yap-6.3/os/pl-fmt.c

/*  Part of SWI-Prolog

    Author:        Jan Wielemaker
    E-mail:        J.Wielemaker@vu.nl
    WWW:           http://www.swi-prolog.org
    Copyright (C): 1985-2013, University of Amsterdam
			      VU University Amsterdam

    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
    License as published by the Free Software Foundation; either
    version 2.1 of the License, 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
    Lesser General Public License for more details.

    You should have received a copy of the GNU Lesser General Public
    License along with this library; if not, write to the Free Software
    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
*/

/**
 * @defgroup Format Formatted Output
 * @ingroup InputOutput
 * @{
 */

/* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Formatted output (Prolog predicates format/[1,2,3]).   One  day,  the  C
source should also use format() to produce error messages, etc.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */

#include "pl-incl.h"
#include "pl-ctype.h"
#include "pl-utf8.h"
#include <ctype.h>

static char *	formatInteger(PL_locale *locale, int div, int radix,
			     bool smll, Number n, Buffer out);
static char *	formatFloat(PL_locale *locale, int how, int arg,
			    Number f, Buffer out);

#define MAXRUBBER 100

struct rubber
{ size_t where;				/* where is rubber in output */
  size_t size;				/* how big should it be */
  pl_wchar_t pad;			/* padding character */
};

typedef struct
{ IOSTREAM *out;			/* our output stream */
  int column;				/* current column */
  tmp_buffer buffer;			/* bin for characters with tabs */
  size_t buffered;			/* characters in buffer */
  int pending_rubber;			/* number of not-filled ~t's */
  struct rubber rub[MAXRUBBER];
} format_state;

#define BUFSIZE		1024
#define DEFAULT		(-1)
#define SHIFT		{ argc--; argv++; }
#define NEED_ARG	{ if ( argc <= 0 ) \
			  { FMT_ERROR("not enough arguments"); \
			  } \
			}
#define FMT_ERROR(fmt)	return (void)Sunlock(fd), \
			  PL_error(NULL, 0, NULL, ERR_FORMAT, fmt)
#define FMT_ARG(c, a)	return (void)Sunlock(fd), \
			       PL_error(NULL, 0, NULL, \
					ERR_FORMAT_ARG, c, a)
#define FMT_EXEPTION()	return (void)Sunlock(fd), FALSE


static PL_locale prolog_locale =
{ 0,0,LOCALE_MAGIC,1,
  L".", NULL
};


static int
update_column(int col, int c)
{ switch(c)
  { case '\n':	return 0;
    case '\r':  return 0;
    case '\t':	return (col + 1) | 0x7;
    case '\b':	return (col <= 0 ? 0 : col - 1);
    default:	return col + 1;
  }
}


/* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Low-level output. If there is pending  rubber   the  output is stored in
UTF-8 format in the state's `buffer'.   The  `buffered' field represents
the number of UTF-8 characters in the buffer.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */

static int
outchr(format_state *state, int chr)
{ if ( state->pending_rubber )
  { if ( chr > 0x7f )
    { char buf[8];
      char *s, *e;

      e = utf8_put_char(buf, chr);
      for(s=buf; s<e; s++)
	addBuffer((Buffer)&state->buffer, *s, char);
    } else
    { char c = chr;

      addBuffer((Buffer)&state->buffer, c, char);
    }

    state->buffered++;
  } else
  { if ( Sputcode(chr, state->out) < 0 )
      return FALSE;
  }

  state->column = update_column(state->column, chr);

  return TRUE;
}


/* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Emit ASCII 0-terminated strings resulting from sprintf() on numeric
arguments.  No fuzz with wide characters here.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */

static int
outstring(format_state *state, const char *s, size_t len)
{ const char *q;
  const char *e = &s[len];

  if ( state->pending_rubber )
  { addMultipleBuffer(&state->buffer, s, len, char);
    state->buffered += len;
  } else
  { for(q=s; q < e; q++)
    { if ( Sputcode(*q&0xff, state->out) < 0 )
	return FALSE;
    }
  }

  for(q=s; q < e; q++)
    state->column = update_column(state->column, *q&0xff);

  return TRUE;
}


static int
oututf8(format_state *state, const char *s, size_t len)
{ const char *e = &s[len];

  while(s<e)
  { int chr;

    s = utf8_get_char(s, &chr);
    if ( !outchr(state, chr) )
      return FALSE;
  }

  return TRUE;
}


static int
oututf80(format_state *state, const char *s)
{ return oututf8(state, s, strlen(s));
}


static int
outtext(format_state *state, PL_chars_t *txt)
{ switch(txt->encoding)
  { case ENC_ISO_LATIN_1:
      return outstring(state, txt->text.t, txt->length);
    case ENC_UTF8:
      return oututf8(state, txt->text.t, txt->length);
    case ENC_WCHAR:
    { const pl_wchar_t *s = txt->text.w;
      const pl_wchar_t *e = &s[txt->length];

      while(s<e)
      { if ( !outchr(state, *s++) )
	  return FALSE;
      }

      return TRUE;
    }
    default:
    { assert(0);
      return FALSE;
    }
  }
}


#define format_predicates (GD->format.predicates)

static int	update_column(int, Char);
static bool	do_format(IOSTREAM *fd, PL_chars_t *fmt,
			  int ac, term_t av, Module m);
static void	distribute_rubber(struct rubber *, int, int);
static int	emit_rubber(format_state *state);


		/********************************
		*       PROLOG CONNECTION	*
		********************************/

word
pl_format_predicate(term_t chr, term_t descr)
{ int c;
  predicate_t proc = NULL;
  Symbol s;
  int arity;

  if ( !PL_get_char_ex(chr, &c, FALSE) )
    fail;

  if ( !get_procedure(descr, &proc, 0, GP_CREATE) )
    fail;
  PL_predicate_info(proc, NULL, &arity, NULL);
  if ( arity == 0 )
    return PL_error(NULL, 0, "arity must be > 0", ERR_DOMAIN,
		    PL_new_atom("format_predicate"),
		    descr);

  if ( !format_predicates )
    format_predicates = newHTable(8);

  if ( (s = lookupHTable(format_predicates, (void *)(intptr_t)c)) )
    s->value = proc;
  else
    addHTable(format_predicates, (void *)(intptr_t)c, proc);

  succeed;
}


word
pl_current_format_predicate(term_t chr, term_t descr, control_t h)
{ GET_LD
  Symbol s = NULL;
  TableEnum e;
  fid_t fid;

  switch( ForeignControl(h) )
  { case FRG_FIRST_CALL:
      if ( !format_predicates )
	fail;
      e = newTableEnum(format_predicates);
      break;
    case FRG_REDO:
      e = ForeignContextPtr(h);
      break;
    case FRG_CUTTED:
      e = ForeignContextPtr(h);
      freeTableEnum(e);
    default:
      succeed;
  }

  if ( !(fid = PL_open_foreign_frame()) )
  { freeTableEnum(e);
    return FALSE;
  }
  while( (s=advanceTableEnum(e)) )
  { if ( PL_unify_integer(chr, (intptr_t)s->name) &&
	 PL_unify_predicate(descr, (predicate_t)s->value, 0) )
    { PL_close_foreign_frame(fid);
      ForeignRedoPtr(e);
    }

    PL_rewind_foreign_frame(fid);
  }

  PL_close_foreign_frame(fid);
  freeTableEnum(e);
  fail;
}


static word
format_impl(IOSTREAM *out, term_t format, term_t Args, Module m)
{ GET_LD
  term_t argv;
  int argc = 0;
  term_t args = PL_copy_term_ref(Args);
  int rval;
  PL_chars_t fmt;

  if ( !PL_get_text(format, &fmt, CVT_ALL|BUF_RING) )
    return PL_error("format", 3, NULL, ERR_TYPE, ATOM_text, format);

  if ( (argc = (int)lengthList(args, FALSE)) >= 0 )
  { term_t head = PL_new_term_ref();
    int n = 0;

    argv = PL_new_term_refs(argc);
    while( PL_get_list(args, head, args) )
      PL_put_term(argv+n++, head);
  } else
  { argc = 1;
    argv = PL_new_term_refs(argc);

    PL_put_term(argv, args);
  }

  startCritical;
  switch(fmt.storage)			/* format can do call-back! */
  { case PL_CHARS_RING:
    case PL_CHARS_STACK:
      PL_save_text(&fmt, BUF_MALLOC);
      break;
    default:
      break;
  }

  rval = do_format(out, &fmt, argc, argv, m);
  PL_free_text(&fmt);
  if ( !endCritical )
    return FALSE;

  return rval;
}


/** @pred  format(+ _S_,+ _T_,+ _L_)

Print formatted output to stream  _S_.


*/
word
pl_format3(term_t out, term_t format, term_t args)
{ GET_LD
  redir_context ctx;
  word rc;
  Module m = NULL;
  term_t list = PL_new_term_ref();

  if ( !PL_strip_module(args, &m, list) )
    return FALSE;

  if ( (rc=setupOutputRedirect(out, &ctx, FALSE)) )
  { if ( (rc = format_impl(ctx.stream, format, list, m)) )
      rc = closeOutputRedirect(&ctx);
    else
      discardOutputRedirect(&ctx);
  }

  return rc;
}

/** @pred  format(+ _T_,+ _L_)


Print formatted output to the current output stream. The arguments in
list  _L_ are output according to the string or atom  _T_.

A control sequence is introduced by a `w`. The following control
sequences are available in YAP:



+ `~~`
Print a single tilde.

+ `~a`
The next argument must be an atom, that will be printed as if by `write`.

+ `~Nc`
The next argument must be an integer, that will be printed as a
character code. The number  _N_ is the number of times to print the
character (default 1).

+ `~Ne`
+ `~NE`
+ `~Nf`
+ `~Ng`
+ `~NG`
The next argument must be a floating point number. The float  _F_, the number
 _N_ and the control code `c` will be passed to `printf` as:

~~~~~{.prolog}
    printf("%s.Nc", F)
~~~~~

As an example:

~~~~~{.prolog}
?- format("~8e, ~8E, ~8f, ~8g, ~8G~w",
          [3.14,3.14,3.14,3.14,3.14,3.14]).
3.140000e+00, 3.140000E+00, 3.140000, 3.14, 3.143.14
~~~~~

+ `~Nd`
The next argument must be an integer, and  _N_ is the number of digits
after the decimal point. If  _N_ is `0` no decimal points will be
printed. The default is  _N = 0_.

~~~~~{.prolog}
?- format("~2d, ~d",[15000, 15000]).
150.00, 15000
~~~~~

+ `~ND`
Identical to `~Nd`, except that commas are used to separate groups
of three digits.

~~~~~{.prolog}
?- format("~2D, ~D",[150000, 150000]).
1,500.00, 150,000
~~~~~

+ `~i`
Ignore the next argument in the list of arguments:

~~~~~{.prolog}
?- format('The ~i met the boregrove',[mimsy]).
The  met the boregrove
~~~~~

+ `~k`
Print the next argument with `write_canonical`:

~~~~~{.prolog}
?- format("Good night ~k",a+[1,2]).
Good night +(a,[1,2])
~~~~~

+ `~Nn`
Print  _N_ newlines (where  _N_ defaults to 1).

+ `~NN`
Print  _N_ newlines if at the beginning of the line (where  _N_
defaults to 1).

+ `~Nr`
The next argument must be an integer, and  _N_ is interpreted as a
radix, such that `2 <= N <= 36` (the default is 8).

~~~~~{.prolog}
?- format("~2r, 0x~16r, ~r",
          [150000, 150000, 150000]).
100100100111110000, 0x249f0, 444760
~~~~~

Note that the letters `a-z` denote digits larger than 9.

+ `~NR`
Similar to `~NR`. The next argument must be an integer, and  _N_ is
interpreted as a radix, such that `2 <= N <= 36` (the default is 8).

~~~~~{.prolog}
?- format("~2r, 0x~16r, ~r",
          [150000, 150000, 150000]).
100100100111110000, 0x249F0, 444760
~~~~~

The only difference is that letters `A-Z` denote digits larger than 9.

+ `~p`
Print the next argument with print/1:

~~~~~{.prolog}
?- format("Good night ~p",a+[1,2]).
Good night a+[1,2]
~~~~~

+ `~q`
Print the next argument with writeq/1:

~~~~~{.prolog}
?- format("Good night ~q",'Hello'+[1,2]).
Good night 'Hello'+[1,2]
~~~~~

+ `~Ns`
The next argument must be a list of character codes. The system then
outputs their representation as a string, where  _N_ is the maximum
number of characters for the string ( _N_ defaults to the length of the
string).

~~~~~{.prolog}
?- format("The ~s are ~4s",["woods","lovely"]).
The woods are love
~~~~~

+ `~w`
Print the next argument with write/1:

~~~~~
?- format("Good night ~w",'Hello'+[1,2]).
Good night Hello+[1,2]
~~~~~


The number of arguments, `N`, may be given as an integer, or it
may be given as an extra argument. The next example shows a small
procedure to write a variable number of `a` characters:

~~~~~
write_many_as(N) :-
        format("~*c",[N,0'a]).
~~~~~

The format/2 built-in also allows for formatted output.  One can
specify column boundaries and fill the intermediate space by a padding
character:

+ `~N|`
Set a column boundary at position  _N_, where  _N_ defaults to the
current position.

+ `~N+`
Set a column boundary at  _N_ characters past the current position, where
 _N_ defaults to `8`.

+ `~Nt`
Set padding for a column, where  _N_ is the fill code (default is
`SPC`).



The next example shows how to align columns and padding. We first show
left-alignment:

~~~~~
   ?- format("~n*Hello~16+*~n",[]).
*Hello          *
~~~~~

Note that we reserve 16 characters for the column.

The following example shows how to do right-alignment:

~~~~~
   ?- format("*~tHello~16+*~n",[]).
*          Hello*

~~~~~

The `~t` escape sequence forces filling before `Hello`.

We next show how to do centering:

~~~~~
   ?- format("*~tHello~t~16+*~n",[]).
*     Hello     *
~~~~~

The two `~t` escape sequence force filling both before and after
`Hello`. Space is then evenly divided between the right and the
left sides.


*/
word
pl_format(term_t fmt, term_t args)
{ return pl_format3(0, fmt, args);
}


static inline int
get_chr_from_text(const PL_chars_t *t, int index)
{ switch(t->encoding)
  { case ENC_ISO_LATIN_1:
      return t->text.t[index]&0xff;
    case ENC_WCHAR:
      return t->text.w[index];
    case ENC_UTF8:
      return t->text.w[index];
    default:
      assert(0);
      return 0;				/* not reached */
  }
}


		/********************************
		*       ACTUAL FORMATTING	*
		********************************/

static bool
do_format(IOSTREAM *fd, PL_chars_t *fmt, int argc, term_t argv, Module m)
{ GET_LD
  format_state state;			/* complete state */
  int tab_stop = 0;			/* padded tab stop */
  Symbol s;
  unsigned int here = 0;
  int rc = TRUE;

  Slock(fd);				/* buffer locally */

  state.out = fd;
  state.pending_rubber = 0;
  initBuffer(&state.buffer);
  state.buffered = 0;

  if ( fd->position )
    state.column = fd->position->linepos;
  else
    state.column = 0;

  while(here < fmt->length)
  { int c = get_chr_from_text(fmt, here);

    switch(c)
    { case '~':
	{ int arg = DEFAULT;		/* Numeric argument */
	  int mod_colon = FALSE;	/* Used colon modifier */
					/* Get the numeric argument */
	  c = get_chr_from_text(fmt, ++here);

	  if ( isDigitW(c) )
	  { arg = c - '0';

	    here++;
	    while(here < fmt->length)
	    { c = get_chr_from_text(fmt, here);

	      if ( isDigitW(c) )
	      { int dw = c - '0';
		int arg2 = arg*10 + dw;

		if ( (arg2 - dw)/10 != arg )	/* see mul64() in pl-arith.c */
		{ FMT_ERROR("argument overflow");
		}
		arg = arg2;
		here++;
	      } else
		break;
	    }
	  } else if ( c == '*' )
	  { NEED_ARG;
	    if ( PL_get_integer(argv, &arg) )
	    { SHIFT;
	    } else
	      FMT_ERROR("no or negative integer for `*' argument");
	    c = get_chr_from_text(fmt, ++here);
	  } else if ( c == '`' && here < fmt->length )
	  { arg = get_chr_from_text(fmt, ++here);
	    c = get_chr_from_text(fmt, ++here);
	  }

	  if ( c == ':' )
	  { mod_colon = TRUE;
	    c = get_chr_from_text(fmt, ++here);
	  }

					/* Check for user defined format */
	  if ( format_predicates &&
	       (s = lookupHTable(format_predicates, (void*)((intptr_t)c))) )
	  { predicate_t proc = (predicate_t) s->value;
	    int arity;
	    term_t av;
	    char buf[BUFSIZE];
	    char *str = buf;
	    size_t bufsize = BUFSIZE;
	    int i;

	    PL_predicate_info(proc, NULL, &arity, NULL);
	    av = PL_new_term_refs(arity);

	    if ( arg == DEFAULT )
	      PL_put_atom(av+0, ATOM_default);
	    else
	      PL_put_integer(av+0, arg);

	    for(i=1; i < arity; i++)
	    { NEED_ARG;
	      PL_put_term(av+i, argv);
	      SHIFT;
	    }

	    tellString(&str, &bufsize, ENC_UTF8);
	    rc = PL_call_predicate(NULL, PL_Q_PASS_EXCEPTION, proc, av);
	    toldString();
	    if ( !rc )
	    { if ( str != buf )
		free(str);
	      goto out;
	    }
	    oututf8(&state, str, bufsize);
	    if ( str != buf )
	      free(str);

	    here++;
	  } else
	  { switch(c)			/* Build in formatting */
	    { case 'a':			/* atomic */
		{ PL_chars_t txt;

		  NEED_ARG;
		  if ( !PL_get_text(argv, &txt, CVT_ATOMIC) )
		    FMT_ARG("a", argv);
		  SHIFT;
		  rc = outtext(&state, &txt);
                  if ( !rc )
		    goto out;
		  here++;
		  break;
		}
	      case 'c':			/* ~c: character code */
		{ int chr;

		  NEED_ARG;
		  if ( PL_get_integer(argv, &chr) && chr >= 0 )
		  { int times = (arg == DEFAULT ? 1 : arg);

		    SHIFT;
		    while(times-- > 0)
		    { rc = outchr(&state, chr);
		      if ( !rc )
		        goto out;
		    }
		  } else
		    FMT_ARG("c", argv);
		  here++;
		  break;
		}
	      case 'e':			/* exponential float */
	      case 'E':			/* Exponential float */
	      case 'f':			/* float */
	      case 'g':			/* shortest of 'f' and 'e' */
	      case 'G':			/* shortest of 'f' and 'E' */
		{ number n;
		  union {
		  tmp_buffer b;
		    buffer b1;
		  } u;
		  PL_locale *l;

		  NEED_ARG;
		  if ( !valueExpression(argv, &n PASS_LD) )
		  { char f[2];

		    f[0] = c;
		    f[1] = EOS;
		    FMT_ARG(f, argv);
		  }
		  SHIFT;

		  if ( c == 'f' && mod_colon )
		    l = fd->locale;
		  else
		    l = &prolog_locale;

		  initBuffer(&u.b);
		  rc = formatFloat(l, c, arg, &n, &u.b1) != NULL;
		  clearNumber(&n);
		  if ( rc )
		    rc = oututf80(&state, baseBuffer(&u.b, char));
		  discardBuffer(&u.b);
                  if ( !rc )
		    goto out;
		  here++;
		  break;
		}
	      case 'd':			/* integer */
	      case 'D':			/* grouped integer */
	      case 'r':			/* radix number */
	      case 'R':			/* Radix number */
	      case 'I':			/* Prolog 1_000_000 */
		{ number i;
		  tmp_buffer b;

		  NEED_ARG;
		  if ( !valueExpression(argv, &i PASS_LD) ||
		       !toIntegerNumber(&i, 0) )
		  { char f[2];

		    f[0] = c;
		    f[1] = EOS;
		    FMT_ARG(f, argv);
		  }
		  SHIFT;
		  initBuffer(&b);
		  if ( c == 'd' || c == 'D' )
		  { PL_locale ltmp;
		    PL_locale *l;
		    static char grouping[] = {3,0};

		    if ( c == 'D' )
		    { ltmp.thousands_sep = L",";
		      ltmp.decimal_point = L".";
		      ltmp.grouping = grouping;
		      l = &ltmp;
		    } else if ( mod_colon )
		    { l = fd->locale;
		    } else
		    { l = NULL;
		    }

		    if ( arg == DEFAULT )
		      arg = 0;
		    if ( !formatInteger(l, arg, 10, TRUE, &i, (Buffer)&b) )
		      FMT_EXEPTION();
		  } else if ( c == 'I' )
		  { PL_locale ltmp;
		    char grouping[2];

		    grouping[0] = (arg == DEFAULT ? 3 : arg);
		    grouping[1] = '\0';
		    ltmp.thousands_sep = L"_";
		    ltmp.grouping = grouping;

		    if ( !formatInteger(&ltmp, 0, 10, TRUE, &i, (Buffer)&b) )
		      FMT_EXEPTION();
		  } else			/* r,R */
		  { if ( arg == DEFAULT )
		      FMT_ERROR("r,R requires radix specifier");
		    if ( arg < 1 || arg > 36 )
		    { term_t r = PL_new_term_ref();

		      PL_put_integer(r, arg);
		      Sunlock(fd);
		      return PL_error(NULL, 0, NULL, ERR_DOMAIN,
				      ATOM_radix, r);
		    }
		    if ( !formatInteger(NULL, 0, arg, c == 'r', &i, (Buffer)&b) )
		      FMT_EXEPTION();
		  }
		  clearNumber(&i);
		  rc = oututf80(&state, baseBuffer(&b, char));
		  discardBuffer(&b);
		  if ( !rc )
		    goto out;
		  here++;
		  break;
		}
	      case 's':			/* string */
		{ PL_chars_t txt;

		  NEED_ARG;
		  if ( !PL_get_text(argv, &txt, CVT_LIST|CVT_STRING) &&
		       !PL_get_text(argv, &txt, CVT_ATOM) ) /* SICStus compat */
		    FMT_ARG("s", argv);
		  rc = outtext(&state, &txt);
		  SHIFT;
		  if ( !rc )
		    goto out;
		  here++;
		  break;
		}
	      case 'i':			/* ignore */
		{ NEED_ARG;
		  SHIFT;
		  here++;
		  break;
		}
		{ Func f;
		  char buf[BUFSIZE];
		  char *str;

	      case 'k':			/* write_canonical */
		  f = pl_write_canonical;
	          goto pl_common;
	      case 'p':			/* print */
		  f = pl_print;
	          goto pl_common;
	      case 'q':			/* writeq */
		  f = pl_writeq;
	          goto pl_common;
	      case 'w':			/* write */
		  f = pl_write;
		  pl_common:

		  NEED_ARG;
		  if ( state.pending_rubber )
		  { size_t bufsize = BUFSIZE;

		    str = buf;
		    tellString(&str, &bufsize, ENC_UTF8);
		    rc = (*f)(argv);
		    toldString();
		    if ( !rc )
		      goto out;
		    oututf8(&state, str, bufsize);
		    if ( str != buf )
		      free(str);
		  } else
		  { if ( fd->position &&
			 fd->position->linepos == state.column )
		    { IOSTREAM *old = Scurout;

		      Scurout = fd;
		      rc = (int)(*f)(argv);
		      Scurout = old;
		      if ( !rc )
			goto out;

		      state.column = fd->position->linepos;
		    } else
		    { size_t bufsize = BUFSIZE;

		      str = buf;
		      tellString(&str, &bufsize, ENC_UTF8);
		      rc = (*f)(argv);
		      toldString();
		      if ( !rc )
		        goto out;
		      oututf8(&state, str, bufsize);
		      if ( str != buf )
			free(str);
		    }
		  }
		  SHIFT;
		  here++;
		  break;
		}
	      case 'W':			/* write_term(Value, Options) */
	       { char buf[BUFSIZE];
		 char *str;

		 if ( argc < 2 )
		 { FMT_ERROR("not enough arguments");
		 }
		 if ( state.pending_rubber )
		  { size_t bufsize = BUFSIZE;

		    str = buf;
		    tellString(&str, &bufsize, ENC_UTF8);
		    rc = (int)pl_write_term(argv, argv+1);
		    toldString();
		    if ( !rc )
		      goto out;
		    oututf8(&state, str, bufsize);
		    if ( str != buf )
		      free(str);
		  } else
		  { if ( fd->position &&
			 fd->position->linepos == state.column )
		    { IOSTREAM *old = Scurout;

		      Scurout = fd;
		      rc = (int)pl_write_term(argv, argv+1);
		      Scurout = old;
		      if ( !rc )
			goto out;

		      state.column = fd->position->linepos;
		    } else
		    { size_t bufsize = BUFSIZE;

		      str = buf;
		      tellString(&str, &bufsize, ENC_UTF8);
		      rc = (int)pl_write_term(argv, argv+1);
		      if ( !rc )
			goto out;
		      toldString();
		      oututf8(&state, str, bufsize);
		      if ( str != buf )
			free(str);
		    }
		  }
		  SHIFT;
		  SHIFT;
		  here++;
		  break;
	       }
	      case '@':
	        { char buf[BUFSIZE];
		  char *str = buf;
		  size_t bufsize = BUFSIZE;
		  term_t ex = 0;
		  int rval;

		  if ( argc < 1 )
		  { FMT_ERROR("not enough arguments");
		  }
		  tellString(&str, &bufsize, ENC_UTF8);
		  rval = callProlog(m, argv, PL_Q_CATCH_EXCEPTION, &ex);
		  toldString();
		  oututf8(&state, str, bufsize);
		  if ( str != buf )
		    free(str);

		  if ( !rval )
		  { Sunlock(fd);

		    if ( ex )
		      return PL_raise_exception(ex);
		    else
		      fail;
		  }

		  SHIFT;
		  here++;
		  break;
	        }
	      case '~':			/* ~ */
		{ rc = outchr(&state, '~');
		  if ( !rc )
		    goto out;
		  here++;
		  break;
		}
	      case 'n':			/* \n */
	      case 'N':			/* \n if not on newline */
		{ if ( arg == DEFAULT )
		    arg = 1;
		  if ( c == 'N' && state.column == 0 )
		    arg--;
		  while( arg-- > 0 )
                  { rc = outchr(&state, '\n');
		    if ( !rc )
		      goto out;
                  }
		  here++;
		  break;
		}
	      case 't':			/* insert tab */
		{ if ( state.pending_rubber >= MAXRUBBER )
		    FMT_ERROR("Too many tab stops");

		  state.rub[state.pending_rubber].where = state.buffered;
		  state.rub[state.pending_rubber].pad   =
					(arg == DEFAULT ? (pl_wchar_t)' '
							: (pl_wchar_t)arg);
		  state.rub[state.pending_rubber].size = 0;
		  state.pending_rubber++;
		  here++;
		  break;
		}
	      case '|':			/* set tab */
		{ int stop;

		  if ( arg == DEFAULT )
		    arg = state.column;
	      case '+':			/* tab relative */
		  if ( arg == DEFAULT )
		    arg = 8;
		  stop = (c == '+' ? tab_stop + arg : arg);

		  if ( state.pending_rubber == 0 ) /* nothing to distribute */
		  { state.rub[0].where = state.buffered;
		    state.rub[0].pad = ' ';
		    state.pending_rubber++;
		  }
		  distribute_rubber(state.rub,
				    state.pending_rubber,
				    stop - state.column);
		  emit_rubber(&state);

		  state.column = tab_stop = stop;
		  here++;
		  break;
		}
	      default:
	      { term_t ex = PL_new_term_ref();

		Sunlock(fd);
		PL_put_atom(ex, codeToAtom(c));
		return PL_error("format", 2, NULL, ERR_EXISTENCE,
				PL_new_atom("format_character"),
				ex);
	      }
	    }
	  }
	  break;			/* the '~' switch */
	}
      default:
	{ rc = outchr(&state, c);
	  if ( !rc )
	    goto out;
	  here++;
	  break;
	}
    }
  }

  if ( state.pending_rubber )		/* not closed ~t: flush out */
    emit_rubber(&state);

out:
  Sunlock(fd);

  return rc;
}


static void
distribute_rubber(struct rubber *r, int rn, int space)
{ if ( space > 0 )
  { int s = space / rn;
    int n, m;

    for(n=0; n < rn; n++)		/* give them equal size */
      r[n].size = s;
					/* distribute from the center */
    space -= s*rn;
    for(m = rn / 2, n = 0; space; n++, space--)
    { r[m + (n % 2 ? n : -n)].size++;
    }
  } else
  { int n;

    for(n=0; n < rn; n++)		/* set all rubber to 0 */
      r[n].size = 0;
  }
}


static int
emit_rubber(format_state *state)
{ const char *s = baseBuffer(&state->buffer, char);
  const char *e = &s[entriesBuffer(&state->buffer, char)];
  struct rubber *r = state->rub;
  int rn = state->pending_rubber;
  size_t j;

  for(j = 0; s <= e; j++)
  { int chr;

    if ( rn && r->where == j )
    { size_t n;

      for(n=0; n<r->size; n++)
      { if ( Sputcode(r->pad, state->out) < 0 )
	  return FALSE;
      }
      r++;
      rn--;
    }

    if ( s < e )
    { s = utf8_get_char(s, &chr);
      if ( Sputcode(chr, state->out) < 0 )
	return FALSE;
    } else
      break;
  }

  discardBuffer(&state->buffer);
  initBuffer(&state->buffer);
  state->buffered = 0;
  state->pending_rubber = 0;

  return TRUE;
}


/*  format an integer according to  a  number  of  modifiers  at various
    radius.   `split'  is a boolean asking to put ',' between each group
    of three digits (e.g. 67,567,288).  `div' askes to divide the number
    by radix^`div' before printing.   `radix'  is  the  radix  used  for
    conversion.  `n' is the number to be converted.

 ** Fri Aug 19 22:26:41 1988  jan@swivax.UUCP (Jan Wielemaker)  */

static void
lappend(const wchar_t *l, int def, Buffer out)
{ if ( l )
  { const wchar_t *e = l+wcslen(l);

    while (--e >= l)
    { int c = *e;

      if ( c < 128 )
      { addBuffer(out, c, char);
      } else
      { char buf[6];
	char *e8, *s;

	e8=utf8_put_char(buf, c);
	for(s=e8; --s>=buf; )		/* must be reversed as we reverse */
	{ addBuffer(out, *s, char);	/* in the end */
	}
      }
    }
  } else
  { addBuffer(out, def, char);
  }
}

static void
revert_string(char *s, size_t len)
{ char *e = &s[len-1];

  for(; e>s; s++,e--)
  { int c = *e;

    *e = *s;
    *s = c;
  }
}

static char *
formatInteger(PL_locale *locale, int div, int radix, bool smll, Number i,
	     Buffer out)
{ const char *grouping = NULL;

  if ( !locale )
  { locale = &prolog_locale;
  } else
  { if ( locale->grouping && locale->grouping[0] &&
	 locale->thousands_sep && locale->thousands_sep[0] )
      grouping = locale->grouping;
  }

  switch(i->type)
  { case V_INTEGER:
    { int64_t n = i->value.i;

      if ( n == 0 && div == 0 )
      { addBuffer(out, '0', char);
      } else
      { int before = FALSE;			/* before decimal point */
	int negative = FALSE;
	int gsize = 0;
	int dweight;

	negative = (n < 0);

	while( n != 0 || div >= 0 )
	{ if ( div-- == 0 && !before )
	  { if ( !isEmptyBuffer(out) )
	      lappend(locale->decimal_point, '.', out);
	    before = TRUE;
	    if ( grouping )
	      gsize = grouping[0];
	  }

	  if ( !negative )
	    dweight = (int)(n % radix);
	  else
	    dweight = -(int)(n % -radix);

	  addBuffer(out, digitName(dweight, smll), char);
	  n /= radix;

	  if ( --gsize == 0 && n != 0 )
	  { lappend(locale->thousands_sep, ',', out);
	    if ( grouping[1] == 0 )
	      gsize = grouping[0];
	    else if ( grouping[1] == CHAR_MAX )
	      gsize = 0;
	    else
	      gsize = *++grouping;
	  }
	}
	if ( negative )
	  addBuffer(out, '-', char);
      }

      revert_string(baseBuffer(out, char), entriesBuffer(out, char));
      addBuffer(out, EOS, char);

      return baseBuffer(out, char);
    }
#ifdef O_GMP
    case V_MPZ:
    { GET_LD
      size_t len = mpz_sizeinbase(i->value.mpz, radix);
      char tmp[256];
      char *buf;
      int rc = TRUE;

      if ( len+2 > sizeof(tmp) )
	buf = PL_malloc(len+2);
      else
	buf = tmp;

      EXCEPTION_GUARDED({ LD->gmp.persistent++;
			  mpz_get_str(buf, radix, i->value.mpz);
			  LD->gmp.persistent--;
			},
			{ LD->gmp.persistent--;
			  rc = PL_rethrow();
			});
      if ( !rc )
	return NULL;

      if ( !smll && radix > 10 )
      { char *s;

	for(s=buf; *s; s++)
	  *s = toupper(*s);
      }

      if ( grouping || div > 0 )
      { int before = FALSE;			/* before decimal point */
	int gsize = 0;
	char *e = buf+strlen(buf)-1;

	while(e >= buf || div >= 0)
	{ if ( div-- == 0 && !before )
	  { if ( !isEmptyBuffer(out) )
	      lappend(locale->decimal_point, '.', out);
	    before = TRUE;
	    if ( grouping )
	      gsize = grouping[0];
	  }

	  addBuffer(out, *e, char);
	  e--;

	  if ( --gsize == 0 && e >= buf && *e != '-' )
	  { lappend(locale->thousands_sep, ',', out);
	    if ( grouping[1] == 0 )
	      gsize = grouping[0];
	    else if ( grouping[1] == CHAR_MAX )
	      gsize = 0;
	    else
	      gsize = *++grouping;
	  }
	}
	revert_string(baseBuffer(out, char), entriesBuffer(out, char));
      } else
      { addMultipleBuffer(out, buf, strlen(buf), char);
      }

      if ( buf != tmp )
	PL_free(buf);

      addBuffer(out, EOS, char);
      return baseBuffer(out, char);
    }
#endif /*O_GMP*/
    default:
      assert(0);
      return NULL;
  }
}

#if O_LOCALE

static int
countGroups(const char *grouping, int len)
{ int groups = 0;
  int gsize = grouping[0];

  while(len>0)
  { len -= gsize;

    if ( len > 0 )
      groups++;

    if ( grouping[1] == 0 )
    { if ( len > 1 )
	groups += (len-1)/grouping[0];
      return groups;
    } else if ( grouping[1] == CHAR_MAX )
    { return groups;
    } else
    { gsize = *++grouping;
    }
  }

  return groups;
}

static int
ths_to_utf8(char *u8, const wchar_t *s, size_t len)
{ char *e = u8+len-7;

  for( ; u8<e && *s; s++)
    u8 = utf8_put_char(u8,*s);
  *u8 = EOS;

  return *s == 0;
}

static int
same_decimal_point(PL_locale *l1, PL_locale *l2)
{ if ( l1->decimal_point && l2->decimal_point &&
       wcscmp(l1->decimal_point, l2->decimal_point) == 0 )
    return TRUE;
  if ( !l1->decimal_point && !l2->decimal_point )
    return TRUE;

  return FALSE;
}

static int
utf8_dp(PL_locale *l, char *s, int *len)
{ if ( l->decimal_point )
  { if ( !ths_to_utf8(s, l->decimal_point, 20) )
      return FALSE;
    *len = strlen(s);
  } else
  { *s++ = '.';
    *s = EOS;
    *len = 1;
  }

  return TRUE;
}


/* localizeDecimalPoint() replaces the decimal point as entered by the
   local sensitive print functions by the one in the specified locale.
   This is overly complicated. Needs more testing, in particular for
   locales with (in UTF-8) multibyte decimal points.
*/

static int
localizeDecimalPoint(PL_locale *locale, Buffer b)
{ if ( locale == GD->locale.default_locale ||
       same_decimal_point(GD->locale.default_locale, locale) )
    return TRUE;

  if ( locale->decimal_point && locale->decimal_point[0] )
  { char *s = baseBuffer(b, char);
    char *e;
    char dp[20];  int dplen;
    char ddp[20]; int ddplen;

    if ( !utf8_dp(locale, dp, &dplen) ||
	 !utf8_dp(GD->locale.default_locale, ddp, &ddplen) )
      return FALSE;

    if ( *s == '-' )
      s++;
    for(e=s; *e && isDigit(*e); e++)
      ;

    if ( strncmp(e, ddp, ddplen) == 0 )
    { if ( dplen == ddplen )
      { memcpy(e, dp, dplen);
      } else
      { char *ob = baseBuffer(b, char);
	if ( dplen > ddplen && !growBuffer(b, dplen-ddplen) )
	  return PL_no_memory();
	e += baseBuffer(b, char) - ob;

	memmove(&e[dplen-ddplen], e, strlen(e)+1);
	memcpy(e, dp, dplen);
      }
    }
  }

  return TRUE;
}

static int
groupDigits(PL_locale *locale, Buffer b)
{ if ( locale->thousands_sep && locale->thousands_sep[0] &&
       locale->grouping && locale->grouping[0] )
  { char *s = baseBuffer(b, char);
    char *e;
    int groups;

    if ( *s == '-' )
      s++;
    for(e=s; *e && isDigit(*e); e++)
      ;

    groups = countGroups(locale->grouping, (int)(e-s));

    if ( groups > 0 )
    { char *o;
      char *grouping = locale->grouping;
      int gsize = grouping[0];
      char ths[20];
      int thslen;

      if ( !ths_to_utf8(ths, locale->thousands_sep, sizeof(ths)) )
	return FALSE;
      thslen = strlen(ths);

      if ( !growBuffer(b, thslen*groups) )
	return PL_no_memory();
      memmove(&e[groups*thslen], e, strlen(e)+1);

      e--;
      for(o=e+groups*thslen; e>=s; )
      { *o-- = *e--;
	if ( --gsize == 0 && e>=s )
	{ o -= thslen-1;
	  memcpy(o, ths, thslen);
	  o--;
	  if ( grouping[1] == 0 )
	    gsize = grouping[0];
	  else if ( grouping[1] == CHAR_MAX )
	    gsize = 0;
	  else
	    gsize = *++grouping;
	}
      }
    }
  }

  return TRUE;
}
#endif


/* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
formatFloat(PL_locale *locale, int how, int arg, Number f, Buffer out)

formats a floating point  number  to  a   buffer.  `How'  is  the format
specifier ([eEfgG]), `arg' the argument.

MPZ/MPQ numbers printed using the format specifier `f' are written using
the following algorithm, courtesy of Jan Burse:

  Given: A rational n/m
  Seeked: The ration rounded to d fractional digits.
  Algorithm: Compute (n*10^d+m//2)//m, and place period at d.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */

static char *
formatFloat(PL_locale *locale, int how, int arg, Number f, Buffer out)
{ if ( arg == DEFAULT )
    arg = 6;

  switch(f->type)
  {
#ifdef O_GMP
    mpf_t mpf;
    mpz_t t1, t2;
    int neg;

    case V_MPZ:
    { switch(how)
      { case 'f':
        { mpz_init(t1);
          mpz_init(t2);
          mpz_ui_pow_ui(t1, 10, arg);
          mpz_mul(t1, f->value.mpz, t1);
          neg = (mpz_cmp_ui(t1, 0) < 0) ? 1 : 0;
          mpz_abs(t1, t1);
          goto print_mpz;
        }
        case 'e':
	case 'E':
        case 'g':
	case 'G':
        { mpf_init2(mpf, arg*4);
          mpf_set_z(mpf, f->value.mpz);
          goto print_mpf;
        }
      }
    }
    case V_MPQ:
    { char tmp[12];
      int size;
      int written = 0;
      int fbits;
      int digits = 0;
      int padding;

      switch(how)
      { case 'f':
        { mpz_init(t1);
          mpz_init(t2);
          mpz_ui_pow_ui(t1, 10, arg);
          mpz_mul(t1, mpq_numref(f->value.mpq), t1);
          mpz_tdiv_q_ui(t2, mpq_denref(f->value.mpq), 2);
          if (mpq_cmp_ui(f->value.mpq, 0, 1) < 0)
          { mpz_sub(t1, t1, t2);
            neg=1;
          } else
          { mpz_add(t1, t1, t2);
            neg=0;
          }
          mpz_tdiv_q(t1, t1, mpq_denref(f->value.mpq));
          mpz_abs(t1, t1);

        print_mpz:

          if (mpz_cmp_ui(t1, 0) != 0)
          { size = mpz_sizeinbase(t1, 10) + 1; /* reserve for <null> */
            if ( !growBuffer(out, size) )
            { PL_no_memory();
              return NULL;
            }
            digits = written = gmp_snprintf(baseBuffer(out, char), size, "%Zd", t1);
          }

          size = digits;
          if (neg) size++;               /* leading - */
          if (arg) size++;               /* decimal point */
          if (digits <= arg)             /* leading '0's */
          { padding = (arg-digits+1);
            size += padding;
          } else
          { padding = 0;
          }
          size++;                        /* NULL terminator */

          if ( !growBuffer(out, size) )
          { PL_no_memory();
            return NULL;
          }

          if (!digits)
          { memset(out->base, '\0', 1);
          }

          if (neg)
          { memmove(out->base+1, out->base, digits+1);
            memset(out->base, '-', 1);
            written++;
          }

          if (padding)
          { memmove(out->base+neg+padding, out->base+neg, written-neg+1);
            memset(out->base+neg, '0', padding);
            written += padding;
          }

          if (arg)
          { memmove(out->base+written-(arg-1), out->base+written-arg, arg+1);
            if ( locale->decimal_point && locale->decimal_point[0] )
              *(out->base+written-arg) = locale->decimal_point[0];
            else
              *(out->base+written-arg) = '.';
            written++;
          }

          out->top = out->base + written;
          mpz_clear(t1);
          mpz_clear(t2);
          break;
        }
        case 'e':
	case 'E':
        case 'g':
	case 'G':
        { switch(how)
          { case 'g':
            case 'G':
            { mpz_t iv;
              mpz_init(iv);
              mpz_set_q(iv, f->value.mpq);
              fbits = (int)mpz_sizeinbase(iv, 2) + 4*arg;
              mpz_clear(iv);
              break;
            }
            default:
              fbits = 4*arg;
          }
          mpf_init2(mpf, fbits);
          mpf_set_q(mpf, f->value.mpq);

        print_mpf:
          Ssprintf(tmp, "%%.%dF%c", arg, how);
          size = 0;
          written = arg+4;
          while(written >= size)
          { size = written+1;

            if ( !growBuffer(out, size) )	/* reserve for -.e<null> */
            { PL_no_memory();
              return NULL;
            }
            written = gmp_snprintf(baseBuffer(out, char), size, tmp, mpf);
          }
          mpf_clear(mpf);
          out->top = out->base + written;

          break;
        }
      }
      break;
    }
#endif
    case V_INTEGER:
      promoteToFloatNumber(f);
      /*FALLTHROUGH*/
    case V_FLOAT:
    { char tmp[12];
      int written = arg+20;
      int size = 0;

      Ssprintf(tmp, "%%.%d%c", arg, how);
      while(written >= size)
      { size = written+1;

	if ( !growBuffer(out, size) )
	{ PL_no_memory();
	  return NULL;
	}
	written = snprintf(baseBuffer(out, char), size, tmp, f->value.f);
      }
      out->top = out->base + written;

      break;
    }
    default:
      assert(0);
      return NULL;
  }

#if O_LOCALE
  if ( locale )
  { if ( !localizeDecimalPoint(locale, out) ||
	 !groupDigits(locale, out) )
      return NULL;
  }
#endif

  return baseBuffer(out, char);
}

//! @}