define template extern template

# endif

# endif

#ifndef BOOST_REGEX_TEMPLATE_DECL

# define BOOST_REGEX_TEMPLATE_DECL BOOST_REGEX_DECL

#endif

# ifdef BOOST_MSVC

# pragma warning(push)

# pragma warning(disable : 4251 4231)

# if BOOST_MSVC < 1600

# pragma warning(disable : 4660)

# endif

# endif

template class BOOST_REGEX_TEMPLATE_DECL basic_regex< BOOST_REGEX_CHAR_T BOOST_REGEX_TRAITS_T >;

template class BOOST_REGEX_TEMPLATE_DECL match_results< const BOOST_REGEX_CHAR_T* >;

#ifndef BOOST_NO_STD_ALLOCATOR

template class BOOST_REGEX_TEMPLATE_DECL ::boost::re_detail::perl_matcher<BOOST_REGEX_CHAR_T const *, match_results< const BOOST_REGEX_CHAR_T* >::allocator_type BOOST_REGEX_TRAITS_T >;

#endif

#if !(defined(BOOST_DINKUMWARE_STDLIB) && (BOOST_DINKUMWARE_STDLIB <= 1))\

&& !(defined(BOOST_INTEL_CXX_VERSION) && (BOOST_INTEL_CXX_VERSION <= 800))\

&& !(defined(__SGI_STL_PORT) || defined(_STLPORT_VERSION))\

&& !defined(BOOST_REGEX_ICU_INSTANCES)

template class BOOST_REGEX_TEMPLATE_DECL match_results< std::basic_string<BOOST_REGEX_CHAR_T>::const_iterator >;

#ifndef BOOST_NO_STD_ALLOCATOR

template class BOOST_REGEX_TEMPLATE_DECL ::boost::re_detail::perl_matcher< std::basic_string<BOOST_REGEX_CHAR_T>::const_iterator, match_results< std::basic_string<BOOST_REGEX_CHAR_T>::const_iterator >::allocator_type, boost::regex_traits<BOOST_REGEX_CHAR_T > >;

#endif

#endif

# ifdef BOOST_MSVC

# pragma warning(pop)

# endif

# ifdef template

# undef template

# endif

#undef BOOST_REGEX_TEMPLATE_DECL

# ifndef BOOST_REGEX_INSTANTIATE

# ifdef __GNUC__

# define template __extension__ extern template

# else

# define template extern template

# endif

# endif

#if !defined(BOOST_NO_STD_LOCALE) && !defined(BOOST_REGEX_ICU_INSTANCES)

namespace re_detail{

template BOOST_REGEX_DECL

std::locale cpp_regex_traits_base<BOOST_REGEX_CHAR_T>::imbue(const std::locale& l);

template BOOST_REGEX_DECL

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::string_type

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::transform_primary(const BOOST_REGEX_CHAR_T* p1, const BOOST_REGEX_CHAR_T* p2) const;

template BOOST_REGEX_DECL

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::string_type

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::transform(const BOOST_REGEX_CHAR_T* p1, const BOOST_REGEX_CHAR_T* p2) const;

template BOOST_REGEX_DECL

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::string_type

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::lookup_collatename(const BOOST_REGEX_CHAR_T* p1, const BOOST_REGEX_CHAR_T* p2) const;

template BOOST_REGEX_DECL

void cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::init();

template BOOST_REGEX_DECL

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::char_class_type

cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::lookup_classname_imp(const BOOST_REGEX_CHAR_T* p1, const BOOST_REGEX_CHAR_T* p2) const;

#ifdef BOOST_REGEX_BUGGY_CTYPE_FACET

template BOOST_REGEX_DECL

bool cpp_regex_traits_implementation<BOOST_REGEX_CHAR_T>::isctype(const BOOST_REGEX_CHAR_T c, char_class_type mask) const;

#endif

} // namespace

template BOOST_REGEX_DECL

int cpp_regex_traits<BOOST_REGEX_CHAR_T>::toi(const BOOST_REGEX_CHAR_T*& first, const BOOST_REGEX_CHAR_T* last, int radix)const;

/*

*

* Copyright (c) 1998-2002

* 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 iterator_traits.cpp

* VERSION see <boost/version.hpp>

* DESCRIPTION: Declares iterator traits workarounds.

*/

#ifndef BOOST_REGEX_V4_ITERATOR_TRAITS_HPP

#define BOOST_REGEX_V4_ITERATOR_TRAITS_HPP

#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

namespace boost{

namespace re_detail{

template <class T>

struct regex_iterator_traits

{

typedef typename T::iterator_category iterator_category;

typedef typename T::value_type value_type;

#if !defined(BOOST_NO_STD_ITERATOR)

typedef typename T::difference_type difference_type;

typedef typename T::pointer pointer;

typedef typename T::reference reference;

#else

typedef std::ptrdiff_t difference_type;

typedef value_type* pointer;

typedef value_type& reference;

#endif

};

template <class T>

struct pointer_iterator_traits

{

typedef std::ptrdiff_t difference_type;

typedef T value_type;

typedef T* pointer;

typedef T& reference;

typedef std::random_access_iterator_tag iterator_category;

};

template <class T>

struct const_pointer_iterator_traits

{

typedef std::ptrdiff_t difference_type;

typedef T value_type;

typedef const T* pointer;

typedef const T& reference;

typedef std::random_access_iterator_tag iterator_category;

};

match_not_eow = match_not_bow << 1, /* last is not end of word */

match_not_dot_newline = match_not_eow << 1, /* \n is not matched by '.' */

match_not_dot_null = match_not_dot_newline << 1, /* '\0' is not matched by '.' */

match_prev_avail = match_not_dot_null << 1, /* *--first is a valid expression */

match_init = match_prev_avail << 1, /* internal use */

match_any = match_init << 1, /* don't care what we match */

match_not_null = match_any << 1, /* string can't be null */

match_continuous = match_not_null << 1, /* each grep match must continue from */

/* uninterupted from the previous one */

match_partial = match_continuous << 1, /* find partial matches */

match_stop = match_partial << 1, /* stop after first match (grep) V3 only */

match_not_initial_null = match_stop, /* don't match initial null, V4 only */

match_all = match_stop << 1, /* must find the whole of input even if match_any is set */

match_perl = match_all << 1, /* Use perl matching rules */

match_posix = match_perl << 1, /* Use POSIX matching rules */

match_nosubs = match_posix << 1, /* don't trap marked subs */

match_extra = match_nosubs << 1, /* include full capture information for repeated captures */

match_single_line = match_extra << 1, /* treat text as single line and ignor any \n's when matching ^ and $. */

match_unused1 = match_single_line << 1, /* unused */

match_unused2 = match_unused1 << 1, /* unused */

match_unused3 = match_unused2 << 1, /* unused */

match_max = match_unused3,

format_perl = 0, /* perl style replacement */

format_default = 0, /* ditto. */

format_sed = match_max << 1, /* sed style replacement. */

format_all = format_sed << 1, /* enable all extentions to sytax. */

format_no_copy = format_all << 1, /* don't copy non-matching segments. */

format_first_only = format_no_copy << 1, /* Only replace first occurance. */

format_is_if = format_first_only << 1, /* internal use only. */

format_literal = format_is_if << 1 /* treat string as a literal */

} match_flags;

typedef unsigned long match_flag_type;

#else

typedef match_flags match_flag_type;

#ifdef __cplusplus

inline match_flags operator&(match_flags m1, match_flags m2)

{ return static_cast<match_flags>(static_cast<boost::int32_t>(m1) & static_cast<boost::int32_t>(m2)); }

inline match_flags operator|(match_flags m1, match_flags m2)

{ return static_cast<match_flags>(static_cast<boost::int32_t>(m1) | static_cast<boost::int32_t>(m2)); }

inline match_flags operator^(match_flags m1, match_flags m2)

{ return static_cast<match_flags>(static_cast<boost::int32_t>(m1) ^ static_cast<boost::int32_t>(m2)); }

inline match_flags operator~(match_flags m1)

{ return static_cast<match_flags>(~static_cast<boost::int32_t>(m1)); }

inline match_flags& operator&=(match_flags& m1, match_flags m2)

{ m1 = m1&m2; return m1; }

inline match_flags& operator|=(match_flags& m1, match_flags m2)

{ m1 = m1|m2; return m1; }

inline match_flags& operator^=(match_flags& m1, match_flags m2)

{ m1 = m1^m2; return m1; }

#endif

#endif

#ifdef __cplusplus

} /* namespace regex_constants */

/*

* import names into boost for backwards compatiblity:

*/

using regex_constants::match_flag_type;

using regex_constants::match_default;

using regex_constants::match_not_bol;

using regex_constants::match_not_eol;

using regex_constants::match_not_bob;

using regex_constants::match_not_eob;

using regex_constants::match_not_bow;

return map[(unsigned char)c] & mask;

}

inline bool can_start(unsigned char c, const unsigned char* map, unsigned char mask)

{

return map[c] & mask;

}

inline bool can_start(unsigned short c, const unsigned char* map, unsigned char mask)

{

return ((c >= (1 << CHAR_BIT)) ? true : map[c] & mask);

}

#if !defined(__hpux) && !defined(__WINSCW__)// WCHAR_MIN not usable in pp-directives.

#if defined(WCHAR_MIN) && (WCHAR_MIN == 0) && !defined(BOOST_NO_INTRINSIC_WCHAR_T)

inline bool can_start(wchar_t c, const unsigned char* map, unsigned char mask)

{

return ((c >= static_cast<wchar_t>(1u << CHAR_BIT)) ? true : map[c] & mask);

}

#endif

#endif

#if !defined(BOOST_NO_INTRINSIC_WCHAR_T)

inline bool can_start(unsigned int c, const unsigned char* map, unsigned char mask)

{

return (((c >= static_cast<unsigned int>(1u << CHAR_BIT)) ? true : map[c] & mask));

}

#endif

//

// Unfortunately Rogue Waves standard library appears to have a bug

// in std::basic_string::compare that results in eroneous answers

// in some cases (tested with Borland C++ 5.1, Rogue Wave lib version

// 0x020101) the test case was:

// {39135,0} < {0xff,0}

// which succeeds when it should not.

//

#ifndef _RWSTD_VER

template <class C, class T, class A>

inline int string_compare(const std::basic_string<C,T,A>& s, const C* p)

{

if(0 == *p)

{

if(s.empty() || ((s.size() == 1) && (s[0] == 0)))

return 0;

}

return s.compare(p);

}

#else

template <class C, class T, class A>

inline int string_compare(const std::basic_string<C,T,A>& s, const C* p)

{

if(0 == *p)

{

if(s.empty() || ((s.size() == 1) && (s[0] == 0)))

return 0;

}

return s.compare(p);

}

inline int string_compare(const std::string& s, const char* p)

{ return std::strcmp(s.c_str(), p); }

# ifndef BOOST_NO_WREGEX

inline int string_compare(const std::wstring& s, const wchar_t* p)

{ return std::wcscmp(s.c_str(), p); }

#endif

#endif

template <class Seq, class C>

inline int string_compare(const Seq& s, const C* p)

{

std::size_t i = 0;

while((i < s.size()) && (p[i] == s[i]))

{

++i;

}

return (i == s.size()) ? -p[i] : s[i] - p[i];

}

# define STR_COMP(s,p) string_compare(s,p)

template<class charT>

inline const charT* re_skip_past_null(const charT* p)

{

while (*p != static_cast<charT>(0)) ++p;

return ++p;

}

template <class iterator, class charT, class traits_type, class char_classT>

iterator BOOST_REGEX_CALL re_is_set_member(iterator next,

iterator last,

const re_set_long<char_classT>* set_,

const regex_data<charT, traits_type>& e, bool icase)

{

const charT* p = reinterpret_cast<const charT*>(set_+1);

iterator ptr;

unsigned int i;

#ifdef __BORLANDC__

# pragma option push -w-8008 -w-8066

#endif

#ifdef BOOST_MSVC

# pragma warning(push)

# pragma warning(disable: 4800)

#endif

namespace boost{

namespace re_detail{

template <class BidiIterator, class Allocator, class traits>

void perl_matcher<BidiIterator, Allocator, traits>::construct_init(const basic_regex<char_type, traits>& e, match_flag_type f)

{

typedef typename regex_iterator_traits<BidiIterator>::iterator_category category;

typedef typename basic_regex<char_type, traits>::flag_type expression_flag_type;

if(e.empty())

{

// precondition failure: e is not a valid regex.

std::invalid_argument ex("Invalid regular expression object");

boost::throw_exception(ex);

}

pstate = 0;

m_match_flags = f;

estimate_max_state_count(static_cast<category*>(0));

expression_flag_type re_f = re.flags();

icase = re_f & regex_constants::icase;

if(!(m_match_flags & (match_perl|match_posix)))

{

if((re_f & (regbase::main_option_type|regbase::no_perl_ex)) == 0)

m_match_flags |= match_perl;

else if((re_f & (regbase::main_option_type|regbase::emacs_ex)) == (regbase::basic_syntax_group|regbase::emacs_ex))

m_match_flags |= match_perl;

else

m_match_flags |= match_posix;

}

if(m_match_flags & match_posix)

{

m_temp_match.reset(new match_results<BidiIterator, Allocator>());

m_presult = m_temp_match.get();

}

else

m_presult = &m_result;

#ifdef BOOST_REGEX_NON_RECURSIVE

m_stack_base = 0;

m_backup_state = 0;

#endif

// find the value to use for matching word boundaries:

m_word_mask = re.get_data().m_word_mask;

// find bitmask to use for matching '.':

match_any_mask = static_cast<unsigned char>((f & match_not_dot_newline) ? re_detail::test_not_newline : re_detail::test_newline);

}

template <class BidiIterator, class Allocator, class traits>

void perl_matcher<BidiIterator, Allocator, traits>::estimate_max_state_count(std::random_access_iterator_tag*)

{

//

// How many states should we allow our machine to visit before giving up?

// This is a heuristic: it takes the greater of O(N^2) and O(NS^2)

// where N is the length of the string, and S is the number of states

// in the machine. It's tempting to up this to O(N^2S) or even O(N^2S^2)

// but these take unreasonably amounts of time to bale out in pathological

// cases.

//

// Calculate NS^2 first:

//

static const std::ptrdiff_t k = 100000;

std::ptrdiff_t dist = boost::re_detail::distance(base, last);

::boost::re_detail::concrete_protected_call

<perl_matcher<BidiIterator, Allocator, traits> >

obj(this, proc);

return obj.execute();

}

#endif

template <class BidiIterator, class Allocator, class traits>

inline bool perl_matcher<BidiIterator, Allocator, traits>::match()

{

#ifdef BOOST_REGEX_HAS_MS_STACK_GUARD

return protected_call(&perl_matcher<BidiIterator, Allocator, traits>::match_imp);

#else

return match_imp();

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_imp()

{

// initialise our stack if we are non-recursive:

#ifdef BOOST_REGEX_NON_RECURSIVE

save_state_init init(&m_stack_base, &m_backup_state);

used_block_count = BOOST_REGEX_MAX_BLOCKS;

#if !defined(BOOST_NO_EXCEPTIONS)

try{

#endif

#endif

// reset our state machine:

position = base;

search_base = base;

state_count = 0;

m_match_flags |= regex_constants::match_all;

m_presult->set_base(base);

m_presult->set_named_subs(this->re.get_named_subs());

if(m_match_flags & match_posix)

m_result = *m_presult;

verify_options(re.flags(), m_match_flags);

if(0 == match_prefix())

return false;

return (m_result[0].second == last) && (m_result[0].first == base);

#if defined(BOOST_REGEX_NON_RECURSIVE) && !defined(BOOST_NO_EXCEPTIONS)

}

catch(...)

{

// unwind all pushed states, apart from anything else this

// ensures that all the states are correctly destructed

// not just the memory freed.

while(unwind(true)){}

throw;

}

#endif

}

template <class BidiIterator, class Allocator, class traits>

inline bool perl_matcher<BidiIterator, Allocator, traits>::find()

{

#ifdef BOOST_REGEX_HAS_MS_STACK_GUARD

return protected_call(&perl_matcher<BidiIterator, Allocator, traits>::find_imp);

#else

return find_imp();

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::find_imp()

{

static matcher_proc_type const s_find_vtable[7] =

{

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_any,

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_word,

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_line,

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_buf,

&perl_matcher<BidiIterator, Allocator, traits>::match_prefix,

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_lit,

&perl_matcher<BidiIterator, Allocator, traits>::find_restart_lit,

};

// initialise our stack if we are non-recursive:

#ifdef BOOST_REGEX_NON_RECURSIVE

save_state_init init(&m_stack_base, &m_backup_state);

used_block_count = BOOST_REGEX_MAX_BLOCKS;

#if !defined(BOOST_NO_EXCEPTIONS)

try{

#endif

#endif

state_count = 0;

if((m_match_flags & regex_constants::match_init) == 0)

{

// reset our state machine:

search_base = position = base;

pstate = re.get_first_state();

m_presult->set_base(base);

m_presult->set_named_subs(this->re.get_named_subs());

m_match_flags |= regex_constants::match_init;

}

else

{

// start again:

search_base = position = m_result[0].second;

// If last match was null and match_not_null was not set then increment

// our start position, otherwise we go into an infinite loop:

if(((m_match_flags & match_not_null) == 0) && (m_result.length() == 0))

{

if(position == last)

return false;

else

++position;

}

// reset $` start:

//if((base != search_base) && (base == backstop))

// m_match_flags |= match_prev_avail;

}

if(m_match_flags & match_posix)

{

m_result.set_base(base);

}

verify_options(re.flags(), m_match_flags);

// find out what kind of expression we have:

unsigned type = (m_match_flags & match_continuous) ?

static_cast<unsigned int>(regbase::restart_continue)

: static_cast<unsigned int>(re.get_restart_type());

// call the appropriate search routine:

matcher_proc_type proc = s_find_vtable[type];

return (this->*proc)();

#if defined(BOOST_REGEX_NON_RECURSIVE) && !defined(BOOST_NO_EXCEPTIONS)

}

catch(...)

{

// unwind all pushed states, apart from anything else this

// ensures that all the states are correctly destructed

// not just the memory freed.

while(unwind(true)){}

throw;

}

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_prefix()

{

m_has_partial_match = false;

m_has_found_match = false;

pstate = re.get_first_state();

m_presult->set_first(position);

restart = position;

match_all_states();

if(!m_has_found_match && m_has_partial_match && (m_match_flags & match_partial))

{

m_has_found_match = true;

m_presult->set_second(last, 0, false);

position = last;

}

#ifdef BOOST_REGEX_MATCH_EXTRA

if(m_has_found_match && (match_extra & m_match_flags))

{

//

// we have a match, reverse the capture information:

//

for(unsigned i = 0; i < m_presult->size(); ++i)

{

typename sub_match<BidiIterator>::capture_sequence_type & seq = ((*m_presult)[i]).get_captures();

std::reverse(seq.begin(), seq.end());

}

}

#endif

if(!m_has_found_match)

position = restart; // reset search postion

return m_has_found_match;

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_literal()

{

unsigned int len = static_cast<const re_literal*>(pstate)->length;

const char_type* what = reinterpret_cast<const char_type*>(static_cast<const re_literal*>(pstate) + 1);

//

// compare string with what we stored in

// our records:

for(unsigned int i = 0; i < len; ++i, ++position)

{

if((position == last) || (traits_inst.translate(*position, icase) != what[i]))

return false;

}

pstate = pstate->next.p;

return true;

}

}

}

pstate = pstate->next.p;

return true;

}

}

else if((m_match_flags & match_not_eol) == 0)

{

pstate = pstate->next.p;

return true;

}

return false;

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_wild()

{

if(position == last)

return false;

if(is_separator(*position) && ((match_any_mask & static_cast<const re_dot*>(pstate)->mask) == 0))

return false;

if((*position == char_type(0)) && (m_match_flags & match_not_dot_null))

return false;

pstate = pstate->next.p;

++position;

return true;

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_word_boundary()

{

bool b; // indcates whether next character is a word character

if(position != last)

{

// prev and this character must be opposites:

b = traits_inst.isctype(*position, m_word_mask);

}

else

{

b = (m_match_flags & match_not_eow) ? true : false;

}

if((position == backstop) && ((m_match_flags & match_prev_avail) == 0))

{

if(m_match_flags & match_not_bow)

b ^= true;

else

b ^= false;

}

else

{

--position;

b ^= traits_inst.isctype(*position, m_word_mask);

++position;

}

if(b)

{

pstate = pstate->next.p;

return true;

}

return false; // no match if we get to here...

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_within_word()

{

if(position == last)

return false;

// both prev and this character must be m_word_mask:

bool prev = traits_inst.isctype(*position, m_word_mask);

{

bool b;

pstate = rep->alt.p;

return true;

}

else

{

// non-greedy, push state and return true if we can skip:

if(count < rep->max)

push_single_repeat(count, rep, position, saved_state_rep_fast_dot);

pstate = rep->alt.p;

return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);

}

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_char_repeat()

{

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable:4127)

#endif

#ifdef __BORLANDC__

#pragma option push -w-8008 -w-8066 -w-8004

#endif

const re_repeat* rep = static_cast<const re_repeat*>(pstate);

BOOST_ASSERT(1 == static_cast<const re_literal*>(rep->next.p)->length);

const char_type what = *reinterpret_cast<const char_type*>(static_cast<const re_literal*>(rep->next.p) + 1);

std::size_t count = 0;

//

// start by working out how much we can skip:

//

bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);

std::size_t desired = greedy ? rep->max : rep->min;

if(::boost::is_random_access_iterator<BidiIterator>::value)

{

BidiIterator end = position;

BidiIterator origin(position);

while((position != end) && (traits_inst.translate(*position, icase) == what))

{

++position;

}

count = (unsigned)::boost::re_detail::distance(origin, position);

}

else

{

while((count < desired) && (position != last) && (traits_inst.translate(*position, icase) == what))

{

++position;

++count;

}

}

if(count < rep->min)

return false;

if(greedy)

{

if((rep->leading) && (count < rep->max))

restart = position;

// push backtrack info if available:

if(count - rep->min)

push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);

// jump to next state:

pstate = rep->alt.p;

return true;

}

else

{

// non-greedy, push state and return true if we can skip:

if(count < rep->max)

push_single_repeat(count, rep, position, saved_state_rep_char);

pstate = rep->alt.p;

return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);

}

#ifdef __BORLANDC__

#pragma option pop

#endif

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_set_repeat()

{

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable:4127)

#endif

#ifdef __BORLANDC__

#pragma option push -w-8008 -w-8066 -w-8004

#endif

const re_repeat* rep = static_cast<const re_repeat*>(pstate);

const unsigned char* map = static_cast<const re_set*>(rep->next.p)->_map;

std::size_t count = 0;

//

// start by working out how much we can skip:

//

bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);

std::size_t desired = greedy ? rep->max : rep->min;

if(::boost::is_random_access_iterator<BidiIterator>::value)

{

BidiIterator end = position;

BidiIterator origin(position);

while((position != end) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])

{

++position;

}

count = (unsigned)::boost::re_detail::distance(origin, position);

}

else

{

while((count < desired) && (position != last) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])

{

++position;

++count;

}

}

if(count < rep->min)

return false;

if(greedy)

{

if((rep->leading) && (count < rep->max))

restart = position;

// push backtrack info if available:

if(count - rep->min)

push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);

// jump to next state:

pstate = rep->alt.p;

return true;

}

else

{

// non-greedy, push state and return true if we can skip:

if(count < rep->max)

push_single_repeat(count, rep, position, saved_state_rep_short_set);

pstate = rep->alt.p;

return (position == last) ? (rep->can_be_null & mask_skip) : can_start(*position, rep->_map, mask_skip);

}

#ifdef __BORLANDC__

#pragma option pop

#endif

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_long_set_repeat()

{

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable:4127)

#endif

#ifdef __BORLANDC__

#pragma option push -w-8008 -w-8066 -w-8004

#endif

typedef typename traits::char_class_type m_type;

const re_repeat* rep = static_cast<const re_repeat*>(pstate);

const re_set_long<m_type>* set = static_cast<const re_set_long<m_type>*>(pstate->next.p);

std::size_t count = 0;

//

// start by working out how much we can skip:

//

bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);

std::size_t desired = greedy ? rep->max : rep->min;

if(::boost::is_random_access_iterator<BidiIterator>::value)

{

BidiIterator end = position;

BidiIterator origin(position);

while((position != end) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))

{

++position;

}

count = (unsigned)::boost::re_detail::distance(origin, position);

}

else

{

while((count < desired) && (position != last) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))

{

++position;

++count;

}

}

if(count < rep->min)

return false;

if(greedy)

{

if((rep->leading) && (count < rep->max))

restart = position;

// push backtrack info if available:

if(count - rep->min)

push_single_repeat(count, rep, position, saved_state_greedy_single_repeat);

// jump to next state:

pstate = rep->alt.p;

return true;

}

else

{

// non-greedy, push state and return true if we can skip:

if(count < rep->max)

push_single_repeat(count, rep, position, saved_state_rep_long_set);

pmp->last_position = position;

}

pstate = rep->alt.p;

return false;

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::unwind_fast_dot_repeat(bool r)

{

saved_single_repeat<BidiIterator>* pmp = static_cast<saved_single_repeat<BidiIterator>*>(m_backup_state);

// if we have a match, just discard this state:

if(r)

{

destroy_single_repeat();

return true;

}

const re_repeat* rep = pmp->rep;

std::size_t count = pmp->count;

BOOST_ASSERT(count < rep->max);

position = pmp->last_position;

if(position != last)

{

// wind forward until we can skip out of the repeat:

do

{

++position;

++count;

++state_count;

}while((count < rep->max) && (position != last) && !can_start(*position, rep->_map, mask_skip));

}

if(position == last)

{

// can't repeat any more, remove the pushed state:

destroy_single_repeat();

if((m_match_flags & match_partial) && (position == last) && (position != search_base))

m_has_partial_match = true;

if(0 == (rep->can_be_null & mask_skip))

return true;

}

else if(count == rep->max)

{

// can't repeat any more, remove the pushed state:

destroy_single_repeat();

if(!can_start(*position, rep->_map, mask_skip))

return true;

}

else

{

pmp->count = count;

pmp->last_position = position;

}

pstate = rep->alt.p;

return false;

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::unwind_char_repeat(bool r)

{

saved_single_repeat<BidiIterator>* pmp = static_cast<saved_single_repeat<BidiIterator>*>(m_backup_state);

// if we have a match, just discard this state:

if(r)

{

destroy_single_repeat();

return true;

}

else

{

return false;

}

}while(true);

#ifdef __BORLANDC__

#pragma option pop

#endif

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_set_repeat()

{

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable:4127)

#endif

#ifdef __BORLANDC__

#pragma option push -w-8008 -w-8066 -w-8004

#endif

const re_repeat* rep = static_cast<const re_repeat*>(pstate);

const unsigned char* map = static_cast<const re_set*>(rep->next.p)->_map;

unsigned count = 0;

//

// start by working out how much we can skip:

//

bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);

std::size_t desired = greedy ? rep->max : rep->min;

if(::boost::is_random_access_iterator<BidiIterator>::value)

{

BidiIterator end = position;

BidiIterator origin(position);

while((position != end) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])

{

++position;

}

count = (unsigned)::boost::re_detail::distance(origin, position);

}

else

{

while((count < desired) && (position != last) && map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])

{

++position;

++count;

}

}

if((rep->leading) && (count < rep->max) && greedy)

restart = position;

if(count < rep->min)

return false;

if(greedy)

return backtrack_till_match(count - rep->min);

// non-greedy, keep trying till we get a match:

BidiIterator save_pos;

do

{

while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))

{

if(map[static_cast<unsigned char>(traits_inst.translate(*position, icase))])

{

++position;

++count;

}

else

else

{

return false;

}

}while(true);

#ifdef __BORLANDC__

#pragma option pop

#endif

#ifdef BOOST_MSVC

#pragma warning(pop)

#endif

}

template <class BidiIterator, class Allocator, class traits>

bool perl_matcher<BidiIterator, Allocator, traits>::match_long_set_repeat()

{

#ifdef BOOST_MSVC

#pragma warning(push)

#pragma warning(disable:4127)

#endif

#ifdef __BORLANDC__

#pragma option push -w-8008 -w-8066 -w-8004

#endif

typedef typename traits::char_class_type char_class_type;

const re_repeat* rep = static_cast<const re_repeat*>(pstate);

const re_set_long<char_class_type>* set = static_cast<const re_set_long<char_class_type>*>(pstate->next.p);

unsigned count = 0;

//

// start by working out how much we can skip:

//

bool greedy = (rep->greedy) && (!(m_match_flags & regex_constants::match_any) || m_independent);

std::size_t desired = greedy ? rep->max : rep->min;

if(::boost::is_random_access_iterator<BidiIterator>::value)

{

BidiIterator end = position;

BidiIterator origin(position);

while((position != end) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))

{

++position;

}

count = (unsigned)::boost::re_detail::distance(origin, position);

}

else

{

while((count < desired) && (position != last) && (position != re_is_set_member(position, last, set, re.get_data(), icase)))

{

++position;

++count;

}

}

if((rep->leading) && (count < rep->max) && greedy)

restart = position;

if(count < rep->min)

return false;

if(greedy)

return backtrack_till_match(count - rep->min);

// non-greedy, keep trying till we get a match:

BidiIterator save_pos;

do

{

while((position != last) && (count < rep->max) && !can_start(*position, rep->_map, mask_skip))

{

if(position != re_is_set_member(position, last, set, re.get_data(), icase))

{

++position;

++count;

}

else

put(*m_position);

++m_position;

break;

case ')':

if(m_flags & boost::regex_constants::format_all)

{

return;

}

put(*m_position);

++m_position;

break;

case ':':

if((m_flags & boost::regex_constants::format_all) && m_have_conditional)

{

return;

}

put(*m_position);

++m_position;

break;

case '?':

if(m_flags & boost::regex_constants::format_all)

{

++m_position;

format_conditional();

break;

}

put(*m_position);

++m_position;

break;

case '$':

if((m_flags & format_sed) == 0)

{

format_perl();

break;

}

default:

put(*m_position);

++m_position;

break;

}

}

}

template <class OutputIterator, class Results, class traits, class ForwardIter>

void basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::format_perl()

{

//

// On entry *m_position points to a '$' character

// output the information that goes with it:

//

BOOST_ASSERT(*m_position == '$');

//

// see if this is a trailing '$':

//

if(++m_position == m_end)

{

--m_position;

put(*m_position);

++m_position;

return;

}

//

// OK find out what kind it is:

//

bool have_brace = false;

ForwardIter save_position = m_position;

switch(*m_position)

{

case '&':

++m_position;

put(this->m_results[0]);

break;

case '`':

++m_position;

put(this->m_results.prefix());

break;

case '\'':

++m_position;

put(this->m_results.suffix());

break;

case '$':

put(*m_position++);

break;

case '+':

if((++m_position != m_end) && (*m_position == '{'))

{

ForwardIter base = ++m_position;

while((m_position != m_end) && (*m_position != '}')) ++m_position;

if(m_position != m_end)

{

// Named sub-expression:

put(get_named_sub(base, m_position));

++m_position;

break;

}

else

{

m_position = --base;

}

}

put((this->m_results)[this->m_results.size() > 1 ? static_cast<int>(this->m_results.size() - 1) : 1]);

break;

case '{':

have_brace = true;

++m_position;

default:

// see if we have a number:

{

std::ptrdiff_t len = ::boost::re_detail::distance(m_position, m_end);

//len = (std::min)(static_cast<std::ptrdiff_t>(2), len);

int v = this->toi(m_position, m_position + len, 10);

if((v < 0) || (have_brace && ((m_position == m_end) || (*m_position != '}'))))

{

// Look for a Perl-5.10 verb:

if(!handle_perl_verb(have_brace))

{

// leave the $ as is, and carry on:

m_position = --save_position;

put(*m_position);

++m_position;

}

break;

}

// otherwise output sub v:

put(this->m_results[v]);

if(have_brace)

++m_position;

}

}

}

template <class OutputIterator, class Results, class traits, class ForwardIter>

bool basic_regex_formatter<OutputIterator, Results, traits, ForwardIter>::handle_perl_verb(bool have_brace)

{

//

// We may have a capitalised string containing a Perl action:

//

static const char_type MATCH[] = { 'M', 'A', 'T', 'C', 'H' };

static const char_type PREMATCH[] = { 'P', 'R', 'E', 'M', 'A', 'T', 'C', 'H' };

static const char_type POSTMATCH[] = { 'P', 'O', 'S', 'T', 'M', 'A', 'T', 'C', 'H' };

template <class S, class OutputIter>

inline OutputIter do_format_string(const S* s, OutputIter i)

{

while(s && *s)

{

*i = *s;

++i;

++s;

}

return i;

}

template <class OutputIter>

OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type /*f*/)

{

return do_format_string(boost::unwrap_ref(func)(m), i);

}

template <class OutputIter, class Traits>

OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits&)

{

return (*this)(m, i, f);

}

private:

Base func;

format_functor1(const format_functor1&);

format_functor1& operator=(const format_functor1&);

};

template <class charT, class Match, class Traits>

struct format_functor_c_string

{

format_functor_c_string(const charT* ps) : func(ps) {}

template <class OutputIter>

OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits& t = Traits())

{

const charT* end = func;

while(*end) ++end;

return regex_format_imp(i, m, func, end, f, t);

}

private:

const charT* func;

format_functor_c_string(const format_functor_c_string&);

format_functor_c_string& operator=(const format_functor_c_string&);

};

template <class Container, class Match, class Traits>

struct format_functor_container

{

format_functor_container(const Container& c) : func(c) {}

template <class OutputIter>

OutputIter operator()(const Match& m, OutputIter i, boost::regex_constants::match_flag_type f, const Traits& t = Traits())

{

return re_detail::regex_format_imp(i, m, func.begin(), func.end(), f, t);

}

private:

const Container& func;

format_functor_container(const format_functor_container&);

format_functor_container& operator=(const format_functor_container&);

};

template <class Func, class Match, class OutputIterator, class Traits = re_detail::trivial_format_traits<typename Match::char_type> >

struct compute_functor_type

{

typedef typename format_traits<Func, Match, OutputIterator>::type tag;

typedef typename boost::remove_cv< typename boost::remove_pointer<Func>::type>::type maybe_char_type;

typedef typename mpl::if_<

::boost::is_same<tag, mpl::int_<0> >, format_functor_c_string<maybe_char_type, Match, Traits>,

typename mpl::if_<

::boost::is_same<tag, mpl::int_<1> >, format_functor_container<Func, Match, Traits>,

typename mpl::if_<

::boost::is_same<tag, mpl::int_<2> >, format_functor1<Func, Match>,

typename mpl::if_<

::boost::is_same<tag, mpl::int_<3> >, format_functor2<Func, Match>,

format_functor3<Func, Match>

>::type

>::type

>::type

>::type type;

};

} // namespace re_detail

template <class OutputIterator, class Iterator, class Allocator, class Functor>

inline OutputIterator regex_format(OutputIterator out,

const match_results<Iterator, Allocator>& m,

Functor fmt,

enum{

padding_size = padding3<sizeof(padding)>::padding_size,

padding_mask = padding3<sizeof(padding)>::padding_mask

};

//

// class raw_storage

// basically this is a simplified vector<unsigned char>

// this is used by basic_regex for expression storage

//

class BOOST_REGEX_DECL raw_storage

{

public:

typedef std::size_t size_type;

typedef unsigned char* pointer;

private:

pointer last, start, end;

public:

raw_storage();

raw_storage(size_type n);

~raw_storage()

{

::operator delete(start);

}

void BOOST_REGEX_CALL resize(size_type n);

void* BOOST_REGEX_CALL extend(size_type n)

{

if(size_type(last - end) < n)

resize(n + (end - start));

end += n;

return result;

}

void* BOOST_REGEX_CALL insert(size_type pos, size_type n);

size_type BOOST_REGEX_CALL size()

{

return end - start;

}

size_type BOOST_REGEX_CALL capacity()

{

return last - start;

}

void* BOOST_REGEX_CALL data()const

{

return start;

}

size_type BOOST_REGEX_CALL index(void* ptr)

{

return static_cast<pointer>(ptr) - static_cast<pointer>(data());

}

void BOOST_REGEX_CALL clear()

{

end = start;

}

void BOOST_REGEX_CALL align()

{

// move end up to a boundary:

end = start + (((end - start) + padding_mask) & ~padding_mask);

// output sub-expressions only:

for(unsigned i = 1; i < what.size(); ++i)

{

*(*p_out) = what.str(i);

++(*p_out);

if(0 == --*p_max) return false;

}

return *p_max != 0;

}

else

{

// output $` only if it's not-null or not at the start of the input:

const sub_match<iterator_type>& sub = what[-1];

if((sub.first != sub.second) || (*p_max != initial_max))

{

*(*p_out) = sub.str();

++(*p_out);

return --*p_max;

}

}

//

// initial null, do nothing:

return true;

}

} // namespace re_detail

template <class OutputIterator, class charT, class Traits1, class Alloc1, class Traits2>

std::size_t regex_split(OutputIterator out,

std::basic_string<charT, Traits1, Alloc1>& s,

const basic_regex<charT, Traits2>& e,

match_flag_type flags,

std::size_t max_split)

{

typedef typename std::basic_string<charT, Traits1, Alloc1>::const_iterator ci_t;

ci_t last = s.begin();

std::size_t init_size = max_split;

re_detail::split_pred<OutputIterator, charT, Traits1, Alloc1> pred(&last, &out, &max_split);

ci_t i, j;

i = s.begin();

j = s.end();

regex_grep(pred, i, j, e, flags);

//

// if there is still input left, do a final push as long as max_split

// is not exhausted, and we're not splitting sub-expressions rather

// than whitespace:

{

*out = std::basic_string<charT, Traits1, Alloc1>((ci_t)last, (ci_t)s.end());

++out;

last = s.end();

--max_split;

}

//

// delete from the string everything that has been processed so far:

s.erase(0, last - s.begin());

//

// return the number of new records pushed:

return init_size - max_split;

}

template <class OutputIterator, class charT, class Traits1, class Alloc1, class Traits2>

inline std::size_t regex_split(OutputIterator out,

std::basic_string<charT, Traits1, Alloc1>& s,

const basic_regex<charT, Traits2>& e,

match_flag_type flags = match_default)

{

return regex_split(out, s, e, flags, UINT_MAX);

}

template <class OutputIterator, class charT, class Traits1, class Alloc1>

inline std::size_t regex_split(OutputIterator out,

std::basic_string<charT, Traits1, Alloc1>& s)

{

return regex_split(out, s, re_detail::get_default_expression(charT(0)), match_default, UINT_MAX);

}

#ifdef BOOST_MSVC

# pragma warning(pop)

#endif

#ifdef BOOST_MSVC

/*

*

* Copyright (c) 2003

* 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 regex_token_iterator.hpp

* VERSION see <boost/version.hpp>

* DESCRIPTION: Provides regex_token_iterator implementation.

*/

#ifndef BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP

#define BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP

#include <boost/shared_ptr.hpp>

#include <boost/detail/workaround.hpp>

#if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003))

//

// Borland C++ Builder 6, and Visual C++ 6,

// can't cope with the array template constructor

// so we have a template member that will accept any type as

// argument, and then assert that is really is an array:

//

#include <boost/static_assert.hpp>

#include <boost/type_traits/is_array.hpp>

#endif

namespace boost{

#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

template <class BidirectionalIterator,

class charT,

class traits>

class regex_token_iterator_implementation

{

typedef basic_regex<charT, traits> regex_type;

typedef sub_match<BidirectionalIterator> value_type;

match_results<BidirectionalIterator> what; // current match

BidirectionalIterator base; // start of search area

BidirectionalIterator end; // end of search area

const regex_type re; // the expression

match_flag_type flags; // match flags

value_type result; // the current string result

int N; // the current sub-expression being enumerated

std::vector<int> subs; // the sub-expressions to enumerate

public:

regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, int sub, match_flag_type f)

: end(last), re(*p), flags(f){ subs.push_back(sub); }

regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const std::vector<int>& v, match_flag_type f)

: end(last), re(*p), flags(f), subs(v){}

#if !BOOST_WORKAROUND(__HP_aCC, < 60700)

#if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \

|| BOOST_WORKAROUND(__HP_aCC, < 60700)

template <class T>

regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const T& submatches, match_flag_type f)

: end(last), re(*p), flags(f)

{

// assert that T really is an array:

BOOST_STATIC_ASSERT(::boost::is_array<T>::value);

const std::size_t array_size = sizeof(T) / sizeof(submatches[0]);

for(std::size_t i = 0; i < array_size; ++i)

{

subs.push_back(submatches[i]);

}

}

#else

template <std::size_t CN>

regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const int (&submatches)[CN], match_flag_type f)

: end(last), re(*p), flags(f)

{

for(std::size_t i = 0; i < CN; ++i)

{

subs.push_back(submatches[i]);

}

}

#endif

#endif

bool init(BidirectionalIterator first)

{

N = 0;

base = first;

if(regex_search(first, end, what, re, flags, base) == true)

{

N = 0;

result = ((subs[N] == -1) ? what.prefix() : what[(int)subs[N]]);

return true;

sub_match<BidirectionalIterator>,

typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type,

const sub_match<BidirectionalIterator>*,

const sub_match<BidirectionalIterator>& >

#endif

{

private:

typedef regex_token_iterator_implementation<BidirectionalIterator, charT, traits> impl;

typedef shared_ptr<impl> pimpl;

public:

typedef basic_regex<charT, traits> regex_type;

typedef sub_match<BidirectionalIterator> value_type;

typedef typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type

difference_type;

typedef const value_type* pointer;

typedef const value_type& reference;

typedef std::forward_iterator_tag iterator_category;

regex_token_iterator(){}

regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

int submatch = 0, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatch, m))

{

if(!pdata->init(a))

pdata.reset();

}

regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const std::vector<int>& submatches, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

#if !BOOST_WORKAROUND(__HP_aCC, < 60700)

#if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \

|| BOOST_WORKAROUND(__HP_aCC, < 60700)

template <class T>

regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const T& submatches, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

#else

template <std::size_t N>

regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const int (&submatches)[N], match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

#endif

#endif

regex_token_iterator(const regex_token_iterator& that)

: pdata(that.pdata) {}

regex_token_iterator& operator=(const regex_token_iterator& that)

{

pdata = that.pdata;

return *this;

}

bool operator==(const regex_token_iterator& that)const

{

if((pdata.get() == 0) || (that.pdata.get() == 0))

return pdata.get() == that.pdata.get();

return pdata->compare(*(that.pdata.get()));

}

bool operator!=(const regex_token_iterator& that)const

regex_token_iterator result(*this);

++(*this);

return result;

}

private:

pimpl pdata;

void cow()

{

// copy-on-write

if(pdata.get() && !pdata.unique())

{

pdata.reset(new impl(*(pdata.get())));

}

}

};

typedef regex_token_iterator<const char*> cregex_token_iterator;

typedef regex_token_iterator<std::string::const_iterator> sregex_token_iterator;

#ifndef BOOST_NO_WREGEX

typedef regex_token_iterator<const wchar_t*> wcregex_token_iterator;

typedef regex_token_iterator<std::wstring::const_iterator> wsregex_token_iterator;

#endif

template <class charT, class traits>

inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);

}

template <class charT, class traits, class ST, class SA>

inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);

}

template <class charT, class traits, std::size_t N>

inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);

}

template <class charT, class traits, class ST, class SA, std::size_t N>

inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);

}

template <class charT, class traits>

inline regex_token_iterator<const charT*, charT, traits> make_regex_token_iterator(const charT* p, const basic_regex<charT, traits>& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<const charT*, charT, traits>(p, p+traits::length(p), e, submatch, m);

}

template <class charT, class traits, class ST, class SA>

inline regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits> make_regex_token_iterator(const std::basic_string<charT, ST, SA>& p, const basic_regex<charT, traits>& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return regex_token_iterator<typename std::basic_string<charT, ST, SA>::const_iterator, charT, traits>(p.begin(), p.end(), e, submatch, m);

}

#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

} // namespace boost

#endif // BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP

#endif

#ifndef BOOST_REGEX_FWD_HPP_INCLUDED

#include <boost/regex_fwd.hpp>

#endif

#include "boost/mpl/has_xxx.hpp"

#include <boost/static_assert.hpp>

#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

namespace boost{

template <class charT, class implementationT >

struct regex_traits : public implementationT

{

regex_traits() : implementationT() {}

};

//

// class regex_traits_wrapper.

// this is what our implementation will actually store;

// it provides default implementations of the "optional"

// interfaces that we support, in addition to the

// required "standard" ones:

//

namespace re_detail{

BOOST_MPL_HAS_XXX_TRAIT_DEF(boost_extensions_tag)

#else

template<class T>

struct has_boost_extensions_tag

{

BOOST_STATIC_CONSTANT(bool, value = false);

};

#endif

template <class BaseT>

struct default_wrapper : public BaseT

{

typedef typename BaseT::char_type char_type;

std::string error_string(::boost::regex_constants::error_type e)const

{

return ::boost::re_detail::get_default_error_string(e);

}

::boost::regex_constants::syntax_type syntax_type(char_type c)const

{

return ((c & 0x7f) == c) ? get_default_syntax_type(static_cast<char>(c)) : ::boost::regex_constants::syntax_char;

}

::boost::regex_constants::escape_syntax_type escape_syntax_type(char_type c)const

{

return ((c & 0x7f) == c) ? get_default_escape_syntax_type(static_cast<char>(c)) : ::boost::regex_constants::escape_type_identity;

}

int toi(const char_type*& p1, const char_type* p2, int radix)const

{

return ::boost::re_detail::global_toi(p1, p2, radix, *this);

}

char_type translate(char_type c, bool icase)const

{

return (icase ? this->translate_nocase(c) : this->translate(c));

}

char_type translate(char_type c)const

{

return BaseT::translate(c);

}

char_type tolower(char_type c)const

{

return ::boost::re_detail::global_lower(c);

}

char_type toupper(char_type c)const

{

return ::boost::re_detail::global_upper(c);

}

};

template <class BaseT, bool has_extensions>

struct compute_wrapper_base

{

typedef BaseT type;

};

template <class BaseT>

struct compute_wrapper_base<BaseT, false>

{

typedef default_wrapper<BaseT> type;

};

#else

template <>

struct compute_wrapper_base<c_regex_traits<char>, false>

{

typedef default_wrapper<c_regex_traits<char> > type;

};

#ifndef BOOST_NO_WREGEX

template <>

struct compute_wrapper_base<c_regex_traits<wchar_t>, false>

{

typedef default_wrapper<c_regex_traits<wchar_t> > type;

};

#endif

#endif

} // namespace re_detail

template <class BaseT>

struct regex_traits_wrapper

: public ::boost::re_detail::compute_wrapper_base<

BaseT,

::boost::re_detail::has_boost_extensions_tag<BaseT>::value

>::type

{

regex_traits_wrapper(){}

private:

regex_traits_wrapper(const regex_traits_wrapper&);

regex_traits_wrapper& operator=(const regex_traits_wrapper&);

};

/*

*

* Copyright (c) 2004

* John Maddock

*

* 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 regex_traits_defaults.hpp

* VERSION see <boost/version.hpp>

* DESCRIPTION: Declares API's for access to regex_traits default properties.

*/

#ifndef BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED

#define BOOST_REGEX_TRAITS_DEFAULTS_HPP_INCLUDED

#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

#ifndef BOOST_REGEX_SYNTAX_TYPE_HPP

#include <boost/regex/v4/syntax_type.hpp>

#endif

#ifndef BOOST_REGEX_ERROR_TYPE_HPP

#include <boost/regex/v4/error_type.hpp>

#endif

#ifdef BOOST_NO_STDC_NAMESPACE

namespace std{

using ::strlen;

}

#endif

namespace boost{ namespace re_detail{

inline bool is_extended(charT c)

{ return c > 256; }

inline bool is_extended(char)

{ return false; }

BOOST_REGEX_DECL const char* BOOST_REGEX_CALL get_default_syntax(regex_constants::syntax_type n);

BOOST_REGEX_DECL const char* BOOST_REGEX_CALL get_default_error_string(regex_constants::error_type n);

BOOST_REGEX_DECL regex_constants::syntax_type BOOST_REGEX_CALL get_default_syntax_type(char c);

BOOST_REGEX_DECL regex_constants::escape_syntax_type BOOST_REGEX_CALL get_default_escape_syntax_type(char c);

// is charT c a combining character?

BOOST_REGEX_DECL bool BOOST_REGEX_CALL is_combining_implementation(uint_least16_t s);

template <class charT>

inline bool is_combining(charT c)

{

return (c <= static_cast<charT>(0)) ? false : ((c >= static_cast<charT>((std::numeric_limits<uint_least16_t>::max)())) ? false : is_combining_implementation(static_cast<unsigned short>(c)));

}

template <>

inline bool is_combining<char>(char)

{

return false;

}

template <>

inline bool is_combining<signed char>(signed char)

{

return false;

}

template <>

inline bool is_combining<unsigned char>(unsigned char)

{

return false;

}

#if !defined(__hpux) && !defined(__WINSCW__) // can't use WCHAR_MAX/MIN in pp-directives

template<>

inline bool is_combining<wchar_t>(wchar_t c)

{

return is_combining_implementation(static_cast<unsigned short>(c));

}

#elif !defined(__DECCXX) && !defined(__osf__) && !defined(__OSF__) && defined(WCHAR_MIN) && (WCHAR_MIN == 0) && !defined(BOOST_NO_INTRINSIC_WCHAR_T)

#if defined(WCHAR_MAX) && (WCHAR_MAX <= USHRT_MAX)

template<>

inline bool is_combining<wchar_t>(wchar_t c)

{

return is_combining_implementation(static_cast<unsigned short>(c));

}

#else

template<>

inline bool is_combining<wchar_t>(wchar_t c)

{

return (c >= (std::numeric_limits<uint_least16_t>::max)()) ? false : is_combining_implementation(static_cast<unsigned short>(c));

}

#endif

#endif

#endif

//

// is a charT c a line separator?

//

template <class charT>

inline bool is_separator(charT c)

{

return BOOST_REGEX_MAKE_BOOL(

|| (static_cast<boost::uint16_t>(c) == 0x85u));

}

template <>

inline bool is_separator<char>(char c)

{

return BOOST_REGEX_MAKE_BOOL((c == '\n') || (c == '\r') || (c == '\f'));

}

//

// get a default collating element:

//

BOOST_REGEX_DECL std::string BOOST_REGEX_CALL lookup_default_collate_name(const std::string& name);

//

// get the state_id of a character clasification, the individual

// traits classes then transform that state_id into a bitmask:

//

template <class charT>

struct character_pointer_range

{

const charT* p1;

const charT* p2;

bool operator < (const character_pointer_range& r)const

{

return std::lexicographical_compare(p1, p2, r.p1, r.p2);

}

bool operator == (const character_pointer_range& r)const

{

// Not only do we check that the ranges are of equal size before

// calling std::equal, but there is no other algorithm available:

// not even a non-standard MS one. So forward to unchecked_equal

// in the MS case.

return ((p2 - p1) == (r.p2 - r.p1)) && re_detail::equal(p1, p2, r.p1);

}

};

template <class charT>

int get_default_class_id(const charT* p1, const charT* p2)

{

static const charT data[73] = {

'a', 'l', 'n', 'u', 'm',

'a', 'l', 'p', 'h', 'a',

'b', 'l', 'a', 'n', 'k',

'c', 'n', 't', 'r', 'l',

'd', 'i', 'g', 'i', 't',

'g', 'r', 'a', 'p', 'h',

'l', 'o', 'w', 'e', 'r',

'p', 'r', 'i', 'n', 't',

'p', 'u', 'n', 'c', 't',

's', 'p', 'a', 'c', 'e',

'u', 'n', 'i', 'c', 'o', 'd', 'e',

'u', 'p', 'p', 'e', 'r',

'v',

'w', 'o', 'r', 'd',

'x', 'd', 'i', 'g', 'i', 't',

};

{

{data+0, data+5,}, // alnum

{data+5, data+10,}, // alpha

{data+10, data+15,}, // blank

{data+15, data+20,}, // cntrl

{data+20, data+21,}, // d

{data+20, data+25,}, // digit

{data+25, data+30,}, // graph

{data+29, data+30,}, // h

{data+30, data+31,}, // l

{data+30, data+35,}, // lower

{data+35, data+40,}, // print

{data+40, data+45,}, // punct

{data+45, data+46,}, // s

{data+45, data+50,}, // space

{data+57, data+58,}, // u

{data+50, data+57,}, // unicode

{data+57, data+62,}, // upper

{data+62, data+63,}, // v

{data+63, data+64,}, // w

{data+63, data+67,}, // word

{data+67, data+73,}, // xdigit

};

static const character_pointer_range<charT>* ranges_begin = ranges;

static const character_pointer_range<charT>* ranges_end = ranges + (sizeof(ranges)/sizeof(ranges[0]));

character_pointer_range<charT> t = { p1, p2, };

const character_pointer_range<charT>* p = std::lower_bound(ranges_begin, ranges_end, t);

if((p != ranges_end) && (t == *p))

return static_cast<int>(p - ranges);

return -1;

}

//

// helper functions:

//

template <class charT>

std::ptrdiff_t global_length(const charT* p)

{

std::ptrdiff_t n = 0;

while(*p)

{

++p;

++n;

}

return n;

}

template<>

inline std::ptrdiff_t global_length<char>(const char* p)

{

return (std::strlen)(p);

}

#ifndef BOOST_NO_WREGEX

template<>

inline std::ptrdiff_t global_length<wchar_t>(const wchar_t* p)

{

return (std::wcslen)(p);

}

#endif

template <class charT>

inline charT BOOST_REGEX_CALL global_lower(charT c)

if(c > f) return -1;

if(c >= a) return 10 + (c - a);

if(c > F) return -1;

if(c >= A) return 10 + (c - A);

if(c > nine) return -1;

if(c >= zero) return c - zero;

return -1;

}

template <class charT, class traits>

int global_toi(const charT*& p1, const charT* p2, int radix, const traits& t)

{

(void)t; // warning suppression

int next_value = t.value(*p1, radix);

if((p1 == p2) || (next_value < 0) || (next_value >= radix))

return -1;

int result = 0;

while(p1 != p2)

{

next_value = t.value(*p1, radix);

if((next_value < 0) || (next_value >= radix))

break;

result *= radix;

result += next_value;

++p1;

}

return result;

}

template <class charT>

inline const charT* get_escape_R_string()

{

#ifdef BOOST_MSVC

# pragma warning(push)

# pragma warning(disable:4309 4245)

#endif

'\\', 'x', '{', '2', '0', '2', '9', '}', ']', ')', '\0' };

charT c = static_cast<charT>(0x2029u);

bool b = (static_cast<unsigned>(c) == 0x2029u);

return (b ? e1 : e2);

#ifdef BOOST_MSVC

# pragma warning(pop)

#endif

}

template <>

inline const char* get_escape_R_string<char>()

{

#ifdef BOOST_MSVC

# pragma warning(push)

# pragma warning(disable:4309)

#endif

'|', '[', '\x0A', '\x0B', '\x0C', '\x85', ']', ')', '\0' };

return e2;

#ifdef BOOST_MSVC

# pragma warning(pop)

#endif

}

} // re_detail

} // boost

#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

* 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 sub_match.cpp

* VERSION see <boost/version.hpp>

* DESCRIPTION: Declares template class sub_match.

*/

#ifndef BOOST_REGEX_V4_SUB_MATCH_HPP

#define BOOST_REGEX_V4_SUB_MATCH_HPP

#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

namespace boost{

template <class BidiIterator>

struct sub_match : public std::pair<BidiIterator, BidiIterator>

{

typedef typename re_detail::regex_iterator_traits<BidiIterator>::value_type value_type;

typedef std::ptrdiff_t difference_type;

#else

typedef typename re_detail::regex_iterator_traits<BidiIterator>::difference_type difference_type;

#endif

typedef BidiIterator iterator_type;

typedef BidiIterator iterator;

typedef BidiIterator const_iterator;

bool matched;

sub_match() : std::pair<BidiIterator, BidiIterator>(), matched(false) {}

sub_match(BidiIterator i) : std::pair<BidiIterator, BidiIterator>(i, i), matched(false) {}

#if !defined(BOOST_NO_TEMPLATED_ITERATOR_CONSTRUCTORS)\

&& !BOOST_WORKAROUND(__BORLANDC__, <= 0x0551)\

&& !BOOST_WORKAROUND(__DECCXX_VER, BOOST_TESTED_AT(60590042))

template <class T, class A>

operator std::basic_string<value_type, T, A> ()const

{

return matched ? std::basic_string<value_type, T, A>(this->first, this->second) : std::basic_string<value_type, T, A>();

}

#else

operator std::basic_string<value_type> ()const

{

return str();

}

#endif

difference_type BOOST_REGEX_CALL length()const

{

difference_type n = matched ? ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second) : 0;

return n;

}

std::basic_string<value_type> str()const

{

std::basic_string<value_type> result;

if(matched)

{

std::size_t len = ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second);

result.reserve(len);

BidiIterator i = this->first;

while(i != this->second)

{

result.append(1, *i);

++i;

}

}

return result;

}

int compare(const sub_match& s)const

/*

*

* Copyright (c) 2003

* 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 u32regex_token_iterator.hpp

* VERSION see <boost/version.hpp>

* DESCRIPTION: Provides u32regex_token_iterator implementation.

*/

#ifndef BOOST_REGEX_V4_U32REGEX_TOKEN_ITERATOR_HPP

#define BOOST_REGEX_V4_U32REGEX_TOKEN_ITERATOR_HPP

#if (BOOST_WORKAROUND(__BORLANDC__, >= 0x560) && BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x570)))\

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003))

//

// Borland C++ Builder 6, and Visual C++ 6,

// can't cope with the array template constructor

// so we have a template member that will accept any type as

// argument, and then assert that is really is an array:

//

#include <boost/static_assert.hpp>

#include <boost/type_traits/is_array.hpp>

#endif

namespace boost{

#ifdef BOOST_HAS_ABI_HEADERS

# include BOOST_ABI_PREFIX

#endif

# pragma warning(push)

# pragma warning(disable:4700)

#endif

template <class BidirectionalIterator>

class u32regex_token_iterator_implementation

{

typedef u32regex regex_type;

typedef sub_match<BidirectionalIterator> value_type;

match_results<BidirectionalIterator> what; // current match

BidirectionalIterator end; // end of search area

BidirectionalIterator base; // start of search area

const regex_type re; // the expression

match_flag_type flags; // match flags

value_type result; // the current string result

int N; // the current sub-expression being enumerated

std::vector<int> subs; // the sub-expressions to enumerate

public:

u32regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, int sub, match_flag_type f)

: end(last), re(*p), flags(f){ subs.push_back(sub); }

u32regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const std::vector<int>& v, match_flag_type f)

: end(last), re(*p), flags(f), subs(v){}

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \

|| BOOST_WORKAROUND(__HP_aCC, < 60700)

template <class T>

u32regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const T& submatches, match_flag_type f)

: end(last), re(*p), flags(f)

{

// assert that T really is an array:

BOOST_STATIC_ASSERT(::boost::is_array<T>::value);

const std::size_t array_size = sizeof(T) / sizeof(submatches[0]);

for(std::size_t i = 0; i < array_size; ++i)

{

subs.push_back(submatches[i]);

}

}

#else

template <std::size_t CN>

u32regex_token_iterator_implementation(const regex_type* p, BidirectionalIterator last, const int (&submatches)[CN], match_flag_type f)

: end(last), re(*p), flags(f)

{

for(std::size_t i = 0; i < CN; ++i)

{

subs.push_back(submatches[i]);

}

}

#endif

bool init(BidirectionalIterator first)

{

base = first;

N = 0;

if(u32regex_search(first, end, what, re, flags, base) == true)

{

N = 0;

result = ((subs[N] == -1) ? what.prefix() : what[(int)subs[N]]);

return true;

: public std::iterator<

std::forward_iterator_tag,

sub_match<BidirectionalIterator>,

typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type,

const sub_match<BidirectionalIterator>*,

const sub_match<BidirectionalIterator>& >

#endif

{

private:

typedef u32regex_token_iterator_implementation<BidirectionalIterator> impl;

typedef shared_ptr<impl> pimpl;

public:

typedef u32regex regex_type;

typedef sub_match<BidirectionalIterator> value_type;

typedef typename re_detail::regex_iterator_traits<BidirectionalIterator>::difference_type

difference_type;

typedef const value_type* pointer;

typedef const value_type& reference;

typedef std::forward_iterator_tag iterator_category;

u32regex_token_iterator(){}

u32regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

int submatch = 0, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatch, m))

{

if(!pdata->init(a))

pdata.reset();

}

u32regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const std::vector<int>& submatches, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

|| BOOST_WORKAROUND(__MWERKS__, BOOST_TESTED_AT(0x3003)) \

|| BOOST_WORKAROUND(__HP_aCC, < 60700)

template <class T>

u32regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const T& submatches, match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

#else

template <std::size_t N>

u32regex_token_iterator(BidirectionalIterator a, BidirectionalIterator b, const regex_type& re,

const int (&submatches)[N], match_flag_type m = match_default)

: pdata(new impl(&re, b, submatches, m))

{

if(!pdata->init(a))

pdata.reset();

}

#endif

u32regex_token_iterator(const u32regex_token_iterator& that)

: pdata(that.pdata) {}

u32regex_token_iterator& operator=(const u32regex_token_iterator& that)

{

pdata = that.pdata;

return *this;

}

bool operator==(const u32regex_token_iterator& that)const

{

if((pdata.get() == 0) || (that.pdata.get() == 0))

return pdata.get() == that.pdata.get();

return pdata->compare(*(that.pdata.get()));

}

bool operator!=(const u32regex_token_iterator& that)const

{ return !(*this == that); }

}

};

typedef u32regex_token_iterator<const char*> utf8regex_token_iterator;

typedef u32regex_token_iterator<const UChar*> utf16regex_token_iterator;

typedef u32regex_token_iterator<const UChar32*> utf32regex_token_iterator;

// construction from an integral sub_match state_id:

inline u32regex_token_iterator<const char*> make_u32regex_token_iterator(const char* p, const u32regex& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const char*>(p, p+std::strlen(p), e, submatch, m);

}

#ifndef BOOST_NO_WREGEX

inline u32regex_token_iterator<const wchar_t*> make_u32regex_token_iterator(const wchar_t* p, const u32regex& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const wchar_t*>(p, p+std::wcslen(p), e, submatch, m);

}

#endif

#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2)

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const UChar* p, const u32regex& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(p, p+u_strlen(p), e, submatch, m);

}

#endif

template <class charT, class Traits, class Alloc>

inline u32regex_token_iterator<typename std::basic_string<charT, Traits, Alloc>::const_iterator> make_u32regex_token_iterator(const std::basic_string<charT, Traits, Alloc>& p, const u32regex& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

typedef typename std::basic_string<charT, Traits, Alloc>::const_iterator iter_type;

return u32regex_token_iterator<iter_type>(p.begin(), p.end(), e, submatch, m);

}

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const U_NAMESPACE_QUALIFIER UnicodeString& s, const u32regex& e, int submatch = 0, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(s.getBuffer(), s.getBuffer() + s.length(), e, submatch, m);

}

// construction from a reference to an array:

template <std::size_t N>

inline u32regex_token_iterator<const char*> make_u32regex_token_iterator(const char* p, const u32regex& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const char*>(p, p+std::strlen(p), e, submatch, m);

}

#ifndef BOOST_NO_WREGEX

template <std::size_t N>

inline u32regex_token_iterator<const wchar_t*> make_u32regex_token_iterator(const wchar_t* p, const u32regex& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const wchar_t*>(p, p+std::wcslen(p), e, submatch, m);

}

#endif

#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2)

template <std::size_t N>

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const UChar* p, const u32regex& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(p, p+u_strlen(p), e, submatch, m);

}

#endif

template <class charT, class Traits, class Alloc, std::size_t N>

inline u32regex_token_iterator<typename std::basic_string<charT, Traits, Alloc>::const_iterator> make_u32regex_token_iterator(const std::basic_string<charT, Traits, Alloc>& p, const u32regex& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

typedef typename std::basic_string<charT, Traits, Alloc>::const_iterator iter_type;

return u32regex_token_iterator<iter_type>(p.begin(), p.end(), e, submatch, m);

}

template <std::size_t N>

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const U_NAMESPACE_QUALIFIER UnicodeString& s, const u32regex& e, const int (&submatch)[N], regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(s.getBuffer(), s.getBuffer() + s.length(), e, submatch, m);

}

// construction from a vector of sub_match state_id's:

inline u32regex_token_iterator<const char*> make_u32regex_token_iterator(const char* p, const u32regex& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const char*>(p, p+std::strlen(p), e, submatch, m);

}

#ifndef BOOST_NO_WREGEX

inline u32regex_token_iterator<const wchar_t*> make_u32regex_token_iterator(const wchar_t* p, const u32regex& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const wchar_t*>(p, p+std::wcslen(p), e, submatch, m);

}

#endif

#if !defined(U_WCHAR_IS_UTF16) && (U_SIZEOF_WCHAR_T != 2)

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const UChar* p, const u32regex& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(p, p+u_strlen(p), e, submatch, m);

}

#endif

template <class charT, class Traits, class Alloc>

inline u32regex_token_iterator<typename std::basic_string<charT, Traits, Alloc>::const_iterator> make_u32regex_token_iterator(const std::basic_string<charT, Traits, Alloc>& p, const u32regex& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

typedef typename std::basic_string<charT, Traits, Alloc>::const_iterator iter_type;

return u32regex_token_iterator<iter_type>(p.begin(), p.end(), e, submatch, m);

}

inline u32regex_token_iterator<const UChar*> make_u32regex_token_iterator(const U_NAMESPACE_QUALIFIER UnicodeString& s, const u32regex& e, const std::vector<int>& submatch, regex_constants::match_flag_type m = regex_constants::match_default)

{

return u32regex_token_iterator<const UChar*>(s.getBuffer(), s.getBuffer() + s.length(), e, submatch, m);

}

# pragma warning(pop)

#endif

#ifdef BOOST_HAS_ABI_HEADERS

# include BOOST_ABI_SUFFIX

#endif

} // namespace boost

#endif // BOOST_REGEX_V4_REGEX_TOKEN_ITERATOR_HPP

{

// get a regular sort key, and then truncate it:

result.assign(this->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->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;

}

if(result.empty())

result = string_type(1, charT(0));

return result;

}

template <class charT>

typename w32_regex_traits_implementation<charT>::string_type

w32_regex_traits_implementation<charT>::lookup_collatename(const charT* p1, const charT* p2) const

{

typedef typename std::map<string_type, string_type>::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(__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(__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 <class charT>

w32_regex_traits_implementation<charT>::w32_regex_traits_implementation(::boost::re_detail::lcid_type l)

: w32_regex_traits_char_layer<charT>(l)

{

cat_type cat;

std::string cat_name(w32_regex_traits<charT>::get_catalog_name());

if(cat_name.size())

{

cat = ::boost::re_detail::w32_cat_open(cat_name);

if(!cat)

{

std::string m("Unable to open message catalog: ");

std::runtime_error err(m + cat_name);

boost::re_detail::raise_runtime_error(err);

}

0x0040u, // C1_BLANK

0x0020u, // C1_CNTRL

0x0004u, // C1_DIGIT

0x0004u, // C1_DIGIT

(~(0x0020u|0x0008u|0x0040) & 0x01ffu) | 0x0400u, // not C1_CNTRL or C1_SPACE or C1_BLANK

w32_regex_traits_implementation<charT>::mask_horizontal,

0x0002u, // C1_LOWER

0x0002u, // C1_LOWER

(~0x0020u & 0x01ffu) | 0x0400, // not C1_CNTRL

0x0010u, // C1_PUNCT

0x0008u, // C1_SPACE

0x0008u, // C1_SPACE

0x0001u, // C1_UPPER

w32_regex_traits_implementation<charT>::mask_unicode,

0x0001u, // C1_UPPER

w32_regex_traits_implementation<charT>::mask_vertical,

0x0104u | w32_regex_traits_implementation<charT>::mask_word,

0x0104u | w32_regex_traits_implementation<charT>::mask_word,

0x0080u, // C1_XDIGIT

};

if(m_custom_class_names.size())

{

typedef typename std::map<std::basic_string<charT>, 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);

if(state_id < sizeof(masks) / sizeof(masks[0]))

return masks[state_id];

return masks[0];

}

template <class charT>

{

// TODO: create a cache for previously constructed objects.

return boost::object_cache< ::boost::re_detail::lcid_type, w32_regex_traits_implementation<charT> >::get(l, 5);

}

} // re_detail

template <class charT>

class w32_regex_traits

{

public:

typedef charT char_type;

typedef std::size_t size_type;

typedef std::basic_string<char_type> string_type;

typedef ::boost::re_detail::lcid_type locale_type;

typedef boost::uint_least32_t char_class_type;

struct boost_extensions_tag{};

w32_regex_traits()

: m_pimpl(re_detail::create_w32_regex_traits<charT>(::boost::re_detail::w32_get_default_locale()))

{ }

static size_type length(const char_type* p)

{

return std::char_traits<charT>::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