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: src/lib/libarchive/archive_write_set_format_ar.c,v 1.7 2008/05/26 17:00:23 kientzle Exp $");
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		*strtab;
52 };
53 
54 /*
55  * Define structure of the "ar" header.
56  */
57 #define AR_name_offset 0
58 #define AR_name_size 16
59 #define AR_date_offset 16
60 #define AR_date_size 12
61 #define AR_uid_offset 28
62 #define AR_uid_size 6
63 #define AR_gid_offset 34
64 #define AR_gid_size 6
65 #define AR_mode_offset 40
66 #define AR_mode_size 8
67 #define AR_size_offset 48
68 #define AR_size_size 10
69 #define AR_fmag_offset 58
70 #define AR_fmag_size 2
71 
72 static int		 archive_write_set_format_ar(struct archive_write *);
73 static int		 archive_write_ar_header(struct archive_write *,
74 			     struct archive_entry *);
75 static ssize_t		 archive_write_ar_data(struct archive_write *,
76 			     const void *buff, size_t s);
77 static int		 archive_write_ar_destroy(struct archive_write *);
78 static int		 archive_write_ar_finish(struct archive_write *);
79 static int		 archive_write_ar_finish_entry(struct archive_write *);
80 static const char	*ar_basename(const char *path);
81 static int		 format_octal(int64_t v, char *p, int s);
82 static int		 format_decimal(int64_t v, char *p, int s);
83 
84 int
85 archive_write_set_format_ar_bsd(struct archive *_a)
86 {
87 	struct archive_write *a = (struct archive_write *)_a;
88 	int r = archive_write_set_format_ar(a);
89 	if (r == ARCHIVE_OK) {
90 		a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD;
91 		a->archive.archive_format_name = "ar (BSD)";
92 	}
93 	return (r);
94 }
95 
96 int
97 archive_write_set_format_ar_svr4(struct archive *_a)
98 {
99 	struct archive_write *a = (struct archive_write *)_a;
100 	int r = archive_write_set_format_ar(a);
101 	if (r == ARCHIVE_OK) {
102 		a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU;
103 		a->archive.archive_format_name = "ar (GNU/SVR4)";
104 	}
105 	return (r);
106 }
107 
108 /*
109  * Generic initialization.
110  */
111 static int
112 archive_write_set_format_ar(struct archive_write *a)
113 {
114 	struct ar_w *ar;
115 
116 	/* If someone else was already registered, unregister them. */
117 	if (a->format_destroy != NULL)
118 		(a->format_destroy)(a);
119 
120 	ar = (struct ar_w *)malloc(sizeof(*ar));
121 	if (ar == NULL) {
122 		archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data");
123 		return (ARCHIVE_FATAL);
124 	}
125 	memset(ar, 0, sizeof(*ar));
126 	a->format_data = ar;
127 
128 	a->format_write_header = archive_write_ar_header;
129 	a->format_write_data = archive_write_ar_data;
130 	a->format_finish = archive_write_ar_finish;
131 	a->format_destroy = archive_write_ar_destroy;
132 	a->format_finish_entry = archive_write_ar_finish_entry;
133 	return (ARCHIVE_OK);
134 }
135 
136 static int
137 archive_write_ar_header(struct archive_write *a, struct archive_entry *entry)
138 {
139 	int ret, append_fn;
140 	char buff[60];
141 	char *ss, *se;
142 	struct ar_w *ar;
143 	const char *pathname;
144 	const char *filename;
145 	int64_t size;
146 
147 	ret = 0;
148 	append_fn = 0;
149 	ar = (struct ar_w *)a->format_data;
150 	ar->is_strtab = 0;
151 	filename = NULL;
152 	size = archive_entry_size(entry);
153 
154 
155 	/*
156 	 * Reject files with empty name.
157 	 */
158 	pathname = archive_entry_pathname(entry);
159 	if (*pathname == '\0') {
160 		archive_set_error(&a->archive, EINVAL,
161 		    "Invalid filename");
162 		return (ARCHIVE_WARN);
163 	}
164 
165 	/*
166 	 * If we are now at the beginning of the archive,
167 	 * we need first write the ar global header.
168 	 */
169 	if (a->archive.file_position == 0)
170 		(a->compressor.write)(a, "!<arch>\n", 8);
171 
172 	memset(buff, ' ', 60);
173 	strncpy(&buff[AR_fmag_offset], "`\n", 2);
174 
175 	if (strcmp(pathname, "/") == 0 ) {
176 		/* Entry is archive symbol table in GNU format */
177 		buff[AR_name_offset] = '/';
178 		goto stat;
179 	}
180 	if (strcmp(pathname, "__.SYMDEF") == 0) {
181 		/* Entry is archive symbol table in BSD format */
182 		strncpy(buff + AR_name_offset, "__.SYMDEF", 9);
183 		goto stat;
184 	}
185 	if (strcmp(pathname, "//") == 0) {
186 		/*
187 		 * Entry is archive filename table, inform that we should
188 		 * collect strtab in next _data call.
189 		 */
190 		ar->is_strtab = 1;
191 		buff[AR_name_offset] = buff[AR_name_offset + 1] = '/';
192 		/*
193 		 * For archive string table, only ar_size filed should
194 		 * be set.
195 		 */
196 		goto size;
197 	}
198 
199 	/*
200 	 * Otherwise, entry is a normal archive member.
201 	 * Strip leading paths from filenames, if any.
202 	 */
203 	if ((filename = ar_basename(pathname)) == NULL) {
204 		/* Reject filenames with trailing "/" */
205 		archive_set_error(&a->archive, EINVAL,
206 		    "Invalid filename");
207 		return (ARCHIVE_WARN);
208 	}
209 
210 	if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) {
211 		/*
212 		 * SVR4/GNU variant use a "/" to mark then end of the filename,
213 		 * make it possible to have embedded spaces in the filename.
214 		 * So, the longest filename here (without extension) is
215 		 * actually 15 bytes.
216 		 */
217 		if (strlen(filename) <= 15) {
218 			strncpy(&buff[AR_name_offset],
219 			    filename, strlen(filename));
220 			buff[AR_name_offset + strlen(filename)] = '/';
221 		} else {
222 			/*
223 			 * For filename longer than 15 bytes, GNU variant
224 			 * makes use of a string table and instead stores the
225 			 * offset of the real filename to in the ar_name field.
226 			 * The string table should have been written before.
227 			 */
228 			if (ar->has_strtab <= 0) {
229 				archive_set_error(&a->archive, EINVAL,
230 				    "Can't find string table");
231 				return (ARCHIVE_WARN);
232 			}
233 
234 			se = (char *)malloc(strlen(filename) + 3);
235 			if (se == NULL) {
236 				archive_set_error(&a->archive, ENOMEM,
237 				    "Can't allocate filename buffer");
238 				return (ARCHIVE_FATAL);
239 			}
240 
241 			strncpy(se, filename, strlen(filename));
242 			strcpy(se + strlen(filename), "/\n");
243 
244 			ss = strstr(ar->strtab, se);
245 			free(se);
246 
247 			if (ss == NULL) {
248 				archive_set_error(&a->archive, EINVAL,
249 				    "Invalid string table");
250 				return (ARCHIVE_WARN);
251 			}
252 
253 			/*
254 			 * GNU variant puts "/" followed by digits into
255 			 * ar_name field. These digits indicates the real
256 			 * filename string's offset to the string table.
257 			 */
258 			buff[AR_name_offset] = '/';
259 			if (format_decimal(ss - ar->strtab,
260 			    buff + AR_name_offset + 1,
261 			    AR_name_size - 1)) {
262 				archive_set_error(&a->archive, ERANGE,
263 				    "string table offset too large");
264 				return (ARCHIVE_WARN);
265 			}
266 		}
267 	} else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) {
268 		/*
269 		 * BSD variant: for any file name which is more than
270 		 * 16 chars or contains one or more embedded space(s), the
271 		 * string "#1/" followed by the ASCII length of the name is
272 		 * put into the ar_name field. The file size (stored in the
273 		 * ar_size field) is incremented by the length of the name.
274 		 * The name is then written immediately following the
275 		 * archive header.
276 		 */
277 		if (strlen(filename) <= 16 && strchr(filename, ' ') == NULL) {
278 			strncpy(&buff[AR_name_offset], filename, strlen(filename));
279 			buff[AR_name_offset + strlen(filename)] = ' ';
280 		}
281 		else {
282 			strncpy(buff + AR_name_offset, "#1/", 3);
283 			if (format_decimal(strlen(filename),
284 			    buff + AR_name_offset + 3,
285 			    AR_name_size - 3)) {
286 				archive_set_error(&a->archive, ERANGE,
287 				    "File name too long");
288 				return (ARCHIVE_WARN);
289 			}
290 			append_fn = 1;
291 			size += strlen(filename);
292 		}
293 	}
294 
295 stat:
296 	if (format_decimal(archive_entry_mtime(entry), buff + AR_date_offset, AR_date_size)) {
297 		archive_set_error(&a->archive, ERANGE,
298 		    "File modification time too large");
299 		return (ARCHIVE_WARN);
300 	}
301 	if (format_decimal(archive_entry_uid(entry), buff + AR_uid_offset, AR_uid_size)) {
302 		archive_set_error(&a->archive, ERANGE,
303 		    "Numeric user ID too large");
304 		return (ARCHIVE_WARN);
305 	}
306 	if (format_decimal(archive_entry_gid(entry), buff + AR_gid_offset, AR_gid_size)) {
307 		archive_set_error(&a->archive, ERANGE,
308 		    "Numeric group ID too large");
309 		return (ARCHIVE_WARN);
310 	}
311 	if (format_octal(archive_entry_mode(entry), buff + AR_mode_offset, AR_mode_size)) {
312 		archive_set_error(&a->archive, ERANGE,
313 		    "Numeric mode too large");
314 		return (ARCHIVE_WARN);
315 	}
316 	/*
317 	 * Sanity Check: A non-pseudo archive member should always be
318 	 * a regular file.
319 	 */
320 	if (filename != NULL && archive_entry_filetype(entry) != AE_IFREG) {
321 		archive_set_error(&a->archive, EINVAL,
322 		    "Regular file required for non-pseudo member");
323 		return (ARCHIVE_WARN);
324 	}
325 
326 size:
327 	if (format_decimal(size, buff + AR_size_offset, AR_size_size)) {
328 		archive_set_error(&a->archive, ERANGE,
329 		    "File size out of range");
330 		return (ARCHIVE_WARN);
331 	}
332 
333 	ret = (a->compressor.write)(a, buff, 60);
334 	if (ret != ARCHIVE_OK)
335 		return (ret);
336 
337 	ar->entry_bytes_remaining = size;
338 	ar->entry_padding = ar->entry_bytes_remaining % 2;
339 
340 	if (append_fn > 0) {
341 		ret = (a->compressor.write)(a, filename, strlen(filename));
342 		if (ret != ARCHIVE_OK)
343 			return (ret);
344 		ar->entry_bytes_remaining -= strlen(filename);
345 	}
346 
347 	return (ARCHIVE_OK);
348 }
349 
350 static ssize_t
351 archive_write_ar_data(struct archive_write *a, const void *buff, size_t s)
352 {
353 	struct ar_w *ar;
354 	int ret;
355 
356 	ar = (struct ar_w *)a->format_data;
357 	if (s > ar->entry_bytes_remaining)
358 		s = ar->entry_bytes_remaining;
359 
360 	if (ar->is_strtab > 0) {
361 		if (ar->has_strtab > 0) {
362 			archive_set_error(&a->archive, EINVAL,
363 			    "More than one string tables exist");
364 			return (ARCHIVE_WARN);
365 		}
366 
367 		ar->strtab = (char *)malloc(s);
368 		if (ar->strtab == NULL) {
369 			archive_set_error(&a->archive, ENOMEM,
370 			    "Can't allocate strtab buffer");
371 			return (ARCHIVE_FATAL);
372 		}
373 		strncpy(ar->strtab, buff, s);
374 		ar->has_strtab = 1;
375 	}
376 
377 	ret = (a->compressor.write)(a, buff, s);
378 	if (ret != ARCHIVE_OK)
379 		return (ret);
380 
381 	ar->entry_bytes_remaining -= s;
382 	return (s);
383 }
384 
385 static int
386 archive_write_ar_destroy(struct archive_write *a)
387 {
388 	struct ar_w *ar;
389 
390 	ar = (struct ar_w *)a->format_data;
391 
392 	if (ar->has_strtab > 0) {
393 		free(ar->strtab);
394 		ar->strtab = NULL;
395 	}
396 
397 	free(ar);
398 	a->format_data = NULL;
399 	return (ARCHIVE_OK);
400 }
401 
402 static int
403 archive_write_ar_finish(struct archive_write *a)
404 {
405 	int ret;
406 
407 	/*
408 	 * If we haven't written anything yet, we need to write
409 	 * the ar global header now to make it a valid ar archive.
410 	 */
411 	if (a->archive.file_position == 0) {
412 		ret = (a->compressor.write)(a, "!<arch>\n", 8);
413 		return (ret);
414 	}
415 
416 	return (ARCHIVE_OK);
417 }
418 
419 static int
420 archive_write_ar_finish_entry(struct archive_write *a)
421 {
422 	struct ar_w *ar;
423 	int ret;
424 
425 	ar = (struct ar_w *)a->format_data;
426 
427 	if (ar->entry_bytes_remaining != 0) {
428 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
429 		    "Entry remaining bytes larger than 0");
430 		return (ARCHIVE_WARN);
431 	}
432 
433 	if (ar->entry_padding == 0) {
434 		return (ARCHIVE_OK);
435 	}
436 
437 	if (ar->entry_padding != 1) {
438 		archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
439 		    "Padding wrong size: %d should be 1 or 0",
440 		    ar->entry_padding);
441 		return (ARCHIVE_WARN);
442 	}
443 
444 	ret = (a->compressor.write)(a, "\n", 1);
445 	return (ret);
446 }
447 
448 /*
449  * Format a number into the specified field using base-8.
450  * NB: This version is slightly different from the one in
451  * _ustar.c
452  */
453 static int
454 format_octal(int64_t v, char *p, int s)
455 {
456 	int len;
457 	char *h;
458 
459 	len = s;
460 	h = p;
461 
462 	/* Octal values can't be negative, so use 0. */
463 	if (v < 0) {
464 		while (len-- > 0)
465 			*p++ = '0';
466 		return (-1);
467 	}
468 
469 	p += s;		/* Start at the end and work backwards. */
470 	do {
471 		*--p = (char)('0' + (v & 7));
472 		v >>= 3;
473 	} while (--s > 0 && v > 0);
474 
475 	if (v == 0) {
476 		memmove(h, p, len - s);
477 		p = h + len - s;
478 		while (s-- > 0)
479 			*p++ = ' ';
480 		return (0);
481 	}
482 	/* If it overflowed, fill field with max value. */
483 	while (len-- > 0)
484 		*p++ = '7';
485 
486 	return (-1);
487 }
488 
489 /*
490  * Format a number into the specified field using base-10.
491  */
492 static int
493 format_decimal(int64_t v, char *p, int s)
494 {
495 	int len;
496 	char *h;
497 
498 	len = s;
499 	h = p;
500 
501 	/* Negative values in ar header are meaningless , so use 0. */
502 	if (v < 0) {
503 		while (len-- > 0)
504 			*p++ = '0';
505 		return (-1);
506 	}
507 
508 	p += s;
509 	do {
510 		*--p = (char)('0' + (v % 10));
511 		v /= 10;
512 	} while (--s > 0 && v > 0);
513 
514 	if (v == 0) {
515 		memmove(h, p, len - s);
516 		p = h + len - s;
517 		while (s-- > 0)
518 			*p++ = ' ';
519 		return (0);
520 	}
521 	/* If it overflowed, fill field with max value. */
522 	while (len-- > 0)
523 		*p++ = '9';
524 
525 	return (-1);
526 }
527 
528 static const char *
529 ar_basename(const char *path)
530 {
531 	const char *endp, *startp;
532 
533 	endp = path + strlen(path) - 1;
534 	/*
535 	 * For filename with trailing slash(es), we return
536 	 * NULL indicating an error.
537 	 */
538 	if (*endp == '/')
539 		return (NULL);
540 
541 	/* Find the start of the base */
542 	startp = endp;
543 	while (startp > path && *(startp - 1) != '/')
544 		startp--;
545 
546 	return (startp);
547 }
548