xref: /freebsd/bin/pax/ftree.c (revision d0b2dbfa0ecf2bbc9709efc5e20baf8e4b44bbbf)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1992 Keith Muller.
5  * Copyright (c) 1992, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * This code is derived from software contributed to Berkeley by
9  * Keith Muller of the University of California, San Diego.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  */
35 
36 #ifndef lint
37 #if 0
38 static char sccsid[] = "@(#)ftree.c	8.2 (Berkeley) 4/18/94";
39 #endif
40 #endif /* not lint */
41 #include <sys/cdefs.h>
42 #include <sys/types.h>
43 #include <sys/time.h>
44 #include <sys/stat.h>
45 #include <unistd.h>
46 #include <string.h>
47 #include <stdio.h>
48 #include <errno.h>
49 #include <stdlib.h>
50 #include <fts.h>
51 #include "pax.h"
52 #include "ftree.h"
53 #include "extern.h"
54 
55 /*
56  * routines to interface with the fts library function.
57  *
58  * file args supplied to pax are stored on a single linked list (of type FTREE)
59  * and given to fts to be processed one at a time. pax "selects" files from
60  * the expansion of each arg into the corresponding file tree (if the arg is a
61  * directory, otherwise the node itself is just passed to pax). The selection
62  * is modified by the -n and -u flags. The user is informed when a specific
63  * file arg does not generate any selected files. -n keeps expanding the file
64  * tree arg until one of its files is selected, then skips to the next file
65  * arg. when the user does not supply the file trees as command line args to
66  * pax, they are read from stdin
67  */
68 
69 static FTS *ftsp = NULL;		/* current FTS handle */
70 static int ftsopts;			/* options to be used on fts_open */
71 static char *farray[2];			/* array for passing each arg to fts */
72 static FTREE *fthead = NULL;		/* head of linked list of file args */
73 static FTREE *fttail = NULL;		/* tail of linked list of file args */
74 static FTREE *ftcur = NULL;		/* current file arg being processed */
75 static FTSENT *ftent = NULL;		/* current file tree entry */
76 static int ftree_skip;			/* when set skip to next file arg */
77 
78 static int ftree_arg(void);
79 
80 /*
81  * ftree_start()
82  *	initialize the options passed to fts_open() during this run of pax
83  *	options are based on the selection of pax options by the user
84  *	fts_start() also calls fts_arg() to open the first valid file arg. We
85  *	also attempt to reset directory access times when -t (tflag) is set.
86  * Return:
87  *	0 if there is at least one valid file arg to process, -1 otherwise
88  */
89 
90 int
91 ftree_start(void)
92 {
93 	/*
94 	 * Set up the operation mode of fts, open the first file arg. We must
95 	 * use FTS_NOCHDIR, as the user may have to open multiple archives and
96 	 * if fts did a chdir off into the boondocks, we may create an archive
97 	 * volume in a place where the user did not expect to.
98 	 */
99 	ftsopts = FTS_NOCHDIR;
100 
101 	/*
102 	 * optional user flags that effect file traversal
103 	 * -H command line symlink follow only (half follow)
104 	 * -L follow symlinks (logical)
105 	 * -P do not follow symlinks (physical). This is the default.
106 	 * -X do not cross over mount points
107 	 * -t preserve access times on files read.
108 	 * -n select only the first member of a file tree when a match is found
109 	 * -d do not extract subtrees rooted at a directory arg.
110 	 */
111 	if (Lflag)
112 		ftsopts |= FTS_LOGICAL;
113 	else
114 		ftsopts |= FTS_PHYSICAL;
115 	if (Hflag)
116 		ftsopts |= FTS_COMFOLLOW;
117 	if (Xflag)
118 		ftsopts |= FTS_XDEV;
119 
120 	if ((fthead == NULL) && ((farray[0] = malloc(PAXPATHLEN+2)) == NULL)) {
121 		paxwarn(1, "Unable to allocate memory for file name buffer");
122 		return(-1);
123 	}
124 
125 	if (ftree_arg() < 0)
126 		return(-1);
127 	if (tflag && (atdir_start() < 0))
128 		return(-1);
129 	return(0);
130 }
131 
132 /*
133  * ftree_add()
134  *	add the arg to the linked list of files to process. Each will be
135  *	processed by fts one at a time
136  * Return:
137  *	0 if added to the linked list, -1 if failed
138  */
139 
140 int
141 ftree_add(char *str, int chflg)
142 {
143 	FTREE *ft;
144 	int len;
145 
146 	/*
147 	 * simple check for bad args
148 	 */
149 	if ((str == NULL) || (*str == '\0')) {
150 		paxwarn(0, "Invalid file name argument");
151 		return(-1);
152 	}
153 
154 	/*
155 	 * allocate FTREE node and add to the end of the linked list (args are
156 	 * processed in the same order they were passed to pax). Get rid of any
157 	 * trailing / the user may pass us. (watch out for / by itself).
158 	 */
159 	if ((ft = (FTREE *)malloc(sizeof(FTREE))) == NULL) {
160 		paxwarn(0, "Unable to allocate memory for filename");
161 		return(-1);
162 	}
163 
164 	if (((len = strlen(str) - 1) > 0) && (str[len] == '/'))
165 		str[len] = '\0';
166 	ft->fname = str;
167 	ft->refcnt = 0;
168 	ft->chflg = chflg;
169 	ft->fow = NULL;
170 	if (fthead == NULL) {
171 		fttail = fthead = ft;
172 		return(0);
173 	}
174 	fttail->fow = ft;
175 	fttail = ft;
176 	return(0);
177 }
178 
179 /*
180  * ftree_sel()
181  *	this entry has been selected by pax. bump up reference count and handle
182  *	-n and -d processing.
183  */
184 
185 void
186 ftree_sel(ARCHD *arcn)
187 {
188 	/*
189 	 * set reference bit for this pattern. This linked list is only used
190 	 * when file trees are supplied pax as args. The list is not used when
191 	 * the trees are read from stdin.
192 	 */
193 	if (ftcur != NULL)
194 		ftcur->refcnt = 1;
195 
196 	/*
197 	 * if -n we are done with this arg, force a skip to the next arg when
198 	 * pax asks for the next file in next_file().
199 	 * if -d we tell fts only to match the directory (if the arg is a dir)
200 	 * and not the entire file tree rooted at that point.
201 	 */
202 	if (nflag)
203 		ftree_skip = 1;
204 
205 	if (!dflag || (arcn->type != PAX_DIR))
206 		return;
207 
208 	if (ftent != NULL)
209 		(void)fts_set(ftsp, ftent, FTS_SKIP);
210 }
211 
212 /*
213  * ftree_notsel()
214  *	this entry has not been selected by pax.
215  */
216 
217 void
218 ftree_notsel(void)
219 {
220 	if (ftent != NULL)
221 		(void)fts_set(ftsp, ftent, FTS_SKIP);
222 }
223 
224 /*
225  * ftree_chk()
226  *	called at end on pax execution. Prints all those file args that did not
227  *	have a selected member (reference count still 0)
228  */
229 
230 void
231 ftree_chk(void)
232 {
233 	FTREE *ft;
234 	int wban = 0;
235 
236 	/*
237 	 * make sure all dir access times were reset.
238 	 */
239 	if (tflag)
240 		atdir_end();
241 
242 	/*
243 	 * walk down list and check reference count. Print out those members
244 	 * that never had a match
245 	 */
246 	for (ft = fthead; ft != NULL; ft = ft->fow) {
247 		if ((ft->refcnt > 0) || ft->chflg)
248 			continue;
249 		if (wban == 0) {
250 			paxwarn(1,"WARNING! These file names were not selected:");
251 			++wban;
252 		}
253 		(void)fprintf(stderr, "%s\n", ft->fname);
254 	}
255 }
256 
257 /*
258  * ftree_arg()
259  *	Get the next file arg for fts to process. Can be from either the linked
260  *	list or read from stdin when the user did not them as args to pax. Each
261  *	arg is processed until the first successful fts_open().
262  * Return:
263  *	0 when the next arg is ready to go, -1 if out of file args (or EOF on
264  *	stdin).
265  */
266 
267 static int
268 ftree_arg(void)
269 {
270 	char *pt;
271 
272 	/*
273 	 * close off the current file tree
274 	 */
275 	if (ftsp != NULL) {
276 		(void)fts_close(ftsp);
277 		ftsp = NULL;
278 	}
279 
280 	/*
281 	 * keep looping until we get a valid file tree to process. Stop when we
282 	 * reach the end of the list (or get an eof on stdin)
283 	 */
284 	for(;;) {
285 		if (fthead == NULL) {
286 			/*
287 			 * the user didn't supply any args, get the file trees
288 			 * to process from stdin;
289 			 */
290 			if (fgets(farray[0], PAXPATHLEN+1, stdin) == NULL)
291 				return(-1);
292 			if ((pt = strchr(farray[0], '\n')) != NULL)
293 				*pt = '\0';
294 		} else {
295 			/*
296 			 * the user supplied the file args as arguments to pax
297 			 */
298 			if (ftcur == NULL)
299 				ftcur = fthead;
300 			else if ((ftcur = ftcur->fow) == NULL)
301 				return(-1);
302 			if (ftcur->chflg) {
303 				/* First fchdir() back... */
304 				if (fchdir(cwdfd) < 0) {
305 					syswarn(1, errno,
306 					  "Can't fchdir to starting directory");
307 					return(-1);
308 				}
309 				if (chdir(ftcur->fname) < 0) {
310 					syswarn(1, errno, "Can't chdir to %s",
311 					    ftcur->fname);
312 					return(-1);
313 				}
314 				continue;
315 			} else
316 				farray[0] = ftcur->fname;
317 		}
318 
319 		/*
320 		 * Watch it, fts wants the file arg stored in an array of char
321 		 * ptrs, with the last one a null. We use a two element array
322 		 * and set farray[0] to point at the buffer with the file name
323 		 * in it. We cannot pass all the file args to fts at one shot
324 		 * as we need to keep a handle on which file arg generates what
325 		 * files (the -n and -d flags need this). If the open is
326 		 * successful, return a 0.
327 		 */
328 		if ((ftsp = fts_open(farray, ftsopts, NULL)) != NULL)
329 			break;
330 	}
331 	return(0);
332 }
333 
334 /*
335  * next_file()
336  *	supplies the next file to process in the supplied archd structure.
337  * Return:
338  *	0 when contents of arcn have been set with the next file, -1 when done.
339  */
340 
341 int
342 next_file(ARCHD *arcn)
343 {
344 	int cnt;
345 	time_t atime;
346 	time_t mtime;
347 
348 	/*
349 	 * ftree_sel() might have set the ftree_skip flag if the user has the
350 	 * -n option and a file was selected from this file arg tree. (-n says
351 	 * only one member is matched for each pattern) ftree_skip being 1
352 	 * forces us to go to the next arg now.
353 	 */
354 	if (ftree_skip) {
355 		/*
356 		 * clear and go to next arg
357 		 */
358 		ftree_skip = 0;
359 		if (ftree_arg() < 0)
360 			return(-1);
361 	}
362 
363 	/*
364 	 * loop until we get a valid file to process
365 	 */
366 	for(;;) {
367 		if ((ftent = fts_read(ftsp)) == NULL) {
368 			/*
369 			 * out of files in this tree, go to next arg, if none
370 			 * we are done
371 			 */
372 			if (ftree_arg() < 0)
373 				return(-1);
374 			continue;
375 		}
376 
377 		/*
378 		 * handle each type of fts_read() flag
379 		 */
380 		switch(ftent->fts_info) {
381 		case FTS_D:
382 		case FTS_DEFAULT:
383 		case FTS_F:
384 		case FTS_SL:
385 		case FTS_SLNONE:
386 			/*
387 			 * these are all ok
388 			 */
389 			break;
390 		case FTS_DP:
391 			/*
392 			 * already saw this directory. If the user wants file
393 			 * access times reset, we use this to restore the
394 			 * access time for this directory since this is the
395 			 * last time we will see it in this file subtree
396 			 * remember to force the time (this is -t on a read
397 			 * directory, not a created directory).
398 			 */
399 			if (!tflag || (get_atdir(ftent->fts_statp->st_dev,
400 			    ftent->fts_statp->st_ino, &mtime, &atime) < 0))
401 				continue;
402 			set_ftime(ftent->fts_path, mtime, atime, 1);
403 			continue;
404 		case FTS_DC:
405 			/*
406 			 * fts claims a file system cycle
407 			 */
408 			paxwarn(1,"File system cycle found at %s",ftent->fts_path);
409 			continue;
410 		case FTS_DNR:
411 			syswarn(1, ftent->fts_errno,
412 			    "Unable to read directory %s", ftent->fts_path);
413 			continue;
414 		case FTS_ERR:
415 			syswarn(1, ftent->fts_errno,
416 			    "File system traversal error");
417 			continue;
418 		case FTS_NS:
419 		case FTS_NSOK:
420 			syswarn(1, ftent->fts_errno,
421 			    "Unable to access %s", ftent->fts_path);
422 			continue;
423 		}
424 
425 		/*
426 		 * ok got a file tree node to process. copy info into arcn
427 		 * structure (initialize as required)
428 		 */
429 		arcn->skip = 0;
430 		arcn->pad = 0;
431 		arcn->ln_nlen = 0;
432 		arcn->ln_name[0] = '\0';
433 		arcn->sb = *(ftent->fts_statp);
434 
435 		/*
436 		 * file type based set up and copy into the arcn struct
437 		 * SIDE NOTE:
438 		 * we try to reset the access time on all files and directories
439 		 * we may read when the -t flag is specified. files are reset
440 		 * when we close them after copying. we reset the directories
441 		 * when we are done with their file tree (we also clean up at
442 		 * end in case we cut short a file tree traversal). However
443 		 * there is no way to reset access times on symlinks.
444 		 */
445 		switch(S_IFMT & arcn->sb.st_mode) {
446 		case S_IFDIR:
447 			arcn->type = PAX_DIR;
448 			if (!tflag)
449 				break;
450 			add_atdir(ftent->fts_path, arcn->sb.st_dev,
451 			    arcn->sb.st_ino, arcn->sb.st_mtime,
452 			    arcn->sb.st_atime);
453 			break;
454 		case S_IFCHR:
455 			arcn->type = PAX_CHR;
456 			break;
457 		case S_IFBLK:
458 			arcn->type = PAX_BLK;
459 			break;
460 		case S_IFREG:
461 			/*
462 			 * only regular files with have data to store on the
463 			 * archive. all others will store a zero length skip.
464 			 * the skip field is used by pax for actual data it has
465 			 * to read (or skip over).
466 			 */
467 			arcn->type = PAX_REG;
468 			arcn->skip = arcn->sb.st_size;
469 			break;
470 		case S_IFLNK:
471 			arcn->type = PAX_SLK;
472 			/*
473 			 * have to read the symlink path from the file
474 			 */
475 			if ((cnt = readlink(ftent->fts_path, arcn->ln_name,
476 			    PAXPATHLEN - 1)) < 0) {
477 				syswarn(1, errno, "Unable to read symlink %s",
478 				    ftent->fts_path);
479 				continue;
480 			}
481 			/*
482 			 * set link name length, watch out readlink does not
483 			 * always NUL terminate the link path
484 			 */
485 			arcn->ln_name[cnt] = '\0';
486 			arcn->ln_nlen = cnt;
487 			break;
488 		case S_IFSOCK:
489 			/*
490 			 * under BSD storing a socket is senseless but we will
491 			 * let the format specific write function make the
492 			 * decision of what to do with it.
493 			 */
494 			arcn->type = PAX_SCK;
495 			break;
496 		case S_IFIFO:
497 			arcn->type = PAX_FIF;
498 			break;
499 		}
500 		break;
501 	}
502 
503 	/*
504 	 * copy file name, set file name length
505 	 */
506 	arcn->nlen = l_strncpy(arcn->name, ftent->fts_path, sizeof(arcn->name) - 1);
507 	arcn->name[arcn->nlen] = '\0';
508 	arcn->org_name = ftent->fts_path;
509 	return(0);
510 }
511