Annotation of /trunk/mkinitrd-magellan/busybox/coreutils/od_bloaty.c
Parent Directory | Revision Log
Revision 1123 -
(hide annotations)
(download)
Wed Aug 18 21:56:57 2010 UTC (13 years, 8 months ago) by niro
File MIME type: text/plain
File size: 37996 byte(s)
Wed Aug 18 21:56:57 2010 UTC (13 years, 8 months ago) by niro
File MIME type: text/plain
File size: 37996 byte(s)
-updated to busybox-1.17.1
1 | niro | 532 | /* od -- dump files in octal and other formats |
2 | Copyright (C) 92, 1995-2004 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 | ||
15 | along with this program; if not, write to the Free Software Foundation, | ||
16 | Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ | ||
17 | |||
18 | /* Written by Jim Meyering. */ | ||
19 | |||
20 | niro | 816 | /* Busyboxed by Denys Vlasenko |
21 | niro | 532 | |
22 | Based on od.c from coreutils-5.2.1 | ||
23 | Top bloat sources: | ||
24 | |||
25 | 00000073 t parse_old_offset | ||
26 | 0000007b t get_lcm | ||
27 | 00000090 r long_options | ||
28 | 00000092 t print_named_ascii | ||
29 | 000000bf t print_ascii | ||
30 | 00000168 t write_block | ||
31 | 00000366 t decode_format_string | ||
32 | 00000a71 T od_main | ||
33 | |||
34 | Tested for compat with coreutils 6.3 | ||
35 | using this script. Minor differences fixed. | ||
36 | |||
37 | #!/bin/sh | ||
38 | echo STD | ||
39 | time /path/to/coreutils/od \ | ||
40 | ...params... \ | ||
41 | >std | ||
42 | echo Exit code $? | ||
43 | echo BBOX | ||
44 | time ./busybox od \ | ||
45 | ...params... \ | ||
46 | >bbox | ||
47 | echo Exit code $? | ||
48 | diff -u -a std bbox >bbox.diff || { echo Different!; sleep 1; } | ||
49 | |||
50 | */ | ||
51 | |||
52 | niro | 816 | #include "libbb.h" |
53 | niro | 532 | |
54 | #define assert(a) ((void)0) | ||
55 | |||
56 | /* Check for 0x7f is a coreutils 6.3 addition */ | ||
57 | #define ISPRINT(c) (((c)>=' ') && (c) != 0x7f) | ||
58 | |||
59 | typedef long double longdouble_t; | ||
60 | typedef unsigned long long ulonglong_t; | ||
61 | typedef long long llong; | ||
62 | |||
63 | #if ENABLE_LFS | ||
64 | # define xstrtooff_sfx xstrtoull_sfx | ||
65 | #else | ||
66 | # define xstrtooff_sfx xstrtoul_sfx | ||
67 | #endif | ||
68 | |||
69 | /* The default number of input bytes per output line. */ | ||
70 | #define DEFAULT_BYTES_PER_BLOCK 16 | ||
71 | |||
72 | /* The number of decimal digits of precision in a float. */ | ||
73 | #ifndef FLT_DIG | ||
74 | # define FLT_DIG 7 | ||
75 | #endif | ||
76 | |||
77 | /* The number of decimal digits of precision in a double. */ | ||
78 | #ifndef DBL_DIG | ||
79 | # define DBL_DIG 15 | ||
80 | #endif | ||
81 | |||
82 | /* The number of decimal digits of precision in a long double. */ | ||
83 | #ifndef LDBL_DIG | ||
84 | # define LDBL_DIG DBL_DIG | ||
85 | #endif | ||
86 | |||
87 | enum size_spec { | ||
88 | NO_SIZE, | ||
89 | CHAR, | ||
90 | SHORT, | ||
91 | INT, | ||
92 | LONG, | ||
93 | LONG_LONG, | ||
94 | FLOAT_SINGLE, | ||
95 | FLOAT_DOUBLE, | ||
96 | FLOAT_LONG_DOUBLE, | ||
97 | N_SIZE_SPECS | ||
98 | }; | ||
99 | |||
100 | enum output_format { | ||
101 | SIGNED_DECIMAL, | ||
102 | UNSIGNED_DECIMAL, | ||
103 | OCTAL, | ||
104 | HEXADECIMAL, | ||
105 | FLOATING_POINT, | ||
106 | NAMED_CHARACTER, | ||
107 | CHARACTER | ||
108 | }; | ||
109 | |||
110 | /* Each output format specification (from '-t spec' or from | ||
111 | old-style options) is represented by one of these structures. */ | ||
112 | struct tspec { | ||
113 | enum output_format fmt; | ||
114 | enum size_spec size; | ||
115 | void (*print_function) (size_t, const char *, const char *); | ||
116 | char *fmt_string; | ||
117 | int hexl_mode_trailer; | ||
118 | int field_width; | ||
119 | }; | ||
120 | |||
121 | /* Convert the number of 8-bit bytes of a binary representation to | ||
122 | the number of characters (digits + sign if the type is signed) | ||
123 | required to represent the same quantity in the specified base/type. | ||
124 | For example, a 32-bit (4-byte) quantity may require a field width | ||
125 | as wide as the following for these types: | ||
126 | 11 unsigned octal | ||
127 | 11 signed decimal | ||
128 | 10 unsigned decimal | ||
129 | 8 unsigned hexadecimal */ | ||
130 | |||
131 | niro | 816 | static const uint8_t bytes_to_oct_digits[] ALIGN1 = |
132 | niro | 532 | {0, 3, 6, 8, 11, 14, 16, 19, 22, 25, 27, 30, 32, 35, 38, 41, 43}; |
133 | |||
134 | niro | 816 | static const uint8_t bytes_to_signed_dec_digits[] ALIGN1 = |
135 | niro | 532 | {1, 4, 6, 8, 11, 13, 16, 18, 20, 23, 25, 28, 30, 33, 35, 37, 40}; |
136 | |||
137 | niro | 816 | static const uint8_t bytes_to_unsigned_dec_digits[] ALIGN1 = |
138 | niro | 532 | {0, 3, 5, 8, 10, 13, 15, 17, 20, 22, 25, 27, 29, 32, 34, 37, 39}; |
139 | |||
140 | niro | 816 | static const uint8_t bytes_to_hex_digits[] ALIGN1 = |
141 | niro | 532 | {0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32}; |
142 | |||
143 | /* Convert enum size_spec to the size of the named type. */ | ||
144 | niro | 816 | static const signed char width_bytes[] ALIGN1 = { |
145 | niro | 532 | -1, |
146 | sizeof(char), | ||
147 | sizeof(short), | ||
148 | sizeof(int), | ||
149 | sizeof(long), | ||
150 | sizeof(ulonglong_t), | ||
151 | sizeof(float), | ||
152 | sizeof(double), | ||
153 | sizeof(longdouble_t) | ||
154 | }; | ||
155 | /* Ensure that for each member of 'enum size_spec' there is an | ||
156 | initializer in the width_bytes array. */ | ||
157 | niro | 816 | struct ERR_width_bytes_has_bad_size { |
158 | char ERR_width_bytes_has_bad_size[ARRAY_SIZE(width_bytes) == N_SIZE_SPECS ? 1 : -1]; | ||
159 | niro | 532 | }; |
160 | |||
161 | niro | 816 | static smallint flag_dump_strings; |
162 | /* Non-zero if an old-style 'pseudo-address' was specified. */ | ||
163 | static smallint flag_pseudo_start; | ||
164 | static smallint limit_bytes_to_format; | ||
165 | /* When zero and two or more consecutive blocks are equal, format | ||
166 | only the first block and output an asterisk alone on the following | ||
167 | line to indicate that identical blocks have been elided. */ | ||
168 | static smallint verbose; | ||
169 | static smallint ioerror; | ||
170 | |||
171 | niro | 532 | static size_t string_min; |
172 | |||
173 | niro | 816 | /* An array of specs describing how to format each input block. */ |
174 | static size_t n_specs; | ||
175 | static struct tspec *spec; | ||
176 | niro | 532 | |
177 | niro | 816 | /* Function that accepts an address and an optional following char, |
178 | and prints the address and char to stdout. */ | ||
179 | static void (*format_address)(off_t, char); | ||
180 | niro | 532 | /* The difference between the old-style pseudo starting address and |
181 | the number of bytes to skip. */ | ||
182 | static off_t pseudo_offset; | ||
183 | /* When zero, MAX_BYTES_TO_FORMAT and END_OFFSET are ignored, and all | ||
184 | input is formatted. */ | ||
185 | |||
186 | /* The number of input bytes formatted per output line. It must be | ||
187 | a multiple of the least common multiple of the sizes associated with | ||
188 | the specified output types. It should be as large as possible, but | ||
189 | no larger than 16 -- unless specified with the -w option. */ | ||
190 | niro | 816 | static unsigned bytes_per_block = 32; /* have to use unsigned, not size_t */ |
191 | niro | 532 | |
192 | /* A NULL-terminated list of the file-arguments from the command line. */ | ||
193 | niro | 816 | static const char *const *file_list; |
194 | niro | 532 | |
195 | /* The input stream associated with the current file. */ | ||
196 | static FILE *in_stream; | ||
197 | |||
198 | #define MAX_INTEGRAL_TYPE_SIZE sizeof(ulonglong_t) | ||
199 | niro | 816 | static const unsigned char integral_type_size[MAX_INTEGRAL_TYPE_SIZE + 1] ALIGN1 = { |
200 | niro | 532 | [sizeof(char)] = CHAR, |
201 | #if USHRT_MAX != UCHAR_MAX | ||
202 | [sizeof(short)] = SHORT, | ||
203 | #endif | ||
204 | #if UINT_MAX != USHRT_MAX | ||
205 | [sizeof(int)] = INT, | ||
206 | #endif | ||
207 | #if ULONG_MAX != UINT_MAX | ||
208 | [sizeof(long)] = LONG, | ||
209 | #endif | ||
210 | #if ULLONG_MAX != ULONG_MAX | ||
211 | [sizeof(ulonglong_t)] = LONG_LONG, | ||
212 | #endif | ||
213 | }; | ||
214 | |||
215 | #define MAX_FP_TYPE_SIZE sizeof(longdouble_t) | ||
216 | niro | 816 | static const unsigned char fp_type_size[MAX_FP_TYPE_SIZE + 1] ALIGN1 = { |
217 | niro | 532 | /* gcc seems to allow repeated indexes. Last one stays */ |
218 | [sizeof(longdouble_t)] = FLOAT_LONG_DOUBLE, | ||
219 | [sizeof(double)] = FLOAT_DOUBLE, | ||
220 | niro | 816 | [sizeof(float)] = FLOAT_SINGLE |
221 | niro | 532 | }; |
222 | |||
223 | |||
224 | static unsigned | ||
225 | gcd(unsigned u, unsigned v) | ||
226 | { | ||
227 | unsigned t; | ||
228 | while (v != 0) { | ||
229 | t = u % v; | ||
230 | u = v; | ||
231 | v = t; | ||
232 | } | ||
233 | return u; | ||
234 | } | ||
235 | |||
236 | /* Compute the least common multiple of U and V. */ | ||
237 | static unsigned | ||
238 | lcm(unsigned u, unsigned v) { | ||
239 | unsigned t = gcd(u, v); | ||
240 | if (t == 0) | ||
241 | return 0; | ||
242 | return u * v / t; | ||
243 | } | ||
244 | |||
245 | static void | ||
246 | print_s_char(size_t n_bytes, const char *block, const char *fmt_string) | ||
247 | { | ||
248 | while (n_bytes--) { | ||
249 | int tmp = *(signed char *) block; | ||
250 | printf(fmt_string, tmp); | ||
251 | block += sizeof(unsigned char); | ||
252 | } | ||
253 | } | ||
254 | |||
255 | static void | ||
256 | print_char(size_t n_bytes, const char *block, const char *fmt_string) | ||
257 | { | ||
258 | while (n_bytes--) { | ||
259 | unsigned tmp = *(unsigned char *) block; | ||
260 | printf(fmt_string, tmp); | ||
261 | block += sizeof(unsigned char); | ||
262 | } | ||
263 | } | ||
264 | |||
265 | static void | ||
266 | print_s_short(size_t n_bytes, const char *block, const char *fmt_string) | ||
267 | { | ||
268 | n_bytes /= sizeof(signed short); | ||
269 | while (n_bytes--) { | ||
270 | int tmp = *(signed short *) block; | ||
271 | printf(fmt_string, tmp); | ||
272 | block += sizeof(unsigned short); | ||
273 | } | ||
274 | } | ||
275 | |||
276 | static void | ||
277 | print_short(size_t n_bytes, const char *block, const char *fmt_string) | ||
278 | { | ||
279 | n_bytes /= sizeof(unsigned short); | ||
280 | while (n_bytes--) { | ||
281 | unsigned tmp = *(unsigned short *) block; | ||
282 | printf(fmt_string, tmp); | ||
283 | block += sizeof(unsigned short); | ||
284 | } | ||
285 | } | ||
286 | |||
287 | static void | ||
288 | print_int(size_t n_bytes, const char *block, const char *fmt_string) | ||
289 | { | ||
290 | n_bytes /= sizeof(unsigned); | ||
291 | while (n_bytes--) { | ||
292 | unsigned tmp = *(unsigned *) block; | ||
293 | printf(fmt_string, tmp); | ||
294 | block += sizeof(unsigned); | ||
295 | } | ||
296 | } | ||
297 | |||
298 | #if UINT_MAX == ULONG_MAX | ||
299 | # define print_long print_int | ||
300 | #else | ||
301 | static void | ||
302 | print_long(size_t n_bytes, const char *block, const char *fmt_string) | ||
303 | { | ||
304 | n_bytes /= sizeof(unsigned long); | ||
305 | while (n_bytes--) { | ||
306 | unsigned long tmp = *(unsigned long *) block; | ||
307 | printf(fmt_string, tmp); | ||
308 | block += sizeof(unsigned long); | ||
309 | } | ||
310 | } | ||
311 | #endif | ||
312 | |||
313 | #if ULONG_MAX == ULLONG_MAX | ||
314 | # define print_long_long print_long | ||
315 | #else | ||
316 | static void | ||
317 | print_long_long(size_t n_bytes, const char *block, const char *fmt_string) | ||
318 | { | ||
319 | n_bytes /= sizeof(ulonglong_t); | ||
320 | while (n_bytes--) { | ||
321 | ulonglong_t tmp = *(ulonglong_t *) block; | ||
322 | printf(fmt_string, tmp); | ||
323 | block += sizeof(ulonglong_t); | ||
324 | } | ||
325 | } | ||
326 | #endif | ||
327 | |||
328 | static void | ||
329 | print_float(size_t n_bytes, const char *block, const char *fmt_string) | ||
330 | { | ||
331 | n_bytes /= sizeof(float); | ||
332 | while (n_bytes--) { | ||
333 | float tmp = *(float *) block; | ||
334 | printf(fmt_string, tmp); | ||
335 | block += sizeof(float); | ||
336 | } | ||
337 | } | ||
338 | |||
339 | static void | ||
340 | print_double(size_t n_bytes, const char *block, const char *fmt_string) | ||
341 | { | ||
342 | n_bytes /= sizeof(double); | ||
343 | while (n_bytes--) { | ||
344 | double tmp = *(double *) block; | ||
345 | printf(fmt_string, tmp); | ||
346 | block += sizeof(double); | ||
347 | } | ||
348 | } | ||
349 | |||
350 | static void | ||
351 | print_long_double(size_t n_bytes, const char *block, const char *fmt_string) | ||
352 | { | ||
353 | n_bytes /= sizeof(longdouble_t); | ||
354 | while (n_bytes--) { | ||
355 | longdouble_t tmp = *(longdouble_t *) block; | ||
356 | printf(fmt_string, tmp); | ||
357 | block += sizeof(longdouble_t); | ||
358 | } | ||
359 | } | ||
360 | |||
361 | /* print_[named]_ascii are optimized for speed. | ||
362 | niro | 816 | * Remember, someday you may want to pump gigabytes through this thing. |
363 | niro | 532 | * Saving a dozen of .text bytes here is counter-productive */ |
364 | |||
365 | static void | ||
366 | print_named_ascii(size_t n_bytes, const char *block, | ||
367 | niro | 816 | const char *unused_fmt_string UNUSED_PARAM) |
368 | niro | 532 | { |
369 | /* Names for some non-printing characters. */ | ||
370 | niro | 816 | static const char charname[33][3] ALIGN1 = { |
371 | niro | 532 | "nul", "soh", "stx", "etx", "eot", "enq", "ack", "bel", |
372 | " bs", " ht", " nl", " vt", " ff", " cr", " so", " si", | ||
373 | "dle", "dc1", "dc2", "dc3", "dc4", "nak", "syn", "etb", | ||
374 | "can", " em", "sub", "esc", " fs", " gs", " rs", " us", | ||
375 | " sp" | ||
376 | }; | ||
377 | // buf[N] pos: 01234 56789 | ||
378 | char buf[12] = " x\0 0xx\0"; | ||
379 | // actually " x\0 xxx\0", but I want to share the string with below. | ||
380 | // [12] because we take three 32bit stack slots anyway, and | ||
381 | // gcc is too dumb to initialize with constant stores, | ||
382 | // it copies initializer from rodata. Oh well. | ||
383 | |||
384 | while (n_bytes--) { | ||
385 | unsigned masked_c = *(unsigned char *) block++; | ||
386 | |||
387 | masked_c &= 0x7f; | ||
388 | if (masked_c == 0x7f) { | ||
389 | fputs(" del", stdout); | ||
390 | continue; | ||
391 | } | ||
392 | if (masked_c > ' ') { | ||
393 | buf[3] = masked_c; | ||
394 | fputs(buf, stdout); | ||
395 | continue; | ||
396 | } | ||
397 | /* Why? Because printf(" %3.3s") is much slower... */ | ||
398 | buf[6] = charname[masked_c][0]; | ||
399 | buf[7] = charname[masked_c][1]; | ||
400 | buf[8] = charname[masked_c][2]; | ||
401 | fputs(buf+5, stdout); | ||
402 | } | ||
403 | } | ||
404 | |||
405 | static void | ||
406 | print_ascii(size_t n_bytes, const char *block, | ||
407 | niro | 816 | const char *unused_fmt_string UNUSED_PARAM) |
408 | niro | 532 | { |
409 | // buf[N] pos: 01234 56789 | ||
410 | char buf[12] = " x\0 0xx\0"; | ||
411 | |||
412 | while (n_bytes--) { | ||
413 | const char *s; | ||
414 | unsigned c = *(unsigned char *) block++; | ||
415 | |||
416 | if (ISPRINT(c)) { | ||
417 | buf[3] = c; | ||
418 | fputs(buf, stdout); | ||
419 | continue; | ||
420 | } | ||
421 | switch (c) { | ||
422 | case '\0': | ||
423 | s = " \\0"; | ||
424 | break; | ||
425 | case '\007': | ||
426 | s = " \\a"; | ||
427 | break; | ||
428 | case '\b': | ||
429 | s = " \\b"; | ||
430 | break; | ||
431 | case '\f': | ||
432 | s = " \\f"; | ||
433 | break; | ||
434 | case '\n': | ||
435 | s = " \\n"; | ||
436 | break; | ||
437 | case '\r': | ||
438 | s = " \\r"; | ||
439 | break; | ||
440 | case '\t': | ||
441 | s = " \\t"; | ||
442 | break; | ||
443 | case '\v': | ||
444 | s = " \\v"; | ||
445 | break; | ||
446 | case '\x7f': | ||
447 | s = " 177"; | ||
448 | break; | ||
449 | default: /* c is never larger than 040 */ | ||
450 | buf[7] = (c >> 3) + '0'; | ||
451 | buf[8] = (c & 7) + '0'; | ||
452 | s = buf + 5; | ||
453 | } | ||
454 | fputs(s, stdout); | ||
455 | } | ||
456 | } | ||
457 | |||
458 | /* Given a list of one or more input filenames FILE_LIST, set the global | ||
459 | file pointer IN_STREAM and the global string INPUT_FILENAME to the | ||
460 | first one that can be successfully opened. Modify FILE_LIST to | ||
461 | reference the next filename in the list. A file name of "-" is | ||
462 | interpreted as standard input. If any file open fails, give an error | ||
463 | message and return nonzero. */ | ||
464 | |||
465 | static void | ||
466 | open_next_file(void) | ||
467 | { | ||
468 | while (1) { | ||
469 | niro | 816 | if (!*file_list) |
470 | niro | 532 | return; |
471 | niro | 816 | in_stream = fopen_or_warn_stdin(*file_list++); |
472 | niro | 532 | if (in_stream) { |
473 | break; | ||
474 | } | ||
475 | ioerror = 1; | ||
476 | } | ||
477 | |||
478 | if (limit_bytes_to_format && !flag_dump_strings) | ||
479 | setbuf(in_stream, NULL); | ||
480 | } | ||
481 | |||
482 | /* Test whether there have been errors on in_stream, and close it if | ||
483 | it is not standard input. Return nonzero if there has been an error | ||
484 | on in_stream or stdout; return zero otherwise. This function will | ||
485 | report more than one error only if both a read and a write error | ||
486 | have occurred. IN_ERRNO, if nonzero, is the error number | ||
487 | corresponding to the most recent action for IN_STREAM. */ | ||
488 | |||
489 | static void | ||
490 | check_and_close(void) | ||
491 | { | ||
492 | if (in_stream) { | ||
493 | if (ferror(in_stream)) { | ||
494 | niro | 816 | bb_error_msg("%s: read error", (in_stream == stdin) |
495 | ? bb_msg_standard_input | ||
496 | : file_list[-1] | ||
497 | ); | ||
498 | niro | 532 | ioerror = 1; |
499 | } | ||
500 | fclose_if_not_stdin(in_stream); | ||
501 | in_stream = NULL; | ||
502 | } | ||
503 | |||
504 | if (ferror(stdout)) { | ||
505 | niro | 1123 | bb_error_msg(bb_msg_write_error); |
506 | niro | 532 | ioerror = 1; |
507 | } | ||
508 | } | ||
509 | |||
510 | /* If S points to a single valid modern od format string, put | ||
511 | niro | 816 | a description of that format in *TSPEC, return pointer to |
512 | character following the just-decoded format. | ||
513 | For example, if S were "d4afL", we will return a rtp to "afL" | ||
514 | and *TSPEC would be | ||
515 | niro | 532 | { |
516 | fmt = SIGNED_DECIMAL; | ||
517 | size = INT or LONG; (whichever integral_type_size[4] resolves to) | ||
518 | print_function = print_int; (assuming size == INT) | ||
519 | niro | 816 | fmt_string = "%011d%c"; |
520 | niro | 532 | } |
521 | S_ORIG is solely for reporting errors. It should be the full format | ||
522 | string argument. */ | ||
523 | |||
524 | niro | 1123 | static NOINLINE const char * |
525 | niro | 816 | decode_one_format(const char *s_orig, const char *s, struct tspec *tspec) |
526 | niro | 532 | { |
527 | enum size_spec size_spec; | ||
528 | unsigned size; | ||
529 | enum output_format fmt; | ||
530 | const char *p; | ||
531 | char *end; | ||
532 | char *fmt_string = NULL; | ||
533 | void (*print_function) (size_t, const char *, const char *); | ||
534 | unsigned c; | ||
535 | unsigned field_width = 0; | ||
536 | int pos; | ||
537 | |||
538 | |||
539 | switch (*s) { | ||
540 | case 'd': | ||
541 | case 'o': | ||
542 | case 'u': | ||
543 | case 'x': { | ||
544 | niro | 816 | static const char CSIL[] ALIGN1 = "CSIL"; |
545 | niro | 532 | |
546 | c = *s++; | ||
547 | p = strchr(CSIL, *s); | ||
548 | niro | 984 | /* if *s == NUL, p != NULL! Testcase: "od -tx" */ |
549 | if (!p || *p == '\0') { | ||
550 | niro | 532 | size = sizeof(int); |
551 | if (isdigit(s[0])) { | ||
552 | size = bb_strtou(s, &end, 0); | ||
553 | if (errno == ERANGE | ||
554 | || MAX_INTEGRAL_TYPE_SIZE < size | ||
555 | || integral_type_size[size] == NO_SIZE | ||
556 | ) { | ||
557 | bb_error_msg_and_die("invalid type string '%s'; " | ||
558 | "%u-byte %s type is not supported", | ||
559 | s_orig, size, "integral"); | ||
560 | } | ||
561 | s = end; | ||
562 | } | ||
563 | } else { | ||
564 | niro | 816 | static const uint8_t CSIL_sizeof[4] = { |
565 | niro | 532 | sizeof(char), |
566 | sizeof(short), | ||
567 | sizeof(int), | ||
568 | sizeof(long), | ||
569 | }; | ||
570 | size = CSIL_sizeof[p - CSIL]; | ||
571 | niro | 816 | s++; /* skip C/S/I/L */ |
572 | niro | 532 | } |
573 | |||
574 | #define ISPEC_TO_FORMAT(Spec, Min_format, Long_format, Max_format) \ | ||
575 | ((Spec) == LONG_LONG ? (Max_format) \ | ||
576 | : ((Spec) == LONG ? (Long_format) : (Min_format))) | ||
577 | |||
578 | #define FMT_BYTES_ALLOCATED 9 | ||
579 | size_spec = integral_type_size[size]; | ||
580 | |||
581 | { | ||
582 | niro | 816 | static const char doux[] ALIGN1 = "doux"; |
583 | niro | 532 | static const char doux_fmt_letter[][4] = { |
584 | "lld", "llo", "llu", "llx" | ||
585 | }; | ||
586 | static const enum output_format doux_fmt[] = { | ||
587 | SIGNED_DECIMAL, | ||
588 | OCTAL, | ||
589 | UNSIGNED_DECIMAL, | ||
590 | HEXADECIMAL, | ||
591 | }; | ||
592 | static const uint8_t *const doux_bytes_to_XXX[] = { | ||
593 | bytes_to_signed_dec_digits, | ||
594 | bytes_to_oct_digits, | ||
595 | bytes_to_unsigned_dec_digits, | ||
596 | bytes_to_hex_digits, | ||
597 | }; | ||
598 | static const char doux_fmtstring[][sizeof(" %%0%u%s")] = { | ||
599 | " %%%u%s", | ||
600 | " %%0%u%s", | ||
601 | " %%%u%s", | ||
602 | " %%0%u%s", | ||
603 | }; | ||
604 | |||
605 | pos = strchr(doux, c) - doux; | ||
606 | fmt = doux_fmt[pos]; | ||
607 | field_width = doux_bytes_to_XXX[pos][size]; | ||
608 | p = doux_fmt_letter[pos] + 2; | ||
609 | if (size_spec == LONG) p--; | ||
610 | if (size_spec == LONG_LONG) p -= 2; | ||
611 | fmt_string = xasprintf(doux_fmtstring[pos], field_width, p); | ||
612 | } | ||
613 | |||
614 | switch (size_spec) { | ||
615 | case CHAR: | ||
616 | print_function = (fmt == SIGNED_DECIMAL | ||
617 | ? print_s_char | ||
618 | : print_char); | ||
619 | break; | ||
620 | case SHORT: | ||
621 | print_function = (fmt == SIGNED_DECIMAL | ||
622 | ? print_s_short | ||
623 | : print_short); | ||
624 | break; | ||
625 | case INT: | ||
626 | print_function = print_int; | ||
627 | break; | ||
628 | case LONG: | ||
629 | print_function = print_long; | ||
630 | break; | ||
631 | default: /* case LONG_LONG: */ | ||
632 | print_function = print_long_long; | ||
633 | break; | ||
634 | } | ||
635 | break; | ||
636 | } | ||
637 | |||
638 | case 'f': { | ||
639 | niro | 816 | static const char FDL[] ALIGN1 = "FDL"; |
640 | niro | 532 | |
641 | fmt = FLOATING_POINT; | ||
642 | ++s; | ||
643 | p = strchr(FDL, *s); | ||
644 | if (!p) { | ||
645 | size = sizeof(double); | ||
646 | if (isdigit(s[0])) { | ||
647 | size = bb_strtou(s, &end, 0); | ||
648 | if (errno == ERANGE || size > MAX_FP_TYPE_SIZE | ||
649 | || fp_type_size[size] == NO_SIZE | ||
650 | ) { | ||
651 | bb_error_msg_and_die("invalid type string '%s'; " | ||
652 | "%u-byte %s type is not supported", | ||
653 | s_orig, size, "floating point"); | ||
654 | } | ||
655 | s = end; | ||
656 | } | ||
657 | } else { | ||
658 | static const uint8_t FDL_sizeof[] = { | ||
659 | sizeof(float), | ||
660 | sizeof(double), | ||
661 | sizeof(longdouble_t), | ||
662 | }; | ||
663 | |||
664 | size = FDL_sizeof[p - FDL]; | ||
665 | } | ||
666 | |||
667 | size_spec = fp_type_size[size]; | ||
668 | |||
669 | switch (size_spec) { | ||
670 | case FLOAT_SINGLE: | ||
671 | print_function = print_float; | ||
672 | field_width = FLT_DIG + 8; | ||
673 | /* Don't use %#e; not all systems support it. */ | ||
674 | fmt_string = xasprintf(" %%%d.%de", field_width, FLT_DIG); | ||
675 | break; | ||
676 | case FLOAT_DOUBLE: | ||
677 | print_function = print_double; | ||
678 | field_width = DBL_DIG + 8; | ||
679 | fmt_string = xasprintf(" %%%d.%de", field_width, DBL_DIG); | ||
680 | break; | ||
681 | default: /* case FLOAT_LONG_DOUBLE: */ | ||
682 | print_function = print_long_double; | ||
683 | field_width = LDBL_DIG + 8; | ||
684 | fmt_string = xasprintf(" %%%d.%dLe", field_width, LDBL_DIG); | ||
685 | break; | ||
686 | } | ||
687 | break; | ||
688 | } | ||
689 | |||
690 | case 'a': | ||
691 | ++s; | ||
692 | fmt = NAMED_CHARACTER; | ||
693 | size_spec = CHAR; | ||
694 | print_function = print_named_ascii; | ||
695 | field_width = 3; | ||
696 | break; | ||
697 | case 'c': | ||
698 | ++s; | ||
699 | fmt = CHARACTER; | ||
700 | size_spec = CHAR; | ||
701 | print_function = print_ascii; | ||
702 | field_width = 3; | ||
703 | break; | ||
704 | default: | ||
705 | bb_error_msg_and_die("invalid character '%c' " | ||
706 | "in type string '%s'", *s, s_orig); | ||
707 | } | ||
708 | |||
709 | tspec->size = size_spec; | ||
710 | tspec->fmt = fmt; | ||
711 | tspec->print_function = print_function; | ||
712 | tspec->fmt_string = fmt_string; | ||
713 | |||
714 | tspec->field_width = field_width; | ||
715 | tspec->hexl_mode_trailer = (*s == 'z'); | ||
716 | if (tspec->hexl_mode_trailer) | ||
717 | s++; | ||
718 | |||
719 | niro | 816 | return s; |
720 | niro | 532 | } |
721 | |||
722 | /* Decode the modern od format string S. Append the decoded | ||
723 | representation to the global array SPEC, reallocating SPEC if | ||
724 | niro | 816 | necessary. */ |
725 | niro | 532 | |
726 | static void | ||
727 | decode_format_string(const char *s) | ||
728 | { | ||
729 | const char *s_orig = s; | ||
730 | |||
731 | while (*s != '\0') { | ||
732 | struct tspec tspec; | ||
733 | const char *next; | ||
734 | |||
735 | niro | 816 | next = decode_one_format(s_orig, s, &tspec); |
736 | niro | 532 | |
737 | assert(s != next); | ||
738 | s = next; | ||
739 | niro | 816 | spec = xrealloc_vector(spec, 4, n_specs); |
740 | memcpy(&spec[n_specs], &tspec, sizeof(spec[0])); | ||
741 | niro | 532 | n_specs++; |
742 | } | ||
743 | } | ||
744 | |||
745 | /* Given a list of one or more input filenames FILE_LIST, set the global | ||
746 | file pointer IN_STREAM to position N_SKIP in the concatenation of | ||
747 | those files. If any file operation fails or if there are fewer than | ||
748 | N_SKIP bytes in the combined input, give an error message and return | ||
749 | nonzero. When possible, use seek rather than read operations to | ||
750 | advance IN_STREAM. */ | ||
751 | |||
752 | static void | ||
753 | skip(off_t n_skip) | ||
754 | { | ||
755 | if (n_skip == 0) | ||
756 | return; | ||
757 | |||
758 | while (in_stream) { /* !EOF */ | ||
759 | struct stat file_stats; | ||
760 | |||
761 | /* First try seeking. For large offsets, this extra work is | ||
762 | worthwhile. If the offset is below some threshold it may be | ||
763 | more efficient to move the pointer by reading. There are two | ||
764 | issues when trying to seek: | ||
765 | - the file must be seekable. | ||
766 | - before seeking to the specified position, make sure | ||
767 | that the new position is in the current file. | ||
768 | Try to do that by getting file's size using fstat. | ||
769 | But that will work only for regular files. */ | ||
770 | |||
771 | /* The st_size field is valid only for regular files | ||
772 | (and for symbolic links, which cannot occur here). | ||
773 | If the number of bytes left to skip is at least | ||
774 | as large as the size of the current file, we can | ||
775 | decrement n_skip and go on to the next file. */ | ||
776 | if (fstat(fileno(in_stream), &file_stats) == 0 | ||
777 | niro | 816 | && S_ISREG(file_stats.st_mode) && file_stats.st_size > 0 |
778 | niro | 532 | ) { |
779 | if (file_stats.st_size < n_skip) { | ||
780 | n_skip -= file_stats.st_size; | ||
781 | niro | 816 | /* take "check & close / open_next" route */ |
782 | niro | 532 | } else { |
783 | if (fseeko(in_stream, n_skip, SEEK_CUR) != 0) | ||
784 | ioerror = 1; | ||
785 | return; | ||
786 | } | ||
787 | } else { | ||
788 | niro | 816 | /* If it's not a regular file with positive size, |
789 | niro | 532 | position the file pointer by reading. */ |
790 | niro | 816 | char buf[1024]; |
791 | size_t n_bytes_to_read = 1024; | ||
792 | size_t n_bytes_read; | ||
793 | niro | 532 | |
794 | while (n_skip > 0) { | ||
795 | if (n_skip < n_bytes_to_read) | ||
796 | n_bytes_to_read = n_skip; | ||
797 | n_bytes_read = fread(buf, 1, n_bytes_to_read, in_stream); | ||
798 | n_skip -= n_bytes_read; | ||
799 | if (n_bytes_read != n_bytes_to_read) | ||
800 | niro | 816 | break; /* EOF on this file or error */ |
801 | niro | 532 | } |
802 | } | ||
803 | if (n_skip == 0) | ||
804 | return; | ||
805 | |||
806 | check_and_close(); | ||
807 | open_next_file(); | ||
808 | } | ||
809 | |||
810 | if (n_skip) | ||
811 | niro | 984 | bb_error_msg_and_die("can't skip past end of combined input"); |
812 | niro | 532 | } |
813 | |||
814 | |||
815 | typedef void FN_format_address(off_t address, char c); | ||
816 | |||
817 | static void | ||
818 | niro | 816 | format_address_none(off_t address UNUSED_PARAM, char c UNUSED_PARAM) |
819 | niro | 532 | { |
820 | } | ||
821 | |||
822 | niro | 816 | static char address_fmt[] ALIGN1 = "%0n"OFF_FMT"xc"; |
823 | niro | 532 | /* Corresponds to 'x' above */ |
824 | #define address_base_char address_fmt[sizeof(address_fmt)-3] | ||
825 | /* Corresponds to 'n' above */ | ||
826 | #define address_pad_len_char address_fmt[2] | ||
827 | |||
828 | static void | ||
829 | format_address_std(off_t address, char c) | ||
830 | { | ||
831 | /* Corresponds to 'c' */ | ||
832 | address_fmt[sizeof(address_fmt)-2] = c; | ||
833 | printf(address_fmt, address); | ||
834 | } | ||
835 | |||
836 | niro | 984 | #if ENABLE_LONG_OPTS |
837 | niro | 532 | /* only used with --traditional */ |
838 | static void | ||
839 | format_address_paren(off_t address, char c) | ||
840 | { | ||
841 | putchar('('); | ||
842 | format_address_std(address, ')'); | ||
843 | if (c) putchar(c); | ||
844 | } | ||
845 | |||
846 | static void | ||
847 | format_address_label(off_t address, char c) | ||
848 | { | ||
849 | format_address_std(address, ' '); | ||
850 | format_address_paren(address + pseudo_offset, c); | ||
851 | } | ||
852 | #endif | ||
853 | |||
854 | static void | ||
855 | dump_hexl_mode_trailer(size_t n_bytes, const char *block) | ||
856 | { | ||
857 | fputs(" >", stdout); | ||
858 | while (n_bytes--) { | ||
859 | unsigned c = *(unsigned char *) block++; | ||
860 | c = (ISPRINT(c) ? c : '.'); | ||
861 | putchar(c); | ||
862 | } | ||
863 | putchar('<'); | ||
864 | } | ||
865 | |||
866 | /* Write N_BYTES bytes from CURR_BLOCK to standard output once for each | ||
867 | of the N_SPEC format specs. CURRENT_OFFSET is the byte address of | ||
868 | CURR_BLOCK in the concatenation of input files, and it is printed | ||
869 | (optionally) only before the output line associated with the first | ||
870 | format spec. When duplicate blocks are being abbreviated, the output | ||
871 | for a sequence of identical input blocks is the output for the first | ||
872 | block followed by an asterisk alone on a line. It is valid to compare | ||
873 | the blocks PREV_BLOCK and CURR_BLOCK only when N_BYTES == BYTES_PER_BLOCK. | ||
874 | That condition may be false only for the last input block -- and then | ||
875 | only when it has not been padded to length BYTES_PER_BLOCK. */ | ||
876 | |||
877 | static void | ||
878 | write_block(off_t current_offset, size_t n_bytes, | ||
879 | const char *prev_block, const char *curr_block) | ||
880 | { | ||
881 | static char first = 1; | ||
882 | static char prev_pair_equal = 0; | ||
883 | size_t i; | ||
884 | |||
885 | niro | 816 | if (!verbose && !first |
886 | niro | 532 | && n_bytes == bytes_per_block |
887 | && memcmp(prev_block, curr_block, bytes_per_block) == 0 | ||
888 | ) { | ||
889 | if (prev_pair_equal) { | ||
890 | /* The two preceding blocks were equal, and the current | ||
891 | block is the same as the last one, so print nothing. */ | ||
892 | } else { | ||
893 | puts("*"); | ||
894 | prev_pair_equal = 1; | ||
895 | } | ||
896 | } else { | ||
897 | first = 0; | ||
898 | prev_pair_equal = 0; | ||
899 | for (i = 0; i < n_specs; i++) { | ||
900 | if (i == 0) | ||
901 | format_address(current_offset, '\0'); | ||
902 | else | ||
903 | printf("%*s", address_pad_len_char - '0', ""); | ||
904 | (*spec[i].print_function) (n_bytes, curr_block, spec[i].fmt_string); | ||
905 | if (spec[i].hexl_mode_trailer) { | ||
906 | /* space-pad out to full line width, then dump the trailer */ | ||
907 | int datum_width = width_bytes[spec[i].size]; | ||
908 | int blank_fields = (bytes_per_block - n_bytes) / datum_width; | ||
909 | int field_width = spec[i].field_width + 1; | ||
910 | printf("%*s", blank_fields * field_width, ""); | ||
911 | dump_hexl_mode_trailer(n_bytes, curr_block); | ||
912 | } | ||
913 | putchar('\n'); | ||
914 | } | ||
915 | } | ||
916 | } | ||
917 | |||
918 | static void | ||
919 | read_block(size_t n, char *block, size_t *n_bytes_in_buffer) | ||
920 | { | ||
921 | assert(0 < n && n <= bytes_per_block); | ||
922 | |||
923 | *n_bytes_in_buffer = 0; | ||
924 | |||
925 | if (n == 0) | ||
926 | return; | ||
927 | |||
928 | while (in_stream != NULL) { /* EOF. */ | ||
929 | size_t n_needed; | ||
930 | size_t n_read; | ||
931 | |||
932 | n_needed = n - *n_bytes_in_buffer; | ||
933 | n_read = fread(block + *n_bytes_in_buffer, 1, n_needed, in_stream); | ||
934 | *n_bytes_in_buffer += n_read; | ||
935 | if (n_read == n_needed) | ||
936 | break; | ||
937 | /* error check is done in check_and_close */ | ||
938 | check_and_close(); | ||
939 | open_next_file(); | ||
940 | } | ||
941 | } | ||
942 | |||
943 | /* Return the least common multiple of the sizes associated | ||
944 | with the format specs. */ | ||
945 | |||
946 | static int | ||
947 | get_lcm(void) | ||
948 | { | ||
949 | size_t i; | ||
950 | int l_c_m = 1; | ||
951 | |||
952 | for (i = 0; i < n_specs; i++) | ||
953 | l_c_m = lcm(l_c_m, width_bytes[(int) spec[i].size]); | ||
954 | return l_c_m; | ||
955 | } | ||
956 | |||
957 | niro | 984 | #if ENABLE_LONG_OPTS |
958 | niro | 532 | /* If S is a valid traditional offset specification with an optional |
959 | leading '+' return nonzero and set *OFFSET to the offset it denotes. */ | ||
960 | |||
961 | static int | ||
962 | parse_old_offset(const char *s, off_t *offset) | ||
963 | { | ||
964 | static const struct suffix_mult Bb[] = { | ||
965 | { "B", 1024 }, | ||
966 | { "b", 512 }, | ||
967 | niro | 984 | { "", 0 } |
968 | niro | 532 | }; |
969 | char *p; | ||
970 | int radix; | ||
971 | |||
972 | /* Skip over any leading '+'. */ | ||
973 | if (s[0] == '+') ++s; | ||
974 | |||
975 | /* Determine the radix we'll use to interpret S. If there is a '.', | ||
976 | * it's decimal, otherwise, if the string begins with '0X'or '0x', | ||
977 | * it's hexadecimal, else octal. */ | ||
978 | p = strchr(s, '.'); | ||
979 | radix = 8; | ||
980 | if (p) { | ||
981 | p[0] = '\0'; /* cheating */ | ||
982 | radix = 10; | ||
983 | } else if (s[0] == '0' && (s[1] == 'x' || s[1] == 'X')) | ||
984 | radix = 16; | ||
985 | |||
986 | *offset = xstrtooff_sfx(s, radix, Bb); | ||
987 | if (p) p[0] = '.'; | ||
988 | |||
989 | return (*offset >= 0); | ||
990 | } | ||
991 | #endif | ||
992 | |||
993 | /* Read a chunk of size BYTES_PER_BLOCK from the input files, write the | ||
994 | formatted block to standard output, and repeat until the specified | ||
995 | maximum number of bytes has been read or until all input has been | ||
996 | processed. If the last block read is smaller than BYTES_PER_BLOCK | ||
997 | and its size is not a multiple of the size associated with a format | ||
998 | spec, extend the input block with zero bytes until its length is a | ||
999 | multiple of all format spec sizes. Write the final block. Finally, | ||
1000 | write on a line by itself the offset of the byte after the last byte | ||
1001 | niro | 816 | read. */ |
1002 | niro | 532 | |
1003 | static void | ||
1004 | niro | 816 | dump(off_t current_offset, off_t end_offset) |
1005 | niro | 532 | { |
1006 | char *block[2]; | ||
1007 | int idx; | ||
1008 | size_t n_bytes_read; | ||
1009 | |||
1010 | block[0] = xmalloc(2*bytes_per_block); | ||
1011 | block[1] = block[0] + bytes_per_block; | ||
1012 | |||
1013 | idx = 0; | ||
1014 | if (limit_bytes_to_format) { | ||
1015 | while (1) { | ||
1016 | size_t n_needed; | ||
1017 | if (current_offset >= end_offset) { | ||
1018 | n_bytes_read = 0; | ||
1019 | break; | ||
1020 | } | ||
1021 | n_needed = MIN(end_offset - current_offset, | ||
1022 | (off_t) bytes_per_block); | ||
1023 | read_block(n_needed, block[idx], &n_bytes_read); | ||
1024 | if (n_bytes_read < bytes_per_block) | ||
1025 | break; | ||
1026 | assert(n_bytes_read == bytes_per_block); | ||
1027 | write_block(current_offset, n_bytes_read, | ||
1028 | block[!idx], block[idx]); | ||
1029 | current_offset += n_bytes_read; | ||
1030 | idx = !idx; | ||
1031 | } | ||
1032 | } else { | ||
1033 | while (1) { | ||
1034 | read_block(bytes_per_block, block[idx], &n_bytes_read); | ||
1035 | if (n_bytes_read < bytes_per_block) | ||
1036 | break; | ||
1037 | assert(n_bytes_read == bytes_per_block); | ||
1038 | write_block(current_offset, n_bytes_read, | ||
1039 | block[!idx], block[idx]); | ||
1040 | current_offset += n_bytes_read; | ||
1041 | idx = !idx; | ||
1042 | } | ||
1043 | } | ||
1044 | |||
1045 | if (n_bytes_read > 0) { | ||
1046 | int l_c_m; | ||
1047 | size_t bytes_to_write; | ||
1048 | |||
1049 | l_c_m = get_lcm(); | ||
1050 | |||
1051 | /* Make bytes_to_write the smallest multiple of l_c_m that | ||
1052 | is at least as large as n_bytes_read. */ | ||
1053 | bytes_to_write = l_c_m * ((n_bytes_read + l_c_m - 1) / l_c_m); | ||
1054 | |||
1055 | memset(block[idx] + n_bytes_read, 0, bytes_to_write - n_bytes_read); | ||
1056 | write_block(current_offset, bytes_to_write, | ||
1057 | block[!idx], block[idx]); | ||
1058 | current_offset += n_bytes_read; | ||
1059 | } | ||
1060 | |||
1061 | format_address(current_offset, '\n'); | ||
1062 | |||
1063 | if (limit_bytes_to_format && current_offset >= end_offset) | ||
1064 | check_and_close(); | ||
1065 | |||
1066 | free(block[0]); | ||
1067 | } | ||
1068 | |||
1069 | /* Read a single byte into *C from the concatenation of the input files | ||
1070 | named in the global array FILE_LIST. On the first call to this | ||
1071 | function, the global variable IN_STREAM is expected to be an open | ||
1072 | stream associated with the input file INPUT_FILENAME. If IN_STREAM | ||
1073 | is at end-of-file, close it and update the global variables IN_STREAM | ||
1074 | and INPUT_FILENAME so they correspond to the next file in the list. | ||
1075 | Then try to read a byte from the newly opened file. Repeat if | ||
1076 | necessary until EOF is reached for the last file in FILE_LIST, then | ||
1077 | niro | 816 | set *C to EOF and return. Subsequent calls do likewise. */ |
1078 | niro | 532 | |
1079 | static void | ||
1080 | read_char(int *c) | ||
1081 | { | ||
1082 | while (in_stream) { /* !EOF */ | ||
1083 | *c = fgetc(in_stream); | ||
1084 | if (*c != EOF) | ||
1085 | return; | ||
1086 | check_and_close(); | ||
1087 | open_next_file(); | ||
1088 | } | ||
1089 | *c = EOF; | ||
1090 | } | ||
1091 | |||
1092 | /* Read N bytes into BLOCK from the concatenation of the input files | ||
1093 | named in the global array FILE_LIST. On the first call to this | ||
1094 | function, the global variable IN_STREAM is expected to be an open | ||
1095 | stream associated with the input file INPUT_FILENAME. If all N | ||
1096 | bytes cannot be read from IN_STREAM, close IN_STREAM and update | ||
1097 | the global variables IN_STREAM and INPUT_FILENAME. Then try to | ||
1098 | read the remaining bytes from the newly opened file. Repeat if | ||
1099 | necessary until EOF is reached for the last file in FILE_LIST. | ||
1100 | On subsequent calls, don't modify BLOCK and return zero. Set | ||
1101 | *N_BYTES_IN_BUFFER to the number of bytes read. If an error occurs, | ||
1102 | it will be detected through ferror when the stream is about to be | ||
1103 | closed. If there is an error, give a message but continue reading | ||
1104 | as usual and return nonzero. Otherwise return zero. */ | ||
1105 | |||
1106 | /* STRINGS mode. Find each "string constant" in the input. | ||
1107 | A string constant is a run of at least 'string_min' ASCII | ||
1108 | graphic (or formatting) characters terminated by a null. | ||
1109 | Based on a function written by Richard Stallman for a | ||
1110 | niro | 816 | traditional version of od. */ |
1111 | niro | 532 | |
1112 | static void | ||
1113 | niro | 816 | dump_strings(off_t address, off_t end_offset) |
1114 | niro | 532 | { |
1115 | size_t bufsize = MAX(100, string_min); | ||
1116 | char *buf = xmalloc(bufsize); | ||
1117 | |||
1118 | while (1) { | ||
1119 | size_t i; | ||
1120 | int c; | ||
1121 | |||
1122 | /* See if the next 'string_min' chars are all printing chars. */ | ||
1123 | tryline: | ||
1124 | if (limit_bytes_to_format && (end_offset - string_min <= address)) | ||
1125 | break; | ||
1126 | i = 0; | ||
1127 | while (!limit_bytes_to_format || address < end_offset) { | ||
1128 | if (i == bufsize) { | ||
1129 | bufsize += bufsize/8; | ||
1130 | buf = xrealloc(buf, bufsize); | ||
1131 | } | ||
1132 | read_char(&c); | ||
1133 | if (c < 0) { /* EOF */ | ||
1134 | free(buf); | ||
1135 | return; | ||
1136 | } | ||
1137 | address++; | ||
1138 | if (!c) | ||
1139 | break; | ||
1140 | if (!ISPRINT(c)) | ||
1141 | goto tryline; /* It isn't; give up on this string. */ | ||
1142 | buf[i++] = c; /* String continues; store it all. */ | ||
1143 | } | ||
1144 | |||
1145 | if (i < string_min) /* Too short! */ | ||
1146 | goto tryline; | ||
1147 | |||
1148 | niro | 816 | /* If we get here, the string is all printable and NUL-terminated, |
1149 | niro | 532 | * so print it. It is all in 'buf' and 'i' is its length. */ |
1150 | buf[i] = 0; | ||
1151 | format_address(address - i - 1, ' '); | ||
1152 | |||
1153 | for (i = 0; (c = buf[i]); i++) { | ||
1154 | switch (c) { | ||
1155 | case '\007': fputs("\\a", stdout); break; | ||
1156 | case '\b': fputs("\\b", stdout); break; | ||
1157 | case '\f': fputs("\\f", stdout); break; | ||
1158 | case '\n': fputs("\\n", stdout); break; | ||
1159 | case '\r': fputs("\\r", stdout); break; | ||
1160 | case '\t': fputs("\\t", stdout); break; | ||
1161 | case '\v': fputs("\\v", stdout); break; | ||
1162 | niro | 816 | default: putchar(c); |
1163 | niro | 532 | } |
1164 | } | ||
1165 | putchar('\n'); | ||
1166 | } | ||
1167 | |||
1168 | /* We reach this point only if we search through | ||
1169 | (max_bytes_to_format - string_min) bytes before reaching EOF. */ | ||
1170 | free(buf); | ||
1171 | |||
1172 | check_and_close(); | ||
1173 | } | ||
1174 | |||
1175 | niro | 816 | int od_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
1176 | int od_main(int argc, char **argv) | ||
1177 | niro | 532 | { |
1178 | static const struct suffix_mult bkm[] = { | ||
1179 | { "b", 512 }, | ||
1180 | { "k", 1024 }, | ||
1181 | { "m", 1024*1024 }, | ||
1182 | niro | 984 | { "", 0 } |
1183 | niro | 532 | }; |
1184 | enum { | ||
1185 | OPT_A = 1 << 0, | ||
1186 | OPT_N = 1 << 1, | ||
1187 | OPT_a = 1 << 2, | ||
1188 | OPT_b = 1 << 3, | ||
1189 | OPT_c = 1 << 4, | ||
1190 | OPT_d = 1 << 5, | ||
1191 | OPT_f = 1 << 6, | ||
1192 | OPT_h = 1 << 7, | ||
1193 | OPT_i = 1 << 8, | ||
1194 | OPT_j = 1 << 9, | ||
1195 | OPT_l = 1 << 10, | ||
1196 | OPT_o = 1 << 11, | ||
1197 | OPT_t = 1 << 12, | ||
1198 | OPT_v = 1 << 13, | ||
1199 | OPT_x = 1 << 14, | ||
1200 | OPT_s = 1 << 15, | ||
1201 | OPT_S = 1 << 16, | ||
1202 | OPT_w = 1 << 17, | ||
1203 | niro | 984 | OPT_traditional = (1 << 18) * ENABLE_LONG_OPTS, |
1204 | niro | 532 | }; |
1205 | niro | 984 | #if ENABLE_LONG_OPTS |
1206 | niro | 816 | static const char od_longopts[] ALIGN1 = |
1207 | "skip-bytes\0" Required_argument "j" | ||
1208 | "address-radix\0" Required_argument "A" | ||
1209 | "read-bytes\0" Required_argument "N" | ||
1210 | "format\0" Required_argument "t" | ||
1211 | "output-duplicates\0" No_argument "v" | ||
1212 | "strings\0" Optional_argument "S" | ||
1213 | "width\0" Optional_argument "w" | ||
1214 | "traditional\0" No_argument "\xff" | ||
1215 | ; | ||
1216 | niro | 532 | #endif |
1217 | char *str_A, *str_N, *str_j, *str_S; | ||
1218 | llist_t *lst_t = NULL; | ||
1219 | niro | 816 | unsigned opt; |
1220 | int l_c_m; | ||
1221 | /* The old-style 'pseudo starting address' to be printed in parentheses | ||
1222 | after any true address. */ | ||
1223 | off_t pseudo_start = pseudo_start; // for gcc | ||
1224 | /* The number of input bytes to skip before formatting and writing. */ | ||
1225 | off_t n_bytes_to_skip = 0; | ||
1226 | /* The offset of the first byte after the last byte to be formatted. */ | ||
1227 | off_t end_offset = 0; | ||
1228 | /* The maximum number of bytes that will be formatted. */ | ||
1229 | off_t max_bytes_to_format = 0; | ||
1230 | niro | 532 | |
1231 | spec = NULL; | ||
1232 | format_address = format_address_std; | ||
1233 | address_base_char = 'o'; | ||
1234 | address_pad_len_char = '7'; | ||
1235 | niro | 816 | /* flag_dump_strings = 0; - already is */ |
1236 | niro | 532 | |
1237 | /* Parse command line */ | ||
1238 | niro | 816 | opt_complementary = "w+:t::"; /* -w N, -t is a list */ |
1239 | niro | 984 | #if ENABLE_LONG_OPTS |
1240 | niro | 816 | applet_long_options = od_longopts; |
1241 | niro | 532 | #endif |
1242 | niro | 816 | opt = getopt32(argv, "A:N:abcdfhij:lot:vxsS:" |
1243 | niro | 532 | "w::", // -w with optional param |
1244 | // -S was -s and also had optional parameter | ||
1245 | // but in coreutils 6.3 it was renamed and now has | ||
1246 | // _mandatory_ parameter | ||
1247 | niro | 816 | &str_A, &str_N, &str_j, &lst_t, &str_S, &bytes_per_block); |
1248 | niro | 532 | argc -= optind; |
1249 | argv += optind; | ||
1250 | if (opt & OPT_A) { | ||
1251 | niro | 816 | static const char doxn[] ALIGN1 = "doxn"; |
1252 | static const char doxn_address_base_char[] ALIGN1 = { | ||
1253 | niro | 532 | 'u', 'o', 'x', /* '?' fourth one is not important */ |
1254 | }; | ||
1255 | niro | 816 | static const uint8_t doxn_address_pad_len_char[] ALIGN1 = { |
1256 | niro | 532 | '7', '7', '6', /* '?' */ |
1257 | }; | ||
1258 | char *p; | ||
1259 | int pos; | ||
1260 | p = strchr(doxn, str_A[0]); | ||
1261 | if (!p) | ||
1262 | bb_error_msg_and_die("bad output address radix " | ||
1263 | "'%c' (must be [doxn])", str_A[0]); | ||
1264 | pos = p - doxn; | ||
1265 | if (pos == 3) format_address = format_address_none; | ||
1266 | address_base_char = doxn_address_base_char[pos]; | ||
1267 | address_pad_len_char = doxn_address_pad_len_char[pos]; | ||
1268 | } | ||
1269 | if (opt & OPT_N) { | ||
1270 | limit_bytes_to_format = 1; | ||
1271 | max_bytes_to_format = xstrtooff_sfx(str_N, 0, bkm); | ||
1272 | } | ||
1273 | if (opt & OPT_a) decode_format_string("a"); | ||
1274 | if (opt & OPT_b) decode_format_string("oC"); | ||
1275 | if (opt & OPT_c) decode_format_string("c"); | ||
1276 | if (opt & OPT_d) decode_format_string("u2"); | ||
1277 | if (opt & OPT_f) decode_format_string("fF"); | ||
1278 | if (opt & OPT_h) decode_format_string("x2"); | ||
1279 | if (opt & OPT_i) decode_format_string("d2"); | ||
1280 | if (opt & OPT_j) n_bytes_to_skip = xstrtooff_sfx(str_j, 0, bkm); | ||
1281 | if (opt & OPT_l) decode_format_string("d4"); | ||
1282 | if (opt & OPT_o) decode_format_string("o2"); | ||
1283 | //if (opt & OPT_t)... | ||
1284 | while (lst_t) { | ||
1285 | niro | 816 | decode_format_string(llist_pop(&lst_t)); |
1286 | niro | 532 | } |
1287 | niro | 816 | if (opt & OPT_v) verbose = 1; |
1288 | niro | 532 | if (opt & OPT_x) decode_format_string("x2"); |
1289 | if (opt & OPT_s) decode_format_string("d2"); | ||
1290 | if (opt & OPT_S) { | ||
1291 | string_min = 3; | ||
1292 | string_min = xstrtou_sfx(str_S, 0, bkm); | ||
1293 | flag_dump_strings = 1; | ||
1294 | } | ||
1295 | //if (opt & OPT_w)... | ||
1296 | //if (opt & OPT_traditional)... | ||
1297 | |||
1298 | if (flag_dump_strings && n_specs > 0) | ||
1299 | bb_error_msg_and_die("no type may be specified when dumping strings"); | ||
1300 | |||
1301 | /* If the --traditional option is used, there may be from | ||
1302 | * 0 to 3 remaining command line arguments; handle each case | ||
1303 | * separately. | ||
1304 | niro | 816 | * od [file] [[+]offset[.][b] [[+]label[.][b]]] |
1305 | niro | 532 | * The offset and pseudo_start have the same syntax. |
1306 | * | ||
1307 | * FIXME: POSIX 1003.1-2001 with XSI requires support for the | ||
1308 | * traditional syntax even if --traditional is not given. */ | ||
1309 | |||
1310 | niro | 984 | #if ENABLE_LONG_OPTS |
1311 | niro | 532 | if (opt & OPT_traditional) { |
1312 | off_t o1, o2; | ||
1313 | |||
1314 | if (argc == 1) { | ||
1315 | if (parse_old_offset(argv[0], &o1)) { | ||
1316 | n_bytes_to_skip = o1; | ||
1317 | --argc; | ||
1318 | ++argv; | ||
1319 | } | ||
1320 | } else if (argc == 2) { | ||
1321 | if (parse_old_offset(argv[0], &o1) | ||
1322 | && parse_old_offset(argv[1], &o2) | ||
1323 | ) { | ||
1324 | n_bytes_to_skip = o1; | ||
1325 | flag_pseudo_start = 1; | ||
1326 | pseudo_start = o2; | ||
1327 | argv += 2; | ||
1328 | argc -= 2; | ||
1329 | } else if (parse_old_offset(argv[1], &o2)) { | ||
1330 | n_bytes_to_skip = o2; | ||
1331 | --argc; | ||
1332 | argv[1] = argv[0]; | ||
1333 | ++argv; | ||
1334 | } else { | ||
1335 | bb_error_msg_and_die("invalid second operand " | ||
1336 | "in compatibility mode '%s'", argv[1]); | ||
1337 | } | ||
1338 | } else if (argc == 3) { | ||
1339 | if (parse_old_offset(argv[1], &o1) | ||
1340 | && parse_old_offset(argv[2], &o2) | ||
1341 | ) { | ||
1342 | n_bytes_to_skip = o1; | ||
1343 | flag_pseudo_start = 1; | ||
1344 | pseudo_start = o2; | ||
1345 | argv[2] = argv[0]; | ||
1346 | argv += 2; | ||
1347 | argc -= 2; | ||
1348 | } else { | ||
1349 | bb_error_msg_and_die("in compatibility mode " | ||
1350 | "the last two arguments must be offsets"); | ||
1351 | } | ||
1352 | } else if (argc > 3) { | ||
1353 | bb_error_msg_and_die("compatibility mode supports " | ||
1354 | "at most three arguments"); | ||
1355 | } | ||
1356 | |||
1357 | if (flag_pseudo_start) { | ||
1358 | if (format_address == format_address_none) { | ||
1359 | address_base_char = 'o'; | ||
1360 | address_pad_len_char = '7'; | ||
1361 | format_address = format_address_paren; | ||
1362 | } else | ||
1363 | format_address = format_address_label; | ||
1364 | } | ||
1365 | } | ||
1366 | #endif | ||
1367 | |||
1368 | if (limit_bytes_to_format) { | ||
1369 | end_offset = n_bytes_to_skip + max_bytes_to_format; | ||
1370 | if (end_offset < n_bytes_to_skip) | ||
1371 | bb_error_msg_and_die("skip-bytes + read-bytes is too large"); | ||
1372 | } | ||
1373 | |||
1374 | if (n_specs == 0) { | ||
1375 | decode_format_string("o2"); | ||
1376 | n_specs = 1; | ||
1377 | } | ||
1378 | |||
1379 | /* If no files were listed on the command line, | ||
1380 | set the global pointer FILE_LIST so that it | ||
1381 | references the null-terminated list of one name: "-". */ | ||
1382 | niro | 816 | file_list = bb_argv_dash; |
1383 | niro | 532 | if (argc > 0) { |
1384 | /* Set the global pointer FILE_LIST so that it | ||
1385 | references the first file-argument on the command-line. */ | ||
1386 | file_list = (char const *const *) argv; | ||
1387 | } | ||
1388 | |||
1389 | /* open the first input file */ | ||
1390 | open_next_file(); | ||
1391 | /* skip over any unwanted header bytes */ | ||
1392 | skip(n_bytes_to_skip); | ||
1393 | if (!in_stream) | ||
1394 | niro | 816 | return EXIT_FAILURE; |
1395 | niro | 532 | |
1396 | pseudo_offset = (flag_pseudo_start ? pseudo_start - n_bytes_to_skip : 0); | ||
1397 | |||
1398 | /* Compute output block length. */ | ||
1399 | l_c_m = get_lcm(); | ||
1400 | |||
1401 | if (opt & OPT_w) { /* -w: width */ | ||
1402 | if (!bytes_per_block || bytes_per_block % l_c_m != 0) { | ||
1403 | niro | 816 | bb_error_msg("warning: invalid width %u; using %d instead", |
1404 | (unsigned)bytes_per_block, l_c_m); | ||
1405 | niro | 532 | bytes_per_block = l_c_m; |
1406 | } | ||
1407 | } else { | ||
1408 | bytes_per_block = l_c_m; | ||
1409 | if (l_c_m < DEFAULT_BYTES_PER_BLOCK) | ||
1410 | bytes_per_block *= DEFAULT_BYTES_PER_BLOCK / l_c_m; | ||
1411 | } | ||
1412 | |||
1413 | #ifdef DEBUG | ||
1414 | for (i = 0; i < n_specs; i++) { | ||
1415 | printf("%d: fmt=\"%s\" width=%d\n", | ||
1416 | i, spec[i].fmt_string, width_bytes[spec[i].size]); | ||
1417 | } | ||
1418 | #endif | ||
1419 | |||
1420 | if (flag_dump_strings) | ||
1421 | niro | 816 | dump_strings(n_bytes_to_skip, end_offset); |
1422 | niro | 532 | else |
1423 | niro | 816 | dump(n_bytes_to_skip, end_offset); |
1424 | niro | 532 | |
1425 | if (fclose(stdin) == EOF) | ||
1426 | bb_perror_msg_and_die(bb_msg_standard_input); | ||
1427 | |||
1428 | niro | 816 | return ioerror; |
1429 | niro | 532 | } |