xref: /freebsd/contrib/libarchive/tar/util.c (revision 6c95142e7977d720e7f1c70672b358aa09fa6bfa)
1caf54c4fSMartin Matuska /*-
2caf54c4fSMartin Matuska  * Copyright (c) 2003-2007 Tim Kientzle
3caf54c4fSMartin Matuska  * All rights reserved.
4caf54c4fSMartin Matuska  *
5caf54c4fSMartin Matuska  * Redistribution and use in source and binary forms, with or without
6caf54c4fSMartin Matuska  * modification, are permitted provided that the following conditions
7caf54c4fSMartin Matuska  * are met:
8caf54c4fSMartin Matuska  * 1. Redistributions of source code must retain the above copyright
9caf54c4fSMartin Matuska  *    notice, this list of conditions and the following disclaimer.
10caf54c4fSMartin Matuska  * 2. Redistributions in binary form must reproduce the above copyright
11caf54c4fSMartin Matuska  *    notice, this list of conditions and the following disclaimer in the
12caf54c4fSMartin Matuska  *    documentation and/or other materials provided with the distribution.
13caf54c4fSMartin Matuska  *
14caf54c4fSMartin Matuska  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
15caf54c4fSMartin Matuska  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16caf54c4fSMartin Matuska  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17caf54c4fSMartin Matuska  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18caf54c4fSMartin Matuska  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19caf54c4fSMartin Matuska  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20caf54c4fSMartin Matuska  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21caf54c4fSMartin Matuska  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22caf54c4fSMartin Matuska  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23caf54c4fSMartin Matuska  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24caf54c4fSMartin Matuska  */
25caf54c4fSMartin Matuska 
26caf54c4fSMartin Matuska #include "bsdtar_platform.h"
276c22d9efSMartin Matuska __FBSDID("$FreeBSD$");
28caf54c4fSMartin Matuska 
29caf54c4fSMartin Matuska #ifdef HAVE_SYS_STAT_H
30caf54c4fSMartin Matuska #include <sys/stat.h>
31caf54c4fSMartin Matuska #endif
32caf54c4fSMartin Matuska #ifdef HAVE_SYS_TYPES_H
33caf54c4fSMartin Matuska #include <sys/types.h>  /* Linux doesn't define mode_t, etc. in sys/stat.h. */
34caf54c4fSMartin Matuska #endif
35caf54c4fSMartin Matuska #include <ctype.h>
36caf54c4fSMartin Matuska #ifdef HAVE_ERRNO_H
37caf54c4fSMartin Matuska #include <errno.h>
38caf54c4fSMartin Matuska #endif
39caf54c4fSMartin Matuska #ifdef HAVE_IO_H
40caf54c4fSMartin Matuska #include <io.h>
41caf54c4fSMartin Matuska #endif
42caf54c4fSMartin Matuska #ifdef HAVE_STDARG_H
43caf54c4fSMartin Matuska #include <stdarg.h>
44caf54c4fSMartin Matuska #endif
45caf54c4fSMartin Matuska #ifdef HAVE_STDINT_H
46caf54c4fSMartin Matuska #include <stdint.h>
47caf54c4fSMartin Matuska #endif
48caf54c4fSMartin Matuska #include <stdio.h>
49caf54c4fSMartin Matuska #ifdef HAVE_STDLIB_H
50caf54c4fSMartin Matuska #include <stdlib.h>
51caf54c4fSMartin Matuska #endif
52caf54c4fSMartin Matuska #ifdef HAVE_STRING_H
53caf54c4fSMartin Matuska #include <string.h>
54caf54c4fSMartin Matuska #endif
55caf54c4fSMartin Matuska #ifdef HAVE_WCTYPE_H
56caf54c4fSMartin Matuska #include <wctype.h>
57caf54c4fSMartin Matuska #else
58caf54c4fSMartin Matuska /* If we don't have wctype, we need to hack up some version of iswprint(). */
59caf54c4fSMartin Matuska #define	iswprint isprint
60caf54c4fSMartin Matuska #endif
61caf54c4fSMartin Matuska 
62caf54c4fSMartin Matuska #include "bsdtar.h"
63caf54c4fSMartin Matuska #include "err.h"
64caf54c4fSMartin Matuska 
65caf54c4fSMartin Matuska static size_t	bsdtar_expand_char(char *, size_t, char);
66caf54c4fSMartin Matuska static const char *strip_components(const char *path, int elements);
67caf54c4fSMartin Matuska 
68caf54c4fSMartin Matuska #if defined(_WIN32) && !defined(__CYGWIN__)
69caf54c4fSMartin Matuska #define	read _read
70caf54c4fSMartin Matuska #endif
71caf54c4fSMartin Matuska 
72caf54c4fSMartin Matuska /* TODO:  Hack up a version of mbtowc for platforms with no wide
73caf54c4fSMartin Matuska  * character support at all.  I think the following might suffice,
74caf54c4fSMartin Matuska  * but it needs careful testing.
75caf54c4fSMartin Matuska  * #if !HAVE_MBTOWC
76caf54c4fSMartin Matuska  * #define	mbtowc(wcp, p, n) ((*wcp = *p), 1)
77caf54c4fSMartin Matuska  * #endif
78caf54c4fSMartin Matuska  */
79caf54c4fSMartin Matuska 
80caf54c4fSMartin Matuska /*
81caf54c4fSMartin Matuska  * Print a string, taking care with any non-printable characters.
82caf54c4fSMartin Matuska  *
83caf54c4fSMartin Matuska  * Note that we use a stack-allocated buffer to receive the formatted
84caf54c4fSMartin Matuska  * string if we can.  This is partly performance (avoiding a call to
85caf54c4fSMartin Matuska  * malloc()), partly out of expedience (we have to call vsnprintf()
86caf54c4fSMartin Matuska  * before malloc() anyway to find out how big a buffer we need; we may
87caf54c4fSMartin Matuska  * as well point that first call at a small local buffer in case it
88caf54c4fSMartin Matuska  * works), but mostly for safety (so we can use this to print messages
89caf54c4fSMartin Matuska  * about out-of-memory conditions).
90caf54c4fSMartin Matuska  */
91caf54c4fSMartin Matuska 
92caf54c4fSMartin Matuska void
93caf54c4fSMartin Matuska safe_fprintf(FILE *f, const char *fmt, ...)
94caf54c4fSMartin Matuska {
95caf54c4fSMartin Matuska 	char fmtbuff_stack[256]; /* Place to format the printf() string. */
96caf54c4fSMartin Matuska 	char outbuff[256]; /* Buffer for outgoing characters. */
97caf54c4fSMartin Matuska 	char *fmtbuff_heap; /* If fmtbuff_stack is too small, we use malloc */
98caf54c4fSMartin Matuska 	char *fmtbuff;  /* Pointer to fmtbuff_stack or fmtbuff_heap. */
99caf54c4fSMartin Matuska 	int fmtbuff_length;
100caf54c4fSMartin Matuska 	int length, n;
101caf54c4fSMartin Matuska 	va_list ap;
102caf54c4fSMartin Matuska 	const char *p;
103caf54c4fSMartin Matuska 	unsigned i;
104caf54c4fSMartin Matuska 	wchar_t wc;
105caf54c4fSMartin Matuska 	char try_wc;
106caf54c4fSMartin Matuska 
107caf54c4fSMartin Matuska 	/* Use a stack-allocated buffer if we can, for speed and safety. */
108caf54c4fSMartin Matuska 	fmtbuff_heap = NULL;
109caf54c4fSMartin Matuska 	fmtbuff_length = sizeof(fmtbuff_stack);
110caf54c4fSMartin Matuska 	fmtbuff = fmtbuff_stack;
111caf54c4fSMartin Matuska 
112caf54c4fSMartin Matuska 	/* Try formatting into the stack buffer. */
113caf54c4fSMartin Matuska 	va_start(ap, fmt);
114caf54c4fSMartin Matuska 	length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap);
115caf54c4fSMartin Matuska 	va_end(ap);
116caf54c4fSMartin Matuska 
117caf54c4fSMartin Matuska 	/* If the result was too large, allocate a buffer on the heap. */
118*6c95142eSMartin Matuska 	while (length < 0 || length >= fmtbuff_length) {
119*6c95142eSMartin Matuska 		if (length >= fmtbuff_length)
120caf54c4fSMartin Matuska 			fmtbuff_length = length+1;
121*6c95142eSMartin Matuska 		else if (fmtbuff_length < 8192)
122*6c95142eSMartin Matuska 			fmtbuff_length *= 2;
123*6c95142eSMartin Matuska 		else {
124*6c95142eSMartin Matuska 			int old_length = fmtbuff_length;
125*6c95142eSMartin Matuska 			fmtbuff_length += fmtbuff_length / 4;
126*6c95142eSMartin Matuska 			if (old_length > fmtbuff_length) {
127*6c95142eSMartin Matuska 				length = old_length;
128*6c95142eSMartin Matuska 				fmtbuff_heap[length-1] = '\0';
129*6c95142eSMartin Matuska 				break;
130*6c95142eSMartin Matuska 			}
131*6c95142eSMartin Matuska 		}
132*6c95142eSMartin Matuska 		free(fmtbuff_heap);
133caf54c4fSMartin Matuska 		fmtbuff_heap = malloc(fmtbuff_length);
134caf54c4fSMartin Matuska 
135caf54c4fSMartin Matuska 		/* Reformat the result into the heap buffer if we can. */
136caf54c4fSMartin Matuska 		if (fmtbuff_heap != NULL) {
137caf54c4fSMartin Matuska 			fmtbuff = fmtbuff_heap;
138caf54c4fSMartin Matuska 			va_start(ap, fmt);
139caf54c4fSMartin Matuska 			length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap);
140caf54c4fSMartin Matuska 			va_end(ap);
141caf54c4fSMartin Matuska 		} else {
142caf54c4fSMartin Matuska 			/* Leave fmtbuff pointing to the truncated
143caf54c4fSMartin Matuska 			 * string in fmtbuff_stack. */
144caf54c4fSMartin Matuska 			length = sizeof(fmtbuff_stack) - 1;
145*6c95142eSMartin Matuska 			break;
146caf54c4fSMartin Matuska 		}
147caf54c4fSMartin Matuska 	}
148caf54c4fSMartin Matuska 
149caf54c4fSMartin Matuska 	/* Note: mbrtowc() has a cleaner API, but mbtowc() seems a bit
150caf54c4fSMartin Matuska 	 * more portable, so we use that here instead. */
151caf54c4fSMartin Matuska 	n = mbtowc(NULL, NULL, 1); /* Reset the shift state. */
152caf54c4fSMartin Matuska 
153caf54c4fSMartin Matuska 	/* Write data, expanding unprintable characters. */
154caf54c4fSMartin Matuska 	p = fmtbuff;
155caf54c4fSMartin Matuska 	i = 0;
156caf54c4fSMartin Matuska 	try_wc = 1;
157caf54c4fSMartin Matuska 	while (*p != '\0') {
158caf54c4fSMartin Matuska 
159caf54c4fSMartin Matuska 		/* Convert to wide char, test if the wide
160caf54c4fSMartin Matuska 		 * char is printable in the current locale. */
161caf54c4fSMartin Matuska 		if (try_wc && (n = mbtowc(&wc, p, length)) != -1) {
162caf54c4fSMartin Matuska 			length -= n;
163caf54c4fSMartin Matuska 			if (iswprint(wc) && wc != L'\\') {
164caf54c4fSMartin Matuska 				/* Printable, copy the bytes through. */
165caf54c4fSMartin Matuska 				while (n-- > 0)
166caf54c4fSMartin Matuska 					outbuff[i++] = *p++;
167caf54c4fSMartin Matuska 			} else {
168caf54c4fSMartin Matuska 				/* Not printable, format the bytes. */
169caf54c4fSMartin Matuska 				while (n-- > 0)
170caf54c4fSMartin Matuska 					i += (unsigned)bsdtar_expand_char(
171caf54c4fSMartin Matuska 					    outbuff, i, *p++);
172caf54c4fSMartin Matuska 			}
173caf54c4fSMartin Matuska 		} else {
174caf54c4fSMartin Matuska 			/* After any conversion failure, don't bother
175caf54c4fSMartin Matuska 			 * trying to convert the rest. */
176caf54c4fSMartin Matuska 			i += (unsigned)bsdtar_expand_char(outbuff, i, *p++);
177caf54c4fSMartin Matuska 			try_wc = 0;
178caf54c4fSMartin Matuska 		}
179caf54c4fSMartin Matuska 
180caf54c4fSMartin Matuska 		/* If our output buffer is full, dump it and keep going. */
181caf54c4fSMartin Matuska 		if (i > (sizeof(outbuff) - 20)) {
182caf54c4fSMartin Matuska 			outbuff[i] = '\0';
183caf54c4fSMartin Matuska 			fprintf(f, "%s", outbuff);
184caf54c4fSMartin Matuska 			i = 0;
185caf54c4fSMartin Matuska 		}
186caf54c4fSMartin Matuska 	}
187caf54c4fSMartin Matuska 	outbuff[i] = '\0';
188caf54c4fSMartin Matuska 	fprintf(f, "%s", outbuff);
189caf54c4fSMartin Matuska 
190caf54c4fSMartin Matuska 	/* If we allocated a heap-based formatting buffer, free it now. */
191caf54c4fSMartin Matuska 	if (fmtbuff_heap != NULL)
192caf54c4fSMartin Matuska 		free(fmtbuff_heap);
193caf54c4fSMartin Matuska }
194caf54c4fSMartin Matuska 
195caf54c4fSMartin Matuska /*
196caf54c4fSMartin Matuska  * Render an arbitrary sequence of bytes into printable ASCII characters.
197caf54c4fSMartin Matuska  */
198caf54c4fSMartin Matuska static size_t
199caf54c4fSMartin Matuska bsdtar_expand_char(char *buff, size_t offset, char c)
200caf54c4fSMartin Matuska {
201caf54c4fSMartin Matuska 	size_t i = offset;
202caf54c4fSMartin Matuska 
203caf54c4fSMartin Matuska 	if (isprint((unsigned char)c) && c != '\\')
204caf54c4fSMartin Matuska 		buff[i++] = c;
205caf54c4fSMartin Matuska 	else {
206caf54c4fSMartin Matuska 		buff[i++] = '\\';
207caf54c4fSMartin Matuska 		switch (c) {
208caf54c4fSMartin Matuska 		case '\a': buff[i++] = 'a'; break;
209caf54c4fSMartin Matuska 		case '\b': buff[i++] = 'b'; break;
210caf54c4fSMartin Matuska 		case '\f': buff[i++] = 'f'; break;
211caf54c4fSMartin Matuska 		case '\n': buff[i++] = 'n'; break;
212caf54c4fSMartin Matuska #if '\r' != '\n'
213caf54c4fSMartin Matuska 		/* On some platforms, \n and \r are the same. */
214caf54c4fSMartin Matuska 		case '\r': buff[i++] = 'r'; break;
215caf54c4fSMartin Matuska #endif
216caf54c4fSMartin Matuska 		case '\t': buff[i++] = 't'; break;
217caf54c4fSMartin Matuska 		case '\v': buff[i++] = 'v'; break;
218caf54c4fSMartin Matuska 		case '\\': buff[i++] = '\\'; break;
219caf54c4fSMartin Matuska 		default:
220caf54c4fSMartin Matuska 			sprintf(buff + i, "%03o", 0xFF & (int)c);
221caf54c4fSMartin Matuska 			i += 3;
222caf54c4fSMartin Matuska 		}
223caf54c4fSMartin Matuska 	}
224caf54c4fSMartin Matuska 
225caf54c4fSMartin Matuska 	return (i - offset);
226caf54c4fSMartin Matuska }
227caf54c4fSMartin Matuska 
228caf54c4fSMartin Matuska int
229caf54c4fSMartin Matuska yes(const char *fmt, ...)
230caf54c4fSMartin Matuska {
231caf54c4fSMartin Matuska 	char buff[32];
232caf54c4fSMartin Matuska 	char *p;
233caf54c4fSMartin Matuska 	ssize_t l;
234caf54c4fSMartin Matuska 
235caf54c4fSMartin Matuska 	va_list ap;
236caf54c4fSMartin Matuska 	va_start(ap, fmt);
237caf54c4fSMartin Matuska 	vfprintf(stderr, fmt, ap);
238caf54c4fSMartin Matuska 	va_end(ap);
239caf54c4fSMartin Matuska 	fprintf(stderr, " (y/N)? ");
240caf54c4fSMartin Matuska 	fflush(stderr);
241caf54c4fSMartin Matuska 
242caf54c4fSMartin Matuska 	l = read(2, buff, sizeof(buff) - 1);
243c5d360f8SMartin Matuska 	if (l < 0) {
244c5d360f8SMartin Matuska 	  fprintf(stderr, "Keyboard read failed\n");
245c5d360f8SMartin Matuska 	  exit(1);
246c5d360f8SMartin Matuska 	}
247c5d360f8SMartin Matuska 	if (l == 0)
248caf54c4fSMartin Matuska 		return (0);
249caf54c4fSMartin Matuska 	buff[l] = 0;
250caf54c4fSMartin Matuska 
251caf54c4fSMartin Matuska 	for (p = buff; *p != '\0'; p++) {
252caf54c4fSMartin Matuska 		if (isspace((unsigned char)*p))
253caf54c4fSMartin Matuska 			continue;
254caf54c4fSMartin Matuska 		switch(*p) {
255caf54c4fSMartin Matuska 		case 'y': case 'Y':
256caf54c4fSMartin Matuska 			return (1);
257caf54c4fSMartin Matuska 		case 'n': case 'N':
258caf54c4fSMartin Matuska 			return (0);
259caf54c4fSMartin Matuska 		default:
260caf54c4fSMartin Matuska 			return (0);
261caf54c4fSMartin Matuska 		}
262caf54c4fSMartin Matuska 	}
263caf54c4fSMartin Matuska 
264caf54c4fSMartin Matuska 	return (0);
265caf54c4fSMartin Matuska }
266caf54c4fSMartin Matuska 
267caf54c4fSMartin Matuska /*-
268caf54c4fSMartin Matuska  * The logic here for -C <dir> attempts to avoid
269caf54c4fSMartin Matuska  * chdir() as long as possible.  For example:
270caf54c4fSMartin Matuska  * "-C /foo -C /bar file"          needs chdir("/bar") but not chdir("/foo")
271caf54c4fSMartin Matuska  * "-C /foo -C bar file"           needs chdir("/foo/bar")
272caf54c4fSMartin Matuska  * "-C /foo -C bar /file1"         does not need chdir()
273caf54c4fSMartin Matuska  * "-C /foo -C bar /file1 file2"   needs chdir("/foo/bar") before file2
274caf54c4fSMartin Matuska  *
275caf54c4fSMartin Matuska  * The only correct way to handle this is to record a "pending" chdir
276caf54c4fSMartin Matuska  * request and combine multiple requests intelligently until we
277caf54c4fSMartin Matuska  * need to process a non-absolute file.  set_chdir() adds the new dir
278caf54c4fSMartin Matuska  * to the pending list; do_chdir() actually executes any pending chdir.
279caf54c4fSMartin Matuska  *
280caf54c4fSMartin Matuska  * This way, programs that build tar command lines don't have to worry
281caf54c4fSMartin Matuska  * about -C with non-existent directories; such requests will only
282caf54c4fSMartin Matuska  * fail if the directory must be accessed.
283caf54c4fSMartin Matuska  *
284caf54c4fSMartin Matuska  */
285caf54c4fSMartin Matuska void
286caf54c4fSMartin Matuska set_chdir(struct bsdtar *bsdtar, const char *newdir)
287caf54c4fSMartin Matuska {
288*6c95142eSMartin Matuska #if defined(_WIN32) && !defined(__CYGWIN__)
289*6c95142eSMartin Matuska 	if (newdir[0] == '/' || newdir[0] == '\\' ||
290*6c95142eSMartin Matuska 	    /* Detect this type, for example, "C:\" or "C:/" */
291*6c95142eSMartin Matuska 	    (((newdir[0] >= 'a' && newdir[0] <= 'z') ||
292*6c95142eSMartin Matuska 	      (newdir[0] >= 'A' && newdir[0] <= 'Z')) &&
293*6c95142eSMartin Matuska 	    newdir[1] == ':' && (newdir[2] == '/' || newdir[2] == '\\'))) {
294*6c95142eSMartin Matuska #else
295caf54c4fSMartin Matuska 	if (newdir[0] == '/') {
296*6c95142eSMartin Matuska #endif
297caf54c4fSMartin Matuska 		/* The -C /foo -C /bar case; dump first one. */
298caf54c4fSMartin Matuska 		free(bsdtar->pending_chdir);
299caf54c4fSMartin Matuska 		bsdtar->pending_chdir = NULL;
300caf54c4fSMartin Matuska 	}
301caf54c4fSMartin Matuska 	if (bsdtar->pending_chdir == NULL)
302caf54c4fSMartin Matuska 		/* Easy case: no previously-saved dir. */
303caf54c4fSMartin Matuska 		bsdtar->pending_chdir = strdup(newdir);
304caf54c4fSMartin Matuska 	else {
305caf54c4fSMartin Matuska 		/* The -C /foo -C bar case; concatenate */
306caf54c4fSMartin Matuska 		char *old_pending = bsdtar->pending_chdir;
307caf54c4fSMartin Matuska 		size_t old_len = strlen(old_pending);
308caf54c4fSMartin Matuska 		bsdtar->pending_chdir = malloc(old_len + strlen(newdir) + 2);
309caf54c4fSMartin Matuska 		if (old_pending[old_len - 1] == '/')
310caf54c4fSMartin Matuska 			old_pending[old_len - 1] = '\0';
311caf54c4fSMartin Matuska 		if (bsdtar->pending_chdir != NULL)
312caf54c4fSMartin Matuska 			sprintf(bsdtar->pending_chdir, "%s/%s",
313caf54c4fSMartin Matuska 			    old_pending, newdir);
314caf54c4fSMartin Matuska 		free(old_pending);
315caf54c4fSMartin Matuska 	}
316caf54c4fSMartin Matuska 	if (bsdtar->pending_chdir == NULL)
317caf54c4fSMartin Matuska 		lafe_errc(1, errno, "No memory");
318caf54c4fSMartin Matuska }
319caf54c4fSMartin Matuska 
320caf54c4fSMartin Matuska void
321caf54c4fSMartin Matuska do_chdir(struct bsdtar *bsdtar)
322caf54c4fSMartin Matuska {
323caf54c4fSMartin Matuska 	if (bsdtar->pending_chdir == NULL)
324caf54c4fSMartin Matuska 		return;
325caf54c4fSMartin Matuska 
326caf54c4fSMartin Matuska 	if (chdir(bsdtar->pending_chdir) != 0) {
327caf54c4fSMartin Matuska 		lafe_errc(1, 0, "could not chdir to '%s'\n",
328caf54c4fSMartin Matuska 		    bsdtar->pending_chdir);
329caf54c4fSMartin Matuska 	}
330caf54c4fSMartin Matuska 	free(bsdtar->pending_chdir);
331caf54c4fSMartin Matuska 	bsdtar->pending_chdir = NULL;
332caf54c4fSMartin Matuska }
333caf54c4fSMartin Matuska 
334caf54c4fSMartin Matuska static const char *
335caf54c4fSMartin Matuska strip_components(const char *p, int elements)
336caf54c4fSMartin Matuska {
337caf54c4fSMartin Matuska 	/* Skip as many elements as necessary. */
338caf54c4fSMartin Matuska 	while (elements > 0) {
339caf54c4fSMartin Matuska 		switch (*p++) {
340caf54c4fSMartin Matuska 		case '/':
341caf54c4fSMartin Matuska #if defined(_WIN32) && !defined(__CYGWIN__)
342caf54c4fSMartin Matuska 		case '\\': /* Support \ path sep on Windows ONLY. */
343caf54c4fSMartin Matuska #endif
344caf54c4fSMartin Matuska 			elements--;
345caf54c4fSMartin Matuska 			break;
346caf54c4fSMartin Matuska 		case '\0':
347caf54c4fSMartin Matuska 			/* Path is too short, skip it. */
348caf54c4fSMartin Matuska 			return (NULL);
349caf54c4fSMartin Matuska 		}
350caf54c4fSMartin Matuska 	}
351caf54c4fSMartin Matuska 
352caf54c4fSMartin Matuska 	/* Skip any / characters.  This handles short paths that have
353caf54c4fSMartin Matuska 	 * additional / termination.  This also handles the case where
354caf54c4fSMartin Matuska 	 * the logic above stops in the middle of a duplicate //
355caf54c4fSMartin Matuska 	 * sequence (which would otherwise get converted to an
356caf54c4fSMartin Matuska 	 * absolute path). */
357caf54c4fSMartin Matuska 	for (;;) {
358caf54c4fSMartin Matuska 		switch (*p) {
359caf54c4fSMartin Matuska 		case '/':
360caf54c4fSMartin Matuska #if defined(_WIN32) && !defined(__CYGWIN__)
361caf54c4fSMartin Matuska 		case '\\': /* Support \ path sep on Windows ONLY. */
362caf54c4fSMartin Matuska #endif
363caf54c4fSMartin Matuska 			++p;
364caf54c4fSMartin Matuska 			break;
365caf54c4fSMartin Matuska 		case '\0':
366caf54c4fSMartin Matuska 			return (NULL);
367caf54c4fSMartin Matuska 		default:
368caf54c4fSMartin Matuska 			return (p);
369caf54c4fSMartin Matuska 		}
370caf54c4fSMartin Matuska 	}
371caf54c4fSMartin Matuska }
372caf54c4fSMartin Matuska 
373caf54c4fSMartin Matuska /*
374caf54c4fSMartin Matuska  * Handle --strip-components and any future path-rewriting options.
375caf54c4fSMartin Matuska  * Returns non-zero if the pathname should not be extracted.
376caf54c4fSMartin Matuska  *
377caf54c4fSMartin Matuska  * TODO: Support pax-style regex path rewrites.
378caf54c4fSMartin Matuska  */
379caf54c4fSMartin Matuska int
380caf54c4fSMartin Matuska edit_pathname(struct bsdtar *bsdtar, struct archive_entry *entry)
381caf54c4fSMartin Matuska {
382caf54c4fSMartin Matuska 	const char *name = archive_entry_pathname(entry);
383caf54c4fSMartin Matuska #if HAVE_REGEX_H
384caf54c4fSMartin Matuska 	char *subst_name;
385caf54c4fSMartin Matuska 	int r;
386caf54c4fSMartin Matuska 
387*6c95142eSMartin Matuska 	r = apply_substitution(bsdtar, name, &subst_name, 0, 0);
388caf54c4fSMartin Matuska 	if (r == -1) {
389caf54c4fSMartin Matuska 		lafe_warnc(0, "Invalid substitution, skipping entry");
390caf54c4fSMartin Matuska 		return 1;
391caf54c4fSMartin Matuska 	}
392caf54c4fSMartin Matuska 	if (r == 1) {
393caf54c4fSMartin Matuska 		archive_entry_copy_pathname(entry, subst_name);
394caf54c4fSMartin Matuska 		if (*subst_name == '\0') {
395caf54c4fSMartin Matuska 			free(subst_name);
396caf54c4fSMartin Matuska 			return -1;
397caf54c4fSMartin Matuska 		} else
398caf54c4fSMartin Matuska 			free(subst_name);
399caf54c4fSMartin Matuska 		name = archive_entry_pathname(entry);
400caf54c4fSMartin Matuska 	}
401caf54c4fSMartin Matuska 
402caf54c4fSMartin Matuska 	if (archive_entry_hardlink(entry)) {
403*6c95142eSMartin Matuska 		r = apply_substitution(bsdtar, archive_entry_hardlink(entry), &subst_name, 0, 1);
404caf54c4fSMartin Matuska 		if (r == -1) {
405caf54c4fSMartin Matuska 			lafe_warnc(0, "Invalid substitution, skipping entry");
406caf54c4fSMartin Matuska 			return 1;
407caf54c4fSMartin Matuska 		}
408caf54c4fSMartin Matuska 		if (r == 1) {
409caf54c4fSMartin Matuska 			archive_entry_copy_hardlink(entry, subst_name);
410caf54c4fSMartin Matuska 			free(subst_name);
411caf54c4fSMartin Matuska 		}
412caf54c4fSMartin Matuska 	}
413caf54c4fSMartin Matuska 	if (archive_entry_symlink(entry) != NULL) {
414*6c95142eSMartin Matuska 		r = apply_substitution(bsdtar, archive_entry_symlink(entry), &subst_name, 1, 0);
415caf54c4fSMartin Matuska 		if (r == -1) {
416caf54c4fSMartin Matuska 			lafe_warnc(0, "Invalid substitution, skipping entry");
417caf54c4fSMartin Matuska 			return 1;
418caf54c4fSMartin Matuska 		}
419caf54c4fSMartin Matuska 		if (r == 1) {
420caf54c4fSMartin Matuska 			archive_entry_copy_symlink(entry, subst_name);
421caf54c4fSMartin Matuska 			free(subst_name);
422caf54c4fSMartin Matuska 		}
423caf54c4fSMartin Matuska 	}
424caf54c4fSMartin Matuska #endif
425caf54c4fSMartin Matuska 
426caf54c4fSMartin Matuska 	/* Strip leading dir names as per --strip-components option. */
427caf54c4fSMartin Matuska 	if (bsdtar->strip_components > 0) {
428caf54c4fSMartin Matuska 		const char *linkname = archive_entry_hardlink(entry);
429caf54c4fSMartin Matuska 
430caf54c4fSMartin Matuska 		name = strip_components(name, bsdtar->strip_components);
431caf54c4fSMartin Matuska 		if (name == NULL)
432caf54c4fSMartin Matuska 			return (1);
433caf54c4fSMartin Matuska 
434caf54c4fSMartin Matuska 		if (linkname != NULL) {
435caf54c4fSMartin Matuska 			linkname = strip_components(linkname,
436caf54c4fSMartin Matuska 			    bsdtar->strip_components);
437caf54c4fSMartin Matuska 			if (linkname == NULL)
438caf54c4fSMartin Matuska 				return (1);
439caf54c4fSMartin Matuska 			archive_entry_copy_hardlink(entry, linkname);
440caf54c4fSMartin Matuska 		}
441caf54c4fSMartin Matuska 	}
442caf54c4fSMartin Matuska 
443caf54c4fSMartin Matuska 	/* By default, don't write or restore absolute pathnames. */
444caf54c4fSMartin Matuska 	if (!bsdtar->option_absolute_paths) {
445caf54c4fSMartin Matuska 		const char *rp, *p = name;
446caf54c4fSMartin Matuska 		int slashonly = 1;
447caf54c4fSMartin Matuska 
448caf54c4fSMartin Matuska 		/* Remove leading "//./" or "//?/" or "//?/UNC/"
449caf54c4fSMartin Matuska 		 * (absolute path prefixes used by Windows API) */
450caf54c4fSMartin Matuska 		if ((p[0] == '/' || p[0] == '\\') &&
451caf54c4fSMartin Matuska 		    (p[1] == '/' || p[1] == '\\') &&
452caf54c4fSMartin Matuska 		    (p[2] == '.' || p[2] == '?') &&
453caf54c4fSMartin Matuska 		    (p[3] == '/' || p[3] == '\\'))
454caf54c4fSMartin Matuska 		{
455caf54c4fSMartin Matuska 			if (p[2] == '?' &&
456caf54c4fSMartin Matuska 			    (p[4] == 'U' || p[4] == 'u') &&
457caf54c4fSMartin Matuska 			    (p[5] == 'N' || p[5] == 'n') &&
458caf54c4fSMartin Matuska 			    (p[6] == 'C' || p[6] == 'c') &&
459caf54c4fSMartin Matuska 			    (p[7] == '/' || p[7] == '\\'))
460caf54c4fSMartin Matuska 				p += 8;
461caf54c4fSMartin Matuska 			else
462caf54c4fSMartin Matuska 				p += 4;
463caf54c4fSMartin Matuska 			slashonly = 0;
464caf54c4fSMartin Matuska 		}
465caf54c4fSMartin Matuska 		do {
466caf54c4fSMartin Matuska 			rp = p;
467caf54c4fSMartin Matuska 			/* Remove leading drive letter from archives created
468caf54c4fSMartin Matuska 			 * on Windows. */
469caf54c4fSMartin Matuska 			if (((p[0] >= 'a' && p[0] <= 'z') ||
470caf54c4fSMartin Matuska 			     (p[0] >= 'A' && p[0] <= 'Z')) &&
471caf54c4fSMartin Matuska 				 p[1] == ':') {
472caf54c4fSMartin Matuska 				p += 2;
473caf54c4fSMartin Matuska 				slashonly = 0;
474caf54c4fSMartin Matuska 			}
475caf54c4fSMartin Matuska 			/* Remove leading "/../", "//", etc. */
476caf54c4fSMartin Matuska 			while (p[0] == '/' || p[0] == '\\') {
477caf54c4fSMartin Matuska 				if (p[1] == '.' && p[2] == '.' &&
478caf54c4fSMartin Matuska 					(p[3] == '/' || p[3] == '\\')) {
479caf54c4fSMartin Matuska 					p += 3; /* Remove "/..", leave "/"
480caf54c4fSMartin Matuska 							 * for next pass. */
481caf54c4fSMartin Matuska 					slashonly = 0;
482caf54c4fSMartin Matuska 				} else
483caf54c4fSMartin Matuska 					p += 1; /* Remove "/". */
484caf54c4fSMartin Matuska 			}
485caf54c4fSMartin Matuska 		} while (rp != p);
486caf54c4fSMartin Matuska 
487caf54c4fSMartin Matuska 		if (p != name && !bsdtar->warned_lead_slash) {
488caf54c4fSMartin Matuska 			/* Generate a warning the first time this happens. */
489caf54c4fSMartin Matuska 			if (slashonly)
490caf54c4fSMartin Matuska 				lafe_warnc(0,
491caf54c4fSMartin Matuska 				    "Removing leading '%c' from member names",
492caf54c4fSMartin Matuska 				    name[0]);
493caf54c4fSMartin Matuska 			else
494caf54c4fSMartin Matuska 				lafe_warnc(0,
495caf54c4fSMartin Matuska 				    "Removing leading drive letter from "
496caf54c4fSMartin Matuska 				    "member names");
497caf54c4fSMartin Matuska 			bsdtar->warned_lead_slash = 1;
498caf54c4fSMartin Matuska 		}
499caf54c4fSMartin Matuska 
500caf54c4fSMartin Matuska 		/* Special case: Stripping everything yields ".". */
501caf54c4fSMartin Matuska 		if (*p == '\0')
502caf54c4fSMartin Matuska 			name = ".";
503caf54c4fSMartin Matuska 		else
504caf54c4fSMartin Matuska 			name = p;
505caf54c4fSMartin Matuska 	} else {
506caf54c4fSMartin Matuska 		/* Strip redundant leading '/' characters. */
507caf54c4fSMartin Matuska 		while (name[0] == '/' && name[1] == '/')
508caf54c4fSMartin Matuska 			name++;
509caf54c4fSMartin Matuska 	}
510caf54c4fSMartin Matuska 
511caf54c4fSMartin Matuska 	/* Safely replace name in archive_entry. */
512caf54c4fSMartin Matuska 	if (name != archive_entry_pathname(entry)) {
513caf54c4fSMartin Matuska 		char *q = strdup(name);
514caf54c4fSMartin Matuska 		archive_entry_copy_pathname(entry, q);
515caf54c4fSMartin Matuska 		free(q);
516caf54c4fSMartin Matuska 	}
517caf54c4fSMartin Matuska 	return (0);
518caf54c4fSMartin Matuska }
519caf54c4fSMartin Matuska 
520caf54c4fSMartin Matuska /*
521caf54c4fSMartin Matuska  * It would be nice to just use printf() for formatting large numbers,
522caf54c4fSMartin Matuska  * but the compatibility problems are quite a headache.  Hence the
523caf54c4fSMartin Matuska  * following simple utility function.
524caf54c4fSMartin Matuska  */
525caf54c4fSMartin Matuska const char *
526caf54c4fSMartin Matuska tar_i64toa(int64_t n0)
527caf54c4fSMartin Matuska {
528caf54c4fSMartin Matuska 	static char buff[24];
529*6c95142eSMartin Matuska 	uint64_t n = n0 < 0 ? -n0 : n0;
530caf54c4fSMartin Matuska 	char *p = buff + sizeof(buff);
531caf54c4fSMartin Matuska 
532caf54c4fSMartin Matuska 	*--p = '\0';
533caf54c4fSMartin Matuska 	do {
534caf54c4fSMartin Matuska 		*--p = '0' + (int)(n % 10);
535*6c95142eSMartin Matuska 	} while (n /= 10);
536caf54c4fSMartin Matuska 	if (n0 < 0)
537caf54c4fSMartin Matuska 		*--p = '-';
538caf54c4fSMartin Matuska 	return p;
539caf54c4fSMartin Matuska }
540caf54c4fSMartin Matuska 
541caf54c4fSMartin Matuska /*
542caf54c4fSMartin Matuska  * Like strcmp(), but try to be a little more aware of the fact that
543caf54c4fSMartin Matuska  * we're comparing two paths.  Right now, it just handles leading
544caf54c4fSMartin Matuska  * "./" and trailing '/' specially, so that "a/b/" == "./a/b"
545caf54c4fSMartin Matuska  *
546caf54c4fSMartin Matuska  * TODO: Make this better, so that "./a//b/./c/" == "a/b/c"
547caf54c4fSMartin Matuska  * TODO: After this works, push it down into libarchive.
548caf54c4fSMartin Matuska  * TODO: Publish the path normalization routines in libarchive so
549caf54c4fSMartin Matuska  * that bsdtar can normalize paths and use fast strcmp() instead
550caf54c4fSMartin Matuska  * of this.
551caf54c4fSMartin Matuska  *
552caf54c4fSMartin Matuska  * Note: This is currently only used within write.c, so should
553caf54c4fSMartin Matuska  * not handle \ path separators.
554caf54c4fSMartin Matuska  */
555caf54c4fSMartin Matuska 
556caf54c4fSMartin Matuska int
557caf54c4fSMartin Matuska pathcmp(const char *a, const char *b)
558caf54c4fSMartin Matuska {
559caf54c4fSMartin Matuska 	/* Skip leading './' */
560caf54c4fSMartin Matuska 	if (a[0] == '.' && a[1] == '/' && a[2] != '\0')
561caf54c4fSMartin Matuska 		a += 2;
562caf54c4fSMartin Matuska 	if (b[0] == '.' && b[1] == '/' && b[2] != '\0')
563caf54c4fSMartin Matuska 		b += 2;
564caf54c4fSMartin Matuska 	/* Find the first difference, or return (0) if none. */
565caf54c4fSMartin Matuska 	while (*a == *b) {
566caf54c4fSMartin Matuska 		if (*a == '\0')
567caf54c4fSMartin Matuska 			return (0);
568caf54c4fSMartin Matuska 		a++;
569caf54c4fSMartin Matuska 		b++;
570caf54c4fSMartin Matuska 	}
571caf54c4fSMartin Matuska 	/*
572caf54c4fSMartin Matuska 	 * If one ends in '/' and the other one doesn't,
573caf54c4fSMartin Matuska 	 * they're the same.
574caf54c4fSMartin Matuska 	 */
575caf54c4fSMartin Matuska 	if (a[0] == '/' && a[1] == '\0' && b[0] == '\0')
576caf54c4fSMartin Matuska 		return (0);
577caf54c4fSMartin Matuska 	if (a[0] == '\0' && b[0] == '/' && b[1] == '\0')
578caf54c4fSMartin Matuska 		return (0);
579caf54c4fSMartin Matuska 	/* They're really different, return the correct sign. */
580caf54c4fSMartin Matuska 	return (*(const unsigned char *)a - *(const unsigned char *)b);
581caf54c4fSMartin Matuska }
582