1 /*-
2 * Copyright (c) 2003-2007 Tim Kientzle
3 * Copyright (c) 2008 Joerg Sonnenberger
4 * Copyright (c) 2011-2012 Michihiro NAKAJIMA
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 #include "archive_platform.h"
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 #ifdef HAVE_CTYPE_H
48 #include <ctype.h>
49 #endif
50
51 #include "archive.h"
52 #include "archive_entry.h"
53 #include "archive_entry_private.h"
54 #include "archive_integer.h"
55 #include "archive_platform_stat.h"
56 #include "archive_private.h"
57 #include "archive_rb.h"
58 #include "archive_read_private.h"
59 #include "archive_string.h"
60 #include "archive_pack_dev.h"
61
62 #ifndef O_BINARY
63 #define O_BINARY 0
64 #endif
65 #ifndef O_CLOEXEC
66 #define O_CLOEXEC 0
67 #endif
68
69 #define MTREE_HAS_DEVICE 0x0001
70 #define MTREE_HAS_FFLAGS 0x0002
71 #define MTREE_HAS_GID 0x0004
72 #define MTREE_HAS_GNAME 0x0008
73 #define MTREE_HAS_MTIME 0x0010
74 #define MTREE_HAS_NLINK 0x0020
75 #define MTREE_HAS_PERM 0x0040
76 #define MTREE_HAS_SIZE 0x0080
77 #define MTREE_HAS_TYPE 0x0100
78 #define MTREE_HAS_UID 0x0200
79 #define MTREE_HAS_UNAME 0x0400
80
81 #define MTREE_HAS_OPTIONAL 0x0800
82 #define MTREE_HAS_NOCHANGE 0x1000 /* FreeBSD specific */
83
84 #define MAX_LINE_LEN (1024 * 1024)
85
86 struct mtree_option {
87 struct mtree_option *next;
88 char *value;
89 };
90
91 struct mtree_entry {
92 struct archive_rb_node rbnode;
93 struct mtree_entry *next_dup;
94 struct mtree_entry *next;
95 struct mtree_option *options;
96 char *name;
97 char full;
98 char used;
99 };
100
101 struct mtree {
102 struct archive_string line;
103 size_t buffsize;
104 char *buff;
105 int64_t offset;
106 int fd;
107 int archive_format;
108 const char *archive_format_name;
109 struct mtree_entry *entries;
110 struct mtree_entry *this_entry;
111 struct archive_rb_tree entry_rbtree;
112 struct archive_string current_dir;
113 struct archive_string contents_name;
114
115 struct archive_entry_linkresolver *resolver;
116 struct archive_rb_tree rbtree;
117
118 int64_t cur_size;
119 char checkfs;
120 };
121
122 static int bid_keycmp(const char *, const char *, ssize_t);
123 static int cleanup(struct archive_read *);
124 static int detect_form(struct archive_read *, int *);
125 static int mtree_bid(struct archive_read *, int);
126 static int parse_file(struct archive_read *, struct archive_entry *,
127 struct mtree *, struct mtree_entry *, int *);
128 static void parse_escapes(char *, struct mtree_entry *);
129 static int parse_line(struct archive_read *, struct archive_entry *,
130 struct mtree *, struct mtree_entry *, int *);
131 static int parse_keyword(struct archive_read *, struct mtree *,
132 struct archive_entry *, struct mtree_option *, int *);
133 static int read_data(struct archive_read *a,
134 const void **buff, size_t *size, int64_t *offset);
135 static ssize_t readline(struct archive_read *, struct mtree *, char **, ssize_t);
136 static int skip(struct archive_read *a);
137 static int read_header(struct archive_read *,
138 struct archive_entry *);
139 static int64_t mtree_atol(char **, int base);
140 #ifndef HAVE_STRNLEN
141 static size_t mtree_strnlen(const char *, size_t);
142 #endif
143
144 #ifdef HAVE_STRNLEN
145 #define mtree_strnlen(a,b) strnlen(a,b)
146 #else
147 static size_t
mtree_strnlen(const char * p,size_t maxlen)148 mtree_strnlen(const char *p, size_t maxlen)
149 {
150 size_t i;
151
152 for (i = 0; i < maxlen; i++) {
153 if (p[i] == 0)
154 break;
155 }
156 return (i);
157 }
158 #endif
159
160 static int
archive_read_format_mtree_options(struct archive_read * a,const char * key,const char * val)161 archive_read_format_mtree_options(struct archive_read *a,
162 const char *key, const char *val)
163 {
164 struct mtree *mtree = a->format->data;
165
166 if (strcmp(key, "checkfs") == 0) {
167 /* Allows to read information missing from the mtree from the file system */
168 if (val == NULL || val[0] == 0) {
169 mtree->checkfs = 0;
170 } else {
171 mtree->checkfs = 1;
172 }
173 return (ARCHIVE_OK);
174 }
175
176 /* Note: The "warn" return is just to inform the options
177 * supervisor that we didn't handle it. It will generate
178 * a suitable error if no one used this option. */
179 return (ARCHIVE_WARN);
180 }
181
182 static void
free_options(struct mtree_option * head)183 free_options(struct mtree_option *head)
184 {
185 struct mtree_option *next;
186
187 for (; head != NULL; head = next) {
188 next = head->next;
189 free(head->value);
190 free(head);
191 }
192 }
193
194 static int
mtree_cmp_node(const struct archive_rb_node * n1,const struct archive_rb_node * n2)195 mtree_cmp_node(const struct archive_rb_node *n1,
196 const struct archive_rb_node *n2)
197 {
198 const struct mtree_entry *e1 = (const struct mtree_entry *)n1;
199 const struct mtree_entry *e2 = (const struct mtree_entry *)n2;
200
201 return (strcmp(e1->name, e2->name));
202 }
203
204 static int
mtree_cmp_key(const struct archive_rb_node * n,const void * key)205 mtree_cmp_key(const struct archive_rb_node *n, const void *key)
206 {
207 const struct mtree_entry *e = (const struct mtree_entry *)n;
208
209 return (strcmp(e->name, key));
210 }
211
212 int
archive_read_support_format_mtree(struct archive * _a)213 archive_read_support_format_mtree(struct archive *_a)
214 {
215 static const struct archive_rb_tree_ops rb_ops = {
216 mtree_cmp_node, mtree_cmp_key,
217 };
218 struct archive_read *a = (struct archive_read *)_a;
219 struct mtree *mtree;
220 int r;
221
222 archive_check_magic(_a, ARCHIVE_READ_MAGIC,
223 ARCHIVE_STATE_NEW, "archive_read_support_format_mtree");
224
225 mtree = calloc(1, sizeof(*mtree));
226 if (mtree == NULL) {
227 archive_set_error(&a->archive, ENOMEM,
228 "Can't allocate mtree data");
229 return (ARCHIVE_FATAL);
230 }
231 mtree->checkfs = 0;
232 mtree->fd = -1;
233
234 __archive_rb_tree_init(&mtree->rbtree, &rb_ops);
235
236 r = __archive_read_register_format(a, mtree, "mtree",
237 mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL);
238
239 if (r != ARCHIVE_OK)
240 free(mtree);
241 return (ARCHIVE_OK);
242 }
243
244 static int
cleanup(struct archive_read * a)245 cleanup(struct archive_read *a)
246 {
247 struct mtree *mtree = a->format->data;
248 struct mtree_entry *p, *q;
249
250 /* Close any dangling file descriptor before freeing */
251 if (mtree->fd >= 0) {
252 close(mtree->fd);
253 mtree->fd = -1;
254 }
255 p = mtree->entries;
256 while (p != NULL) {
257 q = p->next;
258 free(p->name);
259 free_options(p->options);
260 free(p);
261 p = q;
262 }
263 archive_string_free(&mtree->line);
264 archive_string_free(&mtree->current_dir);
265 archive_string_free(&mtree->contents_name);
266 archive_entry_linkresolver_free(mtree->resolver);
267
268 free(mtree->buff);
269 free(mtree);
270 a->format->data = NULL;
271 return (ARCHIVE_OK);
272 }
273
274 static ssize_t
get_line_size(const char * b,ssize_t avail,ssize_t * nlsize)275 get_line_size(const char *b, ssize_t avail, ssize_t *nlsize)
276 {
277 ssize_t len;
278
279 len = 0;
280 while (len < avail) {
281 switch (*b) {
282 case '\0':/* Non-ascii character or control character. */
283 if (nlsize != NULL)
284 *nlsize = 0;
285 return (-1);
286 case '\r':
287 if (avail-len > 1 && b[1] == '\n') {
288 if (nlsize != NULL)
289 *nlsize = 2;
290 return (len+2);
291 }
292 /* FALL THROUGH */
293 case '\n':
294 if (nlsize != NULL)
295 *nlsize = 1;
296 return (len+1);
297 default:
298 b++;
299 len++;
300 break;
301 }
302 }
303 if (nlsize != NULL)
304 *nlsize = 0;
305 return (avail);
306 }
307
308 /*
309 * <---------------- ravail --------------------->
310 * <-- diff ------> <--- avail ----------------->
311 * <---- len ----------->
312 * | Previous lines | line being parsed nl extra |
313 * ^
314 * b
315 *
316 */
317 static ssize_t
next_line(struct archive_read * a,const char ** b,ssize_t * avail,ssize_t * ravail,ssize_t * nl)318 next_line(struct archive_read *a,
319 const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl)
320 {
321 ssize_t len;
322 int quit;
323
324 quit = 0;
325 if (*avail == 0) {
326 *nl = 0;
327 len = 0;
328 } else
329 len = get_line_size(*b, *avail, nl);
330 /*
331 * Read bytes more while it does not reach the end of line.
332 */
333 while (*nl == 0 && len == *avail && !quit) {
334 ssize_t diff = *ravail - *avail;
335 size_t nbytes_req = ((size_t)*ravail + 1023) & ~1023U;
336 ssize_t tested;
337
338 /*
339 * Place an arbitrary limit on the line length.
340 * mtree is almost free-form input and without line length limits,
341 * it can consume a lot of memory.
342 */
343 if (len >= MAX_LINE_LEN)
344 return (-1);
345
346 /* Increase reading bytes if it is not enough for at least
347 * two new lines. */
348 if (nbytes_req < (size_t)*ravail + 160)
349 nbytes_req <<= 1;
350
351 *b = __archive_read_ahead(a, nbytes_req, avail);
352 if (*b == NULL) {
353 if (*ravail >= *avail)
354 return (0);
355 /* Reading bytes reaches the end of file. */
356 *b = __archive_read_ahead(a, *avail, avail);
357 quit = 1;
358 }
359 *ravail = *avail;
360 *b += diff;
361 *avail -= diff;
362 tested = len;/* Skip some bytes we already determined. */
363 len = get_line_size(*b + len, *avail - len, nl);
364 if (len >= 0)
365 len += tested;
366 }
367 return (len);
368 }
369
370 /*
371 * Compare characters with an mtree keyword.
372 * Returns the length of an mtree keyword if matched.
373 * Returns 0 if not matched.
374 */
375 static int
bid_keycmp(const char * p,const char * key,ssize_t len)376 bid_keycmp(const char *p, const char *key, ssize_t len)
377 {
378 int match_len = 0;
379
380 while (len > 0 && *p && *key) {
381 if (*p == *key) {
382 --len;
383 ++p;
384 ++key;
385 ++match_len;
386 continue;
387 }
388 return (0);/* Not match */
389 }
390 if (*key != '\0')
391 return (0);/* Not match */
392
393 /* A following character should be specified characters */
394 if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' ||
395 p[0] == '\n' || p[0] == '\r' ||
396 (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')))
397 return (match_len);
398 return (0);/* Not match */
399 }
400
401 /*
402 * Test whether the characters 'p' has is mtree keyword.
403 * Returns the length of a detected keyword.
404 * Returns 0 if any keywords were not found.
405 */
406 static int
bid_keyword(const char * p,ssize_t len)407 bid_keyword(const char *p, ssize_t len)
408 {
409 static const char * const keys_c[] = {
410 "content", "contents", "cksum", NULL
411 };
412 static const char * const keys_df[] = {
413 "device", "flags", NULL
414 };
415 static const char * const keys_g[] = {
416 "gid", "gname", NULL
417 };
418 static const char * const keys_il[] = {
419 "ignore", "inode", "link", NULL
420 };
421 static const char * const keys_m[] = {
422 "md5", "md5digest", "mode", NULL
423 };
424 static const char * const keys_no[] = {
425 "nlink", "nochange", "optional", NULL
426 };
427 static const char * const keys_r[] = {
428 "resdevice", "rmd160", "rmd160digest", NULL
429 };
430 static const char * const keys_s[] = {
431 "sha1", "sha1digest",
432 "sha256", "sha256digest",
433 "sha384", "sha384digest",
434 "sha512", "sha512digest",
435 "size", NULL
436 };
437 static const char * const keys_t[] = {
438 "tags", "time", "type", NULL
439 };
440 static const char * const keys_u[] = {
441 "uid", "uname", NULL
442 };
443 const char * const *keys;
444 int i;
445
446 switch (*p) {
447 case 'c': keys = keys_c; break;
448 case 'd': case 'f': keys = keys_df; break;
449 case 'g': keys = keys_g; break;
450 case 'i': case 'l': keys = keys_il; break;
451 case 'm': keys = keys_m; break;
452 case 'n': case 'o': keys = keys_no; break;
453 case 'r': keys = keys_r; break;
454 case 's': keys = keys_s; break;
455 case 't': keys = keys_t; break;
456 case 'u': keys = keys_u; break;
457 default: return (0);/* Unknown key */
458 }
459
460 for (i = 0; keys[i] != NULL; i++) {
461 int l = bid_keycmp(p, keys[i], len);
462 if (l > 0)
463 return (l);
464 }
465 return (0);/* Unknown key */
466 }
467
468 /*
469 * Test whether there is a set of mtree keywords.
470 * Returns the number of keywords.
471 * Returns -1 if we got incorrect sequence.
472 * This function expects a set of "<space characters>keyword=value".
473 * When "unset" is specified, expects a set of "<space characters>keyword".
474 */
475 static int
bid_keyword_list(const char * p,ssize_t len,int unset,int last_is_path)476 bid_keyword_list(const char *p, ssize_t len, int unset, int last_is_path)
477 {
478 int l;
479 int keycnt = 0;
480
481 while (len > 0 && *p) {
482 int blank = 0;
483
484 /* Test whether there are blank characters in the line. */
485 while (len >0 && (*p == ' ' || *p == '\t')) {
486 ++p;
487 --len;
488 blank = 1;
489 }
490 if (*p == '\n' || *p == '\r')
491 break;
492 if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))
493 break;
494 if (!blank && !last_is_path) /* No blank character. */
495 return (-1);
496 if (last_is_path && len == 0)
497 return (keycnt);
498
499 if (unset) {
500 l = bid_keycmp(p, "all", len);
501 if (l > 0)
502 return (1);
503 }
504 /* Test whether there is a correct key in the line. */
505 l = bid_keyword(p, len);
506 if (l == 0)
507 return (-1);/* Unknown keyword was found. */
508 p += l;
509 len -= l;
510 keycnt++;
511
512 /* Skip value */
513 if (*p == '=') {
514 int value = 0;
515 ++p;
516 --len;
517 while (len > 0 && *p != ' ' && *p != '\t') {
518 ++p;
519 --len;
520 value = 1;
521 }
522 /* A keyword should have a value unless this is
523 * an "/unset" operation. */
524 if (!unset && value == 0)
525 return (-1);
526 }
527 }
528 return (keycnt);
529 }
530
531 static int
bid_entry(const char * p,ssize_t len,ssize_t nl,int * last_is_path)532 bid_entry(const char *p, ssize_t len, ssize_t nl, int *last_is_path)
533 {
534 int f = 0;
535 static const unsigned char safe_char[256] = {
536 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */
537 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */
538 /* !"$%&'()*+,-./ EXCLUSION:( )(#) */
539 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */
540 /* 0123456789:;<>? EXCLUSION:(=) */
541 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */
542 /* @ABCDEFGHIJKLMNO */
543 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */
544 /* PQRSTUVWXYZ[\]^_ */
545 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */
546 /* `abcdefghijklmno */
547 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */
548 /* pqrstuvwxyz{|}~ */
549 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */
550 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */
551 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */
552 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */
553 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */
554 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */
555 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */
556 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */
557 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */
558 };
559 ssize_t ll;
560 const char *pp = p;
561 const char * const pp_end = pp + len;
562
563 *last_is_path = 0;
564 /*
565 * Skip the path-name which is quoted.
566 */
567 for (;pp < pp_end; ++pp) {
568 if (!safe_char[*(const unsigned char *)pp]) {
569 if (*pp != ' ' && *pp != '\t' && *pp != '\r'
570 && *pp != '\n')
571 f = 0;
572 break;
573 }
574 f = 1;
575 }
576 ll = pp_end - pp;
577
578 /* If a path-name was not found at the first, try to check
579 * a mtree format(a.k.a form D) ``NetBSD's mtree -D'' creates,
580 * which places the path-name at the last. */
581 if (f == 0) {
582 const char *pb = p + len - nl;
583 int name_len = 0;
584 int slash;
585
586 /* The form D accepts only a single line for an entry. */
587 if (pb-2 >= p &&
588 pb[-1] == '\\' && (pb[-2] == ' ' || pb[-2] == '\t'))
589 return (-1);
590 if (pb-1 >= p && pb[-1] == '\\')
591 return (-1);
592
593 slash = 0;
594 while (p <= --pb && *pb != ' ' && *pb != '\t') {
595 if (!safe_char[*(const unsigned char *)pb])
596 return (-1);
597 name_len++;
598 /* The pathname should have a slash in this
599 * format. */
600 if (*pb == '/')
601 slash = 1;
602 }
603 if (name_len == 0 || slash == 0)
604 return (-1);
605 /* If '/' is placed at the first in this field, this is not
606 * a valid filename. */
607 if (pb[1] == '/')
608 return (-1);
609 ll = len - nl - name_len;
610 pp = p;
611 *last_is_path = 1;
612 }
613
614 return (bid_keyword_list(pp, ll, 0, *last_is_path));
615 }
616
617 #define MAX_BID_ENTRY 3
618
619 static int
mtree_bid(struct archive_read * a,int best_bid)620 mtree_bid(struct archive_read *a, int best_bid)
621 {
622 const char *signature = "#mtree";
623 const char *p;
624
625 (void)best_bid; /* UNUSED */
626
627 /* Now let's look at the actual header and see if it matches. */
628 p = __archive_read_ahead(a, strlen(signature), NULL);
629 if (p == NULL)
630 return (-1);
631
632 if (memcmp(p, signature, strlen(signature)) == 0)
633 return (8 * (int)strlen(signature));
634
635 /*
636 * There is not a mtree signature. Let's try to detect mtree format.
637 */
638 return (detect_form(a, NULL));
639 }
640
641 static int
detect_form(struct archive_read * a,int * is_form_d)642 detect_form(struct archive_read *a, int *is_form_d)
643 {
644 const char *p;
645 ssize_t avail, ravail;
646 ssize_t len, nl;
647 int entry_cnt = 0, multiline = 0;
648 int form_D = 0;/* The archive is generated by `NetBSD mtree -D'
649 * (In this source we call it `form D') . */
650
651 if (is_form_d != NULL)
652 *is_form_d = 0;
653 p = __archive_read_ahead(a, 1, &avail);
654 if (p == NULL)
655 return (-1);
656 ravail = avail;
657 for (;;) {
658 len = next_line(a, &p, &avail, &ravail, &nl);
659 /* The terminal character of the line should be
660 * a new line character, '\r\n' or '\n'. */
661 if (len <= 0 || nl == 0)
662 break;
663 if (!multiline) {
664 /* Leading whitespace is never significant,
665 * ignore it. */
666 while (len > 0 && (*p == ' ' || *p == '\t')) {
667 ++p;
668 --avail;
669 --len;
670 }
671 /* Skip comment or empty line. */
672 if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') {
673 p += len;
674 avail -= len;
675 continue;
676 }
677 } else {
678 /* A continuance line; the terminal
679 * character of previous line was '\' character. */
680 if (bid_keyword_list(p, len, 0, 0) <= 0)
681 break;
682 if (p[len-nl-1] != '\\') {
683 if (multiline == 1 &&
684 ++entry_cnt >= MAX_BID_ENTRY)
685 break;
686 multiline = 0;
687 }
688 p += len;
689 avail -= len;
690 continue;
691 }
692 if (p[0] != '/') {
693 int last_is_path, keywords;
694
695 keywords = bid_entry(p, len, nl, &last_is_path);
696 if (keywords >= 0) {
697 if (form_D == 0) {
698 if (last_is_path)
699 form_D = 1;
700 else if (keywords > 0)
701 /* This line is not `form D'. */
702 form_D = -1;
703 } else if (form_D == 1) {
704 if (!last_is_path && keywords > 0)
705 /* This this is not `form D'
706 * and we cannot accept mixed
707 * format. */
708 break;
709 }
710 if (!last_is_path && p[len-nl-1] == '\\')
711 /* This line continues. */
712 multiline = 1;
713 else {
714 /* We've got plenty of correct lines
715 * to assume that this file is an mtree
716 * format. */
717 if (++entry_cnt >= MAX_BID_ENTRY)
718 break;
719 }
720 } else
721 break;
722 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
723 if (bid_keyword_list(p+4, len-4, 0, 0) <= 0)
724 break;
725 /* This line continues. */
726 if (p[len-nl-1] == '\\')
727 multiline = 2;
728 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
729 if (bid_keyword_list(p+6, len-6, 1, 0) <= 0)
730 break;
731 /* This line continues. */
732 if (p[len-nl-1] == '\\')
733 multiline = 2;
734 } else
735 break;
736
737 /* Test next line. */
738 p += len;
739 avail -= len;
740 }
741 if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) {
742 if (is_form_d != NULL) {
743 if (form_D == 1)
744 *is_form_d = 1;
745 }
746 return (32);
747 }
748
749 return (0);
750 }
751
752 /*
753 * The extended mtree format permits multiple lines specifying
754 * attributes for each file. For those entries, only the last line
755 * is actually used. Practically speaking, that means we have
756 * to read the entire mtree file into memory up front.
757 *
758 * The parsing is done in two steps. First, it is decided if a line
759 * changes the global defaults and if it does, it is processed accordingly.
760 * Otherwise, the options of the line are merged with the current
761 * global options.
762 */
763 static int
add_option(struct archive_read * a,struct mtree_option ** global,const char * value,size_t len)764 add_option(struct archive_read *a, struct mtree_option **global,
765 const char *value, size_t len)
766 {
767 struct mtree_option *opt;
768
769 if ((opt = malloc(sizeof(*opt))) == NULL) {
770 archive_set_error(&a->archive, errno, "Can't allocate memory");
771 return (ARCHIVE_FATAL);
772 }
773 if ((opt->value = malloc(len + 1)) == NULL) {
774 free(opt);
775 archive_set_error(&a->archive, errno, "Can't allocate memory");
776 return (ARCHIVE_FATAL);
777 }
778 memcpy(opt->value, value, len);
779 opt->value[len] = '\0';
780 opt->next = *global;
781 *global = opt;
782 return (ARCHIVE_OK);
783 }
784
785 static void
remove_option(struct mtree_option ** global,const char * value,size_t len)786 remove_option(struct mtree_option **global, const char *value, size_t len)
787 {
788 struct mtree_option *iter, *last;
789
790 last = NULL;
791 for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
792 if (strncmp(iter->value, value, len) == 0 &&
793 (iter->value[len] == '\0' ||
794 iter->value[len] == '='))
795 break;
796 }
797 if (iter == NULL)
798 return;
799 if (last == NULL)
800 *global = iter->next;
801 else
802 last->next = iter->next;
803
804 free(iter->value);
805 free(iter);
806 }
807
808 static int
process_global_set(struct archive_read * a,struct mtree_option ** global,const char * line)809 process_global_set(struct archive_read *a,
810 struct mtree_option **global, const char *line)
811 {
812 const char *next, *eq;
813 size_t len;
814 int r;
815
816 line += 4;
817 for (;;) {
818 next = line + strspn(line, " \t\r\n");
819 if (*next == '\0')
820 return (ARCHIVE_OK);
821 line = next;
822 next = line + strcspn(line, " \t\r\n");
823 eq = strchr(line, '=');
824 if (eq == NULL || eq > next)
825 len = next - line;
826 else
827 len = eq - line;
828
829 remove_option(global, line, len);
830 r = add_option(a, global, line, next - line);
831 if (r != ARCHIVE_OK)
832 return (r);
833 line = next;
834 }
835 }
836
837 static int
process_global_unset(struct archive_read * a,struct mtree_option ** global,const char * line)838 process_global_unset(struct archive_read *a,
839 struct mtree_option **global, const char *line)
840 {
841 const char *next;
842 size_t len;
843
844 line += 6;
845 if (strchr(line, '=') != NULL) {
846 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
847 "/unset shall not contain `='");
848 return ARCHIVE_FATAL;
849 }
850
851 for (;;) {
852 next = line + strspn(line, " \t\r\n");
853 if (*next == '\0')
854 return (ARCHIVE_OK);
855 line = next;
856 len = strcspn(line, " \t\r\n");
857
858 if (len == 3 && strncmp(line, "all", 3) == 0) {
859 free_options(*global);
860 *global = NULL;
861 } else {
862 remove_option(global, line, len);
863 }
864
865 line += len;
866 }
867 }
868
869 static int
process_add_entry(struct archive_read * a,struct mtree * mtree,struct mtree_option ** global,const char * line,ssize_t line_len,struct mtree_entry ** last_entry,int is_form_d)870 process_add_entry(struct archive_read *a, struct mtree *mtree,
871 struct mtree_option **global, const char *line, ssize_t line_len,
872 struct mtree_entry **last_entry, int is_form_d)
873 {
874 struct mtree_entry *entry;
875 struct mtree_option *iter;
876 const char *next, *eq, *name, *end;
877 size_t name_len, len;
878 int r, i;
879
880 if ((entry = malloc(sizeof(*entry))) == NULL) {
881 archive_set_error(&a->archive, errno, "Can't allocate memory");
882 return (ARCHIVE_FATAL);
883 }
884 entry->next = NULL;
885 entry->options = NULL;
886 entry->name = NULL;
887 entry->used = 0;
888 entry->full = 0;
889
890 /* Add this entry to list. */
891 if (*last_entry == NULL)
892 mtree->entries = entry;
893 else
894 (*last_entry)->next = entry;
895 *last_entry = entry;
896
897 if (is_form_d) {
898 /* Filename is last item on line. */
899 /* Adjust line_len to trim trailing whitespace */
900 while (line_len > 0) {
901 char last_character = line[line_len - 1];
902 if (last_character == '\r'
903 || last_character == '\n'
904 || last_character == '\t'
905 || last_character == ' ') {
906 line_len--;
907 } else {
908 break;
909 }
910 }
911 /* Name starts after the last whitespace separator */
912 name = line;
913 for (i = 0; i < line_len; i++) {
914 if (line[i] == '\r'
915 || line[i] == '\n'
916 || line[i] == '\t'
917 || line[i] == ' ') {
918 name = line + i + 1;
919 }
920 }
921 name_len = line + line_len - name;
922 end = name;
923 } else {
924 /* Filename is first item on line */
925 name_len = strcspn(line, " \t\r\n");
926 name = line;
927 line += name_len;
928 end = line + line_len;
929 }
930 /* name/name_len is the name within the line. */
931 /* line..end brackets the entire line except the name */
932
933 if ((entry->name = malloc(name_len + 1)) == NULL) {
934 archive_set_error(&a->archive, errno, "Can't allocate memory");
935 return (ARCHIVE_FATAL);
936 }
937
938 memcpy(entry->name, name, name_len);
939 entry->name[name_len] = '\0';
940 parse_escapes(entry->name, entry);
941
942 entry->next_dup = NULL;
943 if (entry->full) {
944 if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) {
945 struct mtree_entry *alt;
946 alt = (struct mtree_entry *)__archive_rb_tree_find_node(
947 &mtree->rbtree, entry->name);
948 if (alt != NULL) {
949 while (alt->next_dup)
950 alt = alt->next_dup;
951 alt->next_dup = entry;
952 }
953 }
954 }
955
956 for (iter = *global; iter != NULL; iter = iter->next) {
957 r = add_option(a, &entry->options, iter->value,
958 strlen(iter->value));
959 if (r != ARCHIVE_OK)
960 return (r);
961 }
962
963 for (;;) {
964 next = line + strspn(line, " \t\r\n");
965 if (*next == '\0')
966 return (ARCHIVE_OK);
967 if (next >= end)
968 return (ARCHIVE_OK);
969 line = next;
970 next = line + strcspn(line, " \t\r\n");
971 eq = strchr(line, '=');
972 if (eq == NULL || eq > next)
973 len = next - line;
974 else
975 len = eq - line;
976
977 remove_option(&entry->options, line, len);
978 r = add_option(a, &entry->options, line, next - line);
979 if (r != ARCHIVE_OK)
980 return (r);
981 line = next;
982 }
983 }
984
985 static int
read_mtree(struct archive_read * a,struct mtree * mtree)986 read_mtree(struct archive_read *a, struct mtree *mtree)
987 {
988 ssize_t len;
989 uintmax_t counter;
990 char *p, *s;
991 struct mtree_option *global;
992 struct mtree_entry *last_entry;
993 int r, is_form_d;
994
995 mtree->archive_format = ARCHIVE_FORMAT_MTREE;
996 mtree->archive_format_name = "mtree";
997
998 global = NULL;
999 last_entry = NULL;
1000
1001 (void)detect_form(a, &is_form_d);
1002
1003 for (counter = 1; ; ++counter) {
1004 r = ARCHIVE_OK;
1005 len = readline(a, mtree, &p, 65536);
1006 if (len == 0) {
1007 mtree->this_entry = mtree->entries;
1008 free_options(global);
1009 return (ARCHIVE_OK);
1010 }
1011 if (len < 0) {
1012 free_options(global);
1013 return ((int)len);
1014 }
1015 /* Leading whitespace is never significant, ignore it. */
1016 while (*p == ' ' || *p == '\t') {
1017 ++p;
1018 --len;
1019 }
1020 /* Skip content lines and blank lines. */
1021 if (*p == '#')
1022 continue;
1023 if (*p == '\r' || *p == '\n' || *p == '\0')
1024 continue;
1025 /* Non-printable characters are not allowed */
1026 for (s = p;s < p + len - 1; s++) {
1027 if (!isprint((unsigned char)*s) && *s != '\t') {
1028 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1029 "Non-printable character 0x%02X", (unsigned char)(*s));
1030 r = ARCHIVE_FATAL;
1031 break;
1032 }
1033 }
1034 if (r != ARCHIVE_OK)
1035 break;
1036 if (*p != '/') {
1037 r = process_add_entry(a, mtree, &global, p, len,
1038 &last_entry, is_form_d);
1039 } else if (len > 4 && strncmp(p, "/set", 4) == 0) {
1040 if (p[4] != ' ' && p[4] != '\t')
1041 break;
1042 r = process_global_set(a, &global, p);
1043 } else if (len > 6 && strncmp(p, "/unset", 6) == 0) {
1044 if (p[6] != ' ' && p[6] != '\t')
1045 break;
1046 r = process_global_unset(a, &global, p);
1047 } else
1048 break;
1049
1050 if (r != ARCHIVE_OK) {
1051 free_options(global);
1052 return r;
1053 }
1054 }
1055
1056 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1057 "Can't parse line %ju", counter);
1058 free_options(global);
1059 return (ARCHIVE_FATAL);
1060 }
1061
1062 /*
1063 * Read in the entire mtree file into memory on the first request.
1064 * Then use the next unused file to satisfy each header request.
1065 */
1066 static int
read_header(struct archive_read * a,struct archive_entry * entry)1067 read_header(struct archive_read *a, struct archive_entry *entry)
1068 {
1069 struct mtree *mtree = a->format->data;
1070 char *p;
1071 int r, use_next;
1072
1073 if (mtree->fd >= 0) {
1074 close(mtree->fd);
1075 mtree->fd = -1;
1076 }
1077
1078 if (mtree->entries == NULL) {
1079 mtree->resolver = archive_entry_linkresolver_new();
1080 if (mtree->resolver == NULL)
1081 return ARCHIVE_FATAL;
1082 archive_entry_linkresolver_set_strategy(mtree->resolver,
1083 ARCHIVE_FORMAT_MTREE);
1084 r = read_mtree(a, mtree);
1085 if (r != ARCHIVE_OK)
1086 return (r);
1087 }
1088
1089 a->archive.archive_format = mtree->archive_format;
1090 a->archive.archive_format_name = mtree->archive_format_name;
1091
1092 for (;;) {
1093 if (mtree->this_entry == NULL)
1094 return (ARCHIVE_EOF);
1095 if (strcmp(mtree->this_entry->name, "..") == 0) {
1096 mtree->this_entry->used = 1;
1097 if (archive_strlen(&mtree->current_dir) > 0) {
1098 /* Roll back current path. */
1099 p = mtree->current_dir.s
1100 + mtree->current_dir.length - 1;
1101 while (p >= mtree->current_dir.s && *p != '/')
1102 --p;
1103 if (p >= mtree->current_dir.s)
1104 --p;
1105 mtree->current_dir.length
1106 = p - mtree->current_dir.s + 1;
1107 }
1108 }
1109 if (!mtree->this_entry->used) {
1110 use_next = 0;
1111 r = parse_file(a, entry, mtree, mtree->this_entry,
1112 &use_next);
1113 if (use_next == 0)
1114 return (r);
1115 }
1116 mtree->this_entry = mtree->this_entry->next;
1117 }
1118 }
1119
1120 /*
1121 * A single file can have multiple lines contribute specifications.
1122 * Parse as many lines as necessary, then pull additional information
1123 * from a backing file on disk as necessary.
1124 */
1125 static int
parse_file(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mentry,int * use_next)1126 parse_file(struct archive_read *a, struct archive_entry *entry,
1127 struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
1128 {
1129 const char *path;
1130 la_seek_stat_t st_storage, *st;
1131 struct mtree_entry *mp;
1132 struct archive_entry *sparse_entry;
1133 int r = ARCHIVE_OK, r1, parsed_kws;
1134
1135 mentry->used = 1;
1136
1137 /* Initialize reasonable defaults. */
1138 archive_entry_set_filetype(entry, AE_IFREG);
1139 archive_entry_set_size(entry, 0);
1140 archive_string_empty(&mtree->contents_name);
1141
1142 /* Parse options from this line. */
1143 parsed_kws = 0;
1144 r = parse_line(a, entry, mtree, mentry, &parsed_kws);
1145
1146 if (mentry->full) {
1147 archive_entry_copy_pathname(entry, mentry->name);
1148 /*
1149 * "Full" entries are allowed to have multiple lines
1150 * and those lines aren't required to be adjacent. We
1151 * don't support multiple lines for "relative" entries
1152 * nor do we make any attempt to merge data from
1153 * separate "relative" and "full" entries. (Merging
1154 * "relative" and "full" entries would require dealing
1155 * with pathname canonicalization, which is a very
1156 * tricky subject.)
1157 */
1158 mp = (struct mtree_entry *)__archive_rb_tree_find_node(
1159 &mtree->rbtree, mentry->name);
1160 for (; mp; mp = mp->next_dup) {
1161 if (mp->full && !mp->used) {
1162 /* Later lines override earlier ones. */
1163 mp->used = 1;
1164 r1 = parse_line(a, entry, mtree, mp, &parsed_kws);
1165 if (r1 < r)
1166 r = r1;
1167 }
1168 }
1169 } else {
1170 /*
1171 * Relative entries require us to construct
1172 * the full path and possibly update the
1173 * current directory.
1174 */
1175 size_t n = archive_strlen(&mtree->current_dir);
1176 if (n > 0)
1177 archive_strcat(&mtree->current_dir, "/");
1178 archive_strcat(&mtree->current_dir, mentry->name);
1179 archive_entry_copy_pathname(entry, mtree->current_dir.s);
1180 if (archive_entry_filetype(entry) != AE_IFDIR)
1181 mtree->current_dir.length = n;
1182 }
1183
1184 if (mtree->checkfs) {
1185 /*
1186 * Try to open and stat the file to get the real size
1187 * and other file info. It would be nice to avoid
1188 * this here so that getting a listing of an mtree
1189 * wouldn't require opening every referenced contents
1190 * file. But then we wouldn't know the actual
1191 * contents size, so I don't see a really viable way
1192 * around this. (Also, we may want to someday pull
1193 * other unspecified info from the contents file on
1194 * disk.)
1195 */
1196 mtree->fd = -1;
1197 if (archive_strlen(&mtree->contents_name) > 0)
1198 path = mtree->contents_name.s;
1199 else
1200 path = archive_entry_pathname(entry);
1201
1202 if (archive_entry_filetype(entry) == AE_IFREG ||
1203 archive_entry_filetype(entry) == AE_IFDIR) {
1204 mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC);
1205 __archive_ensure_cloexec_flag(mtree->fd);
1206 if (mtree->fd == -1 && (
1207 #if defined(_WIN32) && !defined(__CYGWIN__)
1208 /*
1209 * On Windows, attempting to open a file with an
1210 * invalid name result in EINVAL (Error 22)
1211 */
1212 (errno != ENOENT && errno != EINVAL)
1213 #else
1214 errno != ENOENT
1215 #endif
1216 || archive_strlen(&mtree->contents_name) > 0)) {
1217 archive_set_error(&a->archive, errno,
1218 "Can't open %s", path);
1219 r = ARCHIVE_WARN;
1220 }
1221 }
1222
1223 st = &st_storage;
1224 if (mtree->fd >= 0) {
1225 if (la_seek_fstat(mtree->fd, st) == -1) {
1226 archive_set_error(&a->archive, errno,
1227 "Could not fstat %s", path);
1228 r = ARCHIVE_WARN;
1229 /* If we can't stat it, don't keep it open. */
1230 close(mtree->fd);
1231 mtree->fd = -1;
1232 st = NULL;
1233 }
1234 }
1235 #ifdef HAVE_LSTAT
1236 else if (lstat(path, st) == -1)
1237 #else
1238 else if (la_seek_stat(path, st) == -1)
1239 #endif
1240 {
1241 st = NULL;
1242 }
1243
1244 /*
1245 * Check for a mismatch between the type in the specification
1246 * and the type of the contents object on disk.
1247 */
1248 if (st != NULL) {
1249 if (((st->st_mode & S_IFMT) == S_IFREG &&
1250 archive_entry_filetype(entry) == AE_IFREG)
1251 #ifdef S_IFLNK
1252 ||((st->st_mode & S_IFMT) == S_IFLNK &&
1253 archive_entry_filetype(entry) == AE_IFLNK)
1254 #endif
1255 #ifdef S_IFSOCK
1256 ||((st->st_mode & S_IFSOCK) == S_IFSOCK &&
1257 archive_entry_filetype(entry) == AE_IFSOCK)
1258 #endif
1259 #ifdef S_IFCHR
1260 ||((st->st_mode & S_IFMT) == S_IFCHR &&
1261 archive_entry_filetype(entry) == AE_IFCHR)
1262 #endif
1263 #ifdef S_IFBLK
1264 ||((st->st_mode & S_IFMT) == S_IFBLK &&
1265 archive_entry_filetype(entry) == AE_IFBLK)
1266 #endif
1267 ||((st->st_mode & S_IFMT) == S_IFDIR &&
1268 archive_entry_filetype(entry) == AE_IFDIR)
1269 #ifdef S_IFIFO
1270 ||((st->st_mode & S_IFMT) == S_IFIFO &&
1271 archive_entry_filetype(entry) == AE_IFIFO)
1272 #endif
1273 ) {
1274 /* Types match. */
1275 } else {
1276 /* Types don't match; bail out gracefully. */
1277 if (mtree->fd >= 0)
1278 close(mtree->fd);
1279 mtree->fd = -1;
1280 if (parsed_kws & MTREE_HAS_OPTIONAL) {
1281 /* It's not an error for an optional
1282 * entry to not match disk. */
1283 *use_next = 1;
1284 } else if (r == ARCHIVE_OK) {
1285 archive_set_error(&a->archive,
1286 ARCHIVE_ERRNO_MISC,
1287 "mtree specification has different"
1288 " type for %s",
1289 archive_entry_pathname(entry));
1290 r = ARCHIVE_WARN;
1291 }
1292 return (r);
1293 }
1294 }
1295
1296 /*
1297 * If there is a contents file on disk, pick some of the
1298 * metadata from that file. For most of these, we only
1299 * set it from the contents if it wasn't already parsed
1300 * from the specification.
1301 */
1302 if (st != NULL) {
1303 if (((parsed_kws & MTREE_HAS_DEVICE) == 0 ||
1304 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) &&
1305 (archive_entry_filetype(entry) == AE_IFCHR ||
1306 archive_entry_filetype(entry) == AE_IFBLK))
1307 archive_entry_set_rdev(entry, st->st_rdev);
1308 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME))
1309 == 0 ||
1310 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1311 archive_entry_set_gid(entry, st->st_gid);
1312 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME))
1313 == 0 ||
1314 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1315 archive_entry_set_uid(entry, st->st_uid);
1316 if ((parsed_kws & MTREE_HAS_MTIME) == 0 ||
1317 (parsed_kws & MTREE_HAS_NOCHANGE) != 0) {
1318 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
1319 archive_entry_set_mtime(entry, st->st_mtime,
1320 st->st_mtimespec.tv_nsec);
1321 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
1322 archive_entry_set_mtime(entry, st->st_mtime,
1323 st->st_mtim.tv_nsec);
1324 #elif HAVE_STRUCT_STAT_ST_MTIME_N
1325 archive_entry_set_mtime(entry, st->st_mtime,
1326 st->st_mtime_n);
1327 #elif HAVE_STRUCT_STAT_ST_UMTIME
1328 archive_entry_set_mtime(entry, st->st_mtime,
1329 st->st_umtime*1000);
1330 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
1331 archive_entry_set_mtime(entry, st->st_mtime,
1332 st->st_mtime_usec*1000);
1333 #else
1334 archive_entry_set_mtime(entry, st->st_mtime, 0);
1335 #endif
1336 }
1337 if ((parsed_kws & MTREE_HAS_NLINK) == 0 ||
1338 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1339 archive_entry_set_nlink(entry, st->st_nlink);
1340 if ((parsed_kws & MTREE_HAS_PERM) == 0 ||
1341 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1342 archive_entry_set_perm(entry, st->st_mode);
1343 if ((parsed_kws & MTREE_HAS_SIZE) == 0 ||
1344 (parsed_kws & MTREE_HAS_NOCHANGE) != 0)
1345 archive_entry_set_size(entry, st->st_size);
1346 archive_entry_set_ino(entry, st->st_ino);
1347 archive_entry_set_dev(entry, st->st_dev);
1348
1349 archive_entry_linkify(mtree->resolver, &entry,
1350 &sparse_entry);
1351 } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
1352 /*
1353 * Couldn't open the entry, stat it or the on-disk type
1354 * didn't match. If this entry is optional, just
1355 * ignore it and read the next header entry.
1356 */
1357 *use_next = 1;
1358 return ARCHIVE_OK;
1359 }
1360 }
1361
1362 mtree->cur_size = archive_entry_size(entry);
1363 mtree->offset = 0;
1364
1365 return r;
1366 }
1367
1368 /*
1369 * Each line contains a sequence of keywords.
1370 */
1371 static int
parse_line(struct archive_read * a,struct archive_entry * entry,struct mtree * mtree,struct mtree_entry * mp,int * parsed_kws)1372 parse_line(struct archive_read *a, struct archive_entry *entry,
1373 struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
1374 {
1375 struct mtree_option *iter;
1376 int r = ARCHIVE_OK, r1;
1377
1378 for (iter = mp->options; iter != NULL; iter = iter->next) {
1379 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
1380 if (r1 < r)
1381 r = r1;
1382 }
1383 if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) {
1384 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1385 "Missing type keyword in mtree specification");
1386 return (ARCHIVE_WARN);
1387 }
1388 return (r);
1389 }
1390
1391 /*
1392 * Device entries have one of the following forms:
1393 * - raw dev_t
1394 * - format,major,minor[,subdevice]
1395 * When parsing succeeded, `pdev' will contain the appropriate dev_t value.
1396 */
1397
1398 /* strsep() is not in C90, but strcspn() is. */
1399 /* Taken from http://unixpapa.com/incnote/string.html */
1400 static char *
la_strsep(char ** sp,const char * sep)1401 la_strsep(char **sp, const char *sep)
1402 {
1403 char *p, *s;
1404 if (sp == NULL || *sp == NULL || **sp == '\0')
1405 return(NULL);
1406 s = *sp;
1407 p = s + strcspn(s, sep);
1408 if (*p != '\0')
1409 *p++ = '\0';
1410 *sp = p;
1411 return(s);
1412 }
1413
1414 static int
parse_device(dev_t * pdev,struct archive * a,char * val)1415 parse_device(dev_t *pdev, struct archive *a, char *val)
1416 {
1417 #define MAX_PACK_ARGS 3
1418 unsigned long numbers[MAX_PACK_ARGS];
1419 char *p, *dev;
1420 int argc;
1421 pack_t *pack;
1422 dev_t result;
1423 const char *error = NULL;
1424
1425 memset(pdev, 0, sizeof(*pdev));
1426 if ((dev = strchr(val, ',')) != NULL) {
1427 /*
1428 * Device's major/minor are given in a specified format.
1429 * Decode and pack it accordingly.
1430 */
1431 *dev++ = '\0';
1432 if ((pack = pack_find(val)) == NULL) {
1433 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1434 "Unknown format `%s'", val);
1435 return ARCHIVE_WARN;
1436 }
1437 argc = 0;
1438 while ((p = la_strsep(&dev, ",")) != NULL) {
1439 if (*p == '\0') {
1440 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1441 "Missing number");
1442 return ARCHIVE_WARN;
1443 }
1444 if (argc >= MAX_PACK_ARGS) {
1445 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1446 "Too many arguments");
1447 return ARCHIVE_WARN;
1448 }
1449 numbers[argc++] = (unsigned long)mtree_atol(&p, 0);
1450 }
1451 if (argc < 2) {
1452 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1453 "Not enough arguments");
1454 return ARCHIVE_WARN;
1455 }
1456 result = (*pack)(argc, numbers, &error);
1457 if (error != NULL) {
1458 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
1459 "%s", error);
1460 return ARCHIVE_WARN;
1461 }
1462 } else {
1463 /* file system raw value. */
1464 result = (dev_t)mtree_atol(&val, 0);
1465 }
1466 *pdev = result;
1467 return ARCHIVE_OK;
1468 #undef MAX_PACK_ARGS
1469 }
1470
1471 static int
parse_hex_nibble(char c)1472 parse_hex_nibble(char c)
1473 {
1474 if (c >= '0' && c <= '9')
1475 return c - '0';
1476 if (c >= 'a' && c <= 'f')
1477 return 10 + c - 'a';
1478 #if 0
1479 /* XXX: Is uppercase something we should support? */
1480 if (c >= 'A' && c <= 'F')
1481 return 10 + c - 'A';
1482 #endif
1483
1484 return -1;
1485 }
1486
1487 static int
parse_digest(struct archive_read * a,struct archive_entry * entry,const char * digest,int type)1488 parse_digest(struct archive_read *a, struct archive_entry *entry,
1489 const char *digest, int type)
1490 {
1491 unsigned char digest_buf[64];
1492 int high, low;
1493 size_t i, j, len;
1494
1495 switch (type) {
1496 case ARCHIVE_ENTRY_DIGEST_MD5:
1497 len = sizeof(entry->digest.md5);
1498 break;
1499 case ARCHIVE_ENTRY_DIGEST_RMD160:
1500 len = sizeof(entry->digest.rmd160);
1501 break;
1502 case ARCHIVE_ENTRY_DIGEST_SHA1:
1503 len = sizeof(entry->digest.sha1);
1504 break;
1505 case ARCHIVE_ENTRY_DIGEST_SHA256:
1506 len = sizeof(entry->digest.sha256);
1507 break;
1508 case ARCHIVE_ENTRY_DIGEST_SHA384:
1509 len = sizeof(entry->digest.sha384);
1510 break;
1511 case ARCHIVE_ENTRY_DIGEST_SHA512:
1512 len = sizeof(entry->digest.sha512);
1513 break;
1514 default:
1515 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1516 "Internal error: Unknown digest type");
1517 return ARCHIVE_FATAL;
1518 }
1519
1520 if (len > sizeof(digest_buf)) {
1521 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
1522 "Internal error: Digest storage too large");
1523 return ARCHIVE_FATAL;
1524 }
1525
1526 len *= 2;
1527
1528 if (mtree_strnlen(digest, len+1) != len) {
1529 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1530 "incorrect digest length, ignoring");
1531 return ARCHIVE_WARN;
1532 }
1533
1534 for (i = 0, j = 0; i < len; i += 2, j++) {
1535 high = parse_hex_nibble(digest[i]);
1536 low = parse_hex_nibble(digest[i+1]);
1537 if (high == -1 || low == -1) {
1538 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1539 "invalid digest data, ignoring");
1540 return ARCHIVE_WARN;
1541 }
1542
1543 digest_buf[j] = high << 4 | low;
1544 }
1545
1546 return archive_entry_set_digest(entry, type, digest_buf);
1547 }
1548
1549 /*
1550 * Parse a single keyword and its value.
1551 */
1552 static int
parse_keyword(struct archive_read * a,struct mtree * mtree,struct archive_entry * entry,struct mtree_option * opt,int * parsed_kws)1553 parse_keyword(struct archive_read *a, struct mtree *mtree,
1554 struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws)
1555 {
1556 char *val, *key;
1557
1558 key = opt->value;
1559
1560 if (*key == '\0')
1561 return (ARCHIVE_OK);
1562
1563 if (strcmp(key, "nochange") == 0) {
1564 *parsed_kws |= MTREE_HAS_NOCHANGE;
1565 return (ARCHIVE_OK);
1566 }
1567 if (strcmp(key, "optional") == 0) {
1568 *parsed_kws |= MTREE_HAS_OPTIONAL;
1569 return (ARCHIVE_OK);
1570 }
1571 if (strcmp(key, "ignore") == 0) {
1572 /*
1573 * The mtree processing is not recursive, so
1574 * recursion will only happen for explicitly listed
1575 * entries.
1576 */
1577 return (ARCHIVE_OK);
1578 }
1579
1580 val = strchr(key, '=');
1581 if (val == NULL) {
1582 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1583 "Malformed attribute \"%s\" (%d)", key, key[0]);
1584 return (ARCHIVE_WARN);
1585 }
1586
1587 *val = '\0';
1588 ++val;
1589
1590 switch (key[0]) {
1591 case 'c':
1592 if (strcmp(key, "content") == 0
1593 || strcmp(key, "contents") == 0) {
1594 parse_escapes(val, NULL);
1595 archive_strcpy(&mtree->contents_name, val);
1596 return (ARCHIVE_OK);
1597 }
1598 if (strcmp(key, "cksum") == 0)
1599 return (ARCHIVE_OK);
1600 break;
1601 case 'd':
1602 if (strcmp(key, "device") == 0) {
1603 /* stat(2) st_rdev field, e.g. the major/minor IDs
1604 * of a char/block special file */
1605 int r;
1606 dev_t dev;
1607
1608 *parsed_kws |= MTREE_HAS_DEVICE;
1609 r = parse_device(&dev, &a->archive, val);
1610 if (r == ARCHIVE_OK)
1611 archive_entry_set_rdev(entry, dev);
1612 return r;
1613 }
1614 break;
1615 case 'f':
1616 if (strcmp(key, "flags") == 0) {
1617 *parsed_kws |= MTREE_HAS_FFLAGS;
1618 archive_entry_copy_fflags_text(entry, val);
1619 return (ARCHIVE_OK);
1620 }
1621 break;
1622 case 'g':
1623 if (strcmp(key, "gid") == 0) {
1624 *parsed_kws |= MTREE_HAS_GID;
1625 archive_entry_set_gid(entry, mtree_atol(&val, 10));
1626 return (ARCHIVE_OK);
1627 }
1628 if (strcmp(key, "gname") == 0) {
1629 *parsed_kws |= MTREE_HAS_GNAME;
1630 archive_entry_copy_gname(entry, val);
1631 return (ARCHIVE_OK);
1632 }
1633 break;
1634 case 'i':
1635 if (strcmp(key, "inode") == 0) {
1636 archive_entry_set_ino(entry, mtree_atol(&val, 10));
1637 return (ARCHIVE_OK);
1638 }
1639 break;
1640 case 'l':
1641 if (strcmp(key, "link") == 0) {
1642 parse_escapes(val, NULL);
1643 archive_entry_copy_symlink(entry, val);
1644 return (ARCHIVE_OK);
1645 }
1646 break;
1647 case 'm':
1648 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) {
1649 return parse_digest(a, entry, val,
1650 ARCHIVE_ENTRY_DIGEST_MD5);
1651 }
1652 if (strcmp(key, "mode") == 0) {
1653 if (val[0] < '0' || val[0] > '7') {
1654 archive_set_error(&a->archive,
1655 ARCHIVE_ERRNO_FILE_FORMAT,
1656 "Symbolic or non-octal mode \"%s\" unsupported", val);
1657 return (ARCHIVE_WARN);
1658 }
1659 *parsed_kws |= MTREE_HAS_PERM;
1660 archive_entry_set_perm(entry, (mode_t)mtree_atol(&val, 8));
1661 return (ARCHIVE_OK);
1662 }
1663 break;
1664 case 'n':
1665 if (strcmp(key, "nlink") == 0) {
1666 *parsed_kws |= MTREE_HAS_NLINK;
1667 archive_entry_set_nlink(entry,
1668 (unsigned int)mtree_atol(&val, 10));
1669 return (ARCHIVE_OK);
1670 }
1671 break;
1672 case 'r':
1673 if (strcmp(key, "resdevice") == 0) {
1674 /* stat(2) st_dev field, e.g. the device ID where the
1675 * inode resides */
1676 int r;
1677 dev_t dev;
1678
1679 r = parse_device(&dev, &a->archive, val);
1680 if (r == ARCHIVE_OK)
1681 archive_entry_set_dev(entry, dev);
1682 return r;
1683 }
1684 if (strcmp(key, "rmd160") == 0 ||
1685 strcmp(key, "rmd160digest") == 0) {
1686 return parse_digest(a, entry, val,
1687 ARCHIVE_ENTRY_DIGEST_RMD160);
1688 }
1689 break;
1690 case 's':
1691 if (strcmp(key, "sha1") == 0 ||
1692 strcmp(key, "sha1digest") == 0) {
1693 return parse_digest(a, entry, val,
1694 ARCHIVE_ENTRY_DIGEST_SHA1);
1695 }
1696 if (strcmp(key, "sha256") == 0 ||
1697 strcmp(key, "sha256digest") == 0) {
1698 return parse_digest(a, entry, val,
1699 ARCHIVE_ENTRY_DIGEST_SHA256);
1700 }
1701 if (strcmp(key, "sha384") == 0 ||
1702 strcmp(key, "sha384digest") == 0) {
1703 return parse_digest(a, entry, val,
1704 ARCHIVE_ENTRY_DIGEST_SHA384);
1705 }
1706 if (strcmp(key, "sha512") == 0 ||
1707 strcmp(key, "sha512digest") == 0) {
1708 return parse_digest(a, entry, val,
1709 ARCHIVE_ENTRY_DIGEST_SHA512);
1710 }
1711 if (strcmp(key, "size") == 0) {
1712 archive_entry_set_size(entry, mtree_atol(&val, 10));
1713 return (ARCHIVE_OK);
1714 }
1715 break;
1716 case 't':
1717 if (strcmp(key, "tags") == 0) {
1718 /*
1719 * Comma delimited list of tags.
1720 * Ignore the tags for now, but the interface
1721 * should be extended to allow inclusion/exclusion.
1722 */
1723 return (ARCHIVE_OK);
1724 }
1725 if (strcmp(key, "time") == 0) {
1726 int64_t m;
1727 long ns = 0;
1728
1729 *parsed_kws |= MTREE_HAS_MTIME;
1730 m = mtree_atol(&val, 10);
1731 /* Replicate an old mtree bug:
1732 * 123456789.1 represents 123456789
1733 * seconds and 1 nanosecond. */
1734 if (*val == '.') {
1735 int64_t v;
1736
1737 ++val;
1738 v = mtree_atol(&val, 10);
1739 if (v < 0)
1740 ns = 0;
1741 else if (v > 999999999)
1742 ns = 999999999;
1743 else
1744 ns = (long)v;
1745 }
1746 if (m > TIME_MAX)
1747 m = TIME_MAX;
1748 else if (m < TIME_MIN)
1749 m = TIME_MIN;
1750 archive_entry_set_mtime(entry, (time_t)m, ns);
1751 return (ARCHIVE_OK);
1752 }
1753 if (strcmp(key, "type") == 0) {
1754 switch (val[0]) {
1755 case 'b':
1756 if (strcmp(val, "block") == 0) {
1757 *parsed_kws |= MTREE_HAS_TYPE;
1758 archive_entry_set_filetype(entry,
1759 AE_IFBLK);
1760 return (ARCHIVE_OK);
1761 }
1762 break;
1763 case 'c':
1764 if (strcmp(val, "char") == 0) {
1765 *parsed_kws |= MTREE_HAS_TYPE;
1766 archive_entry_set_filetype(entry,
1767 AE_IFCHR);
1768 return (ARCHIVE_OK);
1769 }
1770 break;
1771 case 'd':
1772 if (strcmp(val, "dir") == 0) {
1773 *parsed_kws |= MTREE_HAS_TYPE;
1774 archive_entry_set_filetype(entry,
1775 AE_IFDIR);
1776 return (ARCHIVE_OK);
1777 }
1778 break;
1779 case 'f':
1780 if (strcmp(val, "fifo") == 0) {
1781 *parsed_kws |= MTREE_HAS_TYPE;
1782 archive_entry_set_filetype(entry,
1783 AE_IFIFO);
1784 return (ARCHIVE_OK);
1785 }
1786 if (strcmp(val, "file") == 0) {
1787 *parsed_kws |= MTREE_HAS_TYPE;
1788 archive_entry_set_filetype(entry,
1789 AE_IFREG);
1790 return (ARCHIVE_OK);
1791 }
1792 break;
1793 case 'l':
1794 if (strcmp(val, "link") == 0) {
1795 *parsed_kws |= MTREE_HAS_TYPE;
1796 archive_entry_set_filetype(entry,
1797 AE_IFLNK);
1798 return (ARCHIVE_OK);
1799 }
1800 break;
1801 default:
1802 break;
1803 }
1804 archive_set_error(&a->archive,
1805 ARCHIVE_ERRNO_FILE_FORMAT,
1806 "Unrecognized file type \"%s\"; "
1807 "assuming \"file\"", val);
1808 archive_entry_set_filetype(entry, AE_IFREG);
1809 return (ARCHIVE_WARN);
1810 }
1811 break;
1812 case 'u':
1813 if (strcmp(key, "uid") == 0) {
1814 *parsed_kws |= MTREE_HAS_UID;
1815 archive_entry_set_uid(entry, mtree_atol(&val, 10));
1816 return (ARCHIVE_OK);
1817 }
1818 if (strcmp(key, "uname") == 0) {
1819 *parsed_kws |= MTREE_HAS_UNAME;
1820 archive_entry_copy_uname(entry, val);
1821 return (ARCHIVE_OK);
1822 }
1823 break;
1824 default:
1825 break;
1826 }
1827 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
1828 "Unrecognized key %s=%s", key, val);
1829 return (ARCHIVE_WARN);
1830 }
1831
1832 static int
read_data(struct archive_read * a,const void ** buff,size_t * size,int64_t * offset)1833 read_data(struct archive_read *a, const void **buff, size_t *size,
1834 int64_t *offset)
1835 {
1836 struct mtree *mtree = a->format->data;
1837 size_t bytes_to_read;
1838 ssize_t bytes_read;
1839
1840 if (mtree->fd < 0) {
1841 *buff = NULL;
1842 *offset = 0;
1843 *size = 0;
1844 return (ARCHIVE_EOF);
1845 }
1846 if (mtree->buff == NULL) {
1847 mtree->buffsize = 64 * 1024;
1848 mtree->buff = malloc(mtree->buffsize);
1849 if (mtree->buff == NULL) {
1850 archive_set_error(&a->archive, ENOMEM,
1851 "Can't allocate memory");
1852 return (ARCHIVE_FATAL);
1853 }
1854 }
1855
1856 *buff = mtree->buff;
1857 *offset = mtree->offset;
1858 if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset)
1859 bytes_to_read = (size_t)(mtree->cur_size - mtree->offset);
1860 else
1861 bytes_to_read = mtree->buffsize;
1862 bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
1863 if (bytes_read < 0) {
1864 archive_set_error(&a->archive, errno, "Can't read");
1865 return (ARCHIVE_WARN);
1866 }
1867 if (bytes_read == 0) {
1868 *size = 0;
1869 return (ARCHIVE_EOF);
1870 }
1871 mtree->offset += bytes_read;
1872 *size = bytes_read;
1873 return (ARCHIVE_OK);
1874 }
1875
1876 /* Skip does nothing except possibly close the contents file. */
1877 static int
skip(struct archive_read * a)1878 skip(struct archive_read *a)
1879 {
1880 struct mtree *mtree = a->format->data;
1881
1882 if (mtree->fd >= 0) {
1883 close(mtree->fd);
1884 mtree->fd = -1;
1885 }
1886 return (ARCHIVE_OK);
1887 }
1888
1889 /*
1890 * Since parsing backslash sequences always makes strings shorter,
1891 * we can always do this conversion in-place.
1892 */
1893 static void
parse_escapes(char * src,struct mtree_entry * mentry)1894 parse_escapes(char *src, struct mtree_entry *mentry)
1895 {
1896 char *dest = src;
1897 char c;
1898
1899 if (mentry != NULL && strcmp(src, ".") == 0)
1900 mentry->full = 1;
1901
1902 while (*src != '\0') {
1903 c = *src++;
1904 if (c == '/' && mentry != NULL)
1905 mentry->full = 1;
1906 if (c == '\\') {
1907 switch (src[0]) {
1908 case '0':
1909 if (src[1] < '0' || src[1] > '7') {
1910 c = 0;
1911 ++src;
1912 break;
1913 }
1914 /* FALLTHROUGH */
1915 case '1':
1916 case '2':
1917 case '3':
1918 if (src[1] >= '0' && src[1] <= '7' &&
1919 src[2] >= '0' && src[2] <= '7') {
1920 c = (src[0] - '0') << 6;
1921 c |= (src[1] - '0') << 3;
1922 c |= (src[2] - '0');
1923 src += 3;
1924 }
1925 break;
1926 case 'a':
1927 c = '\a';
1928 ++src;
1929 break;
1930 case 'b':
1931 c = '\b';
1932 ++src;
1933 break;
1934 case 'f':
1935 c = '\f';
1936 ++src;
1937 break;
1938 case 'n':
1939 c = '\n';
1940 ++src;
1941 break;
1942 case 'r':
1943 c = '\r';
1944 ++src;
1945 break;
1946 case 's':
1947 c = ' ';
1948 ++src;
1949 break;
1950 case 't':
1951 c = '\t';
1952 ++src;
1953 break;
1954 case 'v':
1955 c = '\v';
1956 ++src;
1957 break;
1958 case '\\':
1959 c = '\\';
1960 ++src;
1961 break;
1962 }
1963 }
1964 *dest++ = c;
1965 }
1966 *dest = '\0';
1967 }
1968
1969 /* Parse a hex digit. */
1970 static int
parsedigit(char c)1971 parsedigit(char c)
1972 {
1973 if (c >= '0' && c <= '9')
1974 return c - '0';
1975 else if (c >= 'a' && c <= 'f')
1976 return 10 + c - 'a';
1977 else if (c >= 'A' && c <= 'F')
1978 return 10 + c - 'A';
1979 else
1980 return -1;
1981 }
1982
1983 /*
1984 * Note that this implementation does not (and should not!) obey
1985 * locale settings; you cannot simply substitute strtol here, since
1986 * it does obey locale.
1987 */
1988 static int64_t
mtree_atol(char ** p,int base)1989 mtree_atol(char **p, int base)
1990 {
1991 int64_t l;
1992 int digit;
1993
1994 if (base == 0) {
1995 if (**p != '0')
1996 base = 10;
1997 else if ((*p)[1] == 'x' || (*p)[1] == 'X') {
1998 *p += 2;
1999 base = 16;
2000 } else {
2001 base = 8;
2002 }
2003 }
2004
2005 if (**p == '-') {
2006 ++(*p);
2007
2008 l = 0;
2009 digit = parsedigit(**p);
2010 while (digit >= 0 && digit < base) {
2011 if (archive_ckd_mul_i64(&l, l, base) ||
2012 archive_ckd_sub_i64(&l, l, digit))
2013 return INT64_MIN;
2014 digit = parsedigit(*++(*p));
2015 }
2016 return l;
2017 } else {
2018 l = 0;
2019 digit = parsedigit(**p);
2020 while (digit >= 0 && digit < base) {
2021 if (archive_ckd_mul_i64(&l, l, base) ||
2022 archive_ckd_add_i64(&l, l, digit))
2023 return INT64_MAX;
2024 digit = parsedigit(*++(*p));
2025 }
2026 return l;
2027 }
2028 }
2029
2030 /*
2031 * Returns length of line (including trailing newline)
2032 * or negative on error. 'start' argument is updated to
2033 * point to first character of line.
2034 */
2035 static ssize_t
readline(struct archive_read * a,struct mtree * mtree,char ** start,ssize_t limit)2036 readline(struct archive_read *a, struct mtree *mtree, char **start,
2037 ssize_t limit)
2038 {
2039 ssize_t bytes_read;
2040 ssize_t total_size = 0;
2041 ssize_t find_off = 0;
2042 const void *nl, *t;
2043 char *u;
2044
2045 /* Accumulate line in a line buffer. */
2046 for (;;) {
2047 /* Read some more. */
2048 t = __archive_read_ahead(a, 1, &bytes_read);
2049 if (t == NULL)
2050 return (0);
2051 if (bytes_read < 0)
2052 return (ARCHIVE_FATAL);
2053 nl = memchr(t, '\n', bytes_read);
2054 /* If we found '\n', trim the read to end exactly there. */
2055 if (nl != NULL) {
2056 bytes_read = ((const char *)nl) - ((const char *)t) + 1;
2057 }
2058 if (total_size + bytes_read + 1 > limit) {
2059 archive_set_error(&a->archive,
2060 ARCHIVE_ERRNO_FILE_FORMAT,
2061 "Line too long");
2062 return (ARCHIVE_FATAL);
2063 }
2064 if (archive_string_ensure(&mtree->line,
2065 total_size + bytes_read + 1) == NULL) {
2066 archive_set_error(&a->archive, ENOMEM,
2067 "Can't allocate working buffer");
2068 return (ARCHIVE_FATAL);
2069 }
2070 /* Append new bytes to string. */
2071 memcpy(mtree->line.s + total_size, t, bytes_read);
2072 __archive_read_consume(a, bytes_read);
2073 total_size += bytes_read;
2074 mtree->line.s[total_size] = '\0';
2075
2076 for (u = mtree->line.s + find_off; *u; ++u) {
2077 if (u[0] == '\n') {
2078 /* Ends with unescaped newline. */
2079 /* Check if preceded by '\r' for CRLF handling */
2080 if (u > mtree->line.s && u[-1] == '\r') {
2081 /* CRLF ending - remove the '\r' */
2082 u[-1] = '\n';
2083 u[0] = '\0';
2084 total_size--;
2085 }
2086 *start = mtree->line.s;
2087 return total_size;
2088 } else if (u[0] == '#') {
2089 /* Ends with comment sequence #...\n */
2090 if (nl == NULL) {
2091 /* But we've not found the \n yet */
2092 break;
2093 }
2094 } else if (u[0] == '\\') {
2095 if (u[1] == '\n') {
2096 /* Trim escaped newline. */
2097 total_size -= 2;
2098 mtree->line.s[total_size] = '\0';
2099 break;
2100 } else if (u[1] == '\r' && u[2] == '\n') {
2101 /* Trim escaped CRLF. */
2102 total_size -= 3;
2103 mtree->line.s[total_size] = '\0';
2104 break;
2105 } else if (u[1] != '\0') {
2106 /* Skip the two-char escape sequence */
2107 ++u;
2108 }
2109 }
2110 }
2111 find_off = u - mtree->line.s;
2112 }
2113 }
2114