1 /* $OpenBSD: ftree.c,v 1.43 2024/08/15 00:47:44 guenther Exp $ */
2 /* $NetBSD: ftree.c,v 1.4 1995/03/21 09:07:21 cgd Exp $ */
3
4 /*-
5 * Copyright (c) 1992 Keith Muller.
6 * Copyright (c) 1992, 1993
7 * The Regents of the University of California. All rights reserved.
8 *
9 * This code is derived from software contributed to Berkeley by
10 * Keith Muller of the University of California, San Diego.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 */
36
37 #include <sys/types.h>
38 #include <sys/stat.h>
39 #include <errno.h>
40 #include <fts.h>
41 #include <stdio.h>
42 #include <stdlib.h>
43 #include <string.h>
44 #include <unistd.h>
45
46 #include "pax.h"
47 #include "extern.h"
48
49 /*
50 * Data structure used to store the file args to be handed to fts().
51 * It keeps track of which args generated a "selected" member.
52 */
53 typedef struct ftree {
54 char *fname; /* file tree name */
55 int refcnt; /* had a selected (or skipped) file? */
56 int chflg; /* change directory flag */
57 struct ftree *fow; /* pointer to next entry on list */
58 } FTREE;
59
60
61 /*
62 * routines to interface with the fts library function.
63 *
64 * file args supplied to pax are stored on a single linked list (of type FTREE)
65 * and given to fts to be processed one at a time. pax "selects" files from
66 * the expansion of each arg into the corresponding file tree (if the arg is a
67 * directory, otherwise the node itself is just passed to pax). The selection
68 * is modified by the -n and -u flags. The user is informed when a specific
69 * file arg does not generate any selected files. -n keeps expanding the file
70 * tree arg until one of its files is selected, then skips to the next file
71 * arg. when the user does not supply the file trees as command line args to
72 * pax, they are read from stdin
73 */
74
75 static FTS *ftsp = NULL; /* current FTS handle */
76 static int ftsopts; /* options to be used on fts_open */
77 static char *farray[2]; /* array for passing each arg to fts */
78 static FTREE *fthead = NULL; /* head of linked list of file args */
79 static FTREE *fttail = NULL; /* tail of linked list of file args */
80 static FTREE *ftcur = NULL; /* current file arg being processed */
81 static FTSENT *ftent = NULL; /* current file tree entry */
82 static int ftree_skip; /* when set skip to next file arg */
83
84 static int ftree_arg(void);
85 static char *getpathname(char *, int);
86
87 /*
88 * ftree_start()
89 * initialize the options passed to fts_open() during this run of pax
90 * options are based on the selection of pax options by the user
91 * fts_start() also calls fts_arg() to open the first valid file arg. We
92 * also attempt to reset directory access times when -t (tflag) is set.
93 * Return:
94 * 0 if there is at least one valid file arg to process, -1 otherwise
95 */
96
97 int
ftree_start(void)98 ftree_start(void)
99 {
100 /*
101 * set up the operation mode of fts, open the first file arg. We must
102 * use FTS_NOCHDIR, as the user may have to open multiple archives and
103 * if fts did a chdir off into the boondocks, we may create an archive
104 * volume in an place where the user did not expect to.
105 */
106 ftsopts = FTS_NOCHDIR;
107
108 /*
109 * optional user flags that effect file traversal
110 * -H command line symlink follow only (half follow)
111 * -L follow sylinks (logical)
112 * -P do not follow sylinks (physical). This is the default.
113 * -X do not cross over mount points
114 * -t preserve access times on files read.
115 * -n select only the first member of a file tree when a match is found
116 * -d do not extract subtrees rooted at a directory arg.
117 */
118 if (Lflag)
119 ftsopts |= FTS_LOGICAL;
120 else
121 ftsopts |= FTS_PHYSICAL;
122 if (Hflag)
123 ftsopts |= FTS_COMFOLLOW;
124 if (Xflag)
125 ftsopts |= FTS_XDEV;
126
127 if ((fthead == NULL) && ((farray[0] = malloc(PAXPATHLEN+2)) == NULL)) {
128 paxwarn(1, "Unable to allocate memory for file name buffer");
129 return(-1);
130 }
131
132 if (ftree_arg() < 0)
133 return(-1);
134 if (tflag && (atdir_start() < 0))
135 return(-1);
136 return(0);
137 }
138
139 /*
140 * ftree_add()
141 * add the arg to the linked list of files to process. Each will be
142 * processed by fts one at a time
143 * Return:
144 * 0 if added to the linked list, -1 if failed
145 */
146
147 int
ftree_add(char * str,int chflg)148 ftree_add(char *str, int chflg)
149 {
150 FTREE *ft;
151 int len;
152
153 /*
154 * simple check for bad args
155 */
156 if ((str == NULL) || (*str == '\0')) {
157 paxwarn(0, "Invalid file name argument");
158 return(-1);
159 }
160
161 /*
162 * allocate FTREE node and add to the end of the linked list (args are
163 * processed in the same order they were passed to pax). Get rid of any
164 * trailing / the user may pass us. (watch out for / by itself).
165 */
166 if ((ft = malloc(sizeof(FTREE))) == NULL) {
167 paxwarn(0, "Unable to allocate memory for filename");
168 return(-1);
169 }
170
171 if (((len = strlen(str) - 1) > 0) && (str[len] == '/'))
172 str[len] = '\0';
173 ft->fname = str;
174 ft->refcnt = 0;
175 ft->chflg = chflg;
176 ft->fow = NULL;
177 if (fthead == NULL) {
178 fttail = fthead = ft;
179 return(0);
180 }
181 fttail->fow = ft;
182 fttail = ft;
183 return(0);
184 }
185
186 /*
187 * ftree_sel()
188 * this entry has been selected by pax. bump up reference count and handle
189 * -n and -d processing.
190 */
191
192 void
ftree_sel(ARCHD * arcn)193 ftree_sel(ARCHD *arcn)
194 {
195 /*
196 * set reference bit for this pattern. This linked list is only used
197 * when file trees are supplied pax as args. The list is not used when
198 * the trees are read from stdin.
199 */
200 if (ftcur != NULL)
201 ftcur->refcnt = 1;
202
203 /*
204 * if -n we are done with this arg, force a skip to the next arg when
205 * pax asks for the next file in next_file().
206 * if -d we tell fts only to match the directory (if the arg is a dir)
207 * and not the entire file tree rooted at that point.
208 */
209 if (nflag)
210 ftree_skip = 1;
211
212 if (!dflag || (arcn->type != PAX_DIR))
213 return;
214
215 if (ftent != NULL)
216 (void)fts_set(ftsp, ftent, FTS_SKIP);
217 }
218
219 /*
220 * ftree_skipped_newer()
221 * file has been skipped because a newer file exists and -u/-D given
222 */
223
224 void
ftree_skipped_newer(ARCHD * arcn)225 ftree_skipped_newer(ARCHD *arcn)
226 {
227 /* skipped due to -u/-D, mark accordingly */
228 if (ftcur != NULL)
229 ftcur->refcnt = 1;
230 }
231
232 /*
233 * ftree_chk()
234 * called at end on pax execution. Prints all those file args that did not
235 * have a selected member (reference count still 0)
236 */
237
238 void
ftree_chk(void)239 ftree_chk(void)
240 {
241 FTREE *ft;
242 int wban = 0;
243
244 /*
245 * make sure all dir access times were reset.
246 */
247 if (tflag)
248 atdir_end();
249
250 /*
251 * walk down list and check reference count. Print out those members
252 * that never had a match
253 */
254 for (ft = fthead; ft != NULL; ft = ft->fow) {
255 if ((ft->refcnt > 0) || ft->chflg)
256 continue;
257 if (wban == 0) {
258 paxwarn(1,"WARNING! These file names were not selected:");
259 ++wban;
260 }
261 (void)fprintf(stderr, "%s\n", ft->fname);
262 }
263 }
264
265 /*
266 * ftree_arg()
267 * Get the next file arg for fts to process. Can be from either the linked
268 * list or read from stdin when the user did not them as args to pax. Each
269 * arg is processed until the first successful fts_open().
270 * Return:
271 * 0 when the next arg is ready to go, -1 if out of file args (or EOF on
272 * stdin).
273 */
274
275 static int
ftree_arg(void)276 ftree_arg(void)
277 {
278
279 /*
280 * close off the current file tree
281 */
282 if (ftsp != NULL) {
283 (void)fts_close(ftsp);
284 ftsp = NULL;
285 }
286
287 /*
288 * keep looping until we get a valid file tree to process. Stop when we
289 * reach the end of the list (or get an eof on stdin)
290 */
291 for (;;) {
292 if (fthead == NULL) {
293 /*
294 * the user didn't supply any args, get the file trees
295 * to process from stdin;
296 */
297 if (getpathname(farray[0], PAXPATHLEN+1) == NULL)
298 return(-1);
299 } else {
300 /*
301 * the user supplied the file args as arguments to pax
302 */
303 if (ftcur == NULL)
304 ftcur = fthead;
305 else if ((ftcur = ftcur->fow) == NULL)
306 return(-1);
307 if (ftcur->chflg) {
308 /* First fchdir() back... */
309 if (fchdir(cwdfd) == -1) {
310 syswarn(1, errno,
311 "Can't fchdir to starting directory");
312 return(-1);
313 }
314 if (chdir(ftcur->fname) == -1) {
315 syswarn(1, errno, "Can't chdir to %s",
316 ftcur->fname);
317 return(-1);
318 }
319 continue;
320 } else
321 farray[0] = ftcur->fname;
322 }
323
324 /*
325 * watch it, fts wants the file arg stored in a array of char
326 * ptrs, with the last one a null. we use a two element array
327 * and set farray[0] to point at the buffer with the file name
328 * in it. We cannot pass all the file args to fts at one shot
329 * as we need to keep a handle on which file arg generates what
330 * files (the -n and -d flags need this). If the open is
331 * successful, return a 0.
332 */
333 if ((ftsp = fts_open(farray, ftsopts, NULL)) != NULL)
334 break;
335 }
336 return(0);
337 }
338
339 /*
340 * next_file()
341 * supplies the next file to process in the supplied archd structure.
342 * Return:
343 * 0 when contents of arcn have been set with the next file, -1 when done.
344 */
345
346 int
next_file(ARCHD * arcn)347 next_file(ARCHD *arcn)
348 {
349 int cnt;
350
351 /*
352 * ftree_sel() might have set the ftree_skip flag if the user has the
353 * -n option and a file was selected from this file arg tree. (-n says
354 * only one member is matched for each pattern) ftree_skip being 1
355 * forces us to go to the next arg now.
356 */
357 if (ftree_skip) {
358 /*
359 * clear and go to next arg
360 */
361 ftree_skip = 0;
362 if (ftree_arg() < 0)
363 return(-1);
364 }
365
366 /*
367 * loop until we get a valid file to process
368 */
369 for (;;) {
370 if ((ftent = fts_read(ftsp)) == NULL) {
371 if (errno)
372 syswarn(1, errno, "next_file");
373 /*
374 * out of files in this tree, go to next arg, if none
375 * we are done
376 */
377 if (ftree_arg() < 0)
378 return(-1);
379 continue;
380 }
381
382 /*
383 * handle each type of fts_read() flag
384 */
385 switch (ftent->fts_info) {
386 case FTS_D:
387 case FTS_DEFAULT:
388 case FTS_F:
389 case FTS_SL:
390 case FTS_SLNONE:
391 /*
392 * these are all ok
393 */
394 break;
395 case FTS_DP:
396 /*
397 * already saw this directory. If the user wants file
398 * access times reset, we use this to restore the
399 * access time for this directory since this is the
400 * last time we will see it in this file subtree
401 * remember to force the time (this is -t on a read
402 * directory, not a created directory).
403 */
404 if (!tflag)
405 continue;
406 do_atdir(ftent->fts_path, ftent->fts_statp->st_dev,
407 ftent->fts_statp->st_ino);
408 continue;
409 case FTS_DC:
410 /*
411 * fts claims a file system cycle
412 */
413 paxwarn(1,"File system cycle found at %s",ftent->fts_path);
414 continue;
415 case FTS_DNR:
416 syswarn(1, ftent->fts_errno,
417 "Unable to read directory %s", ftent->fts_path);
418 continue;
419 case FTS_ERR:
420 syswarn(1, ftent->fts_errno,
421 "File system traversal error");
422 continue;
423 case FTS_NS:
424 case FTS_NSOK:
425 syswarn(1, ftent->fts_errno,
426 "Unable to access %s", ftent->fts_path);
427 continue;
428 }
429
430 /*
431 * ok got a file tree node to process. copy info into arcn
432 * structure (initialize as required)
433 */
434 arcn->skip = 0;
435 arcn->pad = 0;
436 arcn->ln_nlen = 0;
437 arcn->ln_name[0] = '\0';
438 memcpy(&arcn->sb, ftent->fts_statp, sizeof(arcn->sb));
439
440 /*
441 * file type based set up and copy into the arcn struct
442 * SIDE NOTE:
443 * we try to reset the access time on all files and directories
444 * we may read when the -t flag is specified. files are reset
445 * when we close them after copying. we reset the directories
446 * when we are done with their file tree (we also clean up at
447 * end in case we cut short a file tree traversal). However
448 * there is no way to reset access times on symlinks.
449 */
450 switch (S_IFMT & arcn->sb.st_mode) {
451 case S_IFDIR:
452 arcn->type = PAX_DIR;
453 if (!tflag)
454 break;
455 add_atdir(ftent->fts_path, arcn->sb.st_dev,
456 arcn->sb.st_ino, &arcn->sb.st_mtim,
457 &arcn->sb.st_atim);
458 break;
459 case S_IFCHR:
460 arcn->type = PAX_CHR;
461 break;
462 case S_IFBLK:
463 arcn->type = PAX_BLK;
464 break;
465 case S_IFREG:
466 /*
467 * only regular files with have data to store on the
468 * archive. all others will store a zero length skip.
469 * the skip field is used by pax for actual data it has
470 * to read (or skip over).
471 */
472 arcn->type = PAX_REG;
473 arcn->skip = arcn->sb.st_size;
474 break;
475 case S_IFLNK:
476 arcn->type = PAX_SLK;
477 /*
478 * have to read the symlink path from the file
479 */
480 if ((cnt = readlink(ftent->fts_path, arcn->ln_name,
481 PAXPATHLEN)) == -1) {
482 syswarn(1, errno, "Unable to read symlink %s",
483 ftent->fts_path);
484 continue;
485 }
486 /*
487 * set link name length, watch out readlink does not
488 * NUL terminate the link path
489 */
490 arcn->ln_name[cnt] = '\0';
491 arcn->ln_nlen = cnt;
492 break;
493 case S_IFSOCK:
494 /*
495 * under BSD storing a socket is senseless but we will
496 * let the format specific write function make the
497 * decision of what to do with it.
498 */
499 arcn->type = PAX_SCK;
500 break;
501 case S_IFIFO:
502 arcn->type = PAX_FIF;
503 break;
504 }
505 break;
506 }
507
508 /*
509 * copy file name, set file name length
510 */
511 arcn->nlen = strlcpy(arcn->name, ftent->fts_path, sizeof(arcn->name));
512 if ((size_t)arcn->nlen >= sizeof(arcn->name))
513 arcn->nlen = sizeof(arcn->name) - 1; /* XXX truncate? */
514 arcn->org_name = ftent->fts_path;
515 return(0);
516 }
517
518 /*
519 * getpathname()
520 * Reads a pathname from stdin, handling NUL- or newline-termination.
521 * Return:
522 * NULL at end of file, otherwise the NUL-terminated buffer.
523 */
524
525 static char *
getpathname(char * buf,int buflen)526 getpathname(char *buf, int buflen)
527 {
528 char *bp, *ep;
529 int ch, term;
530
531 if (zeroflag) {
532 /*
533 * Read a NUL-terminated pathname, being especially
534 * paranoid about proper termination and pathname length.
535 */
536 for (bp = buf, ep = buf + buflen; bp < ep; bp++) {
537 if ((ch = getchar()) == EOF) {
538 if (bp != buf)
539 paxwarn(1, "Ignoring unterminated "
540 "pathname at EOF");
541 return(NULL);
542 }
543 if ((*bp = ch) == '\0')
544 return(buf);
545 }
546 /* Too long - skip this path */
547 *--bp = '\0';
548 term = '\0';
549 } else {
550 if (fgets(buf, buflen, stdin) == NULL)
551 return(NULL);
552 if ((bp = strchr(buf, '\n')) != NULL || feof(stdin)) {
553 if (bp != NULL)
554 *bp = '\0';
555 return(buf);
556 }
557 /* Too long - skip this path */
558 term = '\n';
559 }
560 while ((ch = getchar()) != term && ch != EOF)
561 continue;
562 paxwarn(1, "Ignoring too-long pathname: %s", buf);
563 return(NULL);
564 }
565