gcc/libstdc++-v3/include/bits/locale_facets.h
Benjamin Kosnik d8dd52a990 user.cfg.in: Update to Doxygen 1.8.0.
2012-03-08  Benjamin Kosnik  <bkoz@redhat.com>

	    * doc/doxygen/user.cfg.in: Update to Doxygen 1.8.0.
	    * include/bits/locale_facets.h: Adjust markup to avoid warnings.
	    * include/ext/pb_ds/assoc_container.hpp: Same.
	    * include/ext/pb_ds/priority_queue.hpp: Same.
	    * include/std/fstream: Same.
	    * include/std/ratio: Same.
	    * testsuite/20_util/ratio/cons/cons_overflow_neg.cc: Adjust line
	    numbers.
	    * testsuite/20_util/ratio/operations/ops_overflow_neg.cc: Same.

Co-Authored-By: Ramana Radhakrishnan <ramana@gcc.gnu.org>

From-SVN: r185094
2012-03-08 05:20:52 +00:00

2613 lines
89 KiB
C++

// Locale support -*- C++ -*-
// Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
// 2006, 2007, 2008, 2009, 2010, 2011, 2012
// Free Software Foundation, Inc.
//
// This file is part of the GNU ISO C++ Library. This library is free
// software; you can redistribute it and/or modify it under the
// terms of the GNU General Public License as published by the
// Free Software Foundation; either version 3, or (at your option)
// any later version.
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
// Under Section 7 of GPL version 3, you are granted additional
// permissions described in the GCC Runtime Library Exception, version
// 3.1, as published by the Free Software Foundation.
// You should have received a copy of the GNU General Public License and
// a copy of the GCC Runtime Library Exception along with this program;
// see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
// <http://www.gnu.org/licenses/>.
/** @file bits/locale_facets.h
* This is an internal header file, included by other library headers.
* Do not attempt to use it directly. @headername{locale}
*/
//
// ISO C++ 14882: 22.1 Locales
//
#ifndef _LOCALE_FACETS_H
#define _LOCALE_FACETS_H 1
#pragma GCC system_header
#include <cwctype> // For wctype_t
#include <cctype>
#include <bits/ctype_base.h>
#include <iosfwd>
#include <bits/ios_base.h> // For ios_base, ios_base::iostate
#include <streambuf>
#include <bits/cpp_type_traits.h>
#include <ext/type_traits.h>
#include <ext/numeric_traits.h>
#include <bits/streambuf_iterator.h>
namespace std _GLIBCXX_VISIBILITY(default)
{
_GLIBCXX_BEGIN_NAMESPACE_VERSION
// NB: Don't instantiate required wchar_t facets if no wchar_t support.
#ifdef _GLIBCXX_USE_WCHAR_T
# define _GLIBCXX_NUM_FACETS 28
#else
# define _GLIBCXX_NUM_FACETS 14
#endif
// Convert string to numeric value of type _Tp and store results.
// NB: This is specialized for all required types, there is no
// generic definition.
template<typename _Tp>
void
__convert_to_v(const char*, _Tp&, ios_base::iostate&,
const __c_locale&) throw();
// Explicit specializations for required types.
template<>
void
__convert_to_v(const char*, float&, ios_base::iostate&,
const __c_locale&) throw();
template<>
void
__convert_to_v(const char*, double&, ios_base::iostate&,
const __c_locale&) throw();
template<>
void
__convert_to_v(const char*, long double&, ios_base::iostate&,
const __c_locale&) throw();
// NB: __pad is a struct, rather than a function, so it can be
// partially-specialized.
template<typename _CharT, typename _Traits>
struct __pad
{
static void
_S_pad(ios_base& __io, _CharT __fill, _CharT* __news,
const _CharT* __olds, streamsize __newlen, streamsize __oldlen);
};
// Used by both numeric and monetary facets.
// Inserts "group separator" characters into an array of characters.
// It's recursive, one iteration per group. It moves the characters
// in the buffer this way: "xxxx12345" -> "12,345xxx". Call this
// only with __gsize != 0.
template<typename _CharT>
_CharT*
__add_grouping(_CharT* __s, _CharT __sep,
const char* __gbeg, size_t __gsize,
const _CharT* __first, const _CharT* __last);
// This template permits specializing facet output code for
// ostreambuf_iterator. For ostreambuf_iterator, sputn is
// significantly more efficient than incrementing iterators.
template<typename _CharT>
inline
ostreambuf_iterator<_CharT>
__write(ostreambuf_iterator<_CharT> __s, const _CharT* __ws, int __len)
{
__s._M_put(__ws, __len);
return __s;
}
// This is the unspecialized form of the template.
template<typename _CharT, typename _OutIter>
inline
_OutIter
__write(_OutIter __s, const _CharT* __ws, int __len)
{
for (int __j = 0; __j < __len; __j++, ++__s)
*__s = __ws[__j];
return __s;
}
// 22.2.1.1 Template class ctype
// Include host and configuration specific ctype enums for ctype_base.
/**
* @brief Common base for ctype facet
*
* This template class provides implementations of the public functions
* that forward to the protected virtual functions.
*
* This template also provides abstract stubs for the protected virtual
* functions.
*/
template<typename _CharT>
class __ctype_abstract_base : public locale::facet, public ctype_base
{
public:
// Types:
/// Typedef for the template parameter
typedef _CharT char_type;
/**
* @brief Test char_type classification.
*
* This function finds a mask M for @a __c and compares it to
* mask @a __m. It does so by returning the value of
* ctype<char_type>::do_is().
*
* @param __c The char_type to compare the mask of.
* @param __m The mask to compare against.
* @return (M & __m) != 0.
*/
bool
is(mask __m, char_type __c) const
{ return this->do_is(__m, __c); }
/**
* @brief Return a mask array.
*
* This function finds the mask for each char_type in the range [lo,hi)
* and successively writes it to vec. vec must have as many elements
* as the char array. It does so by returning the value of
* ctype<char_type>::do_is().
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __vec Pointer to an array of mask storage.
* @return @a __hi.
*/
const char_type*
is(const char_type *__lo, const char_type *__hi, mask *__vec) const
{ return this->do_is(__lo, __hi, __vec); }
/**
* @brief Find char_type matching a mask
*
* This function searches for and returns the first char_type c in
* [lo,hi) for which is(m,c) is true. It does so by returning
* ctype<char_type>::do_scan_is().
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to matching char_type if found, else @a __hi.
*/
const char_type*
scan_is(mask __m, const char_type* __lo, const char_type* __hi) const
{ return this->do_scan_is(__m, __lo, __hi); }
/**
* @brief Find char_type not matching a mask
*
* This function searches for and returns the first char_type c in
* [lo,hi) for which is(m,c) is false. It does so by returning
* ctype<char_type>::do_scan_not().
*
* @param __m The mask to compare against.
* @param __lo Pointer to first char in range.
* @param __hi Pointer to end of range.
* @return Pointer to non-matching char if found, else @a __hi.
*/
const char_type*
scan_not(mask __m, const char_type* __lo, const char_type* __hi) const
{ return this->do_scan_not(__m, __lo, __hi); }
/**
* @brief Convert to uppercase.
*
* This function converts the argument to uppercase if possible.
* If not possible (for example, '2'), returns the argument. It does
* so by returning ctype<char_type>::do_toupper().
*
* @param __c The char_type to convert.
* @return The uppercase char_type if convertible, else @a __c.
*/
char_type
toupper(char_type __c) const
{ return this->do_toupper(__c); }
/**
* @brief Convert array to uppercase.
*
* This function converts each char_type in the range [lo,hi) to
* uppercase if possible. Other elements remain untouched. It does so
* by returning ctype<char_type>:: do_toupper(lo, hi).
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
const char_type*
toupper(char_type *__lo, const char_type* __hi) const
{ return this->do_toupper(__lo, __hi); }
/**
* @brief Convert to lowercase.
*
* This function converts the argument to lowercase if possible. If
* not possible (for example, '2'), returns the argument. It does so
* by returning ctype<char_type>::do_tolower(c).
*
* @param __c The char_type to convert.
* @return The lowercase char_type if convertible, else @a __c.
*/
char_type
tolower(char_type __c) const
{ return this->do_tolower(__c); }
/**
* @brief Convert array to lowercase.
*
* This function converts each char_type in the range [__lo,__hi) to
* lowercase if possible. Other elements remain untouched. It does so
* by returning ctype<char_type>:: do_tolower(__lo, __hi).
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
const char_type*
tolower(char_type* __lo, const char_type* __hi) const
{ return this->do_tolower(__lo, __hi); }
/**
* @brief Widen char to char_type
*
* This function converts the char argument to char_type using the
* simplest reasonable transformation. It does so by returning
* ctype<char_type>::do_widen(c).
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @return The converted char_type.
*/
char_type
widen(char __c) const
{ return this->do_widen(__c); }
/**
* @brief Widen array to char_type
*
* This function converts each char in the input to char_type using the
* simplest reasonable transformation. It does so by returning
* ctype<char_type>::do_widen(c).
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
const char*
widen(const char* __lo, const char* __hi, char_type* __to) const
{ return this->do_widen(__lo, __hi, __to); }
/**
* @brief Narrow char_type to char
*
* This function converts the char_type to char using the simplest
* reasonable transformation. If the conversion fails, dfault is
* returned instead. It does so by returning
* ctype<char_type>::do_narrow(__c).
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char_type to convert.
* @param __dfault Char to return if conversion fails.
* @return The converted char.
*/
char
narrow(char_type __c, char __dfault) const
{ return this->do_narrow(__c, __dfault); }
/**
* @brief Narrow array to char array
*
* This function converts each char_type in the input to char using the
* simplest reasonable transformation and writes the results to the
* destination array. For any char_type in the input that cannot be
* converted, @a dfault is used instead. It does so by returning
* ctype<char_type>::do_narrow(__lo, __hi, __dfault, __to).
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __dfault Char to use if conversion fails.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
const char_type*
narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const
{ return this->do_narrow(__lo, __hi, __dfault, __to); }
protected:
explicit
__ctype_abstract_base(size_t __refs = 0): facet(__refs) { }
virtual
~__ctype_abstract_base() { }
/**
* @brief Test char_type classification.
*
* This function finds a mask M for @a c and compares it to mask @a m.
*
* do_is() is a hook for a derived facet to change the behavior of
* classifying. do_is() must always return the same result for the
* same input.
*
* @param __c The char_type to find the mask of.
* @param __m The mask to compare against.
* @return (M & __m) != 0.
*/
virtual bool
do_is(mask __m, char_type __c) const = 0;
/**
* @brief Return a mask array.
*
* This function finds the mask for each char_type in the range [lo,hi)
* and successively writes it to vec. vec must have as many elements
* as the input.
*
* do_is() is a hook for a derived facet to change the behavior of
* classifying. do_is() must always return the same result for the
* same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __vec Pointer to an array of mask storage.
* @return @a __hi.
*/
virtual const char_type*
do_is(const char_type* __lo, const char_type* __hi,
mask* __vec) const = 0;
/**
* @brief Find char_type matching mask
*
* This function searches for and returns the first char_type c in
* [__lo,__hi) for which is(__m,c) is true.
*
* do_scan_is() is a hook for a derived facet to change the behavior of
* match searching. do_is() must always return the same result for the
* same input.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a matching char_type if found, else @a __hi.
*/
virtual const char_type*
do_scan_is(mask __m, const char_type* __lo,
const char_type* __hi) const = 0;
/**
* @brief Find char_type not matching mask
*
* This function searches for and returns a pointer to the first
* char_type c of [lo,hi) for which is(m,c) is false.
*
* do_scan_is() is a hook for a derived facet to change the behavior of
* match searching. do_is() must always return the same result for the
* same input.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a non-matching char_type if found, else @a __hi.
*/
virtual const char_type*
do_scan_not(mask __m, const char_type* __lo,
const char_type* __hi) const = 0;
/**
* @brief Convert to uppercase.
*
* This virtual function converts the char_type argument to uppercase
* if possible. If not possible (for example, '2'), returns the
* argument.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __c The char_type to convert.
* @return The uppercase char_type if convertible, else @a __c.
*/
virtual char_type
do_toupper(char_type __c) const = 0;
/**
* @brief Convert array to uppercase.
*
* This virtual function converts each char_type in the range [__lo,__hi)
* to uppercase if possible. Other elements remain untouched.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_toupper(char_type* __lo, const char_type* __hi) const = 0;
/**
* @brief Convert to lowercase.
*
* This virtual function converts the argument to lowercase if
* possible. If not possible (for example, '2'), returns the argument.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __c The char_type to convert.
* @return The lowercase char_type if convertible, else @a __c.
*/
virtual char_type
do_tolower(char_type __c) const = 0;
/**
* @brief Convert array to lowercase.
*
* This virtual function converts each char_type in the range [__lo,__hi)
* to lowercase if possible. Other elements remain untouched.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_tolower(char_type* __lo, const char_type* __hi) const = 0;
/**
* @brief Widen char
*
* This virtual function converts the char to char_type using the
* simplest reasonable transformation.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @return The converted char_type
*/
virtual char_type
do_widen(char __c) const = 0;
/**
* @brief Widen char array
*
* This function converts each char in the input to char_type using the
* simplest reasonable transformation.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start range.
* @param __hi Pointer to end of range.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char*
do_widen(const char* __lo, const char* __hi, char_type* __to) const = 0;
/**
* @brief Narrow char_type to char
*
* This virtual function converts the argument to char using the
* simplest reasonable transformation. If the conversion fails, dfault
* is returned instead.
*
* do_narrow() is a hook for a derived facet to change the behavior of
* narrowing. do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char_type to convert.
* @param __dfault Char to return if conversion fails.
* @return The converted char.
*/
virtual char
do_narrow(char_type __c, char __dfault) const = 0;
/**
* @brief Narrow char_type array to char
*
* This virtual function converts each char_type in the range
* [__lo,__hi) to char using the simplest reasonable
* transformation and writes the results to the destination
* array. For any element in the input that cannot be
* converted, @a __dfault is used instead.
*
* do_narrow() is a hook for a derived facet to change the behavior of
* narrowing. do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __dfault Char to use if conversion fails.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char_type*
do_narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const = 0;
};
/**
* @brief Primary class template ctype facet.
* @ingroup locales
*
* This template class defines classification and conversion functions for
* character sets. It wraps cctype functionality. Ctype gets used by
* streams for many I/O operations.
*
* This template provides the protected virtual functions the developer
* will have to replace in a derived class or specialization to make a
* working facet. The public functions that access them are defined in
* __ctype_abstract_base, to allow for implementation flexibility. See
* ctype<wchar_t> for an example. The functions are documented in
* __ctype_abstract_base.
*
* Note: implementations are provided for all the protected virtual
* functions, but will likely not be useful.
*/
template<typename _CharT>
class ctype : public __ctype_abstract_base<_CharT>
{
public:
// Types:
typedef _CharT char_type;
typedef typename __ctype_abstract_base<_CharT>::mask mask;
/// The facet id for ctype<char_type>
static locale::id id;
explicit
ctype(size_t __refs = 0) : __ctype_abstract_base<_CharT>(__refs) { }
protected:
virtual
~ctype();
virtual bool
do_is(mask __m, char_type __c) const;
virtual const char_type*
do_is(const char_type* __lo, const char_type* __hi, mask* __vec) const;
virtual const char_type*
do_scan_is(mask __m, const char_type* __lo, const char_type* __hi) const;
virtual const char_type*
do_scan_not(mask __m, const char_type* __lo,
const char_type* __hi) const;
virtual char_type
do_toupper(char_type __c) const;
virtual const char_type*
do_toupper(char_type* __lo, const char_type* __hi) const;
virtual char_type
do_tolower(char_type __c) const;
virtual const char_type*
do_tolower(char_type* __lo, const char_type* __hi) const;
virtual char_type
do_widen(char __c) const;
virtual const char*
do_widen(const char* __lo, const char* __hi, char_type* __dest) const;
virtual char
do_narrow(char_type, char __dfault) const;
virtual const char_type*
do_narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const;
};
template<typename _CharT>
locale::id ctype<_CharT>::id;
/**
* @brief The ctype<char> specialization.
* @ingroup locales
*
* This class defines classification and conversion functions for
* the char type. It gets used by char streams for many I/O
* operations. The char specialization provides a number of
* optimizations as well.
*/
template<>
class ctype<char> : public locale::facet, public ctype_base
{
public:
// Types:
/// Typedef for the template parameter char.
typedef char char_type;
protected:
// Data Members:
__c_locale _M_c_locale_ctype;
bool _M_del;
__to_type _M_toupper;
__to_type _M_tolower;
const mask* _M_table;
mutable char _M_widen_ok;
mutable char _M_widen[1 + static_cast<unsigned char>(-1)];
mutable char _M_narrow[1 + static_cast<unsigned char>(-1)];
mutable char _M_narrow_ok; // 0 uninitialized, 1 init,
// 2 memcpy can't be used
public:
/// The facet id for ctype<char>
static locale::id id;
/// The size of the mask table. It is SCHAR_MAX + 1.
static const size_t table_size = 1 + static_cast<unsigned char>(-1);
/**
* @brief Constructor performs initialization.
*
* This is the constructor provided by the standard.
*
* @param __table If non-zero, table is used as the per-char mask.
* Else classic_table() is used.
* @param __del If true, passes ownership of table to this facet.
* @param __refs Passed to the base facet class.
*/
explicit
ctype(const mask* __table = 0, bool __del = false, size_t __refs = 0);
/**
* @brief Constructor performs static initialization.
*
* This constructor is used to construct the initial C locale facet.
*
* @param __cloc Handle to C locale data.
* @param __table If non-zero, table is used as the per-char mask.
* @param __del If true, passes ownership of table to this facet.
* @param __refs Passed to the base facet class.
*/
explicit
ctype(__c_locale __cloc, const mask* __table = 0, bool __del = false,
size_t __refs = 0);
/**
* @brief Test char classification.
*
* This function compares the mask table[c] to @a __m.
*
* @param __c The char to compare the mask of.
* @param __m The mask to compare against.
* @return True if __m & table[__c] is true, false otherwise.
*/
inline bool
is(mask __m, char __c) const;
/**
* @brief Return a mask array.
*
* This function finds the mask for each char in the range [lo, hi) and
* successively writes it to vec. vec must have as many elements as
* the char array.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __vec Pointer to an array of mask storage.
* @return @a __hi.
*/
inline const char*
is(const char* __lo, const char* __hi, mask* __vec) const;
/**
* @brief Find char matching a mask
*
* This function searches for and returns the first char in [lo,hi) for
* which is(m,char) is true.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a matching char if found, else @a __hi.
*/
inline const char*
scan_is(mask __m, const char* __lo, const char* __hi) const;
/**
* @brief Find char not matching a mask
*
* This function searches for and returns a pointer to the first char
* in [__lo,__hi) for which is(m,char) is false.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a non-matching char if found, else @a __hi.
*/
inline const char*
scan_not(mask __m, const char* __lo, const char* __hi) const;
/**
* @brief Convert to uppercase.
*
* This function converts the char argument to uppercase if possible.
* If not possible (for example, '2'), returns the argument.
*
* toupper() acts as if it returns ctype<char>::do_toupper(c).
* do_toupper() must always return the same result for the same input.
*
* @param __c The char to convert.
* @return The uppercase char if convertible, else @a __c.
*/
char_type
toupper(char_type __c) const
{ return this->do_toupper(__c); }
/**
* @brief Convert array to uppercase.
*
* This function converts each char in the range [__lo,__hi) to uppercase
* if possible. Other chars remain untouched.
*
* toupper() acts as if it returns ctype<char>:: do_toupper(__lo, __hi).
* do_toupper() must always return the same result for the same input.
*
* @param __lo Pointer to first char in range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
const char_type*
toupper(char_type *__lo, const char_type* __hi) const
{ return this->do_toupper(__lo, __hi); }
/**
* @brief Convert to lowercase.
*
* This function converts the char argument to lowercase if possible.
* If not possible (for example, '2'), returns the argument.
*
* tolower() acts as if it returns ctype<char>::do_tolower(__c).
* do_tolower() must always return the same result for the same input.
*
* @param __c The char to convert.
* @return The lowercase char if convertible, else @a __c.
*/
char_type
tolower(char_type __c) const
{ return this->do_tolower(__c); }
/**
* @brief Convert array to lowercase.
*
* This function converts each char in the range [lo,hi) to lowercase
* if possible. Other chars remain untouched.
*
* tolower() acts as if it returns ctype<char>:: do_tolower(__lo, __hi).
* do_tolower() must always return the same result for the same input.
*
* @param __lo Pointer to first char in range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
const char_type*
tolower(char_type* __lo, const char_type* __hi) const
{ return this->do_tolower(__lo, __hi); }
/**
* @brief Widen char
*
* This function converts the char to char_type using the simplest
* reasonable transformation. For an underived ctype<char> facet, the
* argument will be returned unchanged.
*
* This function works as if it returns ctype<char>::do_widen(c).
* do_widen() must always return the same result for the same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @return The converted character.
*/
char_type
widen(char __c) const
{
if (_M_widen_ok)
return _M_widen[static_cast<unsigned char>(__c)];
this->_M_widen_init();
return this->do_widen(__c);
}
/**
* @brief Widen char array
*
* This function converts each char in the input to char using the
* simplest reasonable transformation. For an underived ctype<char>
* facet, the argument will be copied unchanged.
*
* This function works as if it returns ctype<char>::do_widen(c).
* do_widen() must always return the same result for the same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to first char in range.
* @param __hi Pointer to end of range.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
const char*
widen(const char* __lo, const char* __hi, char_type* __to) const
{
if (_M_widen_ok == 1)
{
__builtin_memcpy(__to, __lo, __hi - __lo);
return __hi;
}
if (!_M_widen_ok)
_M_widen_init();
return this->do_widen(__lo, __hi, __to);
}
/**
* @brief Narrow char
*
* This function converts the char to char using the simplest
* reasonable transformation. If the conversion fails, dfault is
* returned instead. For an underived ctype<char> facet, @a c
* will be returned unchanged.
*
* This function works as if it returns ctype<char>::do_narrow(c).
* do_narrow() must always return the same result for the same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @param __dfault Char to return if conversion fails.
* @return The converted character.
*/
char
narrow(char_type __c, char __dfault) const
{
if (_M_narrow[static_cast<unsigned char>(__c)])
return _M_narrow[static_cast<unsigned char>(__c)];
const char __t = do_narrow(__c, __dfault);
if (__t != __dfault)
_M_narrow[static_cast<unsigned char>(__c)] = __t;
return __t;
}
/**
* @brief Narrow char array
*
* This function converts each char in the input to char using the
* simplest reasonable transformation and writes the results to the
* destination array. For any char in the input that cannot be
* converted, @a dfault is used instead. For an underived ctype<char>
* facet, the argument will be copied unchanged.
*
* This function works as if it returns ctype<char>::do_narrow(lo, hi,
* dfault, to). do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __dfault Char to use if conversion fails.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
const char_type*
narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const
{
if (__builtin_expect(_M_narrow_ok == 1, true))
{
__builtin_memcpy(__to, __lo, __hi - __lo);
return __hi;
}
if (!_M_narrow_ok)
_M_narrow_init();
return this->do_narrow(__lo, __hi, __dfault, __to);
}
// _GLIBCXX_RESOLVE_LIB_DEFECTS
// DR 695. ctype<char>::classic_table() not accessible.
/// Returns a pointer to the mask table provided to the constructor, or
/// the default from classic_table() if none was provided.
const mask*
table() const throw()
{ return _M_table; }
/// Returns a pointer to the C locale mask table.
static const mask*
classic_table() throw();
protected:
/**
* @brief Destructor.
*
* This function deletes table() if @a del was true in the
* constructor.
*/
virtual
~ctype();
/**
* @brief Convert to uppercase.
*
* This virtual function converts the char argument to uppercase if
* possible. If not possible (for example, '2'), returns the argument.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __c The char to convert.
* @return The uppercase char if convertible, else @a __c.
*/
virtual char_type
do_toupper(char_type __c) const;
/**
* @brief Convert array to uppercase.
*
* This virtual function converts each char in the range [lo,hi) to
* uppercase if possible. Other chars remain untouched.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_toupper(char_type* __lo, const char_type* __hi) const;
/**
* @brief Convert to lowercase.
*
* This virtual function converts the char argument to lowercase if
* possible. If not possible (for example, '2'), returns the argument.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __c The char to convert.
* @return The lowercase char if convertible, else @a __c.
*/
virtual char_type
do_tolower(char_type __c) const;
/**
* @brief Convert array to lowercase.
*
* This virtual function converts each char in the range [lo,hi) to
* lowercase if possible. Other chars remain untouched.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __lo Pointer to first char in range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_tolower(char_type* __lo, const char_type* __hi) const;
/**
* @brief Widen char
*
* This virtual function converts the char to char using the simplest
* reasonable transformation. For an underived ctype<char> facet, the
* argument will be returned unchanged.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @return The converted character.
*/
virtual char_type
do_widen(char __c) const
{ return __c; }
/**
* @brief Widen char array
*
* This function converts each char in the range [lo,hi) to char using
* the simplest reasonable transformation. For an underived
* ctype<char> facet, the argument will be copied unchanged.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char*
do_widen(const char* __lo, const char* __hi, char_type* __to) const
{
__builtin_memcpy(__to, __lo, __hi - __lo);
return __hi;
}
/**
* @brief Narrow char
*
* This virtual function converts the char to char using the simplest
* reasonable transformation. If the conversion fails, dfault is
* returned instead. For an underived ctype<char> facet, @a c will be
* returned unchanged.
*
* do_narrow() is a hook for a derived facet to change the behavior of
* narrowing. do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @param __dfault Char to return if conversion fails.
* @return The converted char.
*/
virtual char
do_narrow(char_type __c, char __dfault) const
{ return __c; }
/**
* @brief Narrow char array to char array
*
* This virtual function converts each char in the range [lo,hi) to
* char using the simplest reasonable transformation and writes the
* results to the destination array. For any char in the input that
* cannot be converted, @a dfault is used instead. For an underived
* ctype<char> facet, the argument will be copied unchanged.
*
* do_narrow() is a hook for a derived facet to change the behavior of
* narrowing. do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __dfault Char to use if conversion fails.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char_type*
do_narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const
{
__builtin_memcpy(__to, __lo, __hi - __lo);
return __hi;
}
private:
void _M_narrow_init() const;
void _M_widen_init() const;
};
#ifdef _GLIBCXX_USE_WCHAR_T
/**
* @brief The ctype<wchar_t> specialization.
* @ingroup locales
*
* This class defines classification and conversion functions for the
* wchar_t type. It gets used by wchar_t streams for many I/O operations.
* The wchar_t specialization provides a number of optimizations as well.
*
* ctype<wchar_t> inherits its public methods from
* __ctype_abstract_base<wchar_t>.
*/
template<>
class ctype<wchar_t> : public __ctype_abstract_base<wchar_t>
{
public:
// Types:
/// Typedef for the template parameter wchar_t.
typedef wchar_t char_type;
typedef wctype_t __wmask_type;
protected:
__c_locale _M_c_locale_ctype;
// Pre-computed narrowed and widened chars.
bool _M_narrow_ok;
char _M_narrow[128];
wint_t _M_widen[1 + static_cast<unsigned char>(-1)];
// Pre-computed elements for do_is.
mask _M_bit[16];
__wmask_type _M_wmask[16];
public:
// Data Members:
/// The facet id for ctype<wchar_t>
static locale::id id;
/**
* @brief Constructor performs initialization.
*
* This is the constructor provided by the standard.
*
* @param __refs Passed to the base facet class.
*/
explicit
ctype(size_t __refs = 0);
/**
* @brief Constructor performs static initialization.
*
* This constructor is used to construct the initial C locale facet.
*
* @param __cloc Handle to C locale data.
* @param __refs Passed to the base facet class.
*/
explicit
ctype(__c_locale __cloc, size_t __refs = 0);
protected:
__wmask_type
_M_convert_to_wmask(const mask __m) const throw();
/// Destructor
virtual
~ctype();
/**
* @brief Test wchar_t classification.
*
* This function finds a mask M for @a c and compares it to mask @a m.
*
* do_is() is a hook for a derived facet to change the behavior of
* classifying. do_is() must always return the same result for the
* same input.
*
* @param __c The wchar_t to find the mask of.
* @param __m The mask to compare against.
* @return (M & __m) != 0.
*/
virtual bool
do_is(mask __m, char_type __c) const;
/**
* @brief Return a mask array.
*
* This function finds the mask for each wchar_t in the range [lo,hi)
* and successively writes it to vec. vec must have as many elements
* as the input.
*
* do_is() is a hook for a derived facet to change the behavior of
* classifying. do_is() must always return the same result for the
* same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __vec Pointer to an array of mask storage.
* @return @a __hi.
*/
virtual const char_type*
do_is(const char_type* __lo, const char_type* __hi, mask* __vec) const;
/**
* @brief Find wchar_t matching mask
*
* This function searches for and returns the first wchar_t c in
* [__lo,__hi) for which is(__m,c) is true.
*
* do_scan_is() is a hook for a derived facet to change the behavior of
* match searching. do_is() must always return the same result for the
* same input.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a matching wchar_t if found, else @a __hi.
*/
virtual const char_type*
do_scan_is(mask __m, const char_type* __lo, const char_type* __hi) const;
/**
* @brief Find wchar_t not matching mask
*
* This function searches for and returns a pointer to the first
* wchar_t c of [__lo,__hi) for which is(__m,c) is false.
*
* do_scan_is() is a hook for a derived facet to change the behavior of
* match searching. do_is() must always return the same result for the
* same input.
*
* @param __m The mask to compare against.
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return Pointer to a non-matching wchar_t if found, else @a __hi.
*/
virtual const char_type*
do_scan_not(mask __m, const char_type* __lo,
const char_type* __hi) const;
/**
* @brief Convert to uppercase.
*
* This virtual function converts the wchar_t argument to uppercase if
* possible. If not possible (for example, '2'), returns the argument.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __c The wchar_t to convert.
* @return The uppercase wchar_t if convertible, else @a __c.
*/
virtual char_type
do_toupper(char_type __c) const;
/**
* @brief Convert array to uppercase.
*
* This virtual function converts each wchar_t in the range [lo,hi) to
* uppercase if possible. Other elements remain untouched.
*
* do_toupper() is a hook for a derived facet to change the behavior of
* uppercasing. do_toupper() must always return the same result for
* the same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_toupper(char_type* __lo, const char_type* __hi) const;
/**
* @brief Convert to lowercase.
*
* This virtual function converts the argument to lowercase if
* possible. If not possible (for example, '2'), returns the argument.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __c The wchar_t to convert.
* @return The lowercase wchar_t if convertible, else @a __c.
*/
virtual char_type
do_tolower(char_type __c) const;
/**
* @brief Convert array to lowercase.
*
* This virtual function converts each wchar_t in the range [lo,hi) to
* lowercase if possible. Other elements remain untouched.
*
* do_tolower() is a hook for a derived facet to change the behavior of
* lowercasing. do_tolower() must always return the same result for
* the same input.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @return @a __hi.
*/
virtual const char_type*
do_tolower(char_type* __lo, const char_type* __hi) const;
/**
* @brief Widen char to wchar_t
*
* This virtual function converts the char to wchar_t using the
* simplest reasonable transformation. For an underived ctype<wchar_t>
* facet, the argument will be cast to wchar_t.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The char to convert.
* @return The converted wchar_t.
*/
virtual char_type
do_widen(char __c) const;
/**
* @brief Widen char array to wchar_t array
*
* This function converts each char in the input to wchar_t using the
* simplest reasonable transformation. For an underived ctype<wchar_t>
* facet, the argument will be copied, casting each element to wchar_t.
*
* do_widen() is a hook for a derived facet to change the behavior of
* widening. do_widen() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start range.
* @param __hi Pointer to end of range.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char*
do_widen(const char* __lo, const char* __hi, char_type* __to) const;
/**
* @brief Narrow wchar_t to char
*
* This virtual function converts the argument to char using
* the simplest reasonable transformation. If the conversion
* fails, dfault is returned instead. For an underived
* ctype<wchar_t> facet, @a c will be cast to char and
* returned.
*
* do_narrow() is a hook for a derived facet to change the
* behavior of narrowing. do_narrow() must always return the
* same result for the same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __c The wchar_t to convert.
* @param __dfault Char to return if conversion fails.
* @return The converted char.
*/
virtual char
do_narrow(char_type __c, char __dfault) const;
/**
* @brief Narrow wchar_t array to char array
*
* This virtual function converts each wchar_t in the range [lo,hi) to
* char using the simplest reasonable transformation and writes the
* results to the destination array. For any wchar_t in the input that
* cannot be converted, @a dfault is used instead. For an underived
* ctype<wchar_t> facet, the argument will be copied, casting each
* element to char.
*
* do_narrow() is a hook for a derived facet to change the behavior of
* narrowing. do_narrow() must always return the same result for the
* same input.
*
* Note: this is not what you want for codepage conversions. See
* codecvt for that.
*
* @param __lo Pointer to start of range.
* @param __hi Pointer to end of range.
* @param __dfault Char to use if conversion fails.
* @param __to Pointer to the destination array.
* @return @a __hi.
*/
virtual const char_type*
do_narrow(const char_type* __lo, const char_type* __hi,
char __dfault, char* __to) const;
// For use at construction time only.
void
_M_initialize_ctype() throw();
};
#endif //_GLIBCXX_USE_WCHAR_T
/// class ctype_byname [22.2.1.2].
template<typename _CharT>
class ctype_byname : public ctype<_CharT>
{
public:
typedef typename ctype<_CharT>::mask mask;
explicit
ctype_byname(const char* __s, size_t __refs = 0);
protected:
virtual
~ctype_byname() { };
};
/// 22.2.1.4 Class ctype_byname specializations.
template<>
class ctype_byname<char> : public ctype<char>
{
public:
explicit
ctype_byname(const char* __s, size_t __refs = 0);
protected:
virtual
~ctype_byname();
};
#ifdef _GLIBCXX_USE_WCHAR_T
template<>
class ctype_byname<wchar_t> : public ctype<wchar_t>
{
public:
explicit
ctype_byname(const char* __s, size_t __refs = 0);
protected:
virtual
~ctype_byname();
};
#endif
_GLIBCXX_END_NAMESPACE_VERSION
} // namespace
// Include host and configuration specific ctype inlines.
#include <bits/ctype_inline.h>
namespace std _GLIBCXX_VISIBILITY(default)
{
_GLIBCXX_BEGIN_NAMESPACE_VERSION
// 22.2.2 The numeric category.
class __num_base
{
public:
// NB: Code depends on the order of _S_atoms_out elements.
// Below are the indices into _S_atoms_out.
enum
{
_S_ominus,
_S_oplus,
_S_ox,
_S_oX,
_S_odigits,
_S_odigits_end = _S_odigits + 16,
_S_oudigits = _S_odigits_end,
_S_oudigits_end = _S_oudigits + 16,
_S_oe = _S_odigits + 14, // For scientific notation, 'e'
_S_oE = _S_oudigits + 14, // For scientific notation, 'E'
_S_oend = _S_oudigits_end
};
// A list of valid numeric literals for output. This array
// contains chars that will be passed through the current locale's
// ctype<_CharT>.widen() and then used to render numbers.
// For the standard "C" locale, this is
// "-+xX0123456789abcdef0123456789ABCDEF".
static const char* _S_atoms_out;
// String literal of acceptable (narrow) input, for num_get.
// "-+xX0123456789abcdefABCDEF"
static const char* _S_atoms_in;
enum
{
_S_iminus,
_S_iplus,
_S_ix,
_S_iX,
_S_izero,
_S_ie = _S_izero + 14,
_S_iE = _S_izero + 20,
_S_iend = 26
};
// num_put
// Construct and return valid scanf format for floating point types.
static void
_S_format_float(const ios_base& __io, char* __fptr, char __mod) throw();
};
template<typename _CharT>
struct __numpunct_cache : public locale::facet
{
const char* _M_grouping;
size_t _M_grouping_size;
bool _M_use_grouping;
const _CharT* _M_truename;
size_t _M_truename_size;
const _CharT* _M_falsename;
size_t _M_falsename_size;
_CharT _M_decimal_point;
_CharT _M_thousands_sep;
// A list of valid numeric literals for output: in the standard
// "C" locale, this is "-+xX0123456789abcdef0123456789ABCDEF".
// This array contains the chars after having been passed
// through the current locale's ctype<_CharT>.widen().
_CharT _M_atoms_out[__num_base::_S_oend];
// A list of valid numeric literals for input: in the standard
// "C" locale, this is "-+xX0123456789abcdefABCDEF"
// This array contains the chars after having been passed
// through the current locale's ctype<_CharT>.widen().
_CharT _M_atoms_in[__num_base::_S_iend];
bool _M_allocated;
__numpunct_cache(size_t __refs = 0)
: facet(__refs), _M_grouping(0), _M_grouping_size(0),
_M_use_grouping(false),
_M_truename(0), _M_truename_size(0), _M_falsename(0),
_M_falsename_size(0), _M_decimal_point(_CharT()),
_M_thousands_sep(_CharT()), _M_allocated(false)
{ }
~__numpunct_cache();
void
_M_cache(const locale& __loc);
private:
__numpunct_cache&
operator=(const __numpunct_cache&);
explicit
__numpunct_cache(const __numpunct_cache&);
};
template<typename _CharT>
__numpunct_cache<_CharT>::~__numpunct_cache()
{
if (_M_allocated)
{
delete [] _M_grouping;
delete [] _M_truename;
delete [] _M_falsename;
}
}
/**
* @brief Primary class template numpunct.
* @ingroup locales
*
* This facet stores several pieces of information related to printing and
* scanning numbers, such as the decimal point character. It takes a
* template parameter specifying the char type. The numpunct facet is
* used by streams for many I/O operations involving numbers.
*
* The numpunct template uses protected virtual functions to provide the
* actual results. The public accessors forward the call to the virtual
* functions. These virtual functions are hooks for developers to
* implement the behavior they require from a numpunct facet.
*/
template<typename _CharT>
class numpunct : public locale::facet
{
public:
// Types:
//@{
/// Public typedefs
typedef _CharT char_type;
typedef basic_string<_CharT> string_type;
//@}
typedef __numpunct_cache<_CharT> __cache_type;
protected:
__cache_type* _M_data;
public:
/// Numpunct facet id.
static locale::id id;
/**
* @brief Numpunct constructor.
*
* @param __refs Refcount to pass to the base class.
*/
explicit
numpunct(size_t __refs = 0)
: facet(__refs), _M_data(0)
{ _M_initialize_numpunct(); }
/**
* @brief Internal constructor. Not for general use.
*
* This is a constructor for use by the library itself to set up the
* predefined locale facets.
*
* @param __cache __numpunct_cache object.
* @param __refs Refcount to pass to the base class.
*/
explicit
numpunct(__cache_type* __cache, size_t __refs = 0)
: facet(__refs), _M_data(__cache)
{ _M_initialize_numpunct(); }
/**
* @brief Internal constructor. Not for general use.
*
* This is a constructor for use by the library itself to set up new
* locales.
*
* @param __cloc The C locale.
* @param __refs Refcount to pass to the base class.
*/
explicit
numpunct(__c_locale __cloc, size_t __refs = 0)
: facet(__refs), _M_data(0)
{ _M_initialize_numpunct(__cloc); }
/**
* @brief Return decimal point character.
*
* This function returns a char_type to use as a decimal point. It
* does so by returning returning
* numpunct<char_type>::do_decimal_point().
*
* @return @a char_type representing a decimal point.
*/
char_type
decimal_point() const
{ return this->do_decimal_point(); }
/**
* @brief Return thousands separator character.
*
* This function returns a char_type to use as a thousands
* separator. It does so by returning returning
* numpunct<char_type>::do_thousands_sep().
*
* @return char_type representing a thousands separator.
*/
char_type
thousands_sep() const
{ return this->do_thousands_sep(); }
/**
* @brief Return grouping specification.
*
* This function returns a string representing groupings for the
* integer part of a number. Groupings indicate where thousands
* separators should be inserted in the integer part of a number.
*
* Each char in the return string is interpret as an integer
* rather than a character. These numbers represent the number
* of digits in a group. The first char in the string
* represents the number of digits in the least significant
* group. If a char is negative, it indicates an unlimited
* number of digits for the group. If more chars from the
* string are required to group a number, the last char is used
* repeatedly.
*
* For example, if the grouping() returns "\003\002" and is
* applied to the number 123456789, this corresponds to
* 12,34,56,789. Note that if the string was "32", this would
* put more than 50 digits into the least significant group if
* the character set is ASCII.
*
* The string is returned by calling
* numpunct<char_type>::do_grouping().
*
* @return string representing grouping specification.
*/
string
grouping() const
{ return this->do_grouping(); }
/**
* @brief Return string representation of bool true.
*
* This function returns a string_type containing the text
* representation for true bool variables. It does so by calling
* numpunct<char_type>::do_truename().
*
* @return string_type representing printed form of true.
*/
string_type
truename() const
{ return this->do_truename(); }
/**
* @brief Return string representation of bool false.
*
* This function returns a string_type containing the text
* representation for false bool variables. It does so by calling
* numpunct<char_type>::do_falsename().
*
* @return string_type representing printed form of false.
*/
string_type
falsename() const
{ return this->do_falsename(); }
protected:
/// Destructor.
virtual
~numpunct();
/**
* @brief Return decimal point character.
*
* Returns a char_type to use as a decimal point. This function is a
* hook for derived classes to change the value returned.
*
* @return @a char_type representing a decimal point.
*/
virtual char_type
do_decimal_point() const
{ return _M_data->_M_decimal_point; }
/**
* @brief Return thousands separator character.
*
* Returns a char_type to use as a thousands separator. This function
* is a hook for derived classes to change the value returned.
*
* @return @a char_type representing a thousands separator.
*/
virtual char_type
do_thousands_sep() const
{ return _M_data->_M_thousands_sep; }
/**
* @brief Return grouping specification.
*
* Returns a string representing groupings for the integer part of a
* number. This function is a hook for derived classes to change the
* value returned. @see grouping() for details.
*
* @return String representing grouping specification.
*/
virtual string
do_grouping() const
{ return _M_data->_M_grouping; }
/**
* @brief Return string representation of bool true.
*
* Returns a string_type containing the text representation for true
* bool variables. This function is a hook for derived classes to
* change the value returned.
*
* @return string_type representing printed form of true.
*/
virtual string_type
do_truename() const
{ return _M_data->_M_truename; }
/**
* @brief Return string representation of bool false.
*
* Returns a string_type containing the text representation for false
* bool variables. This function is a hook for derived classes to
* change the value returned.
*
* @return string_type representing printed form of false.
*/
virtual string_type
do_falsename() const
{ return _M_data->_M_falsename; }
// For use at construction time only.
void
_M_initialize_numpunct(__c_locale __cloc = 0);
};
template<typename _CharT>
locale::id numpunct<_CharT>::id;
template<>
numpunct<char>::~numpunct();
template<>
void
numpunct<char>::_M_initialize_numpunct(__c_locale __cloc);
#ifdef _GLIBCXX_USE_WCHAR_T
template<>
numpunct<wchar_t>::~numpunct();
template<>
void
numpunct<wchar_t>::_M_initialize_numpunct(__c_locale __cloc);
#endif
/// class numpunct_byname [22.2.3.2].
template<typename _CharT>
class numpunct_byname : public numpunct<_CharT>
{
public:
typedef _CharT char_type;
typedef basic_string<_CharT> string_type;
explicit
numpunct_byname(const char* __s, size_t __refs = 0)
: numpunct<_CharT>(__refs)
{
if (__builtin_strcmp(__s, "C") != 0
&& __builtin_strcmp(__s, "POSIX") != 0)
{
__c_locale __tmp;
this->_S_create_c_locale(__tmp, __s);
this->_M_initialize_numpunct(__tmp);
this->_S_destroy_c_locale(__tmp);
}
}
protected:
virtual
~numpunct_byname() { }
};
_GLIBCXX_BEGIN_NAMESPACE_LDBL
/**
* @brief Primary class template num_get.
* @ingroup locales
*
* This facet encapsulates the code to parse and return a number
* from a string. It is used by the istream numeric extraction
* operators.
*
* The num_get template uses protected virtual functions to provide the
* actual results. The public accessors forward the call to the virtual
* functions. These virtual functions are hooks for developers to
* implement the behavior they require from the num_get facet.
*/
template<typename _CharT, typename _InIter>
class num_get : public locale::facet
{
public:
// Types:
//@{
/// Public typedefs
typedef _CharT char_type;
typedef _InIter iter_type;
//@}
/// Numpunct facet id.
static locale::id id;
/**
* @brief Constructor performs initialization.
*
* This is the constructor provided by the standard.
*
* @param __refs Passed to the base facet class.
*/
explicit
num_get(size_t __refs = 0) : facet(__refs) { }
/**
* @brief Numeric parsing.
*
* Parses the input stream into the bool @a v. It does so by calling
* num_get::do_get().
*
* If ios_base::boolalpha is set, attempts to read
* ctype<CharT>::truename() or ctype<CharT>::falsename(). Sets
* @a v to true or false if successful. Sets err to
* ios_base::failbit if reading the string fails. Sets err to
* ios_base::eofbit if the stream is emptied.
*
* If ios_base::boolalpha is not set, proceeds as with reading a long,
* except if the value is 1, sets @a v to true, if the value is 0, sets
* @a v to false, and otherwise set err to ios_base::failbit.
*
* @param __in Start of input stream.
* @param __end End of input stream.
* @param __io Source of locale and flags.
* @param __err Error flags to set.
* @param __v Value to format and insert.
* @return Iterator after reading.
*/
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, bool& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
//@{
/**
* @brief Numeric parsing.
*
* Parses the input stream into the integral variable @a v. It does so
* by calling num_get::do_get().
*
* Parsing is affected by the flag settings in @a io.
*
* The basic parse is affected by the value of io.flags() &
* ios_base::basefield. If equal to ios_base::oct, parses like the
* scanf %o specifier. Else if equal to ios_base::hex, parses like %X
* specifier. Else if basefield equal to 0, parses like the %i
* specifier. Otherwise, parses like %d for signed and %u for unsigned
* types. The matching type length modifier is also used.
*
* Digit grouping is interpreted according to
* numpunct::grouping() and numpunct::thousands_sep(). If the
* pattern of digit groups isn't consistent, sets err to
* ios_base::failbit.
*
* If parsing the string yields a valid value for @a v, @a v is set.
* Otherwise, sets err to ios_base::failbit and leaves @a v unaltered.
* Sets err to ios_base::eofbit if the stream is emptied.
*
* @param __in Start of input stream.
* @param __end End of input stream.
* @param __io Source of locale and flags.
* @param __err Error flags to set.
* @param __v Value to format and insert.
* @return Iterator after reading.
*/
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, long& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned short& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned int& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned long& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
#ifdef _GLIBCXX_USE_LONG_LONG
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, long long& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned long long& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
#endif
//@}
//@{
/**
* @brief Numeric parsing.
*
* Parses the input stream into the integral variable @a v. It does so
* by calling num_get::do_get().
*
* The input characters are parsed like the scanf %g specifier. The
* matching type length modifier is also used.
*
* The decimal point character used is numpunct::decimal_point().
* Digit grouping is interpreted according to
* numpunct::grouping() and numpunct::thousands_sep(). If the
* pattern of digit groups isn't consistent, sets err to
* ios_base::failbit.
*
* If parsing the string yields a valid value for @a v, @a v is set.
* Otherwise, sets err to ios_base::failbit and leaves @a v unaltered.
* Sets err to ios_base::eofbit if the stream is emptied.
*
* @param __in Start of input stream.
* @param __end End of input stream.
* @param __io Source of locale and flags.
* @param __err Error flags to set.
* @param __v Value to format and insert.
* @return Iterator after reading.
*/
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, float& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, double& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, long double& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
//@}
/**
* @brief Numeric parsing.
*
* Parses the input stream into the pointer variable @a v. It does so
* by calling num_get::do_get().
*
* The input characters are parsed like the scanf %p specifier.
*
* Digit grouping is interpreted according to
* numpunct::grouping() and numpunct::thousands_sep(). If the
* pattern of digit groups isn't consistent, sets err to
* ios_base::failbit.
*
* Note that the digit grouping effect for pointers is a bit ambiguous
* in the standard and shouldn't be relied on. See DR 344.
*
* If parsing the string yields a valid value for @a v, @a v is set.
* Otherwise, sets err to ios_base::failbit and leaves @a v unaltered.
* Sets err to ios_base::eofbit if the stream is emptied.
*
* @param __in Start of input stream.
* @param __end End of input stream.
* @param __io Source of locale and flags.
* @param __err Error flags to set.
* @param __v Value to format and insert.
* @return Iterator after reading.
*/
iter_type
get(iter_type __in, iter_type __end, ios_base& __io,
ios_base::iostate& __err, void*& __v) const
{ return this->do_get(__in, __end, __io, __err, __v); }
protected:
/// Destructor.
virtual ~num_get() { }
iter_type
_M_extract_float(iter_type, iter_type, ios_base&, ios_base::iostate&,
string&) const;
template<typename _ValueT>
iter_type
_M_extract_int(iter_type, iter_type, ios_base&, ios_base::iostate&,
_ValueT&) const;
template<typename _CharT2>
typename __gnu_cxx::__enable_if<__is_char<_CharT2>::__value, int>::__type
_M_find(const _CharT2*, size_t __len, _CharT2 __c) const
{
int __ret = -1;
if (__len <= 10)
{
if (__c >= _CharT2('0') && __c < _CharT2(_CharT2('0') + __len))
__ret = __c - _CharT2('0');
}
else
{
if (__c >= _CharT2('0') && __c <= _CharT2('9'))
__ret = __c - _CharT2('0');
else if (__c >= _CharT2('a') && __c <= _CharT2('f'))
__ret = 10 + (__c - _CharT2('a'));
else if (__c >= _CharT2('A') && __c <= _CharT2('F'))
__ret = 10 + (__c - _CharT2('A'));
}
return __ret;
}
template<typename _CharT2>
typename __gnu_cxx::__enable_if<!__is_char<_CharT2>::__value,
int>::__type
_M_find(const _CharT2* __zero, size_t __len, _CharT2 __c) const
{
int __ret = -1;
const char_type* __q = char_traits<_CharT2>::find(__zero, __len, __c);
if (__q)
{
__ret = __q - __zero;
if (__ret > 15)
__ret -= 6;
}
return __ret;
}
//@{
/**
* @brief Numeric parsing.
*
* Parses the input stream into the variable @a v. This function is a
* hook for derived classes to change the value returned. @see get()
* for more details.
*
* @param __beg Start of input stream.
* @param __end End of input stream.
* @param __io Source of locale and flags.
* @param __err Error flags to set.
* @param __v Value to format and insert.
* @return Iterator after reading.
*/
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&, bool&) const;
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, long& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned short& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned int& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned long& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
#ifdef _GLIBCXX_USE_LONG_LONG
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, long long& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
virtual iter_type
do_get(iter_type __beg, iter_type __end, ios_base& __io,
ios_base::iostate& __err, unsigned long long& __v) const
{ return _M_extract_int(__beg, __end, __io, __err, __v); }
#endif
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&, float&) const;
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&,
double&) const;
// XXX GLIBCXX_ABI Deprecated
#if defined _GLIBCXX_LONG_DOUBLE_COMPAT && defined __LONG_DOUBLE_128__
virtual iter_type
__do_get(iter_type, iter_type, ios_base&, ios_base::iostate&,
double&) const;
#else
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&,
long double&) const;
#endif
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&, void*&) const;
// XXX GLIBCXX_ABI Deprecated
#if defined _GLIBCXX_LONG_DOUBLE_COMPAT && defined __LONG_DOUBLE_128__
virtual iter_type
do_get(iter_type, iter_type, ios_base&, ios_base::iostate&,
long double&) const;
#endif
//@}
};
template<typename _CharT, typename _InIter>
locale::id num_get<_CharT, _InIter>::id;
/**
* @brief Primary class template num_put.
* @ingroup locales
*
* This facet encapsulates the code to convert a number to a string. It is
* used by the ostream numeric insertion operators.
*
* The num_put template uses protected virtual functions to provide the
* actual results. The public accessors forward the call to the virtual
* functions. These virtual functions are hooks for developers to
* implement the behavior they require from the num_put facet.
*/
template<typename _CharT, typename _OutIter>
class num_put : public locale::facet
{
public:
// Types:
//@{
/// Public typedefs
typedef _CharT char_type;
typedef _OutIter iter_type;
//@}
/// Numpunct facet id.
static locale::id id;
/**
* @brief Constructor performs initialization.
*
* This is the constructor provided by the standard.
*
* @param __refs Passed to the base facet class.
*/
explicit
num_put(size_t __refs = 0) : facet(__refs) { }
/**
* @brief Numeric formatting.
*
* Formats the boolean @a v and inserts it into a stream. It does so
* by calling num_put::do_put().
*
* If ios_base::boolalpha is set, writes ctype<CharT>::truename() or
* ctype<CharT>::falsename(). Otherwise formats @a v as an int.
*
* @param __s Stream to write to.
* @param __io Source of locale and flags.
* @param __fill Char_type to use for filling.
* @param __v Value to format and insert.
* @return Iterator after writing.
*/
iter_type
put(iter_type __s, ios_base& __io, char_type __fill, bool __v) const
{ return this->do_put(__s, __io, __fill, __v); }
//@{
/**
* @brief Numeric formatting.
*
* Formats the integral value @a v and inserts it into a
* stream. It does so by calling num_put::do_put().
*
* Formatting is affected by the flag settings in @a io.
*
* The basic format is affected by the value of io.flags() &
* ios_base::basefield. If equal to ios_base::oct, formats like the
* printf %o specifier. Else if equal to ios_base::hex, formats like
* %x or %X with ios_base::uppercase unset or set respectively.
* Otherwise, formats like %d, %ld, %lld for signed and %u, %lu, %llu
* for unsigned values. Note that if both oct and hex are set, neither
* will take effect.
*
* If ios_base::showpos is set, '+' is output before positive values.
* If ios_base::showbase is set, '0' precedes octal values (except 0)
* and '0[xX]' precedes hex values.
*
* The decimal point character used is numpunct::decimal_point().
* Thousands separators are inserted according to
* numpunct::grouping() and numpunct::thousands_sep().
*
* If io.width() is non-zero, enough @a fill characters are inserted to
* make the result at least that wide. If
* (io.flags() & ios_base::adjustfield) == ios_base::left, result is
* padded at the end. If ios_base::internal, then padding occurs
* immediately after either a '+' or '-' or after '0x' or '0X'.
* Otherwise, padding occurs at the beginning.
*
* @param __s Stream to write to.
* @param __io Source of locale and flags.
* @param __fill Char_type to use for filling.
* @param __v Value to format and insert.
* @return Iterator after writing.
*/
iter_type
put(iter_type __s, ios_base& __io, char_type __fill, long __v) const
{ return this->do_put(__s, __io, __fill, __v); }
iter_type
put(iter_type __s, ios_base& __io, char_type __fill,
unsigned long __v) const
{ return this->do_put(__s, __io, __fill, __v); }
#ifdef _GLIBCXX_USE_LONG_LONG
iter_type
put(iter_type __s, ios_base& __io, char_type __fill, long long __v) const
{ return this->do_put(__s, __io, __fill, __v); }
iter_type
put(iter_type __s, ios_base& __io, char_type __fill,
unsigned long long __v) const
{ return this->do_put(__s, __io, __fill, __v); }
#endif
//@}
//@{
/**
* @brief Numeric formatting.
*
* Formats the floating point value @a v and inserts it into a stream.
* It does so by calling num_put::do_put().
*
* Formatting is affected by the flag settings in @a io.
*
* The basic format is affected by the value of io.flags() &
* ios_base::floatfield. If equal to ios_base::fixed, formats like the
* printf %f specifier. Else if equal to ios_base::scientific, formats
* like %e or %E with ios_base::uppercase unset or set respectively.
* Otherwise, formats like %g or %G depending on uppercase. Note that
* if both fixed and scientific are set, the effect will also be like
* %g or %G.
*
* The output precision is given by io.precision(). This precision is
* capped at numeric_limits::digits10 + 2 (different for double and
* long double). The default precision is 6.
*
* If ios_base::showpos is set, '+' is output before positive values.
* If ios_base::showpoint is set, a decimal point will always be
* output.
*
* The decimal point character used is numpunct::decimal_point().
* Thousands separators are inserted according to
* numpunct::grouping() and numpunct::thousands_sep().
*
* If io.width() is non-zero, enough @a fill characters are inserted to
* make the result at least that wide. If
* (io.flags() & ios_base::adjustfield) == ios_base::left, result is
* padded at the end. If ios_base::internal, then padding occurs
* immediately after either a '+' or '-' or after '0x' or '0X'.
* Otherwise, padding occurs at the beginning.
*
* @param __s Stream to write to.
* @param __io Source of locale and flags.
* @param __fill Char_type to use for filling.
* @param __v Value to format and insert.
* @return Iterator after writing.
*/
iter_type
put(iter_type __s, ios_base& __io, char_type __fill, double __v) const
{ return this->do_put(__s, __io, __fill, __v); }
iter_type
put(iter_type __s, ios_base& __io, char_type __fill,
long double __v) const
{ return this->do_put(__s, __io, __fill, __v); }
//@}
/**
* @brief Numeric formatting.
*
* Formats the pointer value @a v and inserts it into a stream. It
* does so by calling num_put::do_put().
*
* This function formats @a v as an unsigned long with ios_base::hex
* and ios_base::showbase set.
*
* @param __s Stream to write to.
* @param __io Source of locale and flags.
* @param __fill Char_type to use for filling.
* @param __v Value to format and insert.
* @return Iterator after writing.
*/
iter_type
put(iter_type __s, ios_base& __io, char_type __fill,
const void* __v) const
{ return this->do_put(__s, __io, __fill, __v); }
protected:
template<typename _ValueT>
iter_type
_M_insert_float(iter_type, ios_base& __io, char_type __fill,
char __mod, _ValueT __v) const;
void
_M_group_float(const char* __grouping, size_t __grouping_size,
char_type __sep, const char_type* __p, char_type* __new,
char_type* __cs, int& __len) const;
template<typename _ValueT>
iter_type
_M_insert_int(iter_type, ios_base& __io, char_type __fill,
_ValueT __v) const;
void
_M_group_int(const char* __grouping, size_t __grouping_size,
char_type __sep, ios_base& __io, char_type* __new,
char_type* __cs, int& __len) const;
void
_M_pad(char_type __fill, streamsize __w, ios_base& __io,
char_type* __new, const char_type* __cs, int& __len) const;
/// Destructor.
virtual
~num_put() { };
//@{
/**
* @brief Numeric formatting.
*
* These functions do the work of formatting numeric values and
* inserting them into a stream. This function is a hook for derived
* classes to change the value returned.
*
* @param __s Stream to write to.
* @param __io Source of locale and flags.
* @param __fill Char_type to use for filling.
* @param __v Value to format and insert.
* @return Iterator after writing.
*/
virtual iter_type
do_put(iter_type __s, ios_base& __io, char_type __fill, bool __v) const;
virtual iter_type
do_put(iter_type __s, ios_base& __io, char_type __fill, long __v) const
{ return _M_insert_int(__s, __io, __fill, __v); }
virtual iter_type
do_put(iter_type __s, ios_base& __io, char_type __fill,
unsigned long __v) const
{ return _M_insert_int(__s, __io, __fill, __v); }
#ifdef _GLIBCXX_USE_LONG_LONG
virtual iter_type
do_put(iter_type __s, ios_base& __io, char_type __fill,
long long __v) const
{ return _M_insert_int(__s, __io, __fill, __v); }
virtual iter_type
do_put(iter_type __s, ios_base& __io, char_type __fill,
unsigned long long __v) const
{ return _M_insert_int(__s, __io, __fill, __v); }
#endif
virtual iter_type
do_put(iter_type, ios_base&, char_type, double) const;
// XXX GLIBCXX_ABI Deprecated
#if defined _GLIBCXX_LONG_DOUBLE_COMPAT && defined __LONG_DOUBLE_128__
virtual iter_type
__do_put(iter_type, ios_base&, char_type, double) const;
#else
virtual iter_type
do_put(iter_type, ios_base&, char_type, long double) const;
#endif
virtual iter_type
do_put(iter_type, ios_base&, char_type, const void*) const;
// XXX GLIBCXX_ABI Deprecated
#if defined _GLIBCXX_LONG_DOUBLE_COMPAT && defined __LONG_DOUBLE_128__
virtual iter_type
do_put(iter_type, ios_base&, char_type, long double) const;
#endif
//@}
};
template <typename _CharT, typename _OutIter>
locale::id num_put<_CharT, _OutIter>::id;
_GLIBCXX_END_NAMESPACE_LDBL
// Subclause convenience interfaces, inlines.
// NB: These are inline because, when used in a loop, some compilers
// can hoist the body out of the loop; then it's just as fast as the
// C is*() function.
/// Convenience interface to ctype.is(ctype_base::space, __c).
template<typename _CharT>
inline bool
isspace(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::space, __c); }
/// Convenience interface to ctype.is(ctype_base::print, __c).
template<typename _CharT>
inline bool
isprint(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::print, __c); }
/// Convenience interface to ctype.is(ctype_base::cntrl, __c).
template<typename _CharT>
inline bool
iscntrl(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::cntrl, __c); }
/// Convenience interface to ctype.is(ctype_base::upper, __c).
template<typename _CharT>
inline bool
isupper(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::upper, __c); }
/// Convenience interface to ctype.is(ctype_base::lower, __c).
template<typename _CharT>
inline bool
islower(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::lower, __c); }
/// Convenience interface to ctype.is(ctype_base::alpha, __c).
template<typename _CharT>
inline bool
isalpha(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::alpha, __c); }
/// Convenience interface to ctype.is(ctype_base::digit, __c).
template<typename _CharT>
inline bool
isdigit(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::digit, __c); }
/// Convenience interface to ctype.is(ctype_base::punct, __c).
template<typename _CharT>
inline bool
ispunct(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::punct, __c); }
/// Convenience interface to ctype.is(ctype_base::xdigit, __c).
template<typename _CharT>
inline bool
isxdigit(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::xdigit, __c); }
/// Convenience interface to ctype.is(ctype_base::alnum, __c).
template<typename _CharT>
inline bool
isalnum(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::alnum, __c); }
/// Convenience interface to ctype.is(ctype_base::graph, __c).
template<typename _CharT>
inline bool
isgraph(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).is(ctype_base::graph, __c); }
/// Convenience interface to ctype.toupper(__c).
template<typename _CharT>
inline _CharT
toupper(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).toupper(__c); }
/// Convenience interface to ctype.tolower(__c).
template<typename _CharT>
inline _CharT
tolower(_CharT __c, const locale& __loc)
{ return use_facet<ctype<_CharT> >(__loc).tolower(__c); }
_GLIBCXX_END_NAMESPACE_VERSION
} // namespace std
# include <bits/locale_facets.tcc>
#endif