Annotation of src/usr.bin/sort/sort.c, Revision 1.53
1.53 ! millert 1: /* $OpenBSD: sort.c,v 1.52 2015/03/31 12:16:02 millert Exp $ */
1.1 millert 2:
3: /*-
1.44 millert 4: * Copyright (C) 2009 Gabor Kovesdan <gabor@FreeBSD.org>
5: * Copyright (C) 2012 Oleg Moskalenko <mom040267@gmail.com>
6: * All rights reserved.
1.1 millert 7: *
8: * Redistribution and use in source and binary forms, with or without
9: * modification, are permitted provided that the following conditions
10: * are met:
11: * 1. Redistributions of source code must retain the above copyright
12: * notice, this list of conditions and the following disclaimer.
13: * 2. Redistributions in binary form must reproduce the above copyright
14: * notice, this list of conditions and the following disclaimer in the
15: * documentation and/or other materials provided with the distribution.
16: *
1.44 millert 17: * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
1.1 millert 18: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
1.44 millert 20: * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
1.1 millert 21: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27: * SUCH DAMAGE.
28: */
29:
1.48 millert 30: #include <sys/resource.h>
1.44 millert 31: #include <sys/stat.h>
1.48 millert 32: #include <sys/sysctl.h>
1.44 millert 33: #include <sys/types.h>
1.1 millert 34:
1.44 millert 35: #include <err.h>
36: #include <errno.h>
37: #include <getopt.h>
38: #include <limits.h>
1.16 ericj 39: #include <locale.h>
1.44 millert 40: #include <md5.h>
41: #include <regex.h>
1.1 millert 42: #include <signal.h>
1.44 millert 43: #include <stdbool.h>
44: #include <stdio.h>
1.1 millert 45: #include <stdlib.h>
46: #include <string.h>
47: #include <unistd.h>
1.44 millert 48: #include <wchar.h>
49: #include <wctype.h>
50:
51: #include "coll.h"
52: #include "file.h"
53: #include "sort.h"
54:
55: #define OPTIONS "bCcdfgHhik:Mmno:RrS:st:T:uVz"
56:
57: static bool need_random;
58: static const char *random_source;
59:
60: MD5_CTX md5_ctx;
61:
62: struct sort_opts sort_opts_vals;
63:
64: bool debug_sort;
65: bool need_hint;
66:
67: static bool gnusort_numeric_compatibility;
68:
69: static struct sort_mods default_sort_mods_object;
70: struct sort_mods * const default_sort_mods = &default_sort_mods_object;
71:
72: static bool print_symbols_on_debug;
73:
74: /*
75: * Arguments from file (when file0-from option is used:
76: */
77: static size_t argc_from_file0 = (size_t)-1;
78: static char **argv_from_file0;
79:
80: /*
81: * Placeholder symbols for options which have no single-character equivalent
82: */
83: enum {
84: SORT_OPT = CHAR_MAX + 1,
85: HELP_OPT,
86: FF_OPT,
87: BS_OPT,
88: VERSION_OPT,
89: DEBUG_OPT,
90: RANDOMSOURCE_OPT,
91: COMPRESSPROGRAM_OPT,
92: QSORT_OPT,
93: HEAPSORT_OPT,
94: RADIXSORT_OPT,
95: MMAP_OPT
96: };
97:
98: #define NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS 6
99: static const char mutually_exclusive_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] = { 'M', 'n', 'g', 'R', 'h', 'V' };
100:
101: static const struct option long_options[] = {
102: { "batch-size", required_argument, NULL, BS_OPT },
103: { "buffer-size", required_argument, NULL, 'S' },
104: { "check", optional_argument, NULL, 'c' },
105: { "check=silent|quiet", optional_argument, NULL, 'C' },
106: { "compress-program", required_argument, NULL, COMPRESSPROGRAM_OPT },
107: { "debug", no_argument, NULL, DEBUG_OPT },
108: { "dictionary-order", no_argument, NULL, 'd' },
109: { "field-separator", required_argument, NULL, 't' },
110: { "files0-from", required_argument, NULL, FF_OPT },
111: { "general-numeric-sort", no_argument, NULL, 'g' },
112: { "heapsort", no_argument, NULL, HEAPSORT_OPT },
113: { "help", no_argument, NULL, HELP_OPT },
114: { "human-numeric-sort", no_argument, NULL, 'h' },
115: { "ignore-leading-blanks", no_argument, NULL, 'b' },
116: { "ignore-case", no_argument, NULL, 'f' },
117: { "ignore-nonprinting", no_argument, NULL, 'i' },
118: { "key", required_argument, NULL, 'k' },
119: { "merge", no_argument, NULL, 'm' },
120: { "mergesort", no_argument, NULL, 'H' },
121: { "mmap", no_argument, NULL, MMAP_OPT },
122: { "month-sort", no_argument, NULL, 'M' },
123: { "numeric-sort", no_argument, NULL, 'n' },
124: { "output", required_argument, NULL, 'o' },
125: { "qsort", no_argument, NULL, QSORT_OPT },
126: { "radixsort", no_argument, NULL, RADIXSORT_OPT },
127: { "random-sort", no_argument, NULL, 'R' },
128: { "random-source", required_argument, NULL, RANDOMSOURCE_OPT },
129: { "reverse", no_argument, NULL, 'r' },
130: { "sort", required_argument, NULL, SORT_OPT },
131: { "stable", no_argument, NULL, 's' },
132: { "temporary-directory", required_argument, NULL, 'T' },
133: { "unique", no_argument, NULL, 'u' },
134: { "version", no_argument, NULL, VERSION_OPT },
135: { "version-sort", no_argument, NULL, 'V' },
136: { "zero-terminated", no_argument, NULL, 'z' },
137: { NULL, no_argument, NULL, 0 }
138: };
139:
140: /*
141: * Check where sort modifier is present
142: */
143: static bool
144: sort_modifier_empty(struct sort_mods *sm)
145: {
146:
147: if (sm == NULL)
148: return true;
149: return !(sm->Mflag || sm->Vflag || sm->nflag || sm->gflag ||
150: sm->rflag || sm->Rflag || sm->hflag || sm->dflag || sm->fflag);
151: }
152:
153: /*
154: * Print out usage text.
155: */
156: static __dead void
157: usage(int exit_val)
158: {
1.1 millert 159:
1.44 millert 160: fprintf(exit_val ? stderr : stdout,
1.46 jmc 161: "usage: %s [-bCcdfgHhiMmnRrsuVz] [-k field1[,field2]] [-o output] "
1.45 jmc 162: "[-S size]\n\t[-T dir] [-t char] [file ...]\n", getprogname());
1.44 millert 163: exit(exit_val);
164: }
1.4 millert 165:
1.1 millert 166: /*
1.44 millert 167: * Read input file names from a file (file0-from option).
1.1 millert 168: */
1.44 millert 169: static void
170: read_fns_from_file0(const char *fn)
171: {
1.47 millert 172: FILE *f;
173: char *line = NULL;
174: size_t linesize = 0;
175: ssize_t linelen;
176:
177: if (fn == NULL)
178: return;
179:
180: f = fopen(fn, "r");
181: if (f == NULL)
182: err(2, "%s", fn);
183:
184: while ((linelen = getdelim(&line, &linesize, '\0', f)) != -1) {
185: if (*line != '\0') {
186: if (argc_from_file0 == (size_t)-1)
187: argc_from_file0 = 0;
188: ++argc_from_file0;
189: argv_from_file0 = sort_reallocarray(argv_from_file0,
190: argc_from_file0, sizeof(char *));
191: argv_from_file0[argc_from_file0 - 1] = line;
192: } else {
193: free(line);
1.44 millert 194: }
1.47 millert 195: line = NULL;
196: linesize = 0;
1.44 millert 197: }
1.47 millert 198: if (ferror(f))
199: err(2, "%s: getdelim", fn);
200:
201: closefile(f, fn);
1.44 millert 202: }
1.4 millert 203:
1.1 millert 204: /*
1.44 millert 205: * Check how much RAM is available for the sort.
1.1 millert 206: */
1.44 millert 207: static void
208: set_hw_params(void)
209: {
1.48 millert 210: long long user_memory;
211: struct rlimit rl;
212: size_t len;
213: int mib[] = { CTL_HW, HW_USERMEM64 };
214:
215: /* Get total user (non-kernel) memory. */
216: len = sizeof(user_memory);
217: if (sysctl(mib, 2, &user_memory, &len, NULL, 0) == -1)
218: user_memory = -1;
219:
220: /* Increase our data size to the max */
221: if (getrlimit(RLIMIT_DATA, &rl) == 0) {
222: free_memory = (unsigned long long)rl.rlim_cur;
223: rl.rlim_cur = rl.rlim_max;
224: if (setrlimit(RLIMIT_DATA, &rl) == 0) {
225: free_memory = (unsigned long long)rl.rlim_max;
226: } else {
227: warn("Can't set resource limit to max data size");
228: }
229: } else
230: warn("Can't get resource limit for data size");
1.1 millert 231:
1.48 millert 232: /* We prefer to use temp files rather than swap space. */
233: if (user_memory != -1 && free_memory > user_memory)
234: free_memory = user_memory;
1.44 millert 235:
236: available_free_memory = free_memory / 2;
237: }
238:
239: /*
240: * Convert "plain" symbol to wide symbol, with default value.
241: */
242: static void
243: conv_mbtowc(wchar_t *wc, const char *c, const wchar_t def)
244: {
1.1 millert 245:
1.44 millert 246: if (wc && c) {
247: int res;
1.1 millert 248:
1.44 millert 249: res = mbtowc(wc, c, MB_CUR_MAX);
250: if (res < 1)
251: *wc = def;
1.12 millert 252: }
1.44 millert 253: }
1.12 millert 254:
1.44 millert 255: /*
256: * Set current locale symbols.
257: */
258: static void
259: set_locale(void)
1.1 millert 260: {
1.44 millert 261: struct lconv *lc;
262: const char *locale;
1.4 millert 263:
1.16 ericj 264: setlocale(LC_ALL, "");
265:
1.44 millert 266: lc = localeconv();
267:
268: if (lc) {
269: /* obtain LC_NUMERIC info */
270: /* Convert to wide char form */
271: conv_mbtowc(&symbol_decimal_point, lc->decimal_point,
272: symbol_decimal_point);
273: conv_mbtowc(&symbol_thousands_sep, lc->thousands_sep,
274: symbol_thousands_sep);
275: conv_mbtowc(&symbol_positive_sign, lc->positive_sign,
276: symbol_positive_sign);
277: conv_mbtowc(&symbol_negative_sign, lc->negative_sign,
278: symbol_negative_sign);
279: }
280:
281: if (getenv("GNUSORT_NUMERIC_COMPATIBILITY"))
282: gnusort_numeric_compatibility = true;
283:
284: locale = setlocale(LC_COLLATE, NULL);
285: if (locale != NULL) {
286: char *tmpl;
287: const char *byteclocale;
288:
289: tmpl = sort_strdup(locale);
290: byteclocale = setlocale(LC_COLLATE, "C");
291: if (byteclocale && strcmp(byteclocale, tmpl) == 0) {
292: byte_sort = true;
293: } else {
294: byteclocale = setlocale(LC_COLLATE, "POSIX");
295: if (byteclocale && strcmp(byteclocale, tmpl) == 0)
296: byte_sort = true;
297: else
298: setlocale(LC_COLLATE, tmpl);
299: }
300: sort_free(tmpl);
301: }
302: if (!byte_sort)
303: sort_mb_cur_max = MB_CUR_MAX;
304: }
305:
306: /*
307: * Set directory temporary files.
308: */
309: static void
310: set_tmpdir(void)
311: {
1.53 ! millert 312: if (!issetugid()) {
! 313: char *td;
1.44 millert 314:
1.53 ! millert 315: td = getenv("TMPDIR");
! 316: if (td != NULL)
! 317: tmpdir = sort_strdup(td);
! 318: }
1.44 millert 319: }
320:
321: /*
322: * Parse -S option.
323: */
324: static unsigned long long
325: parse_memory_buffer_value(const char *value)
326: {
327:
328: if (value == NULL)
329: return available_free_memory;
330: else {
331: char *endptr;
332: unsigned long long membuf;
333:
334: endptr = NULL;
335: errno = 0;
336: membuf = strtoll(value, &endptr, 10);
337:
338: if (errno != 0) {
339: warn("Wrong memory buffer specification");
340: membuf = available_free_memory;
341: } else {
342: switch (*endptr){
343: case 'Y':
344: membuf *= 1024;
345: /* FALLTHROUGH */
346: case 'Z':
347: membuf *= 1024;
348: /* FALLTHROUGH */
349: case 'E':
350: membuf *= 1024;
351: /* FALLTHROUGH */
352: case 'P':
353: membuf *= 1024;
354: /* FALLTHROUGH */
355: case 'T':
356: membuf *= 1024;
357: /* FALLTHROUGH */
358: case 'G':
359: membuf *= 1024;
360: /* FALLTHROUGH */
361: case 'M':
362: membuf *= 1024;
363: /* FALLTHROUGH */
364: case '\0':
365: case 'K':
366: membuf *= 1024;
367: /* FALLTHROUGH */
368: case 'b':
369: break;
370: case '%':
371: membuf = (available_free_memory * membuf) /
372: 100;
373: break;
374: default:
375: warnc(EINVAL, "%s", optarg);
376: membuf = available_free_memory;
377: }
378: }
379: return membuf;
380: }
381: }
382:
383: /*
384: * Signal handler that clears the temporary files.
385: */
386: static void
1.49 millert 387: sig_handler(int sig __unused)
1.44 millert 388: {
389:
390: clear_tmp_files();
1.50 millert 391: _exit(2);
1.44 millert 392: }
393:
394: /*
395: * Set signal handler on panic signals.
396: */
397: static void
398: set_signal_handler(void)
399: {
400: struct sigaction sa;
1.49 millert 401: int i, signals[] = {SIGTERM, SIGHUP, SIGINT, SIGQUIT, SIGUSR1, SIGUSR2,
402: SIGPIPE, SIGXCPU, SIGXFSZ, SIGVTALRM, SIGPROF, 0};
1.44 millert 403:
404: memset(&sa, 0, sizeof(sa));
1.49 millert 405: sigfillset(&sa.sa_mask);
406: sa.sa_flags = SA_RESTART;
407: sa.sa_handler = sig_handler;
408:
409: for (i = 0; signals[i] != 0; i++) {
410: if (sigaction(signals[i], &sa, NULL) < 0) {
411: warn("sigaction(%d)", i);
412: continue;
413: }
1.44 millert 414: }
415: }
416:
417: /*
418: * Print "unknown" message and exit with status 2.
419: */
420: static void
421: unknown(const char *what)
422: {
423:
424: errx(2, "Unknown feature: %s", what);
425: }
426:
427: /*
428: * Check whether contradictory input options are used.
429: */
430: static void
431: check_mutually_exclusive_flags(char c, bool *mef_flags)
432: {
433: int i, fo_index, mec;
434: bool found_others, found_this;
435:
436: found_others = found_this =false;
437: fo_index = 0;
438:
439: for (i = 0; i < NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS; i++) {
440: mec = mutually_exclusive_flags[i];
441:
442: if (mec != c) {
443: if (mef_flags[i]) {
444: if (found_this)
445: errx(1, "%c:%c: mutually exclusive flags", c, mec);
446: found_others = true;
447: fo_index = i;
448: }
449: } else {
450: if (found_others)
451: errx(1, "%c:%c: mutually exclusive flags", c, mutually_exclusive_flags[fo_index]);
452: mef_flags[i] = true;
453: found_this = true;
454: }
455: }
456: }
457:
458: /*
459: * Initialise sort opts data.
460: */
461: static void
462: set_sort_opts(void)
463: {
464:
465: memset(&default_sort_mods_object, 0,
466: sizeof(default_sort_mods_object));
467: memset(&sort_opts_vals, 0, sizeof(sort_opts_vals));
468: default_sort_mods_object.func =
469: get_sort_func(&default_sort_mods_object);
470: }
471:
472: /*
473: * Set a sort modifier on a sort modifiers object.
474: */
475: static bool
476: set_sort_modifier(struct sort_mods *sm, int c)
477: {
478:
479: if (sm) {
480: switch (c){
481: case 'b':
482: sm->bflag = true;
1.1 millert 483: break;
484: case 'd':
1.44 millert 485: sm->dflag = true;
486: break;
1.4 millert 487: case 'f':
1.44 millert 488: sm->fflag = true;
489: break;
490: case 'g':
491: sm->gflag = true;
492: need_hint = true;
493: break;
1.1 millert 494: case 'i':
1.44 millert 495: sm->iflag = true;
1.1 millert 496: break;
497: case 'R':
1.44 millert 498: sm->Rflag = true;
499: need_random = true;
1.1 millert 500: break;
1.44 millert 501: case 'M':
502: initialise_months();
503: sm->Mflag = true;
504: need_hint = true;
1.1 millert 505: break;
1.44 millert 506: case 'n':
507: sm->nflag = true;
508: need_hint = true;
509: print_symbols_on_debug = true;
1.1 millert 510: break;
1.44 millert 511: case 'r':
512: sm->rflag = true;
1.1 millert 513: break;
1.44 millert 514: case 'V':
515: sm->Vflag = true;
1.1 millert 516: break;
1.44 millert 517: case 'h':
518: sm->hflag = true;
519: need_hint = true;
520: print_symbols_on_debug = true;
1.1 millert 521: break;
1.8 deraadt 522: default:
1.44 millert 523: return false;
1.1 millert 524: }
1.44 millert 525: sort_opts_vals.complex_sort = true;
526: sm->func = get_sort_func(sm);
527: }
528: return true;
529: }
530:
531: /*
532: * Parse POS in -k option.
533: */
534: static int
535: parse_pos(const char *s, struct key_specs *ks, bool *mef_flags, bool second)
536: {
537: regmatch_t pmatch[4];
538: regex_t re;
539: char *c, *f;
540: const char *sregexp = "^([0-9]+)(\\.[0-9]+)?([bdfirMngRhV]+)?$";
541: size_t len, nmatch;
542: int ret;
543:
544: ret = -1;
545: nmatch = 4;
546: c = f = NULL;
547:
548: if (regcomp(&re, sregexp, REG_EXTENDED) != 0)
549: return -1;
550:
551: if (regexec(&re, s, nmatch, pmatch, 0) != 0)
552: goto end;
553:
554: if (pmatch[0].rm_eo <= pmatch[0].rm_so)
555: goto end;
556:
557: if (pmatch[1].rm_eo <= pmatch[1].rm_so)
558: goto end;
559:
560: len = pmatch[1].rm_eo - pmatch[1].rm_so;
561: f = sort_malloc((len + 1) * sizeof(char));
562:
563: strncpy(f, s + pmatch[1].rm_so, len);
564: f[len] = '\0';
565:
566: if (second) {
567: errno = 0;
568: ks->f2 = (size_t) strtoul(f, NULL, 10);
569: if (errno != 0)
570: err(2, "-k");
571: if (ks->f2 == 0) {
572: warn("0 field in key specs");
573: goto end;
574: }
575: } else {
576: errno = 0;
577: ks->f1 = (size_t) strtoul(f, NULL, 10);
578: if (errno != 0)
579: err(2, "-k");
580: if (ks->f1 == 0) {
581: warn("0 field in key specs");
582: goto end;
583: }
584: }
585:
586: if (pmatch[2].rm_eo > pmatch[2].rm_so) {
587: len = pmatch[2].rm_eo - pmatch[2].rm_so - 1;
588: c = sort_malloc((len + 1) * sizeof(char));
589:
590: strncpy(c, s + pmatch[2].rm_so + 1, len);
591: c[len] = '\0';
592:
593: if (second) {
594: errno = 0;
595: ks->c2 = (size_t) strtoul(c, NULL, 10);
596: if (errno != 0)
597: err(2, "-k");
598: } else {
599: errno = 0;
600: ks->c1 = (size_t) strtoul(c, NULL, 10);
601: if (errno != 0)
602: err(2, "-k");
603: if (ks->c1 == 0) {
604: warn("0 column in key specs");
605: goto end;
606: }
607: }
608: } else {
609: if (second)
610: ks->c2 = 0;
611: else
612: ks->c1 = 1;
613: }
614:
615: if (pmatch[3].rm_eo > pmatch[3].rm_so) {
616: regoff_t i = 0;
617:
618: for (i = pmatch[3].rm_so; i < pmatch[3].rm_eo; i++) {
619: check_mutually_exclusive_flags(s[i], mef_flags);
620: if (s[i] == 'b') {
621: if (second)
622: ks->pos2b = true;
623: else
624: ks->pos1b = true;
625: } else if (!set_sort_modifier(&(ks->sm), s[i]))
626: goto end;
627: }
628: }
629:
630: ret = 0;
631:
632: end:
633:
634: if (c)
635: sort_free(c);
636: if (f)
637: sort_free(f);
638: regfree(&re);
639:
640: return ret;
641: }
642:
643: /*
644: * Parse -k option value.
645: */
646: static int
647: parse_k(const char *s, struct key_specs *ks)
648: {
649: int ret = -1;
650: bool mef_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] =
651: { false, false, false, false, false, false };
652:
653: if (s && *s) {
654: char *sptr;
655:
656: sptr = strchr(s, ',');
657: if (sptr) {
658: size_t size1;
659: char *pos1, *pos2;
660:
661: size1 = sptr - s;
662:
663: if (size1 < 1)
664: return -1;
665: pos1 = sort_malloc((size1 + 1) * sizeof(char));
666:
667: strncpy(pos1, s, size1);
668: pos1[size1] = '\0';
669:
670: ret = parse_pos(pos1, ks, mef_flags, false);
671:
672: sort_free(pos1);
673: if (ret < 0)
674: return ret;
675:
676: pos2 = sort_strdup(sptr + 1);
677: ret = parse_pos(pos2, ks, mef_flags, true);
678: sort_free(pos2);
679: } else
680: ret = parse_pos(s, ks, mef_flags, false);
1.1 millert 681: }
1.4 millert 682:
1.44 millert 683: return ret;
684: }
685:
686: /*
687: * Parse POS in +POS -POS option.
688: */
689: static int
690: parse_pos_obs(const char *s, int *nf, int *nc, char *sopts)
691: {
692: regex_t re;
693: regmatch_t pmatch[4];
694: char *c, *f;
695: const char *sregexp = "^([0-9]+)(\\.[0-9]+)?([A-Za-z]+)?$";
696: int ret;
697: size_t len, nmatch;
698:
699: ret = -1;
700: nmatch = 4;
701: c = f = NULL;
702: *nc = *nf = 0;
703:
704: if (regcomp(&re, sregexp, REG_EXTENDED) != 0)
705: return -1;
706:
707: if (regexec(&re, s, nmatch, pmatch, 0) != 0)
708: goto end;
709:
710: if (pmatch[0].rm_eo <= pmatch[0].rm_so)
711: goto end;
712:
713: if (pmatch[1].rm_eo <= pmatch[1].rm_so)
714: goto end;
715:
716: len = pmatch[1].rm_eo - pmatch[1].rm_so;
717: f = sort_malloc((len + 1) * sizeof(char));
718:
719: strncpy(f, s + pmatch[1].rm_so, len);
720: f[len] = '\0';
721:
722: errno = 0;
723: *nf = (size_t) strtoul(f, NULL, 10);
724: if (errno != 0)
725: errx(2, "Invalid key position");
726:
727: if (pmatch[2].rm_eo > pmatch[2].rm_so) {
728: len = pmatch[2].rm_eo - pmatch[2].rm_so - 1;
729: c = sort_malloc((len + 1) * sizeof(char));
730:
731: strncpy(c, s + pmatch[2].rm_so + 1, len);
732: c[len] = '\0';
733:
734: errno = 0;
735: *nc = (size_t) strtoul(c, NULL, 10);
736: if (errno != 0)
737: errx(2, "Invalid key position");
738: }
739:
740: if (pmatch[3].rm_eo > pmatch[3].rm_so) {
741:
742: len = pmatch[3].rm_eo - pmatch[3].rm_so;
1.4 millert 743:
1.44 millert 744: strncpy(sopts, s + pmatch[3].rm_so, len);
745: sopts[len] = '\0';
1.1 millert 746: }
1.4 millert 747:
1.44 millert 748: ret = 0;
1.4 millert 749:
1.44 millert 750: end:
751: if (c)
752: sort_free(c);
753: if (f)
754: sort_free(f);
755: regfree(&re);
756:
757: return ret;
758: }
759:
760: /*
761: * "Translate" obsolete +POS1 -POS2 syntax into new -kPOS1,POS2 syntax
762: */
763: static void
764: fix_obsolete_keys(int *argc, char **argv)
765: {
766: char sopt[129];
767: int i;
768:
769: for (i = 1; i < *argc; i++) {
770: char *arg1;
771:
772: arg1 = argv[i];
773:
774: if (strlen(arg1) > 1 && arg1[0] == '+') {
775: int c1, f1;
776: char sopts1[128];
777:
778: sopts1[0] = 0;
779: c1 = f1 = 0;
780:
781: if (parse_pos_obs(arg1 + 1, &f1, &c1, sopts1) < 0)
782: continue;
1.1 millert 783: else {
1.44 millert 784: f1 += 1;
785: c1 += 1;
786: if (i + 1 < *argc) {
787: char *arg2 = argv[i + 1];
788:
789: if (strlen(arg2) > 1 &&
790: arg2[0] == '-') {
791: int c2, f2;
792: char sopts2[128];
793:
794: sopts2[0] = 0;
795: c2 = f2 = 0;
796:
797: if (parse_pos_obs(arg2 + 1,
798: &f2, &c2, sopts2) >= 0) {
799: int j;
800: if (c2 > 0)
801: f2 += 1;
802: snprintf(sopt,
803: sizeof(sopt),
804: "-k%d.%d%s,%d.%d%s",
805: f1, c1, sopts1, f2,
806: c2, sopts2);
807: argv[i] = sort_strdup(sopt);
808: for (j = i + 1; j + 1 < *argc; j++)
809: argv[j] = argv[j + 1];
810: *argc -= 1;
811: continue;
812: }
813: }
814: }
815: snprintf(sopt, sizeof(sopt), "-k%d.%d%s",
816: f1, c1, sopts1);
817: argv[i] = sort_strdup(sopt);
1.1 millert 818: }
1.44 millert 819: }
1.1 millert 820: }
1.44 millert 821: }
822:
823: /*
824: * Set random seed
825: */
826: static void
827: set_random_seed(void)
828: {
829: if (!need_random)
830: return;
1.4 millert 831:
1.44 millert 832: MD5Init(&md5_ctx);
833: if (random_source != NULL) {
834: unsigned char buf[BUFSIZ];
835: size_t nr;
836: FILE *fp;
837:
838: if ((fp = fopen(random_source, "r")) == NULL)
839: err(2, "%s", random_source);
840: while ((nr = fread(buf, 1, sizeof(buf), fp)) != 0)
841: MD5Update(&md5_ctx, buf, nr);
842: if (ferror(fp))
843: err(2, "%s", random_source);
844: fclose(fp);
1.1 millert 845: } else {
1.44 millert 846: unsigned char rsd[1024];
847:
848: arc4random_buf(rsd, sizeof(rsd));
849: MD5Update(&md5_ctx, rsd, sizeof(rsd));
850: }
851: }
852:
853: /*
854: * Main function.
855: */
856: int
857: main(int argc, char *argv[])
858: {
859: char *outfile, *real_outfile;
860: int c, result;
861: size_t i;
862: bool mef_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] =
863: { false, false, false, false, false, false };
864:
865: result = 0;
1.51 millert 866: outfile = "-";
1.44 millert 867: real_outfile = NULL;
868:
869: struct sort_mods *sm = &default_sort_mods_object;
870:
871: init_tmp_files();
872:
873: set_signal_handler();
874:
1.51 millert 875: atexit(clear_tmp_files);
876:
1.44 millert 877: set_hw_params();
878: set_locale();
879: set_tmpdir();
880: set_sort_opts();
881:
882: fix_obsolete_keys(&argc, argv);
883:
884: while (((c = getopt_long(argc, argv, OPTIONS, long_options, NULL))
885: != -1)) {
886:
887: check_mutually_exclusive_flags(c, mef_flags);
888:
889: if (!set_sort_modifier(sm, c)) {
890:
891: switch (c) {
892: case 'c':
893: sort_opts_vals.cflag = true;
894: if (optarg) {
895: if (!strcmp(optarg, "diagnose-first"))
896: ;
897: else if (!strcmp(optarg, "silent") ||
898: !strcmp(optarg, "quiet"))
899: sort_opts_vals.csilentflag = true;
900: else if (*optarg)
901: unknown(optarg);
902: }
903: break;
904: case 'C':
905: sort_opts_vals.cflag = true;
906: sort_opts_vals.csilentflag = true;
907: break;
908: case 'k':
909: {
910: sort_opts_vals.complex_sort = true;
911: sort_opts_vals.kflag = true;
912:
913: keys_num++;
914: keys = sort_reallocarray(keys, keys_num,
915: sizeof(struct key_specs));
916: memset(&(keys[keys_num - 1]), 0,
917: sizeof(struct key_specs));
918:
919: if (parse_k(optarg, &(keys[keys_num - 1]))
920: < 0) {
921: errc(2, EINVAL, "-k %s", optarg);
922: }
923:
924: break;
925: }
926: case 'm':
927: sort_opts_vals.mflag = true;
928: break;
929: case 'o':
1.51 millert 930: outfile = optarg;
1.44 millert 931: break;
932: case 's':
933: sort_opts_vals.sflag = true;
934: break;
935: case 'S':
936: available_free_memory =
937: parse_memory_buffer_value(optarg);
938: break;
939: case 'T':
940: tmpdir = sort_strdup(optarg);
941: break;
942: case 't':
943: while (strlen(optarg) > 1) {
944: if (optarg[0] != '\\') {
945: errc(2, EINVAL, "%s", optarg);
946: }
947: optarg += 1;
948: if (*optarg == '0') {
949: *optarg = 0;
950: break;
951: }
952: }
953: sort_opts_vals.tflag = true;
954: sort_opts_vals.field_sep = btowc(optarg[0]);
955: if (sort_opts_vals.field_sep == WEOF) {
956: errno = EINVAL;
957: err(2, NULL);
958: }
959: if (!gnusort_numeric_compatibility) {
960: if (symbol_decimal_point == sort_opts_vals.field_sep)
961: symbol_decimal_point = WEOF;
962: if (symbol_thousands_sep == sort_opts_vals.field_sep)
963: symbol_thousands_sep = WEOF;
964: if (symbol_negative_sign == sort_opts_vals.field_sep)
965: symbol_negative_sign = WEOF;
966: if (symbol_positive_sign == sort_opts_vals.field_sep)
967: symbol_positive_sign = WEOF;
968: }
969: break;
970: case 'u':
971: sort_opts_vals.uflag = true;
972: /* stable sort for the correct unique val */
973: sort_opts_vals.sflag = true;
974: break;
975: case 'z':
976: sort_opts_vals.zflag = true;
977: break;
978: case SORT_OPT:
979: if (optarg) {
980: if (!strcmp(optarg, "general-numeric"))
981: set_sort_modifier(sm, 'g');
982: else if (!strcmp(optarg, "human-numeric"))
983: set_sort_modifier(sm, 'h');
984: else if (!strcmp(optarg, "numeric"))
985: set_sort_modifier(sm, 'n');
986: else if (!strcmp(optarg, "month"))
987: set_sort_modifier(sm, 'M');
988: else if (!strcmp(optarg, "random"))
989: set_sort_modifier(sm, 'R');
990: else
991: unknown(optarg);
992: }
993: break;
994: case QSORT_OPT:
995: sort_opts_vals.sort_method = SORT_QSORT;
996: break;
997: case 'H':
998: sort_opts_vals.sort_method = SORT_MERGESORT;
999: break;
1000: case MMAP_OPT:
1001: use_mmap = true;
1002: break;
1003: case HEAPSORT_OPT:
1004: sort_opts_vals.sort_method = SORT_HEAPSORT;
1005: break;
1006: case RADIXSORT_OPT:
1007: sort_opts_vals.sort_method = SORT_RADIXSORT;
1008: break;
1009: case RANDOMSOURCE_OPT:
1010: random_source = strdup(optarg);
1011: break;
1012: case COMPRESSPROGRAM_OPT:
1013: compress_program = strdup(optarg);
1014: break;
1015: case FF_OPT:
1016: read_fns_from_file0(optarg);
1017: break;
1018: case BS_OPT:
1019: {
1020: errno = 0;
1021: long mof = strtol(optarg, NULL, 10);
1022: if (errno != 0)
1023: err(2, "--batch-size");
1024: if (mof >= 2)
1025: max_open_files = (size_t) mof + 1;
1026: }
1027: break;
1028: case VERSION_OPT:
1029: printf("%s\n", VERSION);
1030: exit(EXIT_SUCCESS);
1031: /* NOTREACHED */
1032: break;
1033: case DEBUG_OPT:
1034: debug_sort = true;
1035: break;
1036: case HELP_OPT:
1037: usage(0);
1038: /* NOTREACHED */
1039: break;
1040: default:
1041: usage(2);
1042: /* NOTREACHED */
1043: }
1044: }
1045: }
1046:
1047: argc -= optind;
1048: argv += optind;
1049:
1050: if (keys_num == 0) {
1051: keys_num = 1;
1052: keys = sort_realloc(keys, sizeof(struct key_specs));
1053: memset(&(keys[0]), 0, sizeof(struct key_specs));
1054: keys[0].c1 = 1;
1055: keys[0].pos1b = default_sort_mods->bflag;
1056: keys[0].pos2b = default_sort_mods->bflag;
1057: memcpy(&(keys[0].sm), default_sort_mods,
1058: sizeof(struct sort_mods));
1059: }
1060:
1061: for (i = 0; i < keys_num; i++) {
1062: struct key_specs *ks;
1063:
1064: ks = &(keys[i]);
1065:
1066: if (sort_modifier_empty(&(ks->sm)) && !(ks->pos1b) &&
1067: !(ks->pos2b)) {
1068: ks->pos1b = sm->bflag;
1069: ks->pos2b = sm->bflag;
1070: memcpy(&(ks->sm), sm, sizeof(struct sort_mods));
1071: }
1072:
1073: ks->sm.func = get_sort_func(&(ks->sm));
1074: }
1075:
1076: if (argv_from_file0) {
1077: argc = argc_from_file0;
1078: argv = argv_from_file0;
1079: }
1080:
1081: if (debug_sort) {
1082: printf("Memory to be used for sorting: %llu\n",
1083: available_free_memory);
1084: printf("Using collate rules of %s locale\n",
1085: setlocale(LC_COLLATE, NULL));
1086: if (byte_sort)
1087: printf("Byte sort is used\n");
1088: if (print_symbols_on_debug) {
1089: printf("Decimal Point: <%lc>\n", symbol_decimal_point);
1090: if (symbol_thousands_sep)
1091: printf("Thousands separator: <%lc>\n",
1092: symbol_thousands_sep);
1093: printf("Positive sign: <%lc>\n", symbol_positive_sign);
1094: printf("Negative sign: <%lc>\n", symbol_negative_sign);
1095: }
1096: }
1097:
1098: set_random_seed();
1.4 millert 1099:
1.44 millert 1100: /* Case when the outfile equals one of the input files: */
1.51 millert 1101: if (strcmp(outfile, "-") != 0) {
1102: struct stat sb;
1103: int fd, i;
1.44 millert 1104:
1105: for (i = 0; i < argc; ++i) {
1106: if (strcmp(argv[i], outfile) == 0) {
1.51 millert 1107: if (stat(outfile, &sb) == -1)
1108: err(2, "%s", outfile);
1109: if (access(outfile, W_OK) == -1)
1110: err(2, "%s", outfile);
1111: real_outfile = outfile;
1112: sort_asprintf(&outfile, "%s.XXXXXXXXXX",
1113: real_outfile);
1114: if ((fd = mkstemp(outfile)) == -1 ||
1115: fchmod(fd, sb.st_mode & ALLPERMS) == -1)
1116: err(2, "%s", outfile);
1117: close(fd);
1.44 millert 1118: tmp_file_atexit(outfile);
1.51 millert 1119: break;
1.44 millert 1120: }
1121: }
1122: }
1123:
1124: if (!sort_opts_vals.cflag && !sort_opts_vals.mflag) {
1125: struct file_list fl;
1126: struct sort_list list;
1127:
1128: sort_list_init(&list);
1129: file_list_init(&fl, true);
1130:
1131: if (argc < 1)
1132: procfile("-", &list, &fl);
1133: else {
1134: while (argc > 0) {
1135: procfile(*argv, &list, &fl);
1136: --argc;
1137: ++argv;
1138: }
1139: }
1140:
1141: if (fl.count < 1)
1142: sort_list_to_file(&list, outfile);
1143: else {
1144: if (list.count > 0) {
1145: char *flast = new_tmp_file_name();
1146:
1147: sort_list_to_file(&list, flast);
1148: file_list_add(&fl, flast, false);
1149: }
1150: merge_files(&fl, outfile);
1151: }
1152:
1153: file_list_clean(&fl);
1154:
1155: /*
1156: * We are about to exit the program, so we can ignore
1157: * the clean-up for speed
1158: *
1159: * sort_list_clean(&list);
1160: */
1161:
1162: } else if (sort_opts_vals.cflag) {
1163: result = (argc == 0) ? (check("-")) : (check(*argv));
1164: } else if (sort_opts_vals.mflag) {
1165: struct file_list fl;
1166:
1167: file_list_init(&fl, false);
1168: file_list_populate(&fl, argc, argv, true);
1169: merge_files(&fl, outfile);
1170: file_list_clean(&fl);
1171: }
1172:
1173: if (real_outfile) {
1174: if (rename(outfile, real_outfile) < 0)
1175: err(2, "%s", real_outfile);
1.51 millert 1176: sort_free(outfile);
1.44 millert 1177: }
1.4 millert 1178:
1.44 millert 1179: return result;
1.1 millert 1180: }