archive_read_support_format_mtree.c revision 228753 
1/*-
2 * Copyright (c) 2003-2007 Tim Kientzle
3 * Copyright (c) 2008 Joerg Sonnenberger
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 *    notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 *    notice, this list of conditions and the following disclaimer in the
13 *    documentation and/or other materials provided with the distribution.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25 */
26
27#include "archive_platform.h"
28__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_mtree.c 201165 2009-12-29 05:52:13Z kientzle $");
29
30#ifdef HAVE_SYS_STAT_H
31#include <sys/stat.h>
32#endif
33#ifdef HAVE_ERRNO_H
34#include <errno.h>
35#endif
36#ifdef HAVE_FCNTL_H
37#include <fcntl.h>
38#endif
39#include <stddef.h>
40/* #include <stdint.h> */ /* See archive_platform.h */
41#ifdef HAVE_STDLIB_H
42#include <stdlib.h>
43#endif
44#ifdef HAVE_STRING_H
45#include <string.h>
46#endif
47
48#include "archive.h"
49#include "archive_entry.h"
50#include "archive_private.h"
51#include "archive_read_private.h"
52#include "archive_string.h"
53
54#ifndef O_BINARY
55#define	O_BINARY 0
56#endif
57
58#define	MTREE_HAS_DEVICE	0x0001
59#define	MTREE_HAS_FFLAGS	0x0002
60#define	MTREE_HAS_GID		0x0004
61#define	MTREE_HAS_GNAME		0x0008
62#define	MTREE_HAS_MTIME		0x0010
63#define	MTREE_HAS_NLINK		0x0020
64#define	MTREE_HAS_PERM		0x0040
65#define	MTREE_HAS_SIZE		0x0080
66#define	MTREE_HAS_TYPE		0x0100
67#define	MTREE_HAS_UID		0x0200
68#define	MTREE_HAS_UNAME		0x0400
69
70#define	MTREE_HAS_OPTIONAL	0x0800
71
72struct mtree_option {
73	struct mtree_option *next;
74	char *value;
75};
76
77struct mtree_entry {
78	struct mtree_entry *next;
79	struct mtree_option *options;
80	char *name;
81	char full;
82	char used;
83};
84
85struct mtree {
86	struct archive_string	 line;
87	size_t			 buffsize;
88	char			*buff;
89	off_t			 offset;
90	int			 fd;
91	int			 filetype;
92	int			 archive_format;
93	const char		*archive_format_name;
94	struct mtree_entry	*entries;
95	struct mtree_entry	*this_entry;
96	struct archive_string	 current_dir;
97	struct archive_string	 contents_name;
98
99	struct archive_entry_linkresolver *resolver;
100
101	off_t			 cur_size, cur_offset;
102};
103
104static int	cleanup(struct archive_read *);
105static int	mtree_bid(struct archive_read *);
106static int	parse_file(struct archive_read *, struct archive_entry *,
107		    struct mtree *, struct mtree_entry *, int *);
108static void	parse_escapes(char *, struct mtree_entry *);
109static int	parse_line(struct archive_read *, struct archive_entry *,
110		    struct mtree *, struct mtree_entry *, int *);
111static int	parse_keyword(struct archive_read *, struct mtree *,
112		    struct archive_entry *, struct mtree_option *, int *);
113static int	read_data(struct archive_read *a,
114		    const void **buff, size_t *size, off_t *offset);
115static ssize_t	readline(struct archive_read *, struct mtree *, char **, ssize_t);
116static int	skip(struct archive_read *a);
117static int	read_header(struct archive_read *,
118		    struct archive_entry *);
119static int64_t	mtree_atol10(char **);
120static int64_t	mtree_atol8(char **);
121static int64_t	mtree_atol(char **);
122
123static void
124free_options(struct mtree_option *head)
125{
126	struct mtree_option *next;
127
128	for (; head != NULL; head = next) {
129		next = head->next;
130		free(head->value);
131		free(head);
132	}
133}
134
135int
136archive_read_support_format_mtree(struct archive *_a)
137{
138	struct archive_read *a = (struct archive_read *)_a;
139	struct mtree *mtree;
140	int r;
141
142	mtree = (struct mtree *)malloc(sizeof(*mtree));
143	if (mtree == NULL) {
144		archive_set_error(&a->archive, ENOMEM,
145		    "Can't allocate mtree data");
146		return (ARCHIVE_FATAL);
147	}
148	memset(mtree, 0, sizeof(*mtree));
149	mtree->fd = -1;
150
151	r = __archive_read_register_format(a, mtree, "mtree",
152	    mtree_bid, NULL, read_header, read_data, skip, cleanup);
153
154	if (r != ARCHIVE_OK)
155		free(mtree);
156	return (ARCHIVE_OK);
157}
158
159static int
160cleanup(struct archive_read *a)
161{
162	struct mtree *mtree;
163	struct mtree_entry *p, *q;
164
165	mtree = (struct mtree *)(a->format->data);
166
167	p = mtree->entries;
168	while (p != NULL) {
169		q = p->next;
170		free(p->name);
171		free_options(p->options);
172		free(p);
173		p = q;
174	}
175	archive_string_free(&mtree->line);
176	archive_string_free(&mtree->current_dir);
177	archive_string_free(&mtree->contents_name);
178	archive_entry_linkresolver_free(mtree->resolver);
179
180	free(mtree->buff);
181	free(mtree);
182	(a->format->data) = NULL;
183	return (ARCHIVE_OK);
184}
185
186
187static int
188mtree_bid(struct archive_read *a)
189{
190	const char *signature = "#mtree";
191	const char *p;
192
193	/* Now let's look at the actual header and see if it matches. */
194	p = __archive_read_ahead(a, strlen(signature), NULL);
195	if (p == NULL)
196		return (-1);
197
198	if (strncmp(p, signature, strlen(signature)) == 0)
199		return (8 * (int)strlen(signature));
200	return (0);
201}
202
203/*
204 * The extended mtree format permits multiple lines specifying
205 * attributes for each file.  For those entries, only the last line
206 * is actually used.  Practically speaking, that means we have
207 * to read the entire mtree file into memory up front.
208 *
209 * The parsing is done in two steps.  First, it is decided if a line
210 * changes the global defaults and if it is, processed accordingly.
211 * Otherwise, the options of the line are merged with the current
212 * global options.
213 */
214static int
215add_option(struct archive_read *a, struct mtree_option **global,
216    const char *value, size_t len)
217{
218	struct mtree_option *option;
219
220	if ((option = malloc(sizeof(*option))) == NULL) {
221		archive_set_error(&a->archive, errno, "Can't allocate memory");
222		return (ARCHIVE_FATAL);
223	}
224	if ((option->value = malloc(len + 1)) == NULL) {
225		free(option);
226		archive_set_error(&a->archive, errno, "Can't allocate memory");
227		return (ARCHIVE_FATAL);
228	}
229	memcpy(option->value, value, len);
230	option->value[len] = '\0';
231	option->next = *global;
232	*global = option;
233	return (ARCHIVE_OK);
234}
235
236static void
237remove_option(struct mtree_option **global, const char *value, size_t len)
238{
239	struct mtree_option *iter, *last;
240
241	last = NULL;
242	for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
243		if (strncmp(iter->value, value, len) == 0 &&
244		    (iter->value[len] == '\0' ||
245		     iter->value[len] == '='))
246			break;
247	}
248	if (iter == NULL)
249		return;
250	if (last == NULL)
251		*global = iter->next;
252	else
253		last->next = iter->next;
254
255	free(iter->value);
256	free(iter);
257}
258
259static int
260process_global_set(struct archive_read *a,
261    struct mtree_option **global, const char *line)
262{
263	const char *next, *eq;
264	size_t len;
265	int r;
266
267	line += 4;
268	for (;;) {
269		next = line + strspn(line, " \t\r\n");
270		if (*next == '\0')
271			return (ARCHIVE_OK);
272		line = next;
273		next = line + strcspn(line, " \t\r\n");
274		eq = strchr(line, '=');
275		if (eq > next)
276			len = next - line;
277		else
278			len = eq - line;
279
280		remove_option(global, line, len);
281		r = add_option(a, global, line, next - line);
282		if (r != ARCHIVE_OK)
283			return (r);
284		line = next;
285	}
286}
287
288static int
289process_global_unset(struct archive_read *a,
290    struct mtree_option **global, const char *line)
291{
292	const char *next;
293	size_t len;
294
295	line += 6;
296	if (strchr(line, '=') != NULL) {
297		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
298		    "/unset shall not contain `='");
299		return ARCHIVE_FATAL;
300	}
301
302	for (;;) {
303		next = line + strspn(line, " \t\r\n");
304		if (*next == '\0')
305			return (ARCHIVE_OK);
306		line = next;
307		len = strcspn(line, " \t\r\n");
308
309		if (len == 3 && strncmp(line, "all", 3) == 0) {
310			free_options(*global);
311			*global = NULL;
312		} else {
313			remove_option(global, line, len);
314		}
315
316		line += len;
317	}
318}
319
320static int
321process_add_entry(struct archive_read *a, struct mtree *mtree,
322    struct mtree_option **global, const char *line,
323    struct mtree_entry **last_entry)
324{
325	struct mtree_entry *entry;
326	struct mtree_option *iter;
327	const char *next, *eq;
328	size_t len;
329	int r;
330
331	if ((entry = malloc(sizeof(*entry))) == NULL) {
332		archive_set_error(&a->archive, errno, "Can't allocate memory");
333		return (ARCHIVE_FATAL);
334	}
335	entry->next = NULL;
336	entry->options = NULL;
337	entry->name = NULL;
338	entry->used = 0;
339	entry->full = 0;
340
341	/* Add this entry to list. */
342	if (*last_entry == NULL)
343		mtree->entries = entry;
344	else
345		(*last_entry)->next = entry;
346	*last_entry = entry;
347
348	len = strcspn(line, " \t\r\n");
349	if ((entry->name = malloc(len + 1)) == NULL) {
350		archive_set_error(&a->archive, errno, "Can't allocate memory");
351		return (ARCHIVE_FATAL);
352	}
353
354	memcpy(entry->name, line, len);
355	entry->name[len] = '\0';
356	parse_escapes(entry->name, entry);
357
358	line += len;
359	for (iter = *global; iter != NULL; iter = iter->next) {
360		r = add_option(a, &entry->options, iter->value,
361		    strlen(iter->value));
362		if (r != ARCHIVE_OK)
363			return (r);
364	}
365
366	for (;;) {
367		next = line + strspn(line, " \t\r\n");
368		if (*next == '\0')
369			return (ARCHIVE_OK);
370		line = next;
371		next = line + strcspn(line, " \t\r\n");
372		eq = strchr(line, '=');
373		if (eq == NULL || eq > next)
374			len = next - line;
375		else
376			len = eq - line;
377
378		remove_option(&entry->options, line, len);
379		r = add_option(a, &entry->options, line, next - line);
380		if (r != ARCHIVE_OK)
381			return (r);
382		line = next;
383	}
384}
385
386static int
387read_mtree(struct archive_read *a, struct mtree *mtree)
388{
389	ssize_t len;
390	uintmax_t counter;
391	char *p;
392	struct mtree_option *global;
393	struct mtree_entry *last_entry;
394	int r;
395
396	mtree->archive_format = ARCHIVE_FORMAT_MTREE;
397	mtree->archive_format_name = "mtree";
398
399	global = NULL;
400	last_entry = NULL;
401
402	for (counter = 1; ; ++counter) {
403		len = readline(a, mtree, &p, 256);
404		if (len == 0) {
405			mtree->this_entry = mtree->entries;
406			free_options(global);
407			return (ARCHIVE_OK);
408		}
409		if (len < 0) {
410			free_options(global);
411			return (len);
412		}
413		/* Leading whitespace is never significant, ignore it. */
414		while (*p == ' ' || *p == '\t') {
415			++p;
416			--len;
417		}
418		/* Skip content lines and blank lines. */
419		if (*p == '#')
420			continue;
421		if (*p == '\r' || *p == '\n' || *p == '\0')
422			continue;
423		if (*p != '/') {
424			r = process_add_entry(a, mtree, &global, p,
425			    &last_entry);
426		} else if (strncmp(p, "/set", 4) == 0) {
427			if (p[4] != ' ' && p[4] != '\t')
428				break;
429			r = process_global_set(a, &global, p);
430		} else if (strncmp(p, "/unset", 6) == 0) {
431			if (p[6] != ' ' && p[6] != '\t')
432				break;
433			r = process_global_unset(a, &global, p);
434		} else
435			break;
436
437		if (r != ARCHIVE_OK) {
438			free_options(global);
439			return r;
440		}
441	}
442
443	archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
444	    "Can't parse line %ju", counter);
445	free_options(global);
446	return (ARCHIVE_FATAL);
447}
448
449/*
450 * Read in the entire mtree file into memory on the first request.
451 * Then use the next unused file to satisfy each header request.
452 */
453static int
454read_header(struct archive_read *a, struct archive_entry *entry)
455{
456	struct mtree *mtree;
457	char *p;
458	int r, use_next;
459
460	mtree = (struct mtree *)(a->format->data);
461
462	if (mtree->fd >= 0) {
463		close(mtree->fd);
464		mtree->fd = -1;
465	}
466
467	if (mtree->entries == NULL) {
468		mtree->resolver = archive_entry_linkresolver_new();
469		if (mtree->resolver == NULL)
470			return ARCHIVE_FATAL;
471		archive_entry_linkresolver_set_strategy(mtree->resolver,
472		    ARCHIVE_FORMAT_MTREE);
473		r = read_mtree(a, mtree);
474		if (r != ARCHIVE_OK)
475			return (r);
476	}
477
478	a->archive.archive_format = mtree->archive_format;
479	a->archive.archive_format_name = mtree->archive_format_name;
480
481	for (;;) {
482		if (mtree->this_entry == NULL)
483			return (ARCHIVE_EOF);
484		if (strcmp(mtree->this_entry->name, "..") == 0) {
485			mtree->this_entry->used = 1;
486			if (archive_strlen(&mtree->current_dir) > 0) {
487				/* Roll back current path. */
488				p = mtree->current_dir.s
489				    + mtree->current_dir.length - 1;
490				while (p >= mtree->current_dir.s && *p != '/')
491					--p;
492				if (p >= mtree->current_dir.s)
493					--p;
494				mtree->current_dir.length
495				    = p - mtree->current_dir.s + 1;
496			}
497		}
498		if (!mtree->this_entry->used) {
499			use_next = 0;
500			r = parse_file(a, entry, mtree, mtree->this_entry, &use_next);
501			if (use_next == 0)
502				return (r);
503		}
504		mtree->this_entry = mtree->this_entry->next;
505	}
506}
507
508/*
509 * A single file can have multiple lines contribute specifications.
510 * Parse as many lines as necessary, then pull additional information
511 * from a backing file on disk as necessary.
512 */
513static int
514parse_file(struct archive_read *a, struct archive_entry *entry,
515    struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
516{
517	const char *path;
518	struct stat st_storage, *st;
519	struct mtree_entry *mp;
520	struct archive_entry *sparse_entry;
521	int r = ARCHIVE_OK, r1, parsed_kws, mismatched_type;
522
523	mentry->used = 1;
524
525	/* Initialize reasonable defaults. */
526	mtree->filetype = AE_IFREG;
527	archive_entry_set_size(entry, 0);
528	archive_string_empty(&mtree->contents_name);
529
530	/* Parse options from this line. */
531	parsed_kws = 0;
532	r = parse_line(a, entry, mtree, mentry, &parsed_kws);
533
534	if (mentry->full) {
535		archive_entry_copy_pathname(entry, mentry->name);
536		/*
537		 * "Full" entries are allowed to have multiple lines
538		 * and those lines aren't required to be adjacent.  We
539		 * don't support multiple lines for "relative" entries
540		 * nor do we make any attempt to merge data from
541		 * separate "relative" and "full" entries.  (Merging
542		 * "relative" and "full" entries would require dealing
543		 * with pathname canonicalization, which is a very
544		 * tricky subject.)
545		 */
546		for (mp = mentry->next; mp != NULL; mp = mp->next) {
547			if (mp->full && !mp->used
548			    && strcmp(mentry->name, mp->name) == 0) {
549				/* Later lines override earlier ones. */
550				mp->used = 1;
551				r1 = parse_line(a, entry, mtree, mp,
552				    &parsed_kws);
553				if (r1 < r)
554					r = r1;
555			}
556		}
557	} else {
558		/*
559		 * Relative entries require us to construct
560		 * the full path and possibly update the
561		 * current directory.
562		 */
563		size_t n = archive_strlen(&mtree->current_dir);
564		if (n > 0)
565			archive_strcat(&mtree->current_dir, "/");
566		archive_strcat(&mtree->current_dir, mentry->name);
567		archive_entry_copy_pathname(entry, mtree->current_dir.s);
568		if (archive_entry_filetype(entry) != AE_IFDIR)
569			mtree->current_dir.length = n;
570	}
571
572	/*
573	 * Try to open and stat the file to get the real size
574	 * and other file info.  It would be nice to avoid
575	 * this here so that getting a listing of an mtree
576	 * wouldn't require opening every referenced contents
577	 * file.  But then we wouldn't know the actual
578	 * contents size, so I don't see a really viable way
579	 * around this.  (Also, we may want to someday pull
580	 * other unspecified info from the contents file on
581	 * disk.)
582	 */
583	mtree->fd = -1;
584	if (archive_strlen(&mtree->contents_name) > 0)
585		path = mtree->contents_name.s;
586	else
587		path = archive_entry_pathname(entry);
588
589	if (archive_entry_filetype(entry) == AE_IFREG ||
590	    archive_entry_filetype(entry) == AE_IFDIR) {
591		mtree->fd = open(path, O_RDONLY | O_BINARY);
592		if (mtree->fd == -1 &&
593		    (errno != ENOENT ||
594		     archive_strlen(&mtree->contents_name) > 0)) {
595			archive_set_error(&a->archive, errno,
596			    "Can't open %s", path);
597			r = ARCHIVE_WARN;
598		}
599	}
600
601	st = &st_storage;
602	if (mtree->fd >= 0) {
603		if (fstat(mtree->fd, st) == -1) {
604			archive_set_error(&a->archive, errno,
605			    "Could not fstat %s", path);
606			r = ARCHIVE_WARN;
607			/* If we can't stat it, don't keep it open. */
608			close(mtree->fd);
609			mtree->fd = -1;
610			st = NULL;
611		}
612	} else if (lstat(path, st) == -1) {
613		st = NULL;
614	}
615
616	/*
617	 * Check for a mismatch between the type in the specification and
618	 * the type of the contents object on disk.
619	 */
620	if (st != NULL) {
621		mismatched_type = 0;
622		if ((st->st_mode & S_IFMT) == S_IFREG &&
623		    archive_entry_filetype(entry) != AE_IFREG)
624			mismatched_type = 1;
625		if ((st->st_mode & S_IFMT) == S_IFLNK &&
626		    archive_entry_filetype(entry) != AE_IFLNK)
627			mismatched_type = 1;
628		if ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
629		    archive_entry_filetype(entry) != AE_IFSOCK)
630			mismatched_type = 1;
631		if ((st->st_mode & S_IFMT) == S_IFCHR &&
632		    archive_entry_filetype(entry) != AE_IFCHR)
633			mismatched_type = 1;
634		if ((st->st_mode & S_IFMT) == S_IFBLK &&
635		    archive_entry_filetype(entry) != AE_IFBLK)
636			mismatched_type = 1;
637		if ((st->st_mode & S_IFMT) == S_IFDIR &&
638		    archive_entry_filetype(entry) != AE_IFDIR)
639			mismatched_type = 1;
640		if ((st->st_mode & S_IFMT) == S_IFIFO &&
641		    archive_entry_filetype(entry) != AE_IFIFO)
642			mismatched_type = 1;
643
644		if (mismatched_type) {
645			if ((parsed_kws & MTREE_HAS_OPTIONAL) == 0) {
646				archive_set_error(&a->archive,
647				    ARCHIVE_ERRNO_MISC,
648				    "mtree specification has different type for %s",
649				    archive_entry_pathname(entry));
650				r = ARCHIVE_WARN;
651			} else {
652				*use_next = 1;
653			}
654			/* Don't hold a non-regular file open. */
655			if (mtree->fd >= 0)
656				close(mtree->fd);
657			mtree->fd = -1;
658			st = NULL;
659			return r;
660		}
661	}
662
663	/*
664	 * If there is a contents file on disk, pick some of the metadata
665	 * from that file.  For most of these, we only set it from the contents
666	 * if it wasn't already parsed from the specification.
667	 */
668	if (st != NULL) {
669		if ((parsed_kws & MTREE_HAS_DEVICE) == 0 &&
670		    (archive_entry_filetype(entry) == AE_IFCHR ||
671		     archive_entry_filetype(entry) == AE_IFBLK))
672			archive_entry_set_rdev(entry, st->st_rdev);
673		if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0)
674			archive_entry_set_gid(entry, st->st_gid);
675		if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0)
676			archive_entry_set_uid(entry, st->st_uid);
677		if ((parsed_kws & MTREE_HAS_MTIME) == 0) {
678#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
679			archive_entry_set_mtime(entry, st->st_mtime,
680			    st->st_mtimespec.tv_nsec);
681#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
682			archive_entry_set_mtime(entry, st->st_mtime,
683			    st->st_mtim.tv_nsec);
684#elif HAVE_STRUCT_STAT_ST_MTIME_N
685			archive_entry_set_mtime(entry, st->st_mtime,
686			    st->st_mtime_n);
687#elif HAVE_STRUCT_STAT_ST_UMTIME
688			archive_entry_set_mtime(entry, st->st_mtime,
689			    st->st_umtime*1000);
690#elif HAVE_STRUCT_STAT_ST_MTIME_USEC
691			archive_entry_set_mtime(entry, st->st_mtime,
692			    st->st_mtime_usec*1000);
693#else
694			archive_entry_set_mtime(entry, st->st_mtime, 0);
695#endif
696		}
697		if ((parsed_kws & MTREE_HAS_NLINK) == 0)
698			archive_entry_set_nlink(entry, st->st_nlink);
699		if ((parsed_kws & MTREE_HAS_PERM) == 0)
700			archive_entry_set_perm(entry, st->st_mode);
701		if ((parsed_kws & MTREE_HAS_SIZE) == 0)
702			archive_entry_set_size(entry, st->st_size);
703		archive_entry_set_ino(entry, st->st_ino);
704		archive_entry_set_dev(entry, st->st_dev);
705
706		archive_entry_linkify(mtree->resolver, &entry, &sparse_entry);
707	} else if (parsed_kws & MTREE_HAS_OPTIONAL) {
708		/*
709		 * Couldn't open the entry, stat it or the on-disk type
710		 * didn't match.  If this entry is optional, just ignore it
711		 * and read the next header entry.
712		 */
713		*use_next = 1;
714		return ARCHIVE_OK;
715	}
716
717	mtree->cur_size = archive_entry_size(entry);
718	mtree->offset = 0;
719
720	return r;
721}
722
723/*
724 * Each line contains a sequence of keywords.
725 */
726static int
727parse_line(struct archive_read *a, struct archive_entry *entry,
728    struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
729{
730	struct mtree_option *iter;
731	int r = ARCHIVE_OK, r1;
732
733	for (iter = mp->options; iter != NULL; iter = iter->next) {
734		r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
735		if (r1 < r)
736			r = r1;
737	}
738	if ((*parsed_kws & MTREE_HAS_TYPE) == 0) {
739		archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
740		    "Missing type keyword in mtree specification");
741		return (ARCHIVE_WARN);
742	}
743	return (r);
744}
745
746/*
747 * Device entries have one of the following forms:
748 * raw dev_t
749 * format,major,minor[,subdevice]
750 *
751 * Just use major and minor, no translation etc is done
752 * between formats.
753 */
754static int
755parse_device(struct archive *a, struct archive_entry *entry, char *val)
756{
757	char *comma1, *comma2;
758
759	comma1 = strchr(val, ',');
760	if (comma1 == NULL) {
761		archive_entry_set_dev(entry, mtree_atol10(&val));
762		return (ARCHIVE_OK);
763	}
764	++comma1;
765	comma2 = strchr(comma1, ',');
766	if (comma2 == NULL) {
767		archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
768		    "Malformed device attribute");
769		return (ARCHIVE_WARN);
770	}
771	++comma2;
772	archive_entry_set_rdevmajor(entry, mtree_atol(&comma1));
773	archive_entry_set_rdevminor(entry, mtree_atol(&comma2));
774	return (ARCHIVE_OK);
775}
776
777/*
778 * Parse a single keyword and its value.
779 */
780static int
781parse_keyword(struct archive_read *a, struct mtree *mtree,
782    struct archive_entry *entry, struct mtree_option *option, int *parsed_kws)
783{
784	char *val, *key;
785
786	key = option->value;
787
788	if (*key == '\0')
789		return (ARCHIVE_OK);
790
791	if (strcmp(key, "optional") == 0) {
792		*parsed_kws |= MTREE_HAS_OPTIONAL;
793		return (ARCHIVE_OK);
794	}
795	if (strcmp(key, "ignore") == 0) {
796		/*
797		 * The mtree processing is not recursive, so
798		 * recursion will only happen for explicitly listed
799		 * entries.
800		 */
801		return (ARCHIVE_OK);
802	}
803
804	val = strchr(key, '=');
805	if (val == NULL) {
806		archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
807		    "Malformed attribute \"%s\" (%d)", key, key[0]);
808		return (ARCHIVE_WARN);
809	}
810
811	*val = '\0';
812	++val;
813
814	switch (key[0]) {
815	case 'c':
816		if (strcmp(key, "content") == 0
817		    || strcmp(key, "contents") == 0) {
818			parse_escapes(val, NULL);
819			archive_strcpy(&mtree->contents_name, val);
820			break;
821		}
822		if (strcmp(key, "cksum") == 0)
823			break;
824	case 'd':
825		if (strcmp(key, "device") == 0) {
826			*parsed_kws |= MTREE_HAS_DEVICE;
827			return parse_device(&a->archive, entry, val);
828		}
829	case 'f':
830		if (strcmp(key, "flags") == 0) {
831			*parsed_kws |= MTREE_HAS_FFLAGS;
832			archive_entry_copy_fflags_text(entry, val);
833			break;
834		}
835	case 'g':
836		if (strcmp(key, "gid") == 0) {
837			*parsed_kws |= MTREE_HAS_GID;
838			archive_entry_set_gid(entry, mtree_atol10(&val));
839			break;
840		}
841		if (strcmp(key, "gname") == 0) {
842			*parsed_kws |= MTREE_HAS_GNAME;
843			archive_entry_copy_gname(entry, val);
844			break;
845		}
846	case 'l':
847		if (strcmp(key, "link") == 0) {
848			archive_entry_copy_symlink(entry, val);
849			break;
850		}
851	case 'm':
852		if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0)
853			break;
854		if (strcmp(key, "mode") == 0) {
855			if (val[0] >= '0' && val[0] <= '9') {
856				*parsed_kws |= MTREE_HAS_PERM;
857				archive_entry_set_perm(entry,
858				    mtree_atol8(&val));
859			} else {
860				archive_set_error(&a->archive,
861				    ARCHIVE_ERRNO_FILE_FORMAT,
862				    "Symbolic mode \"%s\" unsupported", val);
863				return ARCHIVE_WARN;
864			}
865			break;
866		}
867	case 'n':
868		if (strcmp(key, "nlink") == 0) {
869			*parsed_kws |= MTREE_HAS_NLINK;
870			archive_entry_set_nlink(entry, mtree_atol10(&val));
871			break;
872		}
873	case 'r':
874		if (strcmp(key, "rmd160") == 0 ||
875		    strcmp(key, "rmd160digest") == 0)
876			break;
877	case 's':
878		if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0)
879			break;
880		if (strcmp(key, "sha256") == 0 ||
881		    strcmp(key, "sha256digest") == 0)
882			break;
883		if (strcmp(key, "sha384") == 0 ||
884		    strcmp(key, "sha384digest") == 0)
885			break;
886		if (strcmp(key, "sha512") == 0 ||
887		    strcmp(key, "sha512digest") == 0)
888			break;
889		if (strcmp(key, "size") == 0) {
890			archive_entry_set_size(entry, mtree_atol10(&val));
891			break;
892		}
893	case 't':
894		if (strcmp(key, "tags") == 0) {
895			/*
896			 * Comma delimited list of tags.
897			 * Ignore the tags for now, but the interface
898			 * should be extended to allow inclusion/exclusion.
899			 */
900			break;
901		}
902		if (strcmp(key, "time") == 0) {
903			time_t m;
904			long ns;
905
906			*parsed_kws |= MTREE_HAS_MTIME;
907			m = (time_t)mtree_atol10(&val);
908			if (*val == '.') {
909				++val;
910				ns = (long)mtree_atol10(&val);
911			} else
912				ns = 0;
913			archive_entry_set_mtime(entry, m, ns);
914			break;
915		}
916		if (strcmp(key, "type") == 0) {
917			*parsed_kws |= MTREE_HAS_TYPE;
918			switch (val[0]) {
919			case 'b':
920				if (strcmp(val, "block") == 0) {
921					mtree->filetype = AE_IFBLK;
922					break;
923				}
924			case 'c':
925				if (strcmp(val, "char") == 0) {
926					mtree->filetype = AE_IFCHR;
927					break;
928				}
929			case 'd':
930				if (strcmp(val, "dir") == 0) {
931					mtree->filetype = AE_IFDIR;
932					break;
933				}
934			case 'f':
935				if (strcmp(val, "fifo") == 0) {
936					mtree->filetype = AE_IFIFO;
937					break;
938				}
939				if (strcmp(val, "file") == 0) {
940					mtree->filetype = AE_IFREG;
941					break;
942				}
943			case 'l':
944				if (strcmp(val, "link") == 0) {
945					mtree->filetype = AE_IFLNK;
946					break;
947				}
948			default:
949				archive_set_error(&a->archive,
950				    ARCHIVE_ERRNO_FILE_FORMAT,
951				    "Unrecognized file type \"%s\"", val);
952				return (ARCHIVE_WARN);
953			}
954			archive_entry_set_filetype(entry, mtree->filetype);
955			break;
956		}
957	case 'u':
958		if (strcmp(key, "uid") == 0) {
959			*parsed_kws |= MTREE_HAS_UID;
960			archive_entry_set_uid(entry, mtree_atol10(&val));
961			break;
962		}
963		if (strcmp(key, "uname") == 0) {
964			*parsed_kws |= MTREE_HAS_UNAME;
965			archive_entry_copy_uname(entry, val);
966			break;
967		}
968	default:
969		archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
970		    "Unrecognized key %s=%s", key, val);
971		return (ARCHIVE_WARN);
972	}
973	return (ARCHIVE_OK);
974}
975
976static int
977read_data(struct archive_read *a, const void **buff, size_t *size, off_t *offset)
978{
979	size_t bytes_to_read;
980	ssize_t bytes_read;
981	struct mtree *mtree;
982
983	mtree = (struct mtree *)(a->format->data);
984	if (mtree->fd < 0) {
985		*buff = NULL;
986		*offset = 0;
987		*size = 0;
988		return (ARCHIVE_EOF);
989	}
990	if (mtree->buff == NULL) {
991		mtree->buffsize = 64 * 1024;
992		mtree->buff = malloc(mtree->buffsize);
993		if (mtree->buff == NULL) {
994			archive_set_error(&a->archive, ENOMEM,
995			    "Can't allocate memory");
996			return (ARCHIVE_FATAL);
997		}
998	}
999
1000	*buff = mtree->buff;
1001	*offset = mtree->offset;
1002	if ((off_t)mtree->buffsize > mtree->cur_size - mtree->offset)
1003		bytes_to_read = mtree->cur_size - mtree->offset;
1004	else
1005		bytes_to_read = mtree->buffsize;
1006	bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
1007	if (bytes_read < 0) {
1008		archive_set_error(&a->archive, errno, "Can't read");
1009		return (ARCHIVE_WARN);
1010	}
1011	if (bytes_read == 0) {
1012		*size = 0;
1013		return (ARCHIVE_EOF);
1014	}
1015	mtree->offset += bytes_read;
1016	*size = bytes_read;
1017	return (ARCHIVE_OK);
1018}
1019
1020/* Skip does nothing except possibly close the contents file. */
1021static int
1022skip(struct archive_read *a)
1023{
1024	struct mtree *mtree;
1025
1026	mtree = (struct mtree *)(a->format->data);
1027	if (mtree->fd >= 0) {
1028		close(mtree->fd);
1029		mtree->fd = -1;
1030	}
1031	return (ARCHIVE_OK);
1032}
1033
1034/*
1035 * Since parsing backslash sequences always makes strings shorter,
1036 * we can always do this conversion in-place.
1037 */
1038static void
1039parse_escapes(char *src, struct mtree_entry *mentry)
1040{
1041	char *dest = src;
1042	char c;
1043
1044	if (mentry != NULL && strcmp(src, ".") == 0)
1045		mentry->full = 1;
1046
1047	while (*src != '\0') {
1048		c = *src++;
1049		if (c == '/' && mentry != NULL)
1050			mentry->full = 1;
1051		if (c == '\\') {
1052			switch (src[0]) {
1053			case '0':
1054				if (src[1] < '0' || src[1] > '7') {
1055					c = 0;
1056					++src;
1057					break;
1058				}
1059				/* FALLTHROUGH */
1060			case '1':
1061			case '2':
1062			case '3':
1063				if (src[1] >= '0' && src[1] <= '7' &&
1064				    src[2] >= '0' && src[2] <= '7') {
1065					c = (src[0] - '0') << 6;
1066					c |= (src[1] - '0') << 3;
1067					c |= (src[2] - '0');
1068					src += 3;
1069				}
1070				break;
1071			case 'a':
1072				c = '\a';
1073				++src;
1074				break;
1075			case 'b':
1076				c = '\b';
1077				++src;
1078				break;
1079			case 'f':
1080				c = '\f';
1081				++src;
1082				break;
1083			case 'n':
1084				c = '\n';
1085				++src;
1086				break;
1087			case 'r':
1088				c = '\r';
1089				++src;
1090				break;
1091			case 's':
1092				c = ' ';
1093				++src;
1094				break;
1095			case 't':
1096				c = '\t';
1097				++src;
1098				break;
1099			case 'v':
1100				c = '\v';
1101				++src;
1102				break;
1103			}
1104		}
1105		*dest++ = c;
1106	}
1107	*dest = '\0';
1108}
1109
1110/*
1111 * Note that this implementation does not (and should not!) obey
1112 * locale settings; you cannot simply substitute strtol here, since
1113 * it does obey locale.
1114 */
1115static int64_t
1116mtree_atol8(char **p)
1117{
1118	int64_t	l, limit, last_digit_limit;
1119	int digit, base;
1120
1121	base = 8;
1122	limit = INT64_MAX / base;
1123	last_digit_limit = INT64_MAX % base;
1124
1125	l = 0;
1126	digit = **p - '0';
1127	while (digit >= 0 && digit < base) {
1128		if (l>limit || (l == limit && digit > last_digit_limit)) {
1129			l = INT64_MAX; /* Truncate on overflow. */
1130			break;
1131		}
1132		l = (l * base) + digit;
1133		digit = *++(*p) - '0';
1134	}
1135	return (l);
1136}
1137
1138/*
1139 * Note that this implementation does not (and should not!) obey
1140 * locale settings; you cannot simply substitute strtol here, since
1141 * it does obey locale.
1142 */
1143static int64_t
1144mtree_atol10(char **p)
1145{
1146	int64_t l, limit, last_digit_limit;
1147	int base, digit, sign;
1148
1149	base = 10;
1150	limit = INT64_MAX / base;
1151	last_digit_limit = INT64_MAX % base;
1152
1153	if (**p == '-') {
1154		sign = -1;
1155		++(*p);
1156	} else
1157		sign = 1;
1158
1159	l = 0;
1160	digit = **p - '0';
1161	while (digit >= 0 && digit < base) {
1162		if (l > limit || (l == limit && digit > last_digit_limit)) {
1163			l = INT64_MAX; /* Truncate on overflow. */
1164			break;
1165		}
1166		l = (l * base) + digit;
1167		digit = *++(*p) - '0';
1168	}
1169	return (sign < 0) ? -l : l;
1170}
1171
1172/*
1173 * Note that this implementation does not (and should not!) obey
1174 * locale settings; you cannot simply substitute strtol here, since
1175 * it does obey locale.
1176 */
1177static int64_t
1178mtree_atol16(char **p)
1179{
1180	int64_t l, limit, last_digit_limit;
1181	int base, digit, sign;
1182
1183	base = 16;
1184	limit = INT64_MAX / base;
1185	last_digit_limit = INT64_MAX % base;
1186
1187	if (**p == '-') {
1188		sign = -1;
1189		++(*p);
1190	} else
1191		sign = 1;
1192
1193	l = 0;
1194	if (**p >= '0' && **p <= '9')
1195		digit = **p - '0';
1196	else if (**p >= 'a' && **p <= 'f')
1197		digit = **p - 'a' + 10;
1198	else if (**p >= 'A' && **p <= 'F')
1199		digit = **p - 'A' + 10;
1200	else
1201		digit = -1;
1202	while (digit >= 0 && digit < base) {
1203		if (l > limit || (l == limit && digit > last_digit_limit)) {
1204			l = INT64_MAX; /* Truncate on overflow. */
1205			break;
1206		}
1207		l = (l * base) + digit;
1208		if (**p >= '0' && **p <= '9')
1209			digit = **p - '0';
1210		else if (**p >= 'a' && **p <= 'f')
1211			digit = **p - 'a' + 10;
1212		else if (**p >= 'A' && **p <= 'F')
1213			digit = **p - 'A' + 10;
1214		else
1215			digit = -1;
1216	}
1217	return (sign < 0) ? -l : l;
1218}
1219
1220static int64_t
1221mtree_atol(char **p)
1222{
1223	if (**p != '0')
1224		return mtree_atol10(p);
1225	if ((*p)[1] == 'x' || (*p)[1] == 'X') {
1226		*p += 2;
1227		return mtree_atol16(p);
1228	}
1229	return mtree_atol8(p);
1230}
1231
1232/*
1233 * Returns length of line (including trailing newline)
1234 * or negative on error.  'start' argument is updated to
1235 * point to first character of line.
1236 */
1237static ssize_t
1238readline(struct archive_read *a, struct mtree *mtree, char **start, ssize_t limit)
1239{
1240	ssize_t bytes_read;
1241	ssize_t total_size = 0;
1242	ssize_t find_off = 0;
1243	const void *t;
1244	const char *s;
1245	void *p;
1246	char *u;
1247
1248	/* Accumulate line in a line buffer. */
1249	for (;;) {
1250		/* Read some more. */
1251		t = __archive_read_ahead(a, 1, &bytes_read);
1252		if (t == NULL)
1253			return (0);
1254		if (bytes_read < 0)
1255			return (ARCHIVE_FATAL);
1256		s = t;  /* Start of line? */
1257		p = memchr(t, '\n', bytes_read);
1258		/* If we found '\n', trim the read. */
1259		if (p != NULL) {
1260			bytes_read = 1 + ((const char *)p) - s;
1261		}
1262		if (total_size + bytes_read + 1 > limit) {
1263			archive_set_error(&a->archive,
1264			    ARCHIVE_ERRNO_FILE_FORMAT,
1265			    "Line too long");
1266			return (ARCHIVE_FATAL);
1267		}
1268		if (archive_string_ensure(&mtree->line,
1269			total_size + bytes_read + 1) == NULL) {
1270			archive_set_error(&a->archive, ENOMEM,
1271			    "Can't allocate working buffer");
1272			return (ARCHIVE_FATAL);
1273		}
1274		memcpy(mtree->line.s + total_size, t, bytes_read);
1275		__archive_read_consume(a, bytes_read);
1276		total_size += bytes_read;
1277		/* Null terminate. */
1278		mtree->line.s[total_size] = '\0';
1279		/* If we found an unescaped '\n', clean up and return. */
1280		for (u = mtree->line.s + find_off; *u; ++u) {
1281			if (u[0] == '\n') {
1282				*start = mtree->line.s;
1283				return total_size;
1284			}
1285			if (u[0] == '#') {
1286				if (p == NULL)
1287					break;
1288				*start = mtree->line.s;
1289				return total_size;
1290			}
1291			if (u[0] != '\\')
1292				continue;
1293			if (u[1] == '\\') {
1294				++u;
1295				continue;
1296			}
1297			if (u[1] == '\n') {
1298				memmove(u, u + 1,
1299				    total_size - (u - mtree->line.s) + 1);
1300				--total_size;
1301				++u;
1302				break;
1303			}
1304			if (u[1] == '\0')
1305				break;
1306		}
1307		find_off = u - mtree->line.s;
1308	}
1309}
1310