]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/blob - usr.bin/sort/file.c
Add UPDATING entries and bump version.
[FreeBSD/FreeBSD.git] / usr.bin / sort / file.c
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (C) 2009 Gabor Kovesdan <gabor@FreeBSD.org>
5  * Copyright (C) 2012 Oleg Moskalenko <mom040267@gmail.com>
6  * All rights reserved.
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  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
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
30 #include <sys/cdefs.h>
31 __FBSDID("$FreeBSD$");
32
33 #include <sys/mman.h>
34 #include <sys/stat.h>
35 #include <sys/types.h>
36 #include <sys/queue.h>
37
38 #include <err.h>
39 #include <fcntl.h>
40 #if defined(SORT_THREADS)
41 #include <pthread.h>
42 #endif
43 #include <semaphore.h>
44 #include <stdio.h>
45 #include <stdlib.h>
46 #include <string.h>
47 #include <unistd.h>
48 #include <wchar.h>
49 #include <wctype.h>
50
51 #include "coll.h"
52 #include "file.h"
53 #include "radixsort.h"
54
55 unsigned long long free_memory = 1000000;
56 unsigned long long available_free_memory = 1000000;
57
58 bool use_mmap;
59
60 const char *tmpdir = "/var/tmp";
61 const char *compress_program;
62
63 size_t max_open_files = 16;
64
65 /*
66  * How much space we read from file at once
67  */
68 #define READ_CHUNK (4096)
69
70 /*
71  * File reader structure
72  */
73 struct file_reader
74 {
75         struct reader_buffer     rb;
76         FILE                    *file;
77         char                    *fname;
78         unsigned char           *buffer;
79         unsigned char           *mmapaddr;
80         unsigned char           *mmapptr;
81         size_t                   bsz;
82         size_t                   cbsz;
83         size_t                   mmapsize;
84         size_t                   strbeg;
85         int                      fd;
86         char                     elsymb;
87 };
88
89 /*
90  * Structure to be used in file merge process.
91  */
92 struct file_header
93 {
94         struct file_reader              *fr;
95         struct sort_list_item           *si; /* current top line */
96         size_t                           file_pos;
97 };
98
99 /*
100  * List elements of "cleanable" files list.
101  */
102 struct CLEANABLE_FILE
103 {
104         char                            *fn;
105         LIST_ENTRY(CLEANABLE_FILE)       files;
106 };
107
108 /*
109  * List header of "cleanable" files list.
110  */
111 static LIST_HEAD(CLEANABLE_FILES,CLEANABLE_FILE) tmp_files;
112
113 /*
114  * Semaphore to protect the tmp file list.
115  * We use semaphore here because it is signal-safe, according to POSIX.
116  * And semaphore does not require pthread library.
117  */
118 static sem_t tmp_files_sem;
119
120 static void mt_sort(struct sort_list *list,
121     int (*sort_func)(void *, size_t, size_t,
122     int (*)(const void *, const void *)), const char* fn);
123
124 /*
125  * Init tmp files list
126  */
127 void
128 init_tmp_files(void)
129 {
130
131         LIST_INIT(&tmp_files);
132         sem_init(&tmp_files_sem, 0, 1);
133 }
134
135 /*
136  * Save name of a tmp file for signal cleanup
137  */
138 void
139 tmp_file_atexit(const char *tmp_file)
140 {
141
142         if (tmp_file) {
143                 sem_wait(&tmp_files_sem);
144                 struct CLEANABLE_FILE *item =
145                     sort_malloc(sizeof(struct CLEANABLE_FILE));
146                 item->fn = sort_strdup(tmp_file);
147                 LIST_INSERT_HEAD(&tmp_files, item, files);
148                 sem_post(&tmp_files_sem);
149         }
150 }
151
152 /*
153  * Clear tmp files
154  */
155 void
156 clear_tmp_files(void)
157 {
158         struct CLEANABLE_FILE *item;
159
160         sem_wait(&tmp_files_sem);
161         LIST_FOREACH(item,&tmp_files,files) {
162                 if ((item) && (item->fn))
163                         unlink(item->fn);
164         }
165         sem_post(&tmp_files_sem);
166 }
167
168 /*
169  * Check whether a file is a temporary file
170  */
171 static bool
172 file_is_tmp(const char* fn)
173 {
174         struct CLEANABLE_FILE *item;
175         bool ret = false;
176
177         if (fn) {
178                 sem_wait(&tmp_files_sem);
179                 LIST_FOREACH(item,&tmp_files,files) {
180                         if ((item) && (item->fn))
181                                 if (strcmp(item->fn, fn) == 0) {
182                                         ret = true;
183                                         break;
184                                 }
185                 }
186                 sem_post(&tmp_files_sem);
187         }
188
189         return (ret);
190 }
191
192 /*
193  * Generate new temporary file name
194  */
195 char *
196 new_tmp_file_name(void)
197 {
198         static size_t tfcounter = 0;
199         static const char *fn = ".bsdsort.";
200         char *ret;
201         size_t sz;
202
203         sz = strlen(tmpdir) + 1 + strlen(fn) + 32 + 1;
204         ret = sort_malloc(sz);
205
206         sprintf(ret, "%s/%s%d.%lu", tmpdir, fn, (int) getpid(), (unsigned long)(tfcounter++));
207         tmp_file_atexit(ret);
208         return (ret);
209 }
210
211 /*
212  * Initialize file list
213  */
214 void
215 file_list_init(struct file_list *fl, bool tmp)
216 {
217
218         if (fl) {
219                 fl->count = 0;
220                 fl->sz = 0;
221                 fl->fns = NULL;
222                 fl->tmp = tmp;
223         }
224 }
225
226 /*
227  * Add a file name to the list
228  */
229 void
230 file_list_add(struct file_list *fl, const char *fn, bool allocate)
231 {
232
233         if (fl && fn) {
234                 if (fl->count >= fl->sz || (fl->fns == NULL)) {
235                         fl->sz = (fl->sz) * 2 + 1;
236                         fl->fns = sort_realloc(fl->fns, fl->sz *
237                             sizeof(char *));
238                 }
239                 fl->fns[fl->count] = allocate ? sort_strdup(fn) : fn;
240                 fl->count += 1;
241         }
242 }
243
244 /*
245  * Populate file list from array of file names
246  */
247 void
248 file_list_populate(struct file_list *fl, int argc, char **argv, bool allocate)
249 {
250
251         if (fl && argv) {
252                 int i;
253
254                 for (i = 0; i < argc; i++)
255                         file_list_add(fl, argv[i], allocate);
256         }
257 }
258
259 /*
260  * Clean file list data and delete the files,
261  * if this is a list of temporary files
262  */
263 void
264 file_list_clean(struct file_list *fl)
265 {
266
267         if (fl) {
268                 if (fl->fns) {
269                         size_t i;
270
271                         for (i = 0; i < fl->count; i++) {
272                                 if (fl->fns[i]) {
273                                         if (fl->tmp)
274                                                 unlink(fl->fns[i]);
275                                         sort_free(fl->fns[i]);
276                                         fl->fns[i] = 0;
277                                 }
278                         }
279                         sort_free(fl->fns);
280                         fl->fns = NULL;
281                 }
282                 fl->sz = 0;
283                 fl->count = 0;
284                 fl->tmp = false;
285         }
286 }
287
288 /*
289  * Init sort list
290  */
291 void
292 sort_list_init(struct sort_list *l)
293 {
294
295         if (l) {
296                 l->count = 0;
297                 l->size = 0;
298                 l->memsize = sizeof(struct sort_list);
299                 l->list = NULL;
300         }
301 }
302
303 /*
304  * Add string to sort list
305  */
306 void
307 sort_list_add(struct sort_list *l, struct bwstring *str)
308 {
309
310         if (l && str) {
311                 size_t indx = l->count;
312
313                 if ((l->list == NULL) || (indx >= l->size)) {
314                         size_t newsize = (l->size + 1) + 1024;
315
316                         l->list = sort_realloc(l->list,
317                             sizeof(struct sort_list_item*) * newsize);
318                         l->memsize += (newsize - l->size) *
319                             sizeof(struct sort_list_item*);
320                         l->size = newsize;
321                 }
322                 l->list[indx] = sort_list_item_alloc();
323                 sort_list_item_set(l->list[indx], str);
324                 l->memsize += sort_list_item_size(l->list[indx]);
325                 l->count += 1;
326         }
327 }
328
329 /*
330  * Clean sort list data
331  */
332 void
333 sort_list_clean(struct sort_list *l)
334 {
335
336         if (l) {
337                 if (l->list) {
338                         size_t i;
339
340                         for (i = 0; i < l->count; i++) {
341                                 struct sort_list_item *item;
342
343                                 item = l->list[i];
344
345                                 if (item) {
346                                         sort_list_item_clean(item);
347                                         sort_free(item);
348                                         l->list[i] = NULL;
349                                 }
350                         }
351                         sort_free(l->list);
352                         l->list = NULL;
353                 }
354                 l->count = 0;
355                 l->size = 0;
356                 l->memsize = sizeof(struct sort_list);
357         }
358 }
359
360 /*
361  * Write sort list to file
362  */
363 void
364 sort_list_dump(struct sort_list *l, const char *fn)
365 {
366
367         if (l && fn) {
368                 FILE *f;
369
370                 f = openfile(fn, "w");
371                 if (f == NULL)
372                         err(2, NULL);
373
374                 if (l->list) {
375                         size_t i;
376                         if (!(sort_opts_vals.uflag)) {
377                                 for (i = 0; i < l->count; ++i)
378                                         bwsfwrite(l->list[i]->str, f,
379                                             sort_opts_vals.zflag);
380                         } else {
381                                 struct sort_list_item *last_printed_item = NULL;
382                                 struct sort_list_item *item;
383                                 for (i = 0; i < l->count; ++i) {
384                                         item = l->list[i];
385                                         if ((last_printed_item == NULL) ||
386                                             list_coll(&last_printed_item, &item)) {
387                                                 bwsfwrite(item->str, f, sort_opts_vals.zflag);
388                                                 last_printed_item = item;
389                                         }
390                                 }
391                         }
392                 }
393
394                 closefile(f, fn);
395         }
396 }
397
398 /*
399  * Checks if the given file is sorted.  Stops at the first disorder,
400  * prints the disordered line and returns 1.
401  */
402 int
403 check(const char *fn)
404 {
405         struct bwstring *s1, *s2, *s1disorder, *s2disorder;
406         struct file_reader *fr;
407         struct keys_array *ka1, *ka2;
408         int res;
409         size_t pos, posdisorder;
410
411         s1 = s2 = s1disorder = s2disorder = NULL;
412         ka1 = ka2 = NULL;
413
414         fr = file_reader_init(fn);
415
416         res = 0;
417         pos = 1;
418         posdisorder = 1;
419
420         if (fr == NULL) {
421                 err(2, NULL);
422                 goto end;
423         }
424
425         s1 = file_reader_readline(fr);
426         if (s1 == NULL)
427                 goto end;
428
429         ka1 = keys_array_alloc();
430         preproc(s1, ka1);
431
432         s2 = file_reader_readline(fr);
433         if (s2 == NULL)
434                 goto end;
435
436         ka2 = keys_array_alloc();
437         preproc(s2, ka2);
438
439         for (;;) {
440
441                 if (debug_sort) {
442                         bwsprintf(stdout, s2, "s1=<", ">");
443                         bwsprintf(stdout, s1, "s2=<", ">");
444                 }
445                 int cmp = key_coll(ka2, ka1, 0);
446                 if (debug_sort)
447                         printf("; cmp1=%d", cmp);
448
449                 if (!cmp && sort_opts_vals.complex_sort &&
450                     !(sort_opts_vals.uflag) && !(sort_opts_vals.sflag)) {
451                         cmp = top_level_str_coll(s2, s1);
452                         if (debug_sort)
453                                 printf("; cmp2=%d", cmp);
454                 }
455                 if (debug_sort)
456                         printf("\n");
457
458                 if ((sort_opts_vals.uflag && (cmp <= 0)) || (cmp < 0)) {
459                         if (!(sort_opts_vals.csilentflag)) {
460                                 s2disorder = bwsdup(s2);
461                                 posdisorder = pos;
462                                 if (debug_sort)
463                                         s1disorder = bwsdup(s1);
464                         }
465                         res = 1;
466                         goto end;
467                 }
468
469                 pos++;
470
471                 clean_keys_array(s1, ka1);
472                 sort_free(ka1);
473                 ka1 = ka2;
474                 ka2 = NULL;
475
476                 bwsfree(s1);
477                 s1 = s2;
478
479                 s2 = file_reader_readline(fr);
480                 if (s2 == NULL)
481                         goto end;
482
483                 ka2 = keys_array_alloc();
484                 preproc(s2, ka2);
485         }
486
487 end:
488         if (ka1) {
489                 clean_keys_array(s1, ka1);
490                 sort_free(ka1);
491         }
492
493         if (s1)
494                 bwsfree(s1);
495
496         if (ka2) {
497                 clean_keys_array(s2, ka2);
498                 sort_free(ka2);
499         }
500
501         if (s2)
502                 bwsfree(s2);
503
504         if ((fn == NULL) || (*fn == 0) || (strcmp(fn, "-") == 0)) {
505                 for (;;) {
506                         s2 = file_reader_readline(fr);
507                         if (s2 == NULL)
508                                 break;
509                         bwsfree(s2);
510                 }
511         }
512
513         file_reader_free(fr);
514
515         if (s2disorder) {
516                 bws_disorder_warnx(s2disorder, fn, posdisorder);
517                 if (s1disorder) {
518                         bws_disorder_warnx(s1disorder, fn, posdisorder);
519                         if (s1disorder != s2disorder)
520                                 bwsfree(s1disorder);
521                 }
522                 bwsfree(s2disorder);
523                 s1disorder = NULL;
524                 s2disorder = NULL;
525         }
526
527         if (res)
528                 exit(res);
529
530         return (0);
531 }
532
533 /*
534  * Opens a file.  If the given filename is "-", stdout will be
535  * opened.
536  */
537 FILE *
538 openfile(const char *fn, const char *mode)
539 {
540         FILE *file;
541
542         if (strcmp(fn, "-") == 0) {
543                 return ((mode && mode[0] == 'r') ? stdin : stdout);
544         } else {
545                 mode_t orig_file_mask = 0;
546                 int is_tmp = file_is_tmp(fn);
547
548                 if (is_tmp && (mode[0] == 'w'))
549                         orig_file_mask = umask(S_IWGRP | S_IWOTH |
550                             S_IRGRP | S_IROTH);
551
552                 if (is_tmp && (compress_program != NULL)) {
553                         char *cmd;
554                         size_t cmdsz;
555
556                         cmdsz = strlen(fn) + 128;
557                         cmd = sort_malloc(cmdsz);
558
559                         fflush(stdout);
560
561                         if (mode[0] == 'r')
562                                 snprintf(cmd, cmdsz - 1, "cat %s | %s -d",
563                                     fn, compress_program);
564                         else if (mode[0] == 'w')
565                                 snprintf(cmd, cmdsz - 1, "%s > %s",
566                                     compress_program, fn);
567                         else
568                                 err(2, "%s", getstr(7));
569
570                         if ((file = popen(cmd, mode)) == NULL)
571                                 err(2, NULL);
572
573                         sort_free(cmd);
574
575                 } else
576                         if ((file = fopen(fn, mode)) == NULL)
577                                 err(2, NULL);
578
579                 if (is_tmp && (mode[0] == 'w'))
580                         umask(orig_file_mask);
581         }
582
583         return (file);
584 }
585
586 /*
587  * Close file
588  */
589 void
590 closefile(FILE *f, const char *fn)
591 {
592         if (f == NULL) {
593                 ;
594         } else if (f == stdin) {
595                 ;
596         } else if (f == stdout) {
597                 fflush(f);
598         } else {
599                 if (file_is_tmp(fn) && compress_program != NULL) {
600                         if(pclose(f)<0)
601                                 err(2,NULL);
602                 } else
603                         fclose(f);
604         }
605 }
606
607 /*
608  * Reads a file into the internal buffer.
609  */
610 struct file_reader *
611 file_reader_init(const char *fsrc)
612 {
613         struct file_reader *ret;
614
615         if (fsrc == NULL)
616                 fsrc = "-";
617
618         ret = sort_malloc(sizeof(struct file_reader));
619         memset(ret, 0, sizeof(struct file_reader));
620
621         ret->elsymb = '\n';
622         if (sort_opts_vals.zflag)
623                 ret->elsymb = 0;
624
625         ret->fname = sort_strdup(fsrc);
626
627         if (strcmp(fsrc, "-") && (compress_program == NULL) && use_mmap) {
628
629                 do {
630                         struct stat stat_buf;
631                         void *addr;
632                         size_t sz = 0;
633                         int fd, flags;
634
635                         flags = MAP_NOCORE | MAP_NOSYNC;
636                         addr = MAP_FAILED;
637
638                         fd = open(fsrc, O_RDONLY);
639                         if (fd < 0)
640                                 err(2, NULL);
641
642                         if (fstat(fd, &stat_buf) < 0) {
643                                 close(fd);
644                                 break;
645                         }
646
647                         sz = stat_buf.st_size;
648
649 #if defined(MAP_PREFAULT_READ)
650                         flags |= MAP_PREFAULT_READ;
651 #endif
652
653                         addr = mmap(NULL, sz, PROT_READ, flags, fd, 0);
654                         if (addr == MAP_FAILED) {
655                                 close(fd);
656                                 break;
657                         }
658
659                         ret->fd = fd;
660                         ret->mmapaddr = addr;
661                         ret->mmapsize = sz;
662                         ret->mmapptr = ret->mmapaddr;
663
664                 } while (0);
665         }
666
667         if (ret->mmapaddr == NULL) {
668                 ret->file = openfile(fsrc, "r");
669                 if (ret->file == NULL)
670                         err(2, NULL);
671
672                 if (strcmp(fsrc, "-")) {
673                         ret->cbsz = READ_CHUNK;
674                         ret->buffer = sort_malloc(ret->cbsz);
675                         ret->bsz = 0;
676                         ret->strbeg = 0;
677
678                         ret->bsz = fread(ret->buffer, 1, ret->cbsz, ret->file);
679                         if (ret->bsz == 0) {
680                                 if (ferror(ret->file))
681                                         err(2, NULL);
682                         }
683                 }
684         }
685
686         return (ret);
687 }
688
689 struct bwstring *
690 file_reader_readline(struct file_reader *fr)
691 {
692         struct bwstring *ret = NULL;
693
694         if (fr->mmapaddr) {
695                 unsigned char *mmapend;
696
697                 mmapend = fr->mmapaddr + fr->mmapsize;
698                 if (fr->mmapptr >= mmapend)
699                         return (NULL);
700                 else {
701                         unsigned char *strend;
702                         size_t sz;
703
704                         sz = mmapend - fr->mmapptr;
705                         strend = memchr(fr->mmapptr, fr->elsymb, sz);
706
707                         if (strend == NULL) {
708                                 ret = bwscsbdup(fr->mmapptr, sz);
709                                 fr->mmapptr = mmapend;
710                         } else {
711                                 ret = bwscsbdup(fr->mmapptr, strend -
712                                     fr->mmapptr);
713                                 fr->mmapptr = strend + 1;
714                         }
715                 }
716
717         } else if (fr->file != stdin) {
718                 unsigned char *strend;
719                 size_t bsz1, remsz, search_start;
720
721                 search_start = 0;
722                 remsz = 0;
723                 strend = NULL;
724
725                 if (fr->bsz > fr->strbeg)
726                         remsz = fr->bsz - fr->strbeg;
727
728                 /* line read cycle */
729                 for (;;) {
730                         if (remsz > search_start)
731                                 strend = memchr(fr->buffer + fr->strbeg +
732                                     search_start, fr->elsymb, remsz -
733                                     search_start);
734                         else
735                                 strend = NULL;
736
737                         if (strend)
738                                 break;
739                         if (feof(fr->file))
740                                 break;
741
742                         if (fr->bsz != fr->cbsz)
743                                 /* NOTREACHED */
744                                 err(2, "File read software error 1");
745
746                         if (remsz > (READ_CHUNK >> 1)) {
747                                 search_start = fr->cbsz - fr->strbeg;
748                                 fr->cbsz += READ_CHUNK;
749                                 fr->buffer = sort_realloc(fr->buffer,
750                                     fr->cbsz);
751                                 bsz1 = fread(fr->buffer + fr->bsz, 1,
752                                     READ_CHUNK, fr->file);
753                                 if (bsz1 == 0) {
754                                         if (ferror(fr->file))
755                                                 err(2, NULL);
756                                         break;
757                                 }
758                                 fr->bsz += bsz1;
759                                 remsz += bsz1;
760                         } else {
761                                 if (remsz > 0 && fr->strbeg>0)
762                                         bcopy(fr->buffer + fr->strbeg,
763                                             fr->buffer, remsz);
764
765                                 fr->strbeg = 0;
766                                 search_start = remsz;
767                                 bsz1 = fread(fr->buffer + remsz, 1,
768                                     fr->cbsz - remsz, fr->file);
769                                 if (bsz1 == 0) {
770                                         if (ferror(fr->file))
771                                                 err(2, NULL);
772                                         break;
773                                 }
774                                 fr->bsz = remsz + bsz1;
775                                 remsz = fr->bsz;
776                         }
777                 }
778
779                 if (strend == NULL)
780                         strend = fr->buffer + fr->bsz;
781
782                 if ((fr->buffer + fr->strbeg <= strend) &&
783                     (fr->strbeg < fr->bsz) && (remsz>0))
784                         ret = bwscsbdup(fr->buffer + fr->strbeg, strend -
785                             fr->buffer - fr->strbeg);
786
787                 fr->strbeg = (strend - fr->buffer) + 1;
788
789         } else {
790                 size_t len = 0;
791
792                 ret = bwsfgetln(fr->file, &len, sort_opts_vals.zflag,
793                     &(fr->rb));
794         }
795
796         return (ret);
797 }
798
799 static void
800 file_reader_clean(struct file_reader *fr)
801 {
802
803         if (fr) {
804                 if (fr->mmapaddr)
805                         munmap(fr->mmapaddr, fr->mmapsize);
806
807                 if (fr->fd)
808                         close(fr->fd);
809
810                 if (fr->buffer)
811                         sort_free(fr->buffer);
812
813                 if (fr->file)
814                         if (fr->file != stdin)
815                                 closefile(fr->file, fr->fname);
816
817                 if(fr->fname)
818                         sort_free(fr->fname);
819
820                 memset(fr, 0, sizeof(struct file_reader));
821         }
822 }
823
824 void
825 file_reader_free(struct file_reader *fr)
826 {
827
828         if (fr) {
829                 file_reader_clean(fr);
830                 sort_free(fr);
831         }
832 }
833
834 int
835 procfile(const char *fsrc, struct sort_list *list, struct file_list *fl)
836 {
837         struct file_reader *fr;
838
839         fr = file_reader_init(fsrc);
840         if (fr == NULL)
841                 err(2, NULL);
842
843         /* file browse cycle */
844         for (;;) {
845                 struct bwstring *bws;
846
847                 bws = file_reader_readline(fr);
848
849                 if (bws == NULL)
850                         break;
851
852                 sort_list_add(list, bws);
853
854                 if (list->memsize >= available_free_memory) {
855                         char *fn;
856
857                         fn = new_tmp_file_name();
858                         sort_list_to_file(list, fn);
859                         file_list_add(fl, fn, false);
860                         sort_list_clean(list);
861                 }
862         }
863
864         file_reader_free(fr);
865
866         return (0);
867 }
868
869 /*
870  * Compare file headers. Files with EOF always go to the end of the list.
871  */
872 static int
873 file_header_cmp(struct file_header *f1, struct file_header *f2)
874 {
875
876         if (f1 == f2)
877                 return (0);
878         else {
879                 if (f1->fr == NULL) {
880                         return ((f2->fr == NULL) ? 0 : +1);
881                 } else if (f2->fr == NULL)
882                         return (-1);
883                 else {
884                         int ret;
885
886                         ret = list_coll(&(f1->si), &(f2->si));
887                         if (!ret)
888                                 return ((f1->file_pos < f2->file_pos) ? -1 : +1);
889                         return (ret);
890                 }
891         }
892 }
893
894 /*
895  * Allocate and init file header structure
896  */
897 static void
898 file_header_init(struct file_header **fh, const char *fn, size_t file_pos)
899 {
900
901         if (fh && fn) {
902                 struct bwstring *line;
903
904                 *fh = sort_malloc(sizeof(struct file_header));
905                 (*fh)->file_pos = file_pos;
906                 (*fh)->fr = file_reader_init(fn);
907                 if ((*fh)->fr == NULL) {
908                         perror(fn);
909                         err(2, "%s", getstr(8));
910                 }
911                 line = file_reader_readline((*fh)->fr);
912                 if (line == NULL) {
913                         file_reader_free((*fh)->fr);
914                         (*fh)->fr = NULL;
915                         (*fh)->si = NULL;
916                 } else {
917                         (*fh)->si = sort_list_item_alloc();
918                         sort_list_item_set((*fh)->si, line);
919                 }
920         }
921 }
922
923 /*
924  * Close file
925  */
926 static void
927 file_header_close(struct file_header **fh)
928 {
929
930         if (fh && *fh) {
931                 if ((*fh)->fr) {
932                         file_reader_free((*fh)->fr);
933                         (*fh)->fr = NULL;
934                 }
935                 if ((*fh)->si) {
936                         sort_list_item_clean((*fh)->si);
937                         sort_free((*fh)->si);
938                         (*fh)->si = NULL;
939                 }
940                 sort_free(*fh);
941                 *fh = NULL;
942         }
943 }
944
945 /*
946  * Swap two array elements
947  */
948 static void
949 file_header_swap(struct file_header **fh, size_t i1, size_t i2)
950 {
951         struct file_header *tmp;
952
953         tmp = fh[i1];
954         fh[i1] = fh[i2];
955         fh[i2] = tmp;
956 }
957
958 /* heap algorithm ==>> */
959
960 /*
961  * See heap sort algorithm
962  * "Raises" last element to its right place
963  */
964 static void
965 file_header_heap_swim(struct file_header **fh, size_t indx)
966 {
967
968         if (indx > 0) {
969                 size_t parent_index;
970
971                 parent_index = (indx - 1) >> 1;
972
973                 if (file_header_cmp(fh[indx], fh[parent_index]) < 0) {
974                         /* swap child and parent and continue */
975                         file_header_swap(fh, indx, parent_index);
976                         file_header_heap_swim(fh, parent_index);
977                 }
978         }
979 }
980
981 /*
982  * Sink the top element to its correct position
983  */
984 static void
985 file_header_heap_sink(struct file_header **fh, size_t indx, size_t size)
986 {
987         size_t left_child_index;
988         size_t right_child_index;
989
990         left_child_index = indx + indx + 1;
991         right_child_index = left_child_index + 1;
992
993         if (left_child_index < size) {
994                 size_t min_child_index;
995
996                 min_child_index = left_child_index;
997
998                 if ((right_child_index < size) &&
999                     (file_header_cmp(fh[left_child_index],
1000                     fh[right_child_index]) > 0))
1001                         min_child_index = right_child_index;
1002                 if (file_header_cmp(fh[indx], fh[min_child_index]) > 0) {
1003                         file_header_swap(fh, indx, min_child_index);
1004                         file_header_heap_sink(fh, min_child_index, size);
1005                 }
1006         }
1007 }
1008
1009 /* <<== heap algorithm */
1010
1011 /*
1012  * Adds element to the "left" end
1013  */
1014 static void
1015 file_header_list_rearrange_from_header(struct file_header **fh, size_t size)
1016 {
1017
1018         file_header_heap_sink(fh, 0, size);
1019 }
1020
1021 /*
1022  * Adds element to the "right" end
1023  */
1024 static void
1025 file_header_list_push(struct file_header *f, struct file_header **fh, size_t size)
1026 {
1027
1028         fh[size++] = f;
1029         file_header_heap_swim(fh, size - 1);
1030 }
1031
1032 struct last_printed
1033 {
1034         struct bwstring *str;
1035 };
1036
1037 /*
1038  * Prints the current line of the file
1039  */
1040 static void
1041 file_header_print(struct file_header *fh, FILE *f_out, struct last_printed *lp)
1042 {
1043
1044         if (fh && fh->fr && f_out && fh->si && fh->si->str) {
1045                 if (sort_opts_vals.uflag) {
1046                         if ((lp->str == NULL) || (str_list_coll(lp->str, &(fh->si)))) {
1047                                 bwsfwrite(fh->si->str, f_out, sort_opts_vals.zflag);
1048                                 if (lp->str)
1049                                         bwsfree(lp->str);
1050                                 lp->str = bwsdup(fh->si->str);
1051                         }
1052                 } else
1053                         bwsfwrite(fh->si->str, f_out, sort_opts_vals.zflag);
1054         }
1055 }
1056
1057 /*
1058  * Read next line
1059  */
1060 static void
1061 file_header_read_next(struct file_header *fh)
1062 {
1063
1064         if (fh && fh->fr) {
1065                 struct bwstring *tmp;
1066
1067                 tmp = file_reader_readline(fh->fr);
1068                 if (tmp == NULL) {
1069                         file_reader_free(fh->fr);
1070                         fh->fr = NULL;
1071                         if (fh->si) {
1072                                 sort_list_item_clean(fh->si);
1073                                 sort_free(fh->si);
1074                                 fh->si = NULL;
1075                         }
1076                 } else {
1077                         if (fh->si == NULL)
1078                                 fh->si = sort_list_item_alloc();
1079                         sort_list_item_set(fh->si, tmp);
1080                 }
1081         }
1082 }
1083
1084 /*
1085  * Merge array of "files headers"
1086  */
1087 static void
1088 file_headers_merge(size_t fnum, struct file_header **fh, FILE *f_out)
1089 {
1090         struct last_printed lp;
1091         size_t i;
1092
1093         memset(&lp, 0, sizeof(lp));
1094
1095         /*
1096          * construct the initial sort structure
1097          */
1098         for (i = 0; i < fnum; i++)
1099                 file_header_list_push(fh[i], fh, i);
1100
1101         while (fh[0]->fr) { /* unfinished files are always in front */
1102                 /* output the smallest line: */
1103                 file_header_print(fh[0], f_out, &lp);
1104                 /* read a new line, if possible: */
1105                 file_header_read_next(fh[0]);
1106                 /* re-arrange the list: */
1107                 file_header_list_rearrange_from_header(fh, fnum);
1108         }
1109
1110         if (lp.str)
1111                 bwsfree(lp.str);
1112 }
1113
1114 /*
1115  * Merges the given files into the output file, which can be
1116  * stdout.
1117  */
1118 static void
1119 merge_files_array(size_t argc, const char **argv, const char *fn_out)
1120 {
1121
1122         if (argv && fn_out) {
1123                 struct file_header **fh;
1124                 FILE *f_out;
1125                 size_t i;
1126
1127                 f_out = openfile(fn_out, "w");
1128
1129                 if (f_out == NULL)
1130                         err(2, NULL);
1131
1132                 fh = sort_malloc((argc + 1) * sizeof(struct file_header *));
1133
1134                 for (i = 0; i < argc; i++)
1135                         file_header_init(fh + i, argv[i], (size_t) i);
1136
1137                 file_headers_merge(argc, fh, f_out);
1138
1139                 for (i = 0; i < argc; i++)
1140                         file_header_close(fh + i);
1141
1142                 sort_free(fh);
1143
1144                 closefile(f_out, fn_out);
1145         }
1146 }
1147
1148 /*
1149  * Shrinks the file list until its size smaller than max number of opened files
1150  */
1151 static int
1152 shrink_file_list(struct file_list *fl)
1153 {
1154
1155         if ((fl == NULL) || (size_t) (fl->count) < max_open_files)
1156                 return (0);
1157         else {
1158                 struct file_list new_fl;
1159                 size_t indx = 0;
1160
1161                 file_list_init(&new_fl, true);
1162                 while (indx < fl->count) {
1163                         char *fnew;
1164                         size_t num;
1165
1166                         num = fl->count - indx;
1167                         fnew = new_tmp_file_name();
1168
1169                         if ((size_t) num >= max_open_files)
1170                                 num = max_open_files - 1;
1171                         merge_files_array(num, fl->fns + indx, fnew);
1172                         if (fl->tmp) {
1173                                 size_t i;
1174
1175                                 for (i = 0; i < num; i++)
1176                                         unlink(fl->fns[indx + i]);
1177                         }
1178                         file_list_add(&new_fl, fnew, false);
1179                         indx += num;
1180                 }
1181                 fl->tmp = false; /* already taken care of */
1182                 file_list_clean(fl);
1183
1184                 fl->count = new_fl.count;
1185                 fl->fns = new_fl.fns;
1186                 fl->sz = new_fl.sz;
1187                 fl->tmp = new_fl.tmp;
1188
1189                 return (1);
1190         }
1191 }
1192
1193 /*
1194  * Merge list of files
1195  */
1196 void
1197 merge_files(struct file_list *fl, const char *fn_out)
1198 {
1199
1200         if (fl && fn_out) {
1201                 while (shrink_file_list(fl));
1202
1203                 merge_files_array(fl->count, fl->fns, fn_out);
1204         }
1205 }
1206
1207 static const char *
1208 get_sort_method_name(int sm)
1209 {
1210
1211         if (sm == SORT_MERGESORT)
1212                 return "mergesort";
1213         else if (sort_opts_vals.sort_method == SORT_RADIXSORT)
1214                 return "radixsort";
1215         else if (sort_opts_vals.sort_method == SORT_HEAPSORT)
1216                 return "heapsort";
1217         else
1218                 return "quicksort";
1219 }
1220
1221 /*
1222  * Wrapper for qsort
1223  */
1224 static int sort_qsort(void *list, size_t count, size_t elem_size,
1225     int (*cmp_func)(const void *, const void *))
1226 {
1227
1228         qsort(list, count, elem_size, cmp_func);
1229         return (0);
1230 }
1231
1232 /*
1233  * Sort list of lines and writes it to the file
1234  */
1235 void
1236 sort_list_to_file(struct sort_list *list, const char *outfile)
1237 {
1238         struct sort_mods *sm = &(keys[0].sm);
1239
1240         if (!(sm->Mflag) && !(sm->Rflag) && !(sm->Vflag) && !(sm->Vflag) &&
1241             !(sm->gflag) && !(sm->hflag) && !(sm->nflag)) {
1242                 if ((sort_opts_vals.sort_method == SORT_DEFAULT) && byte_sort)
1243                         sort_opts_vals.sort_method = SORT_RADIXSORT;
1244
1245         } else if (sort_opts_vals.sort_method == SORT_RADIXSORT)
1246                 err(2, "%s", getstr(9));
1247
1248         /*
1249          * to handle stable sort and the unique cases in the
1250          * right order, we need stable basic algorithm
1251          */
1252         if (sort_opts_vals.sflag) {
1253                 switch (sort_opts_vals.sort_method){
1254                 case SORT_MERGESORT:
1255                         break;
1256                 case SORT_RADIXSORT:
1257                         break;
1258                 case SORT_DEFAULT:
1259                         sort_opts_vals.sort_method = SORT_MERGESORT;
1260                         break;
1261                 default:
1262                         errx(2, "%s", getstr(10));
1263                 }
1264         }
1265
1266         if (sort_opts_vals.sort_method == SORT_DEFAULT)
1267                 sort_opts_vals.sort_method = DEFAULT_SORT_ALGORITHM;
1268
1269         if (debug_sort)
1270                 printf("sort_method=%s\n",
1271                     get_sort_method_name(sort_opts_vals.sort_method));
1272
1273         switch (sort_opts_vals.sort_method){
1274         case SORT_RADIXSORT:
1275                 rxsort(list->list, list->count);
1276                 sort_list_dump(list, outfile);
1277                 break;
1278         case SORT_MERGESORT:
1279                 mt_sort(list, mergesort, outfile);
1280                 break;
1281         case SORT_HEAPSORT:
1282                 mt_sort(list, heapsort, outfile);
1283                 break;
1284         case SORT_QSORT:
1285                 mt_sort(list, sort_qsort, outfile);
1286                 break;
1287         default:
1288                 mt_sort(list, DEFAULT_SORT_FUNC, outfile);
1289                 break;
1290         }
1291 }
1292
1293 /******************* MT SORT ************************/
1294
1295 #if defined(SORT_THREADS)
1296 /* semaphore to count threads */
1297 static sem_t mtsem;
1298
1299 /* current system sort function */
1300 static int (*g_sort_func)(void *, size_t, size_t,
1301     int(*)(const void *, const void *));
1302
1303 /*
1304  * Sort cycle thread (in multi-threaded mode)
1305  */
1306 static void*
1307 mt_sort_thread(void* arg)
1308 {
1309         struct sort_list *list = arg;
1310
1311         g_sort_func(list->list, list->count, sizeof(struct sort_list_item *),
1312             (int(*)(const void *, const void *)) list_coll);
1313
1314         sem_post(&mtsem);
1315
1316         return (arg);
1317 }
1318
1319 /*
1320  * Compare sub-lists. Empty sub-lists always go to the end of the list.
1321  */
1322 static int
1323 sub_list_cmp(struct sort_list *l1, struct sort_list *l2)
1324 {
1325
1326         if (l1 == l2)
1327                 return (0);
1328         else {
1329                 if (l1->count == 0) {
1330                         return ((l2->count == 0) ? 0 : +1);
1331                 } else if (l2->count == 0) {
1332                         return (-1);
1333                 } else {
1334                         int ret;
1335
1336                         ret = list_coll(&(l1->list[0]), &(l2->list[0]));
1337                         if (!ret)
1338                                 return ((l1->sub_list_pos < l2->sub_list_pos) ?
1339                                     -1 : +1);
1340                         return (ret);
1341                 }
1342         }
1343 }
1344
1345 /*
1346  * Swap two array elements
1347  */
1348 static void
1349 sub_list_swap(struct sort_list **sl, size_t i1, size_t i2)
1350 {
1351         struct sort_list *tmp;
1352
1353         tmp = sl[i1];
1354         sl[i1] = sl[i2];
1355         sl[i2] = tmp;
1356 }
1357
1358 /* heap algorithm ==>> */
1359
1360 /*
1361  * See heap sort algorithm
1362  * "Raises" last element to its right place
1363  */
1364 static void
1365 sub_list_swim(struct sort_list **sl, size_t indx)
1366 {
1367
1368         if (indx > 0) {
1369                 size_t parent_index;
1370
1371                 parent_index = (indx - 1) >> 1;
1372
1373                 if (sub_list_cmp(sl[indx], sl[parent_index]) < 0) {
1374                         /* swap child and parent and continue */
1375                         sub_list_swap(sl, indx, parent_index);
1376                         sub_list_swim(sl, parent_index);
1377                 }
1378         }
1379 }
1380
1381 /*
1382  * Sink the top element to its correct position
1383  */
1384 static void
1385 sub_list_sink(struct sort_list **sl, size_t indx, size_t size)
1386 {
1387         size_t left_child_index;
1388         size_t right_child_index;
1389
1390         left_child_index = indx + indx + 1;
1391         right_child_index = left_child_index + 1;
1392
1393         if (left_child_index < size) {
1394                 size_t min_child_index;
1395
1396                 min_child_index = left_child_index;
1397
1398                 if ((right_child_index < size) &&
1399                     (sub_list_cmp(sl[left_child_index],
1400                     sl[right_child_index]) > 0))
1401                         min_child_index = right_child_index;
1402                 if (sub_list_cmp(sl[indx], sl[min_child_index]) > 0) {
1403                         sub_list_swap(sl, indx, min_child_index);
1404                         sub_list_sink(sl, min_child_index, size);
1405                 }
1406         }
1407 }
1408
1409 /* <<== heap algorithm */
1410
1411 /*
1412  * Adds element to the "right" end
1413  */
1414 static void
1415 sub_list_push(struct sort_list *s, struct sort_list **sl, size_t size)
1416 {
1417
1418         sl[size++] = s;
1419         sub_list_swim(sl, size - 1);
1420 }
1421
1422 struct last_printed_item
1423 {
1424         struct sort_list_item *item;
1425 };
1426
1427 /*
1428  * Prints the current line of the file
1429  */
1430 static void
1431 sub_list_header_print(struct sort_list *sl, FILE *f_out,
1432     struct last_printed_item *lp)
1433 {
1434
1435         if (sl && sl->count && f_out && sl->list[0]->str) {
1436                 if (sort_opts_vals.uflag) {
1437                         if ((lp->item == NULL) || (list_coll(&(lp->item),
1438                             &(sl->list[0])))) {
1439                                 bwsfwrite(sl->list[0]->str, f_out,
1440                                     sort_opts_vals.zflag);
1441                                 lp->item = sl->list[0];
1442                         }
1443                 } else
1444                         bwsfwrite(sl->list[0]->str, f_out,
1445                             sort_opts_vals.zflag);
1446         }
1447 }
1448
1449 /*
1450  * Read next line
1451  */
1452 static void
1453 sub_list_next(struct sort_list *sl)
1454 {
1455
1456         if (sl && sl->count) {
1457                 sl->list += 1;
1458                 sl->count -= 1;
1459         }
1460 }
1461
1462 /*
1463  * Merge sub-lists to a file
1464  */
1465 static void
1466 merge_sub_lists(struct sort_list **sl, size_t n, FILE* f_out)
1467 {
1468         struct last_printed_item lp;
1469         size_t i;
1470
1471         memset(&lp,0,sizeof(lp));
1472
1473         /* construct the initial list: */
1474         for (i = 0; i < n; i++)
1475                 sub_list_push(sl[i], sl, i);
1476
1477         while (sl[0]->count) { /* unfinished lists are always in front */
1478                 /* output the smallest line: */
1479                 sub_list_header_print(sl[0], f_out, &lp);
1480                 /* move to a new line, if possible: */
1481                 sub_list_next(sl[0]);
1482                 /* re-arrange the list: */
1483                 sub_list_sink(sl, 0, n);
1484         }
1485 }
1486
1487 /*
1488  * Merge sub-lists to a file
1489  */
1490 static void
1491 merge_list_parts(struct sort_list **parts, size_t n, const char *fn)
1492 {
1493         FILE* f_out;
1494
1495         f_out = openfile(fn,"w");
1496
1497         merge_sub_lists(parts, n, f_out);
1498
1499         closefile(f_out, fn);
1500 }
1501
1502 #endif /* defined(SORT_THREADS) */
1503 /*
1504  * Multi-threaded sort algorithm "driver"
1505  */
1506 static void
1507 mt_sort(struct sort_list *list,
1508     int(*sort_func)(void *, size_t, size_t, int(*)(const void *, const void *)),
1509     const char* fn)
1510 {
1511 #if defined(SORT_THREADS)
1512         if (nthreads < 2 || list->count < MT_SORT_THRESHOLD) {
1513                 size_t nthreads_save = nthreads;
1514                 nthreads = 1;
1515 #endif
1516                 /* if single thread or small data, do simple sort */
1517                 sort_func(list->list, list->count,
1518                     sizeof(struct sort_list_item *),
1519                     (int(*)(const void *, const void *)) list_coll);
1520                 sort_list_dump(list, fn);
1521 #if defined(SORT_THREADS)
1522                 nthreads = nthreads_save;
1523         } else {
1524                 /* multi-threaded sort */
1525                 struct sort_list **parts;
1526                 size_t avgsize, cstart, i;
1527
1528                 /* array of sub-lists */
1529                 parts = sort_malloc(sizeof(struct sort_list*) * nthreads);
1530                 cstart = 0;
1531                 avgsize = list->count / nthreads;
1532
1533                 /* set global system sort function */
1534                 g_sort_func = sort_func;
1535
1536                 /* set sublists */
1537                 for (i = 0; i < nthreads; ++i) {
1538                         size_t sz = 0;
1539
1540                         parts[i] = sort_malloc(sizeof(struct sort_list));
1541                         parts[i]->list = list->list + cstart;
1542                         parts[i]->memsize = 0;
1543                         parts[i]->sub_list_pos = i;
1544
1545                         sz = (i == nthreads - 1) ? list->count - cstart :
1546                             avgsize;
1547
1548                         parts[i]->count = sz;
1549
1550                         parts[i]->size = parts[i]->count;
1551
1552                         cstart += sz;
1553                 }
1554
1555                 /* init threads counting semaphore */
1556                 sem_init(&mtsem, 0, 0);
1557
1558                 /* start threads */
1559                 for (i = 0; i < nthreads; ++i) {
1560                         pthread_t pth;
1561                         pthread_attr_t attr;
1562
1563                         pthread_attr_init(&attr);
1564                         pthread_attr_setdetachstate(&attr, PTHREAD_DETACHED);
1565
1566                         for (;;) {
1567                                 int res = pthread_create(&pth, &attr,
1568                                     mt_sort_thread, parts[i]);
1569
1570                                 if (res >= 0)
1571                                         break;
1572                                 if (errno == EAGAIN) {
1573                                         pthread_yield();
1574                                         continue;
1575                                 }
1576                                 err(2, NULL);
1577                         }
1578
1579                         pthread_attr_destroy(&attr);
1580                 }
1581
1582                 /* wait for threads completion */
1583                 for (i = 0; i < nthreads; ++i) {
1584                         sem_wait(&mtsem);
1585                 }
1586                 /* destroy the semaphore - we do not need it anymore */
1587                 sem_destroy(&mtsem);
1588
1589                 /* merge sorted sub-lists to the file */
1590                 merge_list_parts(parts, nthreads, fn);
1591
1592                 /* free sub-lists data */
1593                 for (i = 0; i < nthreads; ++i) {
1594                         sort_free(parts[i]);
1595                 }
1596                 sort_free(parts);
1597         }
1598 #endif /* defined(SORT_THREADS) */
1599 }