1 /*-
2  * Copyright (c) 2007 Kai Wang
3  * Copyright (c) 2007 Tim Kientzle
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  *    in this position and unchanged.
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 __FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_ar.c 201108 2009-12-28 03:28:21Z kientzle $");
30 
31 #ifdef HAVE_ERRNO_H
32 #include <errno.h>
33 #endif
34 #ifdef HAVE_STDLIB_H
35 #include <stdlib.h>
36 #endif
37 #ifdef HAVE_STRING_H
38 #include <string.h>
39 #endif
40 
41 #include "archive.h"
42 #include "archive_entry.h"
43 #include "archive_private.h"
44 #include "archive_write_private.h"
45 
46 struct ar_w {
47 	uint64_t	 entry_bytes_remaining;
48 	uint64_t	 entry_padding;
49 	int		 is_strtab;
50 	int		 has_strtab;
51 	char		 wrote_global_header;
52 	char		*strtab;
53 };
54 
55 /*
56  * Define structure of the "ar" header.
57  */
58 #define AR_name_offset 0
59 #define AR_name_size 16
60 #define AR_date_offset 16
61 #define AR_date_size 12
62 #define AR_uid_offset 28
63 #define AR_uid_size 6
64 #define AR_gid_offset 34
65 #define AR_gid_size 6
66 #define AR_mode_offset 40
67 #define AR_mode_size 8
68 #define AR_size_offset 48
69 #define AR_size_size 10
70 #define AR_fmag_offset 58
71 #define AR_fmag_size 2
72 
73 static int		 archive_write_set_format_ar(struct archive_write *);
74 static int		 archive_write_ar_header(struct archive_write *,
75 			     struct archive_entry *);
76 static ssize_t		 archive_write_ar_data(struct archive_write *,
77 			     const void *buff, size_t s);
78 static int		 archive_write_ar_free(struct archive_write *);
79 static int		 archive_write_ar_close(struct archive_write *);
80 static int		 archive_write_ar_finish_entry(struct archive_write *);
81 static const char	*ar_basename(const char *path);
82 static int		 format_octal(int64_t v, char *p, int s);
83 static int		 format_decimal(int64_t v, char *p, int s);
84 
85 int
86 archive_write_set_format_ar_bsd(struct archive *_a)
87 {
88 	struct archive_write *a = (struct archive_write *)_a;
89 	int r;
90 
91 	archive_check_magic(_a, ARCHIVE_WRITE_MAGIC,
92 	    ARCHIVE_STATE_NEW, "archive_write_set_format_ar_bsd");
93 	r = archive_write_set_format_ar(a);
94 	if (r == ARCHIVE_OK) {
95 		a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD;
96 		a->archive.archive_format_name = "ar (BSD)";
97 	}
98 	return (r);
99 }
100 
101 int
102 archive_write_set_format_ar_svr4(struct archive *_a)
103 {
104 	struct archive_write *a = (struct archive_write *)_a;
105 	int r;
106 
107 	archive_check_magic(_a, ARCHIVE_WRITE_MAGIC,
108 	    ARCHIVE_STATE_NEW, "archive_write_set_format_ar_svr4");
109 	r = archive_write_set_format_ar(a);
110 	if (r == ARCHIVE_OK) {
111 		a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU;
112 		a->archive.archive_format_name = "ar (GNU/SVR4)";
113 	}
114 	return (r);
115 }
116 
117 /*
118  * Generic initialization.
119  */
120 static int
121 archive_write_set_format_ar(struct archive_write *a)
122 {
123 	struct ar_w *ar;
124 
125 	/* If someone else was already registered, unregister them. */
126 	if (a->format_free != NULL)
127 		(a->format_free)(a);
128 
129 	ar = (struct ar_w *)malloc(sizeof(*ar));
130 	if (ar == NULL) {
131 		archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data");
132 		return (ARCHIVE_FATAL);
133 	}
134 	memset(ar, 0, sizeof(*ar));
135 	a->format_data = ar;
136 
137 	a->format_name = "ar";
138 	a->format_write_header = archive_write_ar_header;
139 	a->format_write_data = archive_write_ar_data;
140 	a->format_close = archive_write_ar_close;
141 	a->format_free = archive_write_ar_free;
142 	a->format_finish_entry = archive_write_ar_finish_entry;
143 	return (ARCHIVE_OK);
144 }
145 
146 static int
147 archive_write_ar_header(struct archive_write *a, struct archive_entry *entry)
148 {
149 	int ret, append_fn;
150 	char buff[60];
151 	char *ss, *se;
152 	struct ar_w *ar;
153 	const char *pathname;
154 	const char *filename;
155 	int64_t size;
156 
157 	append_fn = 0;
158 	ar = (struct ar_w *)a->format_data;
159 	ar->is_strtab = 0;
160 	filename = NULL;
161 	size = archive_entry_size(entry);
162 
163 
164 	/*
165 	 * Reject files with empty name.
166 	 */
167 	pathname = archive_entry_pathname(entry);
168 	if (pathname == NULL || *pathname == '\0') {
169 		archive_set_error(&a->archive, EINVAL,
170 		    "Invalid filename");
171 		return (ARCHIVE_WARN);
172 	}
173 
174 	/*
175 	 * If we are now at the beginning of the archive,
176 	 * we need first write the ar global header.
177 	 */
178 	if (!ar->wrote_global_header) {
179 		__archive_write_output(a, "!<arch>\n", 8);
180 		ar->wrote_global_header = 1;
181 	}
182 
183 	memset(buff, ' ', 60);
184 	strncpy(&buff[AR_fmag_offset], "`\n", 2);
185 
186 	if (strcmp(pathname, "/") == 0 ) {
187 		/* Entry is archive symbol table in GNU format */
188 		buff[AR_name_offset] = '/';
189 		goto stat;
190 	}
191 	if (strcmp(pathname, "__.SYMDEF") == 0) {
192 		/* Entry is archive symbol table in BSD format */
193 		strncpy(buff + AR_name_offset, "__.SYMDEF", 9);
194 		goto stat;
195 	}
196 	if (strcmp(pathname, "//") == 0) {
197 		/*
198 		 * Entry is archive filename table, inform that we should
199 		 * collect strtab in next _data call.
200 		 */
201 		ar->is_strtab = 1;
202 		buff[AR_name_offset] = buff[AR_name_offset + 1] = '/';
203 		/*
204 		 * For archive string table, only ar_size field should
205 		 * be set.
206 		 */
207 		goto size;
208 	}
209 
210 	/*
211 	 * Otherwise, entry is a normal archive member.
212 	 * Strip leading paths from filenames, if any.
213 	 */
214 	if ((filename = ar_basename(pathname)) == NULL) {
215 		/* Reject filenames with trailing "/" */
216 		archive_set_error(&a->archive, EINVAL,
217 		    "Invalid filename");
218 		return (ARCHIVE_WARN);
219 	}
220 
221 	if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) {
222 		/*
223 		 * SVR4/GNU variant use a "/" to mark then end of the filename,
224 		 * make it possible to have embedded spaces in the filename.
225 		 * So, the longest filename here (without extension) is
226 		 * actually 15 bytes.
227 		 */
228 		if (strlen(filename) <= 15) {
229 			strncpy(&buff[AR_name_offset],
230 			    filename, strlen(filename));
231 			buff[AR_name_offset + strlen(filename)] = '/';
232 		} else {
233 			/*
234 			 * For filename longer than 15 bytes, GNU variant
235 			 * makes use of a string table and instead stores the
236 			 * offset of the real filename to in the ar_name field.
237 			 * The string table should have been written before.
238 			 */
239 			if (ar->has_strtab <= 0) {
240 				archive_set_error(&a->archive, EINVAL,
241 				    "Can't find string table");
242 				return (ARCHIVE_WARN);
243 			}
244 
245 			se = (char *)malloc(strlen(filename) + 3);
246 			if (se == NULL) {
247 				archive_set_error(&a->archive, ENOMEM,
248 				    "Can't allocate filename buffer");
249 				return (ARCHIVE_FATAL);
250 			}
251 
252 			strncpy(se, filename, strlen(filename));
253 			strcpy(se + strlen(filename), "/\n");
254 
255 			ss = strstr(ar->strtab, se);
256 			free(se);
257 
258 			if (ss == NULL) {
259 				archive_set_error(&a->archive, EINVAL,
260 				    "Invalid string table");
261 				return (ARCHIVE_WARN);
262 			}
263 
264 			/*
265 			 * GNU variant puts "/" followed by digits into
266 			 * ar_name field. These digits indicates the real
267 			 * filename string's offset to the string table.
268 			 */
269 			buff[AR_name_offset] = '/';
270 			if (format_decimal(ss - ar->strtab,
271 			    buff + AR_name_offset + 1,
272 			    AR_name_size - 1)) {
273 				archive_set_error(&a->archive, ERANGE,
274 				    "string table offset too large");
275 				return (ARCHIVE_WARN);
276 			}
277 		}
278 	} else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) {
279 		/*
280 		 * BSD variant: for any file name which is more than
281 		 * 16 chars or contains one or more embedded space(s), the
282 		 * string "#1/" followed by the ASCII length of the name is
283 		 * put into the ar_name field. The file size (stored in the
284 		 * ar_size field) is incremented by the length of the name.
285 		 * The name is then written immediately following the
286 		 * archive header.
287 		 */
288 		if (strlen(filename) <= 16 && strchr(filename, ' ') == NULL) {
289 			strncpy(&buff[AR_name_offset], filename, strlen(filename));
290 			buff[AR_name_offset + strlen(filename)] = ' ';
291 		}
292 		else {
293 			strncpy(buff + AR_name_offset, "#1/", 3);
294 			if (format_decimal(strlen(filename),
295 			    buff + AR_name_offset + 3,
296 			    AR_name_size - 3)) {
297 				archive_set_error(&a->archive, ERANGE,
298 				    "File name too long");
299 				return (ARCHIVE_WARN);
300 			}
301 			append_fn = 1;
302 			size += strlen(filename);
303 		}
304 	}
305 
306 stat:
307 	if (format_decimal(archive_entry_mtime(entry), buff + AR_date_offset, AR_date_size)) {
308 		archive_set_error(&a->archive, ERANGE,
309 		    "File modification time too large");
310 		return (ARCHIVE_WARN);
311 	}
312 	if (format_decimal(archive_entry_uid(entry), buff + AR_uid_offset, AR_uid_size)) {
313 		archive_set_error(&a->archive, ERANGE,
314 		    "Numeric user ID too large");
315 		return (ARCHIVE_WARN);
316 	}
317 	if (format_decimal(archive_entry_gid(entry), buff + AR_gid_offset, AR_gid_size)) {
318 		archive_set_error(&a->archive, ERANGE,
319 		    "Numeric group ID too large");
320 		return (ARCHIVE_WARN);
321 	}
322 	if (format_octal(archive_entry_mode(entry), buff + AR_mode_offset, AR_mode_size)) {
323 		archive_set_error(&a->archive, ERANGE,
324 		    "Numeric mode too large");
325 		return (ARCHIVE_WARN);
326 	}
327 	/*
328 	 * Sanity Check: A non-pseudo archive member should always be
329 	 * a regular file.
330 	 */
331 	if (filename != NULL && archive_entry_filetype(entry) != AE_IFREG) {
332 		archive_set_error(&a->archive, EINVAL,
333 		    "Regular file required for non-pseudo member");
334 		return (ARCHIVE_WARN);
335 	}
336 
337 size:
338 	if (format_decimal(size, buff + AR_size_offset, AR_size_size)) {
339 		archive_set_error(&a->archive, ERANGE,
340 		    "File size out of range");
341 		return (ARCHIVE_WARN);
342 	}
343 
344 	ret = __archive_write_output(a, buff, 60);
345 	if (ret != ARCHIVE_OK)
346 		return (ret);
347 
348 	ar->entry_bytes_remaining = size;
349 	ar->entry_padding = ar->entry_bytes_remaining % 2;
350 
351 	if (append_fn > 0) {
352 		ret = __archive_write_output(a, filename, strlen(filename));
353 		if (ret != ARCHIVE_OK)
354 			return (ret);
355 		ar->entry_bytes_remaining -= strlen(filename);
356 	}
357 
358 	return (ARCHIVE_OK);
359 }
360 
361 static ssize_t
362 archive_write_ar_data(struct archive_write *a, const void *buff, size_t s)
363 {
364 	struct ar_w *ar;
365 	int ret;
366 
367 	ar = (struct ar_w *)a->format_data;
368 	if (s > ar->entry_bytes_remaining)
369 		s = (size_t)ar->entry_bytes_remaining;
370 
371 	if (ar->is_strtab > 0) {
372 		if (ar->has_strtab > 0) {
373 			archive_set_error(&a->archive, EINVAL,
374 			    "More than one string tables exist");
375 			return (ARCHIVE_WARN);
376 		}
377 
378 		ar->strtab = (char *)malloc(s);
379 		if (ar->strtab == NULL) {
380 			archive_set_error(&a->archive, ENOMEM,
381 			    "Can't allocate strtab buffer");
382 			return (ARCHIVE_FATAL);
383 		}
384 		strncpy(ar->strtab, buff, s);
385 		ar->has_strtab = 1;
386 	}
387 
388 	ret = __archive_write_output(a, buff, s);
389 	if (ret != ARCHIVE_OK)
390 		return (ret);
391 
392 	ar->entry_bytes_remaining -= s;
393 	return (s);
394 }
395 
396 static int
397 archive_write_ar_free(struct archive_write *a)
398 {
399 	struct ar_w *ar;
400 
401 	ar = (struct ar_w *)a->format_data;
402 
403 	if (ar == NULL)
404 		return (ARCHIVE_OK);
405 
406 	if (ar->has_strtab > 0) {
407 		free(ar->strtab);
408 		ar->strtab = NULL;
409 	}
410 
411 	free(ar);
412 	a->format_data = NULL;
413 	return (ARCHIVE_OK);
414 }
415 
416 static int
417 archive_write_ar_close(struct archive_write *a)
418 {
419 	struct ar_w *ar;
420 	int ret;
421 
422 	/*
423 	 * If we haven't written anything yet, we need to write
424 	 * the ar global header now to make it a valid ar archive.
425 	 */
426 	ar = (struct ar_w *)a->format_data;
427 	if (!ar->wrote_global_header) {
428 		ar->wrote_global_header = 1;
429 		ret = __archive_write_output(a, "!<arch>\n", 8);
430 		return (ret);
431 	}
432 
433 	return (ARCHIVE_OK);
434 }
435 
436 static int
437 archive_write_ar_finish_entry(struct archive_write *a)
438 {
439 	struct ar_w *ar;
440 	int ret;
441 
442 	ar = (struct ar_w *)a->format_data;
443 
444 	if (ar->entry_bytes_remaining != 0) {
445 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
446 		    "Entry remaining bytes larger than 0");
447 		return (ARCHIVE_WARN);
448 	}
449 
450 	if (ar->entry_padding == 0) {
451 		return (ARCHIVE_OK);
452 	}
453 
454 	if (ar->entry_padding != 1) {
455 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
456 		    "Padding wrong size: %ju should be 1 or 0",
457 		    (uintmax_t)ar->entry_padding);
458 		return (ARCHIVE_WARN);
459 	}
460 
461 	ret = __archive_write_output(a, "\n", 1);
462 	return (ret);
463 }
464 
465 /*
466  * Format a number into the specified field using base-8.
467  * NB: This version is slightly different from the one in
468  * _ustar.c
469  */
470 static int
471 format_octal(int64_t v, char *p, int s)
472 {
473 	int len;
474 	char *h;
475 
476 	len = s;
477 	h = p;
478 
479 	/* Octal values can't be negative, so use 0. */
480 	if (v < 0) {
481 		while (len-- > 0)
482 			*p++ = '0';
483 		return (-1);
484 	}
485 
486 	p += s;		/* Start at the end and work backwards. */
487 	do {
488 		*--p = (char)('0' + (v & 7));
489 		v >>= 3;
490 	} while (--s > 0 && v > 0);
491 
492 	if (v == 0) {
493 		memmove(h, p, len - s);
494 		p = h + len - s;
495 		while (s-- > 0)
496 			*p++ = ' ';
497 		return (0);
498 	}
499 	/* If it overflowed, fill field with max value. */
500 	while (len-- > 0)
501 		*p++ = '7';
502 
503 	return (-1);
504 }
505 
506 /*
507  * Format a number into the specified field using base-10.
508  */
509 static int
510 format_decimal(int64_t v, char *p, int s)
511 {
512 	int len;
513 	char *h;
514 
515 	len = s;
516 	h = p;
517 
518 	/* Negative values in ar header are meaningless, so use 0. */
519 	if (v < 0) {
520 		while (len-- > 0)
521 			*p++ = '0';
522 		return (-1);
523 	}
524 
525 	p += s;
526 	do {
527 		*--p = (char)('0' + (v % 10));
528 		v /= 10;
529 	} while (--s > 0 && v > 0);
530 
531 	if (v == 0) {
532 		memmove(h, p, len - s);
533 		p = h + len - s;
534 		while (s-- > 0)
535 			*p++ = ' ';
536 		return (0);
537 	}
538 	/* If it overflowed, fill field with max value. */
539 	while (len-- > 0)
540 		*p++ = '9';
541 
542 	return (-1);
543 }
544 
545 static const char *
546 ar_basename(const char *path)
547 {
548 	const char *endp, *startp;
549 
550 	endp = path + strlen(path) - 1;
551 	/*
552 	 * For filename with trailing slash(es), we return
553 	 * NULL indicating an error.
554 	 */
555 	if (*endp == '/')
556 		return (NULL);
557 
558 	/* Find the start of the base */
559 	startp = endp;
560 	while (startp > path && *(startp - 1) != '/')
561 		startp--;
562 
563 	return (startp);
564 }
565