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 *)calloc(1, sizeof(*ar));
130 	if (ar == NULL) {
131 		archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data");
132 		return (ARCHIVE_FATAL);
133 	}
134 	a->format_data = ar;
135 
136 	a->format_name = "ar";
137 	a->format_write_header = archive_write_ar_header;
138 	a->format_write_data = archive_write_ar_data;
139 	a->format_close = archive_write_ar_close;
140 	a->format_free = archive_write_ar_free;
141 	a->format_finish_entry = archive_write_ar_finish_entry;
142 	return (ARCHIVE_OK);
143 }
144 
145 static int
146 archive_write_ar_header(struct archive_write *a, struct archive_entry *entry)
147 {
148 	int ret, append_fn;
149 	char buff[60];
150 	char *ss, *se;
151 	struct ar_w *ar;
152 	const char *pathname;
153 	const char *filename;
154 	int64_t size;
155 
156 	append_fn = 0;
157 	ar = (struct ar_w *)a->format_data;
158 	ar->is_strtab = 0;
159 	filename = NULL;
160 	size = archive_entry_size(entry);
161 
162 
163 	/*
164 	 * Reject files with empty name.
165 	 */
166 	pathname = archive_entry_pathname(entry);
167 	if (pathname == NULL || *pathname == '\0') {
168 		archive_set_error(&a->archive, EINVAL,
169 		    "Invalid filename");
170 		return (ARCHIVE_WARN);
171 	}
172 
173 	/*
174 	 * If we are now at the beginning of the archive,
175 	 * we need first write the ar global header.
176 	 */
177 	if (!ar->wrote_global_header) {
178 		__archive_write_output(a, "!<arch>\n", 8);
179 		ar->wrote_global_header = 1;
180 	}
181 
182 	memset(buff, ' ', 60);
183 	memcpy(&buff[AR_fmag_offset], "`\n", 2);
184 
185 	if (strcmp(pathname, "/") == 0 ) {
186 		/* Entry is archive symbol table in GNU format */
187 		buff[AR_name_offset] = '/';
188 		goto stat;
189 	}
190 	if (strcmp(pathname, "__.SYMDEF") == 0) {
191 		/* Entry is archive symbol table in BSD format */
192 		memcpy(buff + AR_name_offset, "__.SYMDEF", 9);
193 		goto stat;
194 	}
195 	if (strcmp(pathname, "//") == 0) {
196 		/*
197 		 * Entry is archive filename table, inform that we should
198 		 * collect strtab in next _data call.
199 		 */
200 		ar->is_strtab = 1;
201 		buff[AR_name_offset] = buff[AR_name_offset + 1] = '/';
202 		/*
203 		 * For archive string table, only ar_size field should
204 		 * be set.
205 		 */
206 		goto size;
207 	}
208 
209 	/*
210 	 * Otherwise, entry is a normal archive member.
211 	 * Strip leading paths from filenames, if any.
212 	 */
213 	if ((filename = ar_basename(pathname)) == NULL) {
214 		/* Reject filenames with trailing "/" */
215 		archive_set_error(&a->archive, EINVAL,
216 		    "Invalid filename");
217 		return (ARCHIVE_WARN);
218 	}
219 
220 	if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) {
221 		/*
222 		 * SVR4/GNU variant use a "/" to mark then end of the filename,
223 		 * make it possible to have embedded spaces in the filename.
224 		 * So, the longest filename here (without extension) is
225 		 * actually 15 bytes.
226 		 */
227 		if (strlen(filename) <= 15) {
228 			memcpy(&buff[AR_name_offset],
229 			    filename, strlen(filename));
230 			buff[AR_name_offset + strlen(filename)] = '/';
231 		} else {
232 			/*
233 			 * For filename longer than 15 bytes, GNU variant
234 			 * makes use of a string table and instead stores the
235 			 * offset of the real filename to in the ar_name field.
236 			 * The string table should have been written before.
237 			 */
238 			if (ar->has_strtab <= 0) {
239 				archive_set_error(&a->archive, EINVAL,
240 				    "Can't find string table");
241 				return (ARCHIVE_WARN);
242 			}
243 
244 			se = (char *)malloc(strlen(filename) + 3);
245 			if (se == NULL) {
246 				archive_set_error(&a->archive, ENOMEM,
247 				    "Can't allocate filename buffer");
248 				return (ARCHIVE_FATAL);
249 			}
250 
251 			memcpy(se, filename, strlen(filename));
252 			strcpy(se + strlen(filename), "/\n");
253 
254 			ss = strstr(ar->strtab, se);
255 			free(se);
256 
257 			if (ss == NULL) {
258 				archive_set_error(&a->archive, EINVAL,
259 				    "Invalid string table");
260 				return (ARCHIVE_WARN);
261 			}
262 
263 			/*
264 			 * GNU variant puts "/" followed by digits into
265 			 * ar_name field. These digits indicates the real
266 			 * filename string's offset to the string table.
267 			 */
268 			buff[AR_name_offset] = '/';
269 			if (format_decimal(ss - ar->strtab,
270 			    buff + AR_name_offset + 1,
271 			    AR_name_size - 1)) {
272 				archive_set_error(&a->archive, ERANGE,
273 				    "string table offset too large");
274 				return (ARCHIVE_WARN);
275 			}
276 		}
277 	} else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) {
278 		/*
279 		 * BSD variant: for any file name which is more than
280 		 * 16 chars or contains one or more embedded space(s), the
281 		 * string "#1/" followed by the ASCII length of the name is
282 		 * put into the ar_name field. The file size (stored in the
283 		 * ar_size field) is incremented by the length of the name.
284 		 * The name is then written immediately following the
285 		 * archive header.
286 		 */
287 		if (strlen(filename) <= 16 && strchr(filename, ' ') == NULL) {
288 			memcpy(&buff[AR_name_offset], filename, strlen(filename));
289 			buff[AR_name_offset + strlen(filename)] = ' ';
290 		}
291 		else {
292 			memcpy(buff + AR_name_offset, "#1/", 3);
293 			if (format_decimal(strlen(filename),
294 			    buff + AR_name_offset + 3,
295 			    AR_name_size - 3)) {
296 				archive_set_error(&a->archive, ERANGE,
297 				    "File name too long");
298 				return (ARCHIVE_WARN);
299 			}
300 			append_fn = 1;
301 			size += strlen(filename);
302 		}
303 	}
304 
305 stat:
306 	if (format_decimal(archive_entry_mtime(entry), buff + AR_date_offset, AR_date_size)) {
307 		archive_set_error(&a->archive, ERANGE,
308 		    "File modification time too large");
309 		return (ARCHIVE_WARN);
310 	}
311 	if (format_decimal(archive_entry_uid(entry), buff + AR_uid_offset, AR_uid_size)) {
312 		archive_set_error(&a->archive, ERANGE,
313 		    "Numeric user ID too large");
314 		return (ARCHIVE_WARN);
315 	}
316 	if (format_decimal(archive_entry_gid(entry), buff + AR_gid_offset, AR_gid_size)) {
317 		archive_set_error(&a->archive, ERANGE,
318 		    "Numeric group ID too large");
319 		return (ARCHIVE_WARN);
320 	}
321 	if (format_octal(archive_entry_mode(entry), buff + AR_mode_offset, AR_mode_size)) {
322 		archive_set_error(&a->archive, ERANGE,
323 		    "Numeric mode too large");
324 		return (ARCHIVE_WARN);
325 	}
326 	/*
327 	 * Sanity Check: A non-pseudo archive member should always be
328 	 * a regular file.
329 	 */
330 	if (filename != NULL && archive_entry_filetype(entry) != AE_IFREG) {
331 		archive_set_error(&a->archive, EINVAL,
332 		    "Regular file required for non-pseudo member");
333 		return (ARCHIVE_WARN);
334 	}
335 
336 size:
337 	if (format_decimal(size, buff + AR_size_offset, AR_size_size)) {
338 		archive_set_error(&a->archive, ERANGE,
339 		    "File size out of range");
340 		return (ARCHIVE_WARN);
341 	}
342 
343 	ret = __archive_write_output(a, buff, 60);
344 	if (ret != ARCHIVE_OK)
345 		return (ret);
346 
347 	ar->entry_bytes_remaining = size;
348 	ar->entry_padding = ar->entry_bytes_remaining % 2;
349 
350 	if (append_fn > 0) {
351 		ret = __archive_write_output(a, filename, strlen(filename));
352 		if (ret != ARCHIVE_OK)
353 			return (ret);
354 		ar->entry_bytes_remaining -= strlen(filename);
355 	}
356 
357 	return (ARCHIVE_OK);
358 }
359 
360 static ssize_t
361 archive_write_ar_data(struct archive_write *a, const void *buff, size_t s)
362 {
363 	struct ar_w *ar;
364 	int ret;
365 
366 	ar = (struct ar_w *)a->format_data;
367 	if (s > ar->entry_bytes_remaining)
368 		s = (size_t)ar->entry_bytes_remaining;
369 
370 	if (ar->is_strtab > 0) {
371 		if (ar->has_strtab > 0) {
372 			archive_set_error(&a->archive, EINVAL,
373 			    "More than one string tables exist");
374 			return (ARCHIVE_WARN);
375 		}
376 
377 		ar->strtab = (char *)malloc(s + 1);
378 		if (ar->strtab == NULL) {
379 			archive_set_error(&a->archive, ENOMEM,
380 			    "Can't allocate strtab buffer");
381 			return (ARCHIVE_FATAL);
382 		}
383 		memcpy(ar->strtab, buff, s);
384 		ar->strtab[s] = '\0';
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