coreutils/printf.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * printf - format and print data
   4  *
   5  * Copyright 1999 Dave Cinege
   6  * Portions copyright (C) 1990-1996 Free Software Foundation, Inc.
   7  *
   8  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
   9  */
  10 /* Usage: printf format [argument...]
  11  *
  12  * A front end to the printf function that lets it be used from the shell.
  13  *
  14  * Backslash escapes:
  15  *
  16  * \" = double quote
  17  * \\ = backslash
  18  * \a = alert (bell)
  19  * \b = backspace
  20  * \c = produce no further output
  21  * \f = form feed
  22  * \n = new line
  23  * \r = carriage return
  24  * \t = horizontal tab
  25  * \v = vertical tab
  26  * \0ooo = octal number (ooo is 0 to 3 digits)
  27  * \xhhh = hexadecimal number (hhh is 1 to 3 digits)
  28  *
  29  * Additional directive:
  30  *
  31  * %b = print an argument string, interpreting backslash escapes
  32  *
  33  * The 'format' argument is re-used as many times as necessary
  34  * to convert all of the given arguments.
  35  *
  36  * David MacKenzie <djm@gnu.ai.mit.edu>
  37  */
  38 /* 19990508 Busy Boxed! Dave Cinege */
  39
  40 //config:config PRINTF
  41 //config:       bool "printf (4.1 kb)"
  42 //config:       default y
  43 //config:       help
  44 //config:       printf is used to format and print specified strings.
  45 //config:       It's similar to 'echo' except it has more options.
  46
  47 //applet:IF_PRINTF(APPLET_NOFORK(printf, printf, BB_DIR_USR_BIN, BB_SUID_DROP, printf))
  48
  49 //kbuild:lib-$(CONFIG_PRINTF) += printf.o
  50 //kbuild:lib-$(CONFIG_ASH_PRINTF)  += printf.o
  51 //kbuild:lib-$(CONFIG_HUSH_PRINTF) += printf.o
  52
  53 //usage:#define printf_trivial_usage
  54 //usage:       "FORMAT [ARG]..."
  55 //usage:#define printf_full_usage "\n\n"
  56 //usage:       "Format and print ARG(s) according to FORMAT (a-la C printf)"
  57 //usage:
  58 //usage:#define printf_example_usage
  59 //usage:       "$ printf \"Val=%d\\n\" 5\n"
  60 //usage:       "Val=5\n"
  61
  62 #include "libbb.h"
  63
  64 /* A note on bad input: neither bash 3.2 nor coreutils 6.10 stop on it.
  65  * They report it:
  66  *  bash: printf: XXX: invalid number
  67  *  printf: XXX: expected a numeric value
  68  *  bash: printf: 123XXX: invalid number
  69  *  printf: 123XXX: value not completely converted
  70  * but then they use 0 (or partially converted numeric prefix) as a value
  71  * and continue. They exit with 1 in this case.
  72  * Both accept insane field width/precision (e.g. %9999999999.9999999999d).
  73  * Both print error message and assume 0 if %*.*f width/precision is "bad"
  74  *  (but negative numbers are not "bad").
  75  * Both accept negative numbers for %u specifier.
  76  *
  77  * We try to be compatible.
  78  */
  79
  80 typedef void FAST_FUNC (*converter)(const char *arg, void *result);
  81
  82 static int multiconvert(const char *arg, void *result, converter convert)
  83 {
  84         if (*arg == '"' || *arg == '\'') {
  85                 arg = utoa((unsigned char)arg[1]);
  86         }
  87         errno = 0;
  88         convert(arg, result);
  89         if (errno) {
  90                 bb_error_msg("invalid number '%s'", arg);
  91                 return 1;
  92         }
  93         return 0;
  94 }
  95
  96 static void FAST_FUNC conv_strtoull(const char *arg, void *result)
  97 {
  98         /* Allow leading '+' - bb_strtoull() by itself does not allow it,
  99          * and probably shouldn't (other callers might require purely numeric
 100          * inputs to be allowed.
 101          */
 102         if (arg[0] == '+')
 103                 arg++;
 104         *(unsigned long long*)result = bb_strtoull(arg, NULL, 0);
 105         /* both coreutils 6.10 and bash 3.2:
 106          * $ printf '%x\n' -2
 107          * fffffffffffffffe
 108          * Mimic that:
 109          */
 110         if (errno) {
 111                 *(unsigned long long*)result = bb_strtoll(arg, NULL, 0);
 112         }
 113 }
 114 static void FAST_FUNC conv_strtoll(const char *arg, void *result)
 115 {
 116         if (arg[0] == '+')
 117                 arg++;
 118         *(long long*)result = bb_strtoll(arg, NULL, 0);
 119 }
 120 static void FAST_FUNC conv_strtod(const char *arg, void *result)
 121 {
 122         char *end;
 123         /* Well, this one allows leading whitespace... so what? */
 124         /* What I like much less is that "-" accepted too! :( */
 125 //TODO: needs setlocale(LC_NUMERIC, "C")?
 126         *(double*)result = strtod(arg, &end);
 127         if (end[0]) {
 128                 errno = ERANGE;
 129                 *(double*)result = 0;
 130         }
 131 }
 132
 133 /* Callers should check errno to detect errors */
 134 static unsigned long long my_xstrtoull(const char *arg)
 135 {
 136         unsigned long long result;
 137         if (multiconvert(arg, &result, conv_strtoull))
 138                 result = 0;
 139         return result;
 140 }
 141 static long long my_xstrtoll(const char *arg)
 142 {
 143         long long result;
 144         if (multiconvert(arg, &result, conv_strtoll))
 145                 result = 0;
 146         return result;
 147 }
 148 static double my_xstrtod(const char *arg)
 149 {
 150         double result;
 151         multiconvert(arg, &result, conv_strtod);
 152         return result;
 153 }
 154
 155 /* Handles %b; return 1 if output is to be short-circuited by \c */
 156 static int print_esc_string(const char *str)
 157 {
 158         char c;
 159         while ((c = *str) != '\0') {
 160                 str++;
 161                 if (c == '\\') {
 162                         /* %b also accepts 4-digit octals of the form \0### */
 163                         if (*str == '0') {
 164                                 if ((unsigned char)(str[1] - '0') < 8) {
 165                                         /* 2nd char is 0..7: skip leading '0' */
 166                                         str++;
 167                                 }
 168                         }
 169                         else if (*str == 'c') {
 170                                 return 1;
 171                         }
 172                         {
 173                                 /* optimization: don't force arg to be on-stack,
 174                                  * use another variable for that. */
 175                                 const char *z = str;
 176                                 c = bb_process_escape_sequence(&z);
 177                                 str = z;
 178                         }
 179                 }
 180                 putchar(c);
 181         }
 182
 183         return 0;
 184 }
 185
 186 static void print_direc(char *format, unsigned fmt_length,
 187                 int field_width, int precision,
 188                 const char *argument)
 189 {
 190         long long llv;
 191         double dv;
 192         char saved;
 193         char *have_prec, *have_width;
 194
 195         saved = format[fmt_length];
 196         format[fmt_length] = '\0';
 197
 198         have_prec = strstr(format, ".*");
 199         have_width = strchr(format, '*');
 200         if (have_width - 1 == have_prec)
 201                 have_width = NULL;
 202
 203         /* multiconvert sets errno = 0, but %s needs it cleared */
 204         errno = 0;
 205
 206         switch (format[fmt_length - 1]) {
 207         case 'c':
 208                 printf(format, *argument);
 209                 break;
 210         case 'd':
 211         case 'i':
 212                 llv = my_xstrtoll(skip_whitespace(argument));
 213  print_long:
 214                 if (!have_width) {
 215                         if (!have_prec)
 216                                 printf(format, llv);
 217                         else
 218                                 printf(format, precision, llv);
 219                 } else {
 220                         if (!have_prec)
 221                                 printf(format, field_width, llv);
 222                         else
 223                                 printf(format, field_width, precision, llv);
 224                 }
 225                 break;
 226         case 'o':
 227         case 'u':
 228         case 'x':
 229         case 'X':
 230                 llv = my_xstrtoull(skip_whitespace(argument));
 231                 /* cheat: unsigned long and long have same width, so... */
 232                 goto print_long;
 233         case 's':
 234                 /* Are char* and long long the same? */
 235                 if (sizeof(argument) == sizeof(llv)) {
 236                         llv = (long long)(ptrdiff_t)argument;
 237                         goto print_long;
 238                 } else {
 239                         /* Hope compiler will optimize it out by moving call
 240                          * instruction after the ifs... */
 241                         if (!have_width) {
 242                                 if (!have_prec)
 243                                         printf(format, argument, /*unused:*/ argument, argument);
 244                                 else
 245                                         printf(format, precision, argument, /*unused:*/ argument);
 246                         } else {
 247                                 if (!have_prec)
 248                                         printf(format, field_width, argument, /*unused:*/ argument);
 249                                 else
 250                                         printf(format, field_width, precision, argument);
 251                         }
 252                         break;
 253                 }
 254         case 'f':
 255         case 'e':
 256         case 'E':
 257         case 'g':
 258         case 'G':
 259                 dv = my_xstrtod(argument);
 260                 if (!have_width) {
 261                         if (!have_prec)
 262                                 printf(format, dv);
 263                         else
 264                                 printf(format, precision, dv);
 265                 } else {
 266                         if (!have_prec)
 267                                 printf(format, field_width, dv);
 268                         else
 269                                 printf(format, field_width, precision, dv);
 270                 }
 271                 break;
 272         } /* switch */
 273
 274         format[fmt_length] = saved;
 275 }
 276
 277 /* Handle params for "%*.*f". Negative numbers are ok (compat). */
 278 static int get_width_prec(const char *str)
 279 {
 280         int v = bb_strtoi(str, NULL, 10);
 281         if (errno) {
 282                 bb_error_msg("invalid number '%s'", str);
 283                 v = 0;
 284         }
 285         return v;
 286 }
 287
 288 /* Print the text in FORMAT, using ARGV for arguments to any '%' directives.
 289    Return advanced ARGV.  */
 290 static char **print_formatted(char *f, char **argv, int *conv_err)
 291 {
 292         char *direc_start;      /* Start of % directive.  */
 293         unsigned direc_length;  /* Length of % directive.  */
 294         int field_width;        /* Arg to first '*' */
 295         int precision;          /* Arg to second '*' */
 296         char **saved_argv = argv;
 297
 298         for (; *f; ++f) {
 299                 switch (*f) {
 300                 case '%':
 301                         direc_start = f++;
 302                         direc_length = 1;
 303                         field_width = precision = 0;
 304                         if (*f == '%') {
 305                                 bb_putchar('%');
 306                                 break;
 307                         }
 308                         if (*f == 'b') {
 309                                 if (*argv) {
 310                                         if (print_esc_string(*argv))
 311                                                 return saved_argv; /* causes main() to exit */
 312                                         ++argv;
 313                                 }
 314                                 break;
 315                         }
 316                         while (*f && strchr("-+ #0", *f)) {
 317                                 ++f;
 318                                 ++direc_length;
 319                         }
 320                         if (*f == '*') {
 321                                 ++f;
 322                                 ++direc_length;
 323                                 if (*argv)
 324                                         field_width = get_width_prec(*argv++);
 325                         } else {
 326                                 while (isdigit(*f)) {
 327                                         ++f;
 328                                         ++direc_length;
 329                                 }
 330                         }
 331                         if (*f == '.') {
 332                                 ++f;
 333                                 ++direc_length;
 334                                 if (*f == '*') {
 335                                         ++f;
 336                                         ++direc_length;
 337                                         if (*argv)
 338                                                 precision = get_width_prec(*argv++);
 339                                 } else {
 340                                         while (isdigit(*f)) {
 341                                                 ++f;
 342                                                 ++direc_length;
 343                                         }
 344                                 }
 345                         }
 346
 347                         /* Remove "lLhz" size modifiers, repeatedly.
 348                          * bash does not like "%lld", but coreutils
 349                          * happily takes even "%Llllhhzhhzd"!
 350                          * We are permissive like coreutils */
 351                         while ((*f | 0x20) == 'l' || *f == 'h' || *f == 'z') {
 352                                 overlapping_strcpy(f, f + 1);
 353                         }
 354                         /* Add "ll" if integer modifier, then print */
 355                         {
 356                                 static const char format_chars[] ALIGN1 = "diouxXfeEgGcs";
 357                                 char *p = strchr(format_chars, *f);
 358                                 /* needed - try "printf %" without it */
 359                                 if (p == NULL || *f == '\0') {
 360                                         bb_error_msg("%s: invalid format", direc_start);
 361                                         /* causes main() to exit with error */
 362                                         return saved_argv - 1;
 363                                 }
 364                                 ++direc_length;
 365                                 if (p - format_chars <= 5) {
 366                                         /* it is one of "diouxX" */
 367                                         p = xmalloc(direc_length + 3);
 368                                         memcpy(p, direc_start, direc_length);
 369                                         p[direc_length + 1] = p[direc_length - 1];
 370                                         p[direc_length - 1] = 'l';
 371                                         p[direc_length] = 'l';
 372                                         //bb_error_msg("<%s>", p);
 373                                         direc_length += 2;
 374                                         direc_start = p;
 375                                 } else {
 376                                         p = NULL;
 377                                 }
 378                                 if (*argv) {
 379                                         print_direc(direc_start, direc_length, field_width,
 380                                                                 precision, *argv++);
 381                                 } else {
 382                                         print_direc(direc_start, direc_length, field_width,
 383                                                                 precision, "");
 384                                 }
 385                                 *conv_err |= errno;
 386                                 free(p);
 387                         }
 388                         break;
 389                 case '\\':
 390                         if (*++f == 'c') {
 391                                 return saved_argv; /* causes main() to exit */
 392                         }
 393                         bb_putchar(bb_process_escape_sequence((const char **)&f));
 394                         f--;
 395                         break;
 396                 default:
 397                         putchar(*f);
 398                 }
 399         }
 400
 401         return argv;
 402 }
 403
 404 int printf_main(int argc UNUSED_PARAM, char **argv)
 405 {
 406         int conv_err;
 407         char *format;
 408         char **argv2;
 409
 410         /* We must check that stdout is not closed.
 411          * The reason for this is highly non-obvious.
 412          * printf_main is used from shell.
 413          * Shell must correctly handle 'printf "%s" foo'
 414          * if stdout is closed. With stdio, output gets shoveled into
 415          * stdout buffer, and even fflush cannot clear it out. It seems that
 416          * even if libc receives EBADF on write attempts, it feels determined
 417          * to output data no matter what. So it will try later,
 418          * and possibly will clobber future output. Not good. */
 419 // TODO: check fcntl() & O_ACCMODE == O_WRONLY or O_RDWR?
 420         if (fcntl(1, F_GETFL) == -1)
 421                 return 1; /* match coreutils 6.10 (sans error msg to stderr) */
 422         //if (dup2(1, 1) != 1) - old way
 423         //      return 1;
 424
 425         /* bash builtin errors out on "printf '-%s-\n' foo",
 426          * coreutils-6.9 works. Both work with "printf -- '-%s-\n' foo".
 427          * We will mimic coreutils. */
 428         argv = skip_dash_dash(argv);
 429
 430         if (!argv[0]) {
 431                 if ((ENABLE_ASH_PRINTF || ENABLE_HUSH_PRINTF)
 432                  && applet_name[0] != 'p'
 433                 ) {
 434                         bb_simple_error_msg("usage: printf FORMAT [ARGUMENT...]");
 435                         return 2; /* bash compat */
 436                 }
 437                 bb_show_usage();
 438         }
 439
 440         format = argv[0];
 441         argv2 = argv + 1;
 442
 443         conv_err = 0;
 444         do {
 445                 argv = argv2;
 446                 argv2 = print_formatted(format, argv, &conv_err);
 447         } while (argv2 > argv && *argv2);
 448
 449         /* coreutils compat (bash doesn't do this):
 450         if (*argv)
 451                 fprintf(stderr, "excess args ignored");
 452         */
 453
 454         return (argv2 < argv) /* if true, print_formatted errored out */
 455                 || conv_err; /* print_formatted saw invalid number */
 456 }