LXR busybox/coreutils/printf.c

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