|
|
Boost : |
From: Alisdair Meredith (alisdair.meredith_at_[hidden])
Date: 2002-12-20 07:40:03
Ok, now I have access to BCB6 I have tracked down the failures to some
problems with std::isdigit.
For some reason isdigit is trying to instantiate the ctype facet for
const char, ctype<const char>. I haven't worked out the reason this
instantiation is required, but by replacing all calls to isdigit in
parsing.hpp with the 'equivalent' code below ( borrowed straight from
Josutti's The Standard C++ Library ) CBuilder6 passes all format tests
(with compiler warnings for binding non-const references to temporaries)
#if defined( __BORLANDC__ )
while ( i1 < buf.size()
&& std::use_facet<std::ctype<Ch> >( oss_.rdbuf()->getloc() ).is(
std::ctype_base::digit, buf[i1] )
)
#else
while(i1 < buf.size() && isdigit(buf[i1],oss_.rdbuf()->getloc()))
#endif
++i1;
Attached is the fully updated parsing.hpp file
-- AlisdairM
// -*- C++ -*-
// Boost general library 'format' ---------------------------
// See http://www.boost.org for updates, documentation, and revision history.
// (C) Samuel Krempp 2001
// krempp_at_[hidden]
// Permission to copy, use, modify, sell and
// distribute this software is granted provided this copyright notice appears
// in all copies. This software is provided "as is" without express or implied
// warranty, and with no claim as to its suitability for any purpose.
// ideas taken from Rudiger Loos's format class
// and Karl Nelson's ofstream (also took its parsing code as basis for printf parsing)
// ------------------------------------------------------------------------------
// parsing.hpp : implementation of the parsing member functions
// ( parse, parse_printf_directive)
// ------------------------------------------------------------------------------
#ifndef BOOST_FORMAT_PARSING_HPP
#define BOOST_FORMAT_PARSING_HPP
#ifdef BOOST_MSVC
#ifdef isdigit
#undef isdigit
#endif
#endif // MSVC workaround
#include "boost/format/format_class.hpp"
#include "boost/throw_exception.hpp"
namespace boost {
namespace io {
namespace detail {
template<class Res, class Ch, class Tr> inline
Res str2int(const std::basic_string<Ch, Tr>& s,
typename std::basic_string<Ch, Tr>::size_type start,
BOOST_IO_STD basic_ios<Ch,Tr> &os,
const Res = Res(0) )
// Input : char string, with starting index
// a basic_ios& merely to call its widen/narrow member function in the desired locale.
// Effects : reads s[start:] and converts digits into an integral n, of type Res
// Returns : n
{
using namespace std; // isdigit is better without 'std::' for borland 0x560
Res n = 0;
#if defined( __BORLANDC__ )
while ( start<s.size()
&& std::use_facet<std::ctype<Ch> >( os.rdbuf()->getloc() ).is( std::ctype_base::digit, s[start] )
)
#else
while(start<s.size() && isdigit(s[start], os.rdbuf()->getloc() ) )
#endif
{
char cur_ch = os.narrow( s[start], 0);
assert(cur_ch != 0 ); // since we called isdigit, this should not happen.
n *= 10;
n += cur_ch - '0'; // 22.2.1.1.2 of the C++ standard
++start;
}
return n;
}
template<class Ch, class Tr>
void skip_asterisk(const std::basic_string<Ch,Tr> & buf,
typename std::basic_string<Ch,Tr>::size_type * pos_p,
BOOST_IO_STD basic_ios<Ch, Tr> &os)
// skip printf's "asterisk-fields" directives in the format-string buf
// Input : char string, with starting index *pos_p
// a basic_ios& merely to call its widen/narrow member function in the desired locale.
// Effects : advance *pos_p by skipping printf's asterisk fields.
// Returns : nothing
{
using namespace std; // isdigit is better without 'std::' for borland 0x560
assert( pos_p != 0);
if(*pos_p >= buf.size() ) return;
if(buf[ *pos_p]==os.widen('*')) {
++ (*pos_p);
#if defined( __BORLANDC__ )
while ( *pos_p < buf.size()
&& std::use_facet<std::ctype<Ch> >( os.rdbuf()->getloc() ).is( std::ctype_base::digit, buf[*pos_p] )
)
#else
while (*pos_p < buf.size() && isdigit(buf[*pos_p],os.rdbuf()->getloc()))
#endif
++(*pos_p);
if(buf[*pos_p]==os.widen('$')) ++(*pos_p);
}
}
inline void maybe_throw_exception( unsigned char exceptions)
// auxiliary func called by parse_printf_directive
// for centralising error handling
// it either throws if user sets the corresponding flag, or does nothing.
{
if(exceptions & io::bad_format_string_bit)
boost::throw_exception(io::bad_format_string());
}
template<class Ch, class Tr>
bool parse_printf_directive(const std::basic_string<Ch, Tr> & buf,
typename std::basic_string<Ch, Tr>::size_type * pos_p,
detail::format_item<Ch, Tr> * fpar,
BOOST_IO_STD basic_ios<Ch,Tr> &os,
unsigned char exceptions)
// Input : a 'printf-directive' in the format-string, starting at buf[ *pos_p ]
// a basic_ios& merely to call its widen/narrow member function in the desired locale.
// a bitset'excpetions' telling whether to throw exceptions on errors.
// Returns : true if parse somehow succeeded (possibly ignoring errors if exceptions disabled)
// false if it failed so bad that the directive should be printed verbatim
// Effects : - *pos_p is incremented so that buf[*pos_p] is the first char after the directive
// - *fpar is set with the parameters read in the directive
{
using namespace std; // isdigit is better without 'std::' for borland 0x560
typedef format_item<Ch, Tr> format_item_t;
assert( pos_p != 0);
typename std::basic_string<Ch, Tr>::size_type &i1 = *pos_p,
i0;
fpar->argN_ = format_item_t::argN_no_posit; // if no positional-directive
bool in_brackets=false;
if(buf[i1]==os.widen('|'))
{
in_brackets=true;
if( ++i1 >= buf.size() ) {
maybe_throw_exception(exceptions);
return false;
}
}
// the flag '0' would be picked as a digit for argument order, but here it's a flag :
if(buf[i1]==os.widen('0'))
goto parse_flags;
// handle argument order (%2$d) or possibly width specification: %2d
i0 = i1; // save position before digits
#if defined( __BORLANDC__ )
while ( i1 < buf.size()
&& std::use_facet<std::ctype<Ch> >( os.rdbuf()->getloc() ).is( std::ctype_base::digit, buf[i1] )
)
#else
while (i1 < buf.size() && isdigit(buf[i1], os.rdbuf()->getloc()))
#endif
++i1;
if (i1!=i0)
{
if( i1 >= buf.size() ) {
maybe_throw_exception(exceptions);
return false;
}
int n=str2int(buf,i0, os, int(0) );
// %N% case : this is already the end of the directive
if( buf[i1] == os.widen('%') )
{
fpar->argN_ = n-1;
++i1;
if( in_brackets)
maybe_throw_exception(exceptions);
// but don't return. maybe "%" was used in lieu of '$', so we go on.
else return true;
}
if ( buf[i1]==os.widen('$') )
{
fpar->argN_ = n-1;
++i1;
}
else
{
// non-positionnal directive
fpar->ref_state_.width_ = n;
fpar->argN_ = format_item_t::argN_no_posit;
goto parse_precision;
}
}
parse_flags:
// handle flags
while ( i1 <buf.size()) // as long as char is one of + - = # 0 l h or ' '
{
// misc switches
switch (os.narrow(buf[i1], 0))
{
case '\'' : break; // no effect yet. (painful to implement)
case 'l':
case 'h': // short/long modifier : for printf-comaptibility (no action needed)
break;
case '-':
fpar->ref_state_.flags_ |= std::ios_base::left;
break;
case '=':
fpar->pad_scheme_ |= format_item_t::centered;
break;
case ' ':
fpar->pad_scheme_ |= format_item_t::spacepad;
break;
case '+':
fpar->ref_state_.flags_ |= std::ios_base::showpos;
break;
case '0':
fpar->pad_scheme_ |= format_item_t::zeropad;
// need to know alignment before really setting flags,
// so just add 'zeropad' flag for now, it will be processed later.
break;
case '#':
fpar->ref_state_.flags_ |= std::ios_base::showpoint | std::ios_base::showbase;
break;
default:
goto parse_width;
}
++i1;
} // loop on flag.
if( i1>=buf.size()) {
maybe_throw_exception(exceptions);
return true;
}
parse_width:
// handle width spec
skip_asterisk(buf, &i1, os); // skips 'asterisk fields' : *, or *N$
i0 = i1; // save position before digits
#if defined( __BORLANDC__ )
while ( i1 < buf.size()
&& std::use_facet<std::ctype<Ch> >( os.rdbuf()->getloc() ).is( std::ctype_base::digit, buf[i1] )
)
#else
while (i1<buf.size() && isdigit(buf[i1], os.rdbuf()->getloc()))
#endif
i1++;
if (i1!=i0)
{ fpar->ref_state_.width_ = str2int( buf,i0, os, std::streamsize(0) ); }
parse_precision:
if( i1>=buf.size()) {
maybe_throw_exception(exceptions);
return true;
}
// handle precision spec
if (buf[i1]==os.widen('.'))
{
++i1;
skip_asterisk(buf, &i1, os);
i0 = i1; // save position before digits
#if defined( __BORLANDC__ )
while ( i1 < buf.size()
&& std::use_facet<std::ctype<Ch> >( os.rdbuf()->getloc() ).is( std::ctype_base::digit, buf[i1] )
)
#else
while (i1<buf.size() && isdigit(buf[i1], os.rdbuf()->getloc()))
#endif
++i1;
if(i1==i0)
fpar->ref_state_.precision_ = 0;
else
fpar->ref_state_.precision_ = str2int(buf,i0, os, std::streamsize(0) );
}
// handle formatting-type flags :
while( i1<buf.size() &&
( buf[i1]==os.widen('l') || buf[i1]==os.widen('L') || buf[i1]==os.widen('h')) )
++i1;
if( i1>=buf.size()) {
maybe_throw_exception(exceptions);
return true;
}
if( in_brackets && buf[i1]==os.widen('|') )
{
++i1;
return true;
}
switch (os.narrow(buf[i1], 0) )
{
case 'X':
fpar->ref_state_.flags_ |= std::ios_base::uppercase;
case 'p': // pointer => set hex.
case 'x':
fpar->ref_state_.flags_ &= ~std::ios_base::basefield;
fpar->ref_state_.flags_ |= std::ios_base::hex;
break;
case 'o':
fpar->ref_state_.flags_ &= ~std::ios_base::basefield;
fpar->ref_state_.flags_ |= std::ios_base::oct;
break;
case 'E':
fpar->ref_state_.flags_ |= std::ios_base::uppercase;
case 'e':
fpar->ref_state_.flags_ &= ~std::ios_base::floatfield;
fpar->ref_state_.flags_ |= std::ios_base::scientific;
fpar->ref_state_.flags_ &= ~std::ios_base::basefield;
fpar->ref_state_.flags_ |= std::ios_base::dec;
break;
case 'f':
fpar->ref_state_.flags_ &= ~std::ios_base::floatfield;
fpar->ref_state_.flags_ |= std::ios_base::fixed;
case 'u':
case 'd':
case 'i':
fpar->ref_state_.flags_ &= ~std::ios_base::basefield;
fpar->ref_state_.flags_ |= std::ios_base::dec;
break;
case 'T':
++i1;
if( i1 >= buf.size())
maybe_throw_exception(exceptions);
else
fpar->ref_state_.fill_ = buf[i1];
fpar->pad_scheme_ |= format_item_t::tabulation;
fpar->argN_ = format_item_t::argN_tabulation;
break;
case 't':
fpar->ref_state_.fill_ = os.widen(' ');
fpar->pad_scheme_ |= format_item_t::tabulation;
fpar->argN_ = format_item_t::argN_tabulation;
break;
case 'G':
fpar->ref_state_.flags_ |= std::ios_base::uppercase;
break;
case 'g': // 'g' conversion is default for floats.
fpar->ref_state_.flags_ &= ~std::ios_base::basefield;
fpar->ref_state_.flags_ |= std::ios_base::dec;
// CLEAR all floatield flags, so stream will CHOOSE
fpar->ref_state_.flags_ &= ~std::ios_base::floatfield;
break;
case 'C':
case 'c':
fpar->truncate_ = 1;
break;
case 'S':
case 's':
fpar->truncate_ = fpar->ref_state_.precision_;
fpar->ref_state_.precision_ = -1;
break;
case 'n' :
fpar->argN_ = format_item_t::argN_ignored;
break;
default:
maybe_throw_exception(exceptions);
}
++i1;
if( in_brackets )
{
if( i1<buf.size() && buf[i1]==os.widen('|') )
{
++i1;
return true;
}
else maybe_throw_exception(exceptions);
}
return true;
}
} // detail namespace
} // io namespace
// -----------------------------------------------
// format :: parse(..)
template<class Ch, class Traits>
void basic_format<Ch, Traits> ::parse(const string_t & buf)
// parse the format-string
{
using namespace std;
const Ch arg_mark = oss_.widen('%');
bool ordered_args=true;
int max_argN=-1;
typename string_t::size_type i1=0;
int num_items=0;
// A: find upper_bound on num_items and allocates arrays
i1=0;
while( (i1=buf.find(arg_mark,i1)) != string::npos )
{
if( i1+1 >= buf.size() ) {
if(exceptions() & io::bad_format_string_bit)
boost::throw_exception(io::bad_format_string()); // must not end in "bla bla %"
else break; // stop there, ignore last '%'
}
if(buf[i1+1] == buf[i1] ) { i1+=2; continue; } // escaped "%%" / "##"
++i1;
// in case of %N% directives, dont count it double (wastes allocations..) :
#if defined( __BORLANDC__ )
while ( i1 < buf.size()
&& std::use_facet<std::ctype<Ch> >( oss_.rdbuf()->getloc() ).is( std::ctype_base::digit, buf[i1] )
)
#else
while(i1 < buf.size() && isdigit(buf[i1],oss_.rdbuf()->getloc()))
#endif
++i1;
if( i1 < buf.size() && buf[i1] == arg_mark ) ++ i1;
++num_items;
}
items_.assign( num_items, format_item_t() );
// B: Now the real parsing of the format string :
num_items=0;
i1 = 0;
typename string_t::size_type i0 = i1;
bool special_things=false;
int cur_it=0;
while( (i1=buf.find(arg_mark,i1)) != string::npos )
{
string_t & piece = (cur_it==0) ? prefix_ : items_[cur_it-1].appendix_;
if( buf[i1+1] == buf[i1] ) // escaped mark, '%%'
{
piece += buf.substr(i0, i1-i0) + buf[i1];
i1+=2; i0=i1;
continue;
}
assert( static_cast<unsigned int>(cur_it) < items_.size() || cur_it==0);
if(i1!=i0) piece += buf.substr(i0, i1-i0);
++i1;
bool parse_ok;
parse_ok = io::detail::parse_printf_directive(buf, &i1, &items_[cur_it], oss_, exceptions());
if( ! parse_ok ) continue; // the directive will be printed verbatim
i0=i1;
items_[cur_it].compute_states(); // process complex options, like zeropad, into stream params.
int argN=items_[cur_it].argN_;
if(argN == format_item_t::argN_ignored)
continue;
if(argN ==format_item_t::argN_no_posit)
ordered_args=false;
else if(argN == format_item_t::argN_tabulation) special_things=true;
else if(argN > max_argN) max_argN = argN;
++num_items;
++cur_it;
} // loop on %'s
assert(cur_it == num_items);
// store the final piece of string
string_t & piece = (cur_it==0) ? prefix_ : items_[cur_it-1].appendix_;
piece += buf.substr(i0);
if( !ordered_args)
{
if(max_argN >= 0 ) // dont mix positional with non-positionnal directives
{
if(exceptions() & io::bad_format_string_bit)
boost::throw_exception(io::bad_format_string());
// else do nothing. => positionnal arguments are processed as non-positionnal
}
// set things like it would have been with positional directives :
int non_ordered_items = 0;
for(int i=0; i< num_items; ++i)
if(items_[i].argN_ == format_item_t::argN_no_posit)
{
items_[i].argN_ = non_ordered_items;
++non_ordered_items;
}
max_argN = non_ordered_items-1;
}
// C: set some member data :
items_.resize(num_items);
if(special_things) style_ |= special_needs;
num_args_ = max_argN + 1;
if(ordered_args) style_ |= ordered;
else style_ &= ~ordered;
}
} // namespace boost
#endif // BOOST_FORMAT_PARSING_HPP
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk