lib/printf-parse.c

   1 /* Formatted output to strings.
   2    Copyright (C) 1999-2000, 2002-2003, 2006-2010 Free Software Foundation, Inc.
   3
   4    This program is free software; you can redistribute it and/or modify
   5    it under the terms of the GNU General Public License as published by
   6    the Free Software Foundation; either version 2, or (at your option)
   7    any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12    GNU General Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License along
  15    with this program; if not, write to the Free Software Foundation,
  16    Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.  */
  17
  18 /* This file can be parametrized with the following macros:
  19      CHAR_T             The element type of the format string.
  20      CHAR_T_ONLY_ASCII  Set to 1 to enable verification that all characters
  21                         in the format string are ASCII.
  22      DIRECTIVE          Structure denoting a format directive.
  23                         Depends on CHAR_T.
  24      DIRECTIVES         Structure denoting the set of format directives of a
  25                         format string.  Depends on CHAR_T.
  26      PRINTF_PARSE       Function that parses a format string.
  27                         Depends on CHAR_T.
  28      STATIC             Set to 'static' to declare the function static.
  29      ENABLE_UNISTDIO    Set to 1 to enable the unistdio extensions.  */
  30
  31 #ifndef PRINTF_PARSE
  32 # include <config.h>
  33 #endif
  34
  35 /* Specification.  */
  36 #ifndef PRINTF_PARSE
  37 # include "printf-parse.h"
  38 #endif
  39
  40 /* Default parameters.  */
  41 #ifndef PRINTF_PARSE
  42 # define PRINTF_PARSE printf_parse
  43 # define CHAR_T char
  44 # define DIRECTIVE char_directive
  45 # define DIRECTIVES char_directives
  46 #endif
  47
  48 /* Get size_t, NULL.  */
  49 #include <stddef.h>
  50
  51 /* Get intmax_t.  */
  52 #if defined IN_LIBINTL || defined IN_LIBASPRINTF
  53 # if HAVE_STDINT_H_WITH_UINTMAX
  54 #  include <stdint.h>
  55 # endif
  56 # if HAVE_INTTYPES_H_WITH_UINTMAX
  57 #  include <inttypes.h>
  58 # endif
  59 #else
  60 # include <stdint.h>
  61 #endif
  62
  63 /* malloc(), realloc(), free().  */
  64 #include <stdlib.h>
  65
  66 /* errno.  */
  67 #include <errno.h>
  68
  69 /* Checked size_t computations.  */
  70 #include "xsize.h"
  71
  72 #if CHAR_T_ONLY_ASCII
  73 /* c_isascii().  */
  74 # include "c-ctype.h"
  75 #endif
  76
  77 #ifdef STATIC
  78 STATIC
  79 #endif
  80 int
  81 PRINTF_PARSE (const CHAR_T *format, DIRECTIVES *d, arguments *a)
  82 {
  83   const CHAR_T *cp = format;            /* pointer into format */
  84   size_t arg_posn = 0;          /* number of regular arguments consumed */
  85   size_t d_allocated;                   /* allocated elements of d->dir */
  86   size_t a_allocated;                   /* allocated elements of a->arg */
  87   size_t max_width_length = 0;
  88   size_t max_precision_length = 0;
  89
  90   d->count = 0;
  91   d_allocated = 1;
  92   d->dir = (DIRECTIVE *) malloc (d_allocated * sizeof (DIRECTIVE));
  93   if (d->dir == NULL)
  94     /* Out of memory.  */
  95     goto out_of_memory_1;
  96
  97   a->count = 0;
  98   a_allocated = 0;
  99   a->arg = NULL;
 100
 101 #define REGISTER_ARG(_index_,_type_) \
 102   {                                                                     \
 103     size_t n = (_index_);                                               \
 104     if (n >= a_allocated)                                               \
 105       {                                                                 \
 106         size_t memory_size;                                             \
 107         argument *memory;                                               \
 108                                                                         \
 109         a_allocated = xtimes (a_allocated, 2);                          \
 110         if (a_allocated <= n)                                           \
 111           a_allocated = xsum (n, 1);                                    \
 112         memory_size = xtimes (a_allocated, sizeof (argument));          \
 113         if (size_overflow_p (memory_size))                              \
 114           /* Overflow, would lead to out of memory.  */                 \
 115           goto out_of_memory;                                           \
 116         memory = (argument *) (a->arg                                   \
 117                                ? realloc (a->arg, memory_size)          \
 118                                : malloc (memory_size));                 \
 119         if (memory == NULL)                                             \
 120           /* Out of memory.  */                                         \
 121           goto out_of_memory;                                           \
 122         a->arg = memory;                                                \
 123       }                                                                 \
 124     while (a->count <= n)                                               \
 125       a->arg[a->count++].type = TYPE_NONE;                              \
 126     if (a->arg[n].type == TYPE_NONE)                                    \
 127       a->arg[n].type = (_type_);                                        \
 128     else if (a->arg[n].type != (_type_))                                \
 129       /* Ambiguous type for positional argument.  */                    \
 130       goto error;                                                       \
 131   }
 132
 133   while (*cp != '\0')
 134     {
 135       CHAR_T c = *cp++;
 136       if (c == '%')
 137         {
 138           size_t arg_index = ARG_NONE;
 139           DIRECTIVE *dp = &d->dir[d->count]; /* pointer to next directive */
 140
 141           /* Initialize the next directive.  */
 142           dp->dir_start = cp - 1;
 143           dp->flags = 0;
 144           dp->width_start = NULL;
 145           dp->width_end = NULL;
 146           dp->width_arg_index = ARG_NONE;
 147           dp->precision_start = NULL;
 148           dp->precision_end = NULL;
 149           dp->precision_arg_index = ARG_NONE;
 150           dp->arg_index = ARG_NONE;
 151
 152           /* Test for positional argument.  */
 153           if (*cp >= '0' && *cp <= '9')
 154             {
 155               const CHAR_T *np;
 156
 157               for (np = cp; *np >= '0' && *np <= '9'; np++)
 158                 ;
 159               if (*np == '$')
 160                 {
 161                   size_t n = 0;
 162
 163                   for (np = cp; *np >= '0' && *np <= '9'; np++)
 164                     n = xsum (xtimes (n, 10), *np - '0');
 165                   if (n == 0)
 166                     /* Positional argument 0.  */
 167                     goto error;
 168                   if (size_overflow_p (n))
 169                     /* n too large, would lead to out of memory later.  */
 170                     goto error;
 171                   arg_index = n - 1;
 172                   cp = np + 1;
 173                 }
 174             }
 175
 176           /* Read the flags.  */
 177           for (;;)
 178             {
 179               if (*cp == '\'')
 180                 {
 181                   dp->flags |= FLAG_GROUP;
 182                   cp++;
 183                 }
 184               else if (*cp == '-')
 185                 {
 186                   dp->flags |= FLAG_LEFT;
 187                   cp++;
 188                 }
 189               else if (*cp == '+')
 190                 {
 191                   dp->flags |= FLAG_SHOWSIGN;
 192                   cp++;
 193                 }
 194               else if (*cp == ' ')
 195                 {
 196                   dp->flags |= FLAG_SPACE;
 197                   cp++;
 198                 }
 199               else if (*cp == '#')
 200                 {
 201                   dp->flags |= FLAG_ALT;
 202                   cp++;
 203                 }
 204               else if (*cp == '0')
 205                 {
 206                   dp->flags |= FLAG_ZERO;
 207                   cp++;
 208                 }
 209 #if __GLIBC__ >= 2 && !defined __UCLIBC__
 210               else if (*cp == 'I')
 211                 {
 212                   dp->flags |= FLAG_LOCALIZED;
 213                   cp++;
 214                 }
 215 #endif
 216               else
 217                 break;
 218             }
 219
 220           /* Parse the field width.  */
 221           if (*cp == '*')
 222             {
 223               dp->width_start = cp;
 224               cp++;
 225               dp->width_end = cp;
 226               if (max_width_length < 1)
 227                 max_width_length = 1;
 228
 229               /* Test for positional argument.  */
 230               if (*cp >= '0' && *cp <= '9')
 231                 {
 232                   const CHAR_T *np;
 233
 234                   for (np = cp; *np >= '0' && *np <= '9'; np++)
 235                     ;
 236                   if (*np == '$')
 237                     {
 238                       size_t n = 0;
 239
 240                       for (np = cp; *np >= '0' && *np <= '9'; np++)
 241                         n = xsum (xtimes (n, 10), *np - '0');
 242                       if (n == 0)
 243                         /* Positional argument 0.  */
 244                         goto error;
 245                       if (size_overflow_p (n))
 246                         /* n too large, would lead to out of memory later.  */
 247                         goto error;
 248                       dp->width_arg_index = n - 1;
 249                       cp = np + 1;
 250                     }
 251                 }
 252               if (dp->width_arg_index == ARG_NONE)
 253                 {
 254                   dp->width_arg_index = arg_posn++;
 255                   if (dp->width_arg_index == ARG_NONE)
 256                     /* arg_posn wrapped around.  */
 257                     goto error;
 258                 }
 259               REGISTER_ARG (dp->width_arg_index, TYPE_INT);
 260             }
 261           else if (*cp >= '0' && *cp <= '9')
 262             {
 263               size_t width_length;
 264
 265               dp->width_start = cp;
 266               for (; *cp >= '0' && *cp <= '9'; cp++)
 267                 ;
 268               dp->width_end = cp;
 269               width_length = dp->width_end - dp->width_start;
 270               if (max_width_length < width_length)
 271                 max_width_length = width_length;
 272             }
 273
 274           /* Parse the precision.  */
 275           if (*cp == '.')
 276             {
 277               cp++;
 278               if (*cp == '*')
 279                 {
 280                   dp->precision_start = cp - 1;
 281                   cp++;
 282                   dp->precision_end = cp;
 283                   if (max_precision_length < 2)
 284                     max_precision_length = 2;
 285
 286                   /* Test for positional argument.  */
 287                   if (*cp >= '0' && *cp <= '9')
 288                     {
 289                       const CHAR_T *np;
 290
 291                       for (np = cp; *np >= '0' && *np <= '9'; np++)
 292                         ;
 293                       if (*np == '$')
 294                         {
 295                           size_t n = 0;
 296
 297                           for (np = cp; *np >= '0' && *np <= '9'; np++)
 298                             n = xsum (xtimes (n, 10), *np - '0');
 299                           if (n == 0)
 300                             /* Positional argument 0.  */
 301                             goto error;
 302                           if (size_overflow_p (n))
 303                             /* n too large, would lead to out of memory
 304                                later.  */
 305                             goto error;
 306                           dp->precision_arg_index = n - 1;
 307                           cp = np + 1;
 308                         }
 309                     }
 310                   if (dp->precision_arg_index == ARG_NONE)
 311                     {
 312                       dp->precision_arg_index = arg_posn++;
 313                       if (dp->precision_arg_index == ARG_NONE)
 314                         /* arg_posn wrapped around.  */
 315                         goto error;
 316                     }
 317                   REGISTER_ARG (dp->precision_arg_index, TYPE_INT);
 318                 }
 319               else
 320                 {
 321                   size_t precision_length;
 322
 323                   dp->precision_start = cp - 1;
 324                   for (; *cp >= '0' && *cp <= '9'; cp++)
 325                     ;
 326                   dp->precision_end = cp;
 327                   precision_length = dp->precision_end - dp->precision_start;
 328                   if (max_precision_length < precision_length)
 329                     max_precision_length = precision_length;
 330                 }
 331             }
 332
 333           {
 334             arg_type type;
 335
 336             /* Parse argument type/size specifiers.  */
 337             {
 338               int flags = 0;
 339
 340               for (;;)
 341                 {
 342                   if (*cp == 'h')
 343                     {
 344                       flags |= (1 << (flags & 1));
 345                       cp++;
 346                     }
 347                   else if (*cp == 'L')
 348                     {
 349                       flags |= 4;
 350                       cp++;
 351                     }
 352                   else if (*cp == 'l')
 353                     {
 354                       flags += 8;
 355                       cp++;
 356                     }
 357                   else if (*cp == 'j')
 358                     {
 359                       if (sizeof (intmax_t) > sizeof (long))
 360                         {
 361                           /* intmax_t = long long */
 362                           flags += 16;
 363                         }
 364                       else if (sizeof (intmax_t) > sizeof (int))
 365                         {
 366                           /* intmax_t = long */
 367                           flags += 8;
 368                         }
 369                       cp++;
 370                     }
 371                   else if (*cp == 'z' || *cp == 'Z')
 372                     {
 373                       /* 'z' is standardized in ISO C 99, but glibc uses 'Z'
 374                          because the warning facility in gcc-2.95.2 understands
 375                          only 'Z' (see gcc-2.95.2/gcc/c-common.c:1784).  */
 376                       if (sizeof (size_t) > sizeof (long))
 377                         {
 378                           /* size_t = long long */
 379                           flags += 16;
 380                         }
 381                       else if (sizeof (size_t) > sizeof (int))
 382                         {
 383                           /* size_t = long */
 384                           flags += 8;
 385                         }
 386                       cp++;
 387                     }
 388                   else if (*cp == 't')
 389                     {
 390                       if (sizeof (ptrdiff_t) > sizeof (long))
 391                         {
 392                           /* ptrdiff_t = long long */
 393                           flags += 16;
 394                         }
 395                       else if (sizeof (ptrdiff_t) > sizeof (int))
 396                         {
 397                           /* ptrdiff_t = long */
 398                           flags += 8;
 399                         }
 400                       cp++;
 401                     }
 402 #if defined __APPLE__ && defined __MACH__
 403                   /* On MacOS X 10.3, PRIdMAX is defined as "qd".
 404                      We cannot change it to "lld" because PRIdMAX must also
 405                      be understood by the system's printf routines.  */
 406                   else if (*cp == 'q')
 407                     {
 408                       if (64 / 8 > sizeof (long))
 409                         {
 410                           /* int64_t = long long */
 411                           flags += 16;
 412                         }
 413                       else
 414                         {
 415                           /* int64_t = long */
 416                           flags += 8;
 417                         }
 418                       cp++;
 419                     }
 420 #endif
 421 #if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__
 422                   /* On native Win32, PRIdMAX is defined as "I64d".
 423                      We cannot change it to "lld" because PRIdMAX must also
 424                      be understood by the system's printf routines.  */
 425                   else if (*cp == 'I' && cp[1] == '6' && cp[2] == '4')
 426                     {
 427                       if (64 / 8 > sizeof (long))
 428                         {
 429                           /* __int64 = long long */
 430                           flags += 16;
 431                         }
 432                       else
 433                         {
 434                           /* __int64 = long */
 435                           flags += 8;
 436                         }
 437                       cp += 3;
 438                     }
 439 #endif
 440                   else
 441                     break;
 442                 }
 443
 444               /* Read the conversion character.  */
 445               c = *cp++;
 446               switch (c)
 447                 {
 448                 case 'd': case 'i':
 449 #if HAVE_LONG_LONG_INT
 450                   /* If 'long long' exists and is larger than 'long':  */
 451                   if (flags >= 16 || (flags & 4))
 452                     type = TYPE_LONGLONGINT;
 453                   else
 454 #endif
 455                   /* If 'long long' exists and is the same as 'long', we parse
 456                      "lld" into TYPE_LONGINT.  */
 457                   if (flags >= 8)
 458                     type = TYPE_LONGINT;
 459                   else if (flags & 2)
 460                     type = TYPE_SCHAR;
 461                   else if (flags & 1)
 462                     type = TYPE_SHORT;
 463                   else
 464                     type = TYPE_INT;
 465                   break;
 466                 case 'o': case 'u': case 'x': case 'X':
 467 #if HAVE_LONG_LONG_INT
 468                   /* If 'long long' exists and is larger than 'long':  */
 469                   if (flags >= 16 || (flags & 4))
 470                     type = TYPE_ULONGLONGINT;
 471                   else
 472 #endif
 473                   /* If 'unsigned long long' exists and is the same as
 474                      'unsigned long', we parse "llu" into TYPE_ULONGINT.  */
 475                   if (flags >= 8)
 476                     type = TYPE_ULONGINT;
 477                   else if (flags & 2)
 478                     type = TYPE_UCHAR;
 479                   else if (flags & 1)
 480                     type = TYPE_USHORT;
 481                   else
 482                     type = TYPE_UINT;
 483                   break;
 484                 case 'f': case 'F': case 'e': case 'E': case 'g': case 'G':
 485                 case 'a': case 'A':
 486                   if (flags >= 16 || (flags & 4))
 487                     type = TYPE_LONGDOUBLE;
 488                   else
 489                     type = TYPE_DOUBLE;
 490                   break;
 491                 case 'c':
 492                   if (flags >= 8)
 493 #if HAVE_WINT_T
 494                     type = TYPE_WIDE_CHAR;
 495 #else
 496                     goto error;
 497 #endif
 498                   else
 499                     type = TYPE_CHAR;
 500                   break;
 501 #if HAVE_WINT_T
 502                 case 'C':
 503                   type = TYPE_WIDE_CHAR;
 504                   c = 'c';
 505                   break;
 506 #endif
 507                 case 's':
 508                   if (flags >= 8)
 509 #if HAVE_WCHAR_T
 510                     type = TYPE_WIDE_STRING;
 511 #else
 512                     goto error;
 513 #endif
 514                   else
 515                     type = TYPE_STRING;
 516                   break;
 517 #if HAVE_WCHAR_T
 518                 case 'S':
 519                   type = TYPE_WIDE_STRING;
 520                   c = 's';
 521                   break;
 522 #endif
 523                 case 'p':
 524                   type = TYPE_POINTER;
 525                   break;
 526                 case 'n':
 527 #if HAVE_LONG_LONG_INT
 528                   /* If 'long long' exists and is larger than 'long':  */
 529                   if (flags >= 16 || (flags & 4))
 530                     type = TYPE_COUNT_LONGLONGINT_POINTER;
 531                   else
 532 #endif
 533                   /* If 'long long' exists and is the same as 'long', we parse
 534                      "lln" into TYPE_COUNT_LONGINT_POINTER.  */
 535                   if (flags >= 8)
 536                     type = TYPE_COUNT_LONGINT_POINTER;
 537                   else if (flags & 2)
 538                     type = TYPE_COUNT_SCHAR_POINTER;
 539                   else if (flags & 1)
 540                     type = TYPE_COUNT_SHORT_POINTER;
 541                   else
 542                     type = TYPE_COUNT_INT_POINTER;
 543                   break;
 544 #if ENABLE_UNISTDIO
 545                 /* The unistdio extensions.  */
 546                 case 'U':
 547                   if (flags >= 16)
 548                     type = TYPE_U32_STRING;
 549                   else if (flags >= 8)
 550                     type = TYPE_U16_STRING;
 551                   else
 552                     type = TYPE_U8_STRING;
 553                   break;
 554 #endif
 555                 case '%':
 556                   type = TYPE_NONE;
 557                   break;
 558                 default:
 559                   /* Unknown conversion character.  */
 560                   goto error;
 561                 }
 562             }
 563
 564             if (type != TYPE_NONE)
 565               {
 566                 dp->arg_index = arg_index;
 567                 if (dp->arg_index == ARG_NONE)
 568                   {
 569                     dp->arg_index = arg_posn++;
 570                     if (dp->arg_index == ARG_NONE)
 571                       /* arg_posn wrapped around.  */
 572                       goto error;
 573                   }
 574                 REGISTER_ARG (dp->arg_index, type);
 575               }
 576             dp->conversion = c;
 577             dp->dir_end = cp;
 578           }
 579
 580           d->count++;
 581           if (d->count >= d_allocated)
 582             {
 583               size_t memory_size;
 584               DIRECTIVE *memory;
 585
 586               d_allocated = xtimes (d_allocated, 2);
 587               memory_size = xtimes (d_allocated, sizeof (DIRECTIVE));
 588               if (size_overflow_p (memory_size))
 589                 /* Overflow, would lead to out of memory.  */
 590                 goto out_of_memory;
 591               memory = (DIRECTIVE *) realloc (d->dir, memory_size);
 592               if (memory == NULL)
 593                 /* Out of memory.  */
 594                 goto out_of_memory;
 595               d->dir = memory;
 596             }
 597         }
 598 #if CHAR_T_ONLY_ASCII
 599       else if (!c_isascii (c))
 600         {
 601           /* Non-ASCII character.  Not supported.  */
 602           goto error;
 603         }
 604 #endif
 605     }
 606   d->dir[d->count].dir_start = cp;
 607
 608   d->max_width_length = max_width_length;
 609   d->max_precision_length = max_precision_length;
 610   return 0;
 611
 612 error:
 613   if (a->arg)
 614     free (a->arg);
 615   if (d->dir)
 616     free (d->dir);
 617   errno = EINVAL;
 618   return -1;
 619
 620 out_of_memory:
 621   if (a->arg)
 622     free (a->arg);
 623   if (d->dir)
 624     free (d->dir);
 625 out_of_memory_1:
 626   errno = ENOMEM;
 627   return -1;
 628 }
 629
 630 #undef PRINTF_PARSE
 631 #undef DIRECTIVES
 632 #undef DIRECTIVE
 633 #undef CHAR_T_ONLY_ASCII
 634 #undef CHAR_T