1/* Internal header for parsing printf format strings.
2   Copyright (C) 1995-1999, 2000 Free Software Foundation, Inc.
3   This file is part of th GNU C Library.
4
5   The GNU C Library is free software; you can redistribute it and/or
6   modify it under the terms of the GNU Lesser General Public
7   License as published by the Free Software Foundation; either
8   version 2.1 of the License, or (at your option) any later version.
9
10   The GNU C Library is distributed in the hope that it will be useful,
11   but WITHOUT ANY WARRANTY; without even the implied warranty of
12   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13   Lesser General Public License for more details.
14
15   You should have received a copy of the GNU Lesser General Public
16   License along with the GNU C Library; if not, write to the Free
17   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
18   02111-1307 USA.  */
19
20#include <ctype.h>
21#include <limits.h>
22#include <printf.h>
23#include <stdint.h>
24#include <stddef.h>
25#include <string.h>
26
27#define NDEBUG 1
28#include <assert.h>
29
30
31struct printf_spec
32  {
33    /* Information parsed from the format spec.  */
34    struct printf_info info;
35
36    /* Pointers into the format string for the end of this format
37       spec and the next (or to the end of the string if no more).  */
38    const UCHAR_T *end_of_fmt, *next_fmt;
39
40    /* Position of arguments for precision and width, or -1 if `info' has
41       the constant value.  */
42    int prec_arg, width_arg;
43
44    int data_arg;		/* Position of data argument.  */
45    int data_arg_type;		/* Type of first argument.  */
46    /* Number of arguments consumed by this format specifier.  */
47    size_t ndata_args;
48  };
49
50
51/* The various kinds off arguments that can be passed to printf.  */
52union printf_arg
53  {
54    unsigned char pa_char;
55    wchar_t pa_wchar;
56    short int pa_short_int;
57    int pa_int;
58    long int pa_long_int;
59    long long int pa_long_long_int;
60    unsigned short int pa_u_short_int;
61    unsigned int pa_u_int;
62    unsigned long int pa_u_long_int;
63    unsigned long long int pa_u_long_long_int;
64    float pa_float;
65    double pa_double;
66    long double pa_long_double;
67    const char *pa_string;
68    const wchar_t *pa_wstring;
69    void *pa_pointer;
70  };
71
72
73/* Read a simple integer from a string and update the string pointer.
74   It is assumed that the first character is a digit.  */
75static inline unsigned int
76read_int (const UCHAR_T * *pstr)
77{
78  unsigned int retval = **pstr - L_('0');
79
80  while (ISDIGIT (*++(*pstr)))
81    {
82      retval *= 10;
83      retval += **pstr - L_('0');
84    }
85
86  return retval;
87}
88
89
90
91/* Find the next spec in FORMAT, or the end of the string.  Returns
92   a pointer into FORMAT, to a '%' or a '\0'.  */
93static inline const UCHAR_T *
94#ifdef COMPILE_WPRINTF
95find_spec (const UCHAR_T *format)
96#else
97find_spec (const UCHAR_T *format, mbstate_t *ps)
98#endif
99{
100#ifdef COMPILE_WPRINTF
101  return (const UCHAR_T *) __wcschrnul ((const CHAR_T *) format, L'%');
102#else
103  while (*format != L_('\0') && *format != L_('%'))
104    {
105      int len;
106
107      /* Remove any hints of a wrong encoding.  */
108      ps->__count = 0;
109      if (! isascii (*format) && (len = __mbrlen (format, MB_CUR_MAX, ps)) > 0)
110	format += len;
111      else
112	++format;
113    }
114  return format;
115#endif
116}
117
118
119/* These are defined in reg-printf.c.  */
120extern printf_arginfo_function *__printf_arginfo_table[];
121extern printf_function **__printf_function_table;
122
123
124/* FORMAT must point to a '%' at the beginning of a spec.  Fills in *SPEC
125   with the parsed details.  POSN is the number of arguments already
126   consumed.  At most MAXTYPES - POSN types are filled in TYPES.  Return
127   the number of args consumed by this spec; *MAX_REF_ARG is updated so it
128   remains the highest argument index used.  */
129static inline size_t
130#ifdef COMPILE_WPRINTF
131parse_one_spec (const UCHAR_T *format, size_t posn, struct printf_spec *spec,
132		size_t *max_ref_arg)
133#else
134parse_one_spec (const UCHAR_T *format, size_t posn, struct printf_spec *spec,
135		size_t *max_ref_arg, mbstate_t *ps)
136#endif
137{
138  unsigned int n;
139  size_t nargs = 0;
140
141  /* Skip the '%'.  */
142  ++format;
143
144  /* Clear information structure.  */
145  spec->data_arg = -1;
146  spec->info.alt = 0;
147  spec->info.space = 0;
148  spec->info.left = 0;
149  spec->info.showsign = 0;
150  spec->info.group = 0;
151  spec->info.i18n = 0;
152  spec->info.pad = ' ';
153  spec->info.wide = sizeof (UCHAR_T) > 1;
154
155  /* Test for positional argument.  */
156  if (ISDIGIT (*format))
157    {
158      const UCHAR_T *begin = format;
159
160      n = read_int (&format);
161
162      if (n > 0 && *format == L_('$'))
163	/* Is positional parameter.  */
164	{
165	  ++format;		/* Skip the '$'.  */
166	  spec->data_arg = n - 1;
167	  *max_ref_arg = MAX (*max_ref_arg, n);
168	}
169      else
170	/* Oops; that was actually the width and/or 0 padding flag.
171	   Step back and read it again.  */
172	format = begin;
173    }
174
175  /* Check for spec modifiers.  */
176  do
177    {
178      switch (*format)
179	{
180	case L_(' '):
181	  /* Output a space in place of a sign, when there is no sign.  */
182	  spec->info.space = 1;
183	  continue;
184	case L_('+'):
185	  /* Always output + or - for numbers.  */
186	  spec->info.showsign = 1;
187	  continue;
188	case L_('-'):
189	  /* Left-justify things.  */
190	  spec->info.left = 1;
191	  continue;
192	case L_('#'):
193	  /* Use the "alternate form":
194	     Hex has 0x or 0X, FP always has a decimal point.  */
195	  spec->info.alt = 1;
196	  continue;
197	case L_('0'):
198	  /* Pad with 0s.  */
199	  spec->info.pad = '0';
200	  continue;
201	case L_('\''):
202	  /* Show grouping in numbers if the locale information
203	     indicates any.  */
204	  spec->info.group = 1;
205	  continue;
206	case L_('I'):
207	  /* Use the internationalized form of the output.  Currently
208	     means to use the `outdigits' of the current locale.  */
209	  spec->info.i18n = 1;
210	  continue;
211	default:
212	  break;
213	}
214      break;
215    }
216  while (*++format);
217
218  if (spec->info.left)
219    spec->info.pad = ' ';
220
221  /* Get the field width.  */
222  spec->width_arg = -1;
223  spec->info.width = 0;
224  if (*format == L_('*'))
225    {
226      /* The field width is given in an argument.
227	 A negative field width indicates left justification.  */
228      const UCHAR_T *begin = ++format;
229
230      if (ISDIGIT (*format))
231	{
232	  /* The width argument might be found in a positional parameter.  */
233	  n = read_int (&format);
234
235	  if (n > 0 && *format == L_('$'))
236	    {
237	      spec->width_arg = n - 1;
238	      *max_ref_arg = MAX (*max_ref_arg, n);
239	      ++format;		/* Skip '$'.  */
240	    }
241	}
242
243      if (spec->width_arg < 0)
244	{
245	  /* Not in a positional parameter.  Consume one argument.  */
246	  spec->width_arg = posn++;
247	  ++nargs;
248	  format = begin;	/* Step back and reread.  */
249	}
250    }
251  else if (ISDIGIT (*format))
252    /* Constant width specification.  */
253    spec->info.width = read_int (&format);
254
255  /* Get the precision.  */
256  spec->prec_arg = -1;
257  /* -1 means none given; 0 means explicit 0.  */
258  spec->info.prec = -1;
259  if (*format == L_('.'))
260    {
261      ++format;
262      if (*format == L_('*'))
263	{
264	  /* The precision is given in an argument.  */
265	  const UCHAR_T *begin = ++format;
266
267	  if (ISDIGIT (*format))
268	    {
269	      n = read_int (&format);
270
271	      if (n > 0 && *format == L_('$'))
272		{
273		  spec->prec_arg = n - 1;
274		  *max_ref_arg = MAX (*max_ref_arg, n);
275		  ++format;
276		}
277	    }
278
279	  if (spec->prec_arg < 0)
280	    {
281	      /* Not in a positional parameter.  */
282	      spec->prec_arg = posn++;
283	      ++nargs;
284	      format = begin;
285	    }
286	}
287      else if (ISDIGIT (*format))
288	spec->info.prec = read_int (&format);
289      else
290	/* "%.?" is treated like "%.0?".  */
291	spec->info.prec = 0;
292    }
293
294  /* Check for type modifiers.  */
295  spec->info.is_long_double = 0;
296  spec->info.is_short = 0;
297  spec->info.is_long = 0;
298  spec->info.is_char = 0;
299
300  switch (*format++)
301    {
302    case L_('h'):
303      /* ints are short ints or chars.  */
304      if (*format != L_('h'))
305	spec->info.is_short = 1;
306      else
307	{
308	  ++format;
309	  spec->info.is_char = 1;
310	}
311      break;
312    case L_('l'):
313      /* ints are long ints.  */
314      spec->info.is_long = 1;
315      if (*format != L_('l'))
316	break;
317      ++format;
318      /* FALLTHROUGH */
319    case L_('L'):
320      /* doubles are long doubles, and ints are long long ints.  */
321    case L_('q'):
322      /* 4.4 uses this for long long.  */
323      spec->info.is_long_double = 1;
324      break;
325    case L_('z'):
326    case L_('Z'):
327      /* ints are size_ts.  */
328      assert (sizeof (size_t) <= sizeof (unsigned long long int));
329#if LONG_MAX != LONG_LONG_MAX
330      spec->info.is_long_double = sizeof (size_t) > sizeof (unsigned long int);
331#endif
332      spec->info.is_long = sizeof (size_t) > sizeof (unsigned int);
333      break;
334    case L_('t'):
335      assert (sizeof (ptrdiff_t) <= sizeof (long long int));
336#if LONG_MAX != LONG_LONG_MAX
337      spec->info.is_long_double = (sizeof (ptrdiff_t) > sizeof (long int));
338#endif
339      spec->info.is_long = sizeof (ptrdiff_t) > sizeof (int);
340      break;
341    case L_('j'):
342      assert (sizeof (uintmax_t) <= sizeof (unsigned long long int));
343#if LONG_MAX != LONG_LONG_MAX
344      spec->info.is_long_double = (sizeof (uintmax_t)
345				   > sizeof (unsigned long int));
346#endif
347      spec->info.is_long = sizeof (uintmax_t) > sizeof (unsigned int);
348      break;
349    default:
350      /* Not a recognized modifier.  Backup.  */
351      --format;
352      break;
353    }
354
355  /* Get the format specification.  */
356  spec->info.spec = (wchar_t) *format++;
357  if (__printf_function_table != NULL
358      && spec->info.spec <= UCHAR_MAX
359      && __printf_arginfo_table[spec->info.spec] != NULL)
360    /* We don't try to get the types for all arguments if the format
361       uses more than one.  The normal case is covered though.  */
362    spec->ndata_args = (*__printf_arginfo_table[spec->info.spec])
363      (&spec->info, 1, &spec->data_arg_type);
364  else
365    {
366      /* Find the data argument types of a built-in spec.  */
367      spec->ndata_args = 1;
368
369      switch (spec->info.spec)
370	{
371	case L'i':
372	case L'd':
373	case L'u':
374	case L'o':
375	case L'X':
376	case L'x':
377#if LONG_MAX != LONG_LONG_MAX
378	  if (spec->info.is_long_double)
379	    spec->data_arg_type = PA_INT|PA_FLAG_LONG_LONG;
380	  else
381#endif
382	    if (spec->info.is_long)
383	      spec->data_arg_type = PA_INT|PA_FLAG_LONG;
384	    else if (spec->info.is_short)
385	      spec->data_arg_type = PA_INT|PA_FLAG_SHORT;
386	    else if (spec->info.is_char)
387	      spec->data_arg_type = PA_CHAR;
388	    else
389	      spec->data_arg_type = PA_INT;
390	  break;
391	case L'e':
392	case L'E':
393	case L'f':
394	case L'F':
395	case L'g':
396	case L'G':
397	case L'a':
398	case L'A':
399	  if (spec->info.is_long_double)
400	    spec->data_arg_type = PA_DOUBLE|PA_FLAG_LONG_DOUBLE;
401	  else
402	    spec->data_arg_type = PA_DOUBLE;
403	  break;
404	case L'c':
405	  spec->data_arg_type = PA_CHAR;
406	  break;
407	case L'C':
408	  spec->data_arg_type = PA_WCHAR;
409	  break;
410	case L's':
411	  spec->data_arg_type = PA_STRING;
412	  break;
413	case L'S':
414	  spec->data_arg_type = PA_WSTRING;
415	  break;
416	case L'p':
417	  spec->data_arg_type = PA_POINTER;
418	  break;
419	case L'n':
420	  spec->data_arg_type = PA_INT|PA_FLAG_PTR;
421	  break;
422
423	case L'm':
424	default:
425	  /* An unknown spec will consume no args.  */
426	  spec->ndata_args = 0;
427	  break;
428	}
429    }
430
431  if (spec->data_arg == -1 && spec->ndata_args > 0)
432    {
433      /* There are args consumed, but no positional spec.  Use the
434	 next sequential arg position.  */
435      spec->data_arg = posn;
436      nargs += spec->ndata_args;
437    }
438
439  if (spec->info.spec == L'\0')
440    /* Format ended before this spec was complete.  */
441    spec->end_of_fmt = spec->next_fmt = format - 1;
442  else
443    {
444      /* Find the next format spec.  */
445      spec->end_of_fmt = format;
446#ifdef COMPILE_WPRINTF
447      spec->next_fmt = find_spec (format);
448#else
449      spec->next_fmt = find_spec (format, ps);
450#endif
451    }
452
453  return nargs;
454}
455