xref: /freebsd/bin/pax/ftree.c (revision 007d33500e91b61271d9374c5994f1bdf89b0890)
1 /*-
2  * Copyright (c) 1992 Keith Muller.
3  * Copyright (c) 1992, 1993
4  *	The Regents of the University of California.  All rights reserved.
5  *
6  * This code is derived from software contributed to Berkeley by
7  * Keith Muller of the University of California, San Diego.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  *
37  *	$Id: ftree.c,v 1.8 1997/08/29 16:12:24 sos Exp $
38  */
39 
40 #ifndef lint
41 static char const sccsid[] = "@(#)ftree.c	8.2 (Berkeley) 4/18/94";
42 #endif /* not lint */
43 
44 #include <sys/types.h>
45 #include <sys/time.h>
46 #include <sys/stat.h>
47 #include <unistd.h>
48 #include <string.h>
49 #include <stdio.h>
50 #include <errno.h>
51 #include <stdlib.h>
52 #include <fts.h>
53 #include "pax.h"
54 #include "ftree.h"
55 #include "extern.h"
56 
57 /*
58  * routines to interface with the fts library function.
59  *
60  * file args supplied to pax are stored on a single linked list (of type FTREE)
61  * and given to fts to be processed one at a time. pax "selects" files from
62  * the expansion of each arg into the corresponding file tree (if the arg is a
63  * directory, otherwise the node itself is just passed to pax). The selection
64  * is modified by the -n and -u flags. The user is informed when a specific
65  * file arg does not generate any selected files. -n keeps expanding the file
66  * tree arg until one of its files is selected, then skips to the next file
67  * arg. when the user does not supply the file trees as command line args to
68  * pax, they are read from stdin
69  */
70 
71 static FTS *ftsp = NULL;		/* curent FTS handle */
72 static int ftsopts;			/* options to be used on fts_open */
73 static char *farray[2];			/* array for passing each arg to fts */
74 static FTREE *fthead = NULL;		/* head of linked list of file args */
75 static FTREE *fttail = NULL;		/* tail of linked list of file args */
76 static FTREE *ftcur = NULL;		/* current file arg being processed */
77 static FTSENT *ftent = NULL;		/* current file tree entry */
78 static int ftree_skip;			/* when set skip to next file arg */
79 
80 static int ftree_arg __P((void));
81 
82 /*
83  * ftree_start()
84  *	initialize the options passed to fts_open() during this run of pax
85  *	options are based on the selection of pax options by the user
86  *	fts_start() also calls fts_arg() to open the first valid file arg. We
87  *	also attempt to reset directory access times when -t (tflag) is set.
88  * Return:
89  *	0 if there is at least one valid file arg to process, -1 otherwise
90  */
91 
92 #if __STDC__
93 int
94 ftree_start(void)
95 #else
96 int
97 ftree_start()
98 #endif
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 #	ifdef NET2_FTS
124 		pax_warn(0, "The -H flag is not supported on this version");
125 #	else
126 		ftsopts |= FTS_COMFOLLOW;
127 #	endif
128 	if (Xflag)
129 		ftsopts |= FTS_XDEV;
130 
131 	if ((fthead == NULL) && ((farray[0] = malloc(PAXPATHLEN+2)) == NULL)) {
132 		pax_warn(1, "Unable to allocate memory for file name buffer");
133 		return(-1);
134 	}
135 
136 	if (ftree_arg() < 0)
137 		return(-1);
138 	if (tflag && (atdir_start() < 0))
139 		return(-1);
140 	return(0);
141 }
142 
143 /*
144  * ftree_add()
145  *	add the arg to the linked list of files to process. Each will be
146  *	processed by fts one at a time
147  * Return:
148  *	0 if added to the linked list, -1 if failed
149  */
150 
151 #if __STDC__
152 int
153 ftree_add(register char *str)
154 #else
155 int
156 ftree_add(str)
157 	register char *str;
158 #endif
159 {
160 	register FTREE *ft;
161 	register int len;
162 
163 	/*
164 	 * simple check for bad args
165 	 */
166 	if ((str == NULL) || (*str == '\0')) {
167 		pax_warn(0, "Invalid file name arguement");
168 		return(-1);
169 	}
170 
171 	/*
172 	 * allocate FTREE node and add to the end of the linked list (args are
173 	 * processed in the same order they were passed to pax). Get rid of any
174 	 * trailing / the user may pass us. (watch out for / by itself).
175 	 */
176 	if ((ft = (FTREE *)malloc(sizeof(FTREE))) == NULL) {
177 		pax_warn(0, "Unable to allocate memory for filename");
178 		return(-1);
179 	}
180 
181 	if (((len = strlen(str) - 1) > 0) && (str[len] == '/'))
182 		str[len] = '\0';
183 	ft->fname = str;
184 	ft->refcnt = 0;
185 	ft->fow = NULL;
186 	if (fthead == NULL) {
187 		fttail = fthead = ft;
188 		return(0);
189 	}
190 	fttail->fow = ft;
191 	fttail = ft;
192 	return(0);
193 }
194 
195 /*
196  * ftree_sel()
197  *	this entry has been selected by pax. bump up reference count and handle
198  *	-n and -d processing.
199  */
200 
201 #if __STDC__
202 void
203 ftree_sel(register ARCHD *arcn)
204 #else
205 void
206 ftree_sel(arcn)
207 	register ARCHD *arcn;
208 #endif
209 {
210 	/*
211 	 * set reference bit for this pattern. This linked list is only used
212 	 * when file trees are supplied pax as args. The list is not used when
213 	 * the trees are read from stdin.
214 	 */
215 	if (ftcur != NULL)
216 		ftcur->refcnt = 1;
217 
218 	/*
219 	 * if -n we are done with this arg, force a skip to the next arg when
220 	 * pax asks for the next file in next_file().
221 	 * if -d we tell fts only to match the directory (if the arg is a dir)
222 	 * and not the entire file tree rooted at that point.
223 	 */
224 	if (nflag)
225 		ftree_skip = 1;
226 
227 	if (!dflag || (arcn->type != PAX_DIR))
228 		return;
229 
230 	if (ftent != NULL)
231 		(void)fts_set(ftsp, ftent, FTS_SKIP);
232 }
233 
234 /*
235  * ftree_chk()
236  *	called at end on pax execution. Prints all those file args that did not
237  *	have a selected member (reference count still 0)
238  */
239 
240 #if __STDC__
241 void
242 ftree_chk(void)
243 #else
244 void
245 ftree_chk()
246 #endif
247 {
248 	register FTREE *ft;
249 	register int wban = 0;
250 
251 	/*
252 	 * make sure all dir access times were reset.
253 	 */
254 	if (tflag)
255 		atdir_end();
256 
257 	/*
258 	 * walk down list and check reference count. Print out those members
259 	 * that never had a match
260 	 */
261 	for (ft = fthead; ft != NULL; ft = ft->fow) {
262 		if (ft->refcnt > 0)
263 			continue;
264 		if (wban == 0) {
265 			pax_warn(1,"WARNING! These file names were not selected:");
266 			++wban;
267 		}
268 		(void)fprintf(stderr, "%s\n", ft->fname);
269 	}
270 }
271 
272 /*
273  * ftree_arg()
274  *	Get the next file arg for fts to process. Can be from either the linked
275  *	list or read from stdin when the user did not them as args to pax. Each
276  *	arg is processed until the first successful fts_open().
277  * Return:
278  *	0 when the next arg is ready to go, -1 if out of file args (or EOF on
279  *	stdin).
280  */
281 
282 #if __STDC__
283 static int
284 ftree_arg(void)
285 #else
286 static int
287 ftree_arg()
288 #endif
289 {
290 	register char *pt;
291 
292 	/*
293 	 * close off the current file tree
294 	 */
295 	if (ftsp != NULL) {
296 		(void)fts_close(ftsp);
297 		ftsp = NULL;
298 	}
299 
300 	/*
301 	 * keep looping until we get a valid file tree to process. Stop when we
302 	 * reach the end of the list (or get an eof on stdin)
303 	 */
304 	for(;;) {
305 		if (fthead == NULL) {
306 			/*
307 			 * the user didn't supply any args, get the file trees
308 			 * to process from stdin;
309 			 */
310 			if (fgets(farray[0], PAXPATHLEN+1, stdin) == NULL)
311 				return(-1);
312 			if ((pt = strchr(farray[0], '\n')) != NULL)
313 				*pt = '\0';
314 		} else {
315 			/*
316 			 * the user supplied the file args as arguements to pax
317 			 */
318 			if (ftcur == NULL)
319 				ftcur = fthead;
320 			else if ((ftcur = ftcur->fow) == NULL)
321 				return(-1);
322 			farray[0] = ftcur->fname;
323 		}
324 
325 		/*
326 		 * watch it, fts wants the file arg stored in a array of char
327 		 * ptrs, with the last one a null. we use a two element array
328 		 * and set farray[0] to point at the buffer with the file name
329 		 * in it. We cannnot pass all the file args to fts at one shot
330 		 * as we need to keep a handle on which file arg generates what
331 		 * files (the -n and -d flags need this). If the open is
332 		 * successful, return a 0.
333 		 */
334 		if ((ftsp = fts_open(farray, ftsopts, NULL)) != NULL)
335 			break;
336 	}
337 	return(0);
338 }
339 
340 /*
341  * next_file()
342  *	supplies the next file to process in the supplied archd structure.
343  * Return:
344  *	0 when contents of arcn have been set with the next file, -1 when done.
345  */
346 
347 #if __STDC__
348 int
349 next_file(register ARCHD *arcn)
350 #else
351 int
352 next_file(arcn)
353 	register ARCHD *arcn;
354 #endif
355 {
356 	register int cnt;
357 	time_t atime;
358 	time_t mtime;
359 
360 	/*
361 	 * ftree_sel() might have set the ftree_skip flag if the user has the
362 	 * -n option and a file was selected from this file arg tree. (-n says
363 	 * only one member is matched for each pattern) ftree_skip being 1
364 	 * forces us to go to the next arg now.
365 	 */
366 	if (ftree_skip) {
367 		/*
368 		 * clear and go to next arg
369 		 */
370 		ftree_skip = 0;
371 		if (ftree_arg() < 0)
372 			return(-1);
373 	}
374 
375 	/*
376 	 * loop until we get a valid file to process
377 	 */
378 	for(;;) {
379 		if ((ftent = fts_read(ftsp)) == NULL) {
380 			/*
381 			 * out of files in this tree, go to next arg, if none
382 			 * we are done
383 			 */
384 			if (ftree_arg() < 0)
385 				return(-1);
386 			continue;
387 		}
388 
389 		/*
390 		 * handle each type of fts_read() flag
391 		 */
392 		switch(ftent->fts_info) {
393 		case FTS_D:
394 		case FTS_DEFAULT:
395 		case FTS_F:
396 		case FTS_SL:
397 		case FTS_SLNONE:
398 			/*
399 			 * these are all ok
400 			 */
401 			break;
402 		case FTS_DP:
403 			/*
404 			 * already saw this directory. If the user wants file
405 			 * access times reset, we use this to restore the
406 			 * access time for this directory since this is the
407 			 * last time we will see it in this file subtree
408 			 * remember to force the time (this is -t on a read
409 			 * directory, not a created directory).
410 			 */
411 #			ifdef NET2_FTS
412 			if (!tflag || (get_atdir(ftent->fts_statb.st_dev,
413 			    ftent->fts_statb.st_ino, &mtime, &atime) < 0))
414 #			else
415 			if (!tflag || (get_atdir(ftent->fts_statp->st_dev,
416 			    ftent->fts_statp->st_ino, &mtime, &atime) < 0))
417 #			endif
418 				continue;
419 			set_ftime(ftent->fts_path, mtime, atime, 1);
420 			continue;
421 		case FTS_DC:
422 			/*
423 			 * fts claims a file system cycle
424 			 */
425 			pax_warn(1,"File system cycle found at %s",ftent->fts_path);
426 			continue;
427 		case FTS_DNR:
428 #			ifdef NET2_FTS
429 			sys_warn(1, errno,
430 #			else
431 			sys_warn(1, ftent->fts_errno,
432 #			endif
433 			    "Unable to read directory %s", ftent->fts_path);
434 			continue;
435 		case FTS_ERR:
436 #			ifdef NET2_FTS
437 			sys_warn(1, errno,
438 #			else
439 			sys_warn(1, ftent->fts_errno,
440 #			endif
441 			    "File system traversal error");
442 			continue;
443 		case FTS_NS:
444 		case FTS_NSOK:
445 #			ifdef NET2_FTS
446 			sys_warn(1, errno,
447 #			else
448 			sys_warn(1, ftent->fts_errno,
449 #			endif
450 			    "Unable to access %s", ftent->fts_path);
451 			continue;
452 		}
453 
454 		/*
455 		 * ok got a file tree node to process. copy info into arcn
456 		 * structure (initialize as required)
457 		 */
458 		arcn->skip = 0;
459 		arcn->pad = 0;
460 		arcn->ln_nlen = 0;
461 		arcn->ln_name[0] = '\0';
462 #		ifdef NET2_FTS
463 		arcn->sb = ftent->fts_statb;
464 #		else
465 		arcn->sb = *(ftent->fts_statp);
466 #		endif
467 
468 		/*
469 		 * file type based set up and copy into the arcn struct
470 		 * SIDE NOTE:
471 		 * we try to reset the access time on all files and directories
472 		 * we may read when the -t flag is specified. files are reset
473 		 * when we close them after copying. we reset the directories
474 		 * when we are done with their file tree (we also clean up at
475 		 * end in case we cut short a file tree traversal). However
476 		 * there is no way to reset access times on symlinks.
477 		 */
478 		switch(S_IFMT & arcn->sb.st_mode) {
479 		case S_IFDIR:
480 			arcn->type = PAX_DIR;
481 			if (!tflag)
482 				break;
483 			add_atdir(ftent->fts_path, arcn->sb.st_dev,
484 			    arcn->sb.st_ino, arcn->sb.st_mtime,
485 			    arcn->sb.st_atime);
486 			break;
487 		case S_IFCHR:
488 			arcn->type = PAX_CHR;
489 			break;
490 		case S_IFBLK:
491 			arcn->type = PAX_BLK;
492 			break;
493 		case S_IFREG:
494 			/*
495 			 * only regular files with have data to store on the
496 			 * archive. all others will store a zero length skip.
497 			 * the skip field is used by pax for actual data it has
498 			 * to read (or skip over).
499 			 */
500 			arcn->type = PAX_REG;
501 			arcn->skip = arcn->sb.st_size;
502 			break;
503 		case S_IFLNK:
504 			arcn->type = PAX_SLK;
505 			/*
506 			 * have to read the symlink path from the file
507 			 */
508 			if ((cnt = readlink(ftent->fts_path, arcn->ln_name,
509 			    PAXPATHLEN)) < 0) {
510 				sys_warn(1, errno, "Unable to read symlink %s",
511 				    ftent->fts_path);
512 				continue;
513 			}
514 			/*
515 			 * set link name length, watch out readlink does not
516 			 * allways null terminate the link path
517 			 */
518 			arcn->ln_name[cnt] = '\0';
519 			arcn->ln_nlen = cnt;
520 			break;
521 		case S_IFSOCK:
522 			/*
523 			 * under BSD storing a socket is senseless but we will
524 			 * let the format specific write function make the
525 			 * decision of what to do with it.
526 			 */
527 			arcn->type = PAX_SCK;
528 			break;
529 		case S_IFIFO:
530 			arcn->type = PAX_FIF;
531 			break;
532 		}
533 		break;
534 	}
535 
536 	/*
537 	 * copy file name, set file name length
538 	 */
539 	arcn->nlen = l_strncpy(arcn->name, ftent->fts_path, PAXPATHLEN+1);
540 	arcn->name[arcn->nlen] = '\0';
541 	arcn->org_name = ftent->fts_path;
542 	return(0);
543 }
544