freebsd-skq/usr.bin/tar/util.c
kientzle 3a229bf1d0 When deciding which characters (of a filename, for instance)
are safe to print, try to take into account the current locale.
This iterates over output strings using mbtowc() to identify
multi-byte sequences.  If iswprint() claims the corresponding
wide character is printable, the original bytes are passed
through.  Otherwise, we expand characters into C-style
\-escape sequences.

Submitted by:	Michihiro NAKAJIMA
MFC after:	30 days
2008-11-29 20:06:53 +00:00

581 lines
15 KiB
C

/*-
* Copyright (c) 2003-2007 Tim Kientzle
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "bsdtar_platform.h"
__FBSDID("$FreeBSD$");
#ifdef HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h> /* Linux doesn't define mode_t, etc. in sys/stat.h. */
#endif
#include <ctype.h>
#ifdef HAVE_ERRNO_H
#include <errno.h>
#endif
#ifdef HAVE_STDARG_H
#include <stdarg.h>
#endif
#include <stdio.h>
#ifdef HAVE_STDLIB_H
#include <stdlib.h>
#endif
#ifdef HAVE_STRING_H
#include <string.h>
#endif
#ifdef HAVE_WCTYPE_H
#include <wctype.h>
#else
/* If we don't have wctype, we need to hack up some version of iswprint(). */
#define iswprint isprint
#endif
#include "bsdtar.h"
static void bsdtar_vwarnc(struct bsdtar *, int code,
const char *fmt, va_list ap);
static size_t bsdtar_expand_char(char *, size_t, char);
static const char *strip_components(const char *path, int elements);
/* TODO: Hack up a version of mbtowc for platforms with no wide
* character support at all. I think the following might suffice,
* but it needs careful testing.
* #if !HAVE_MBTOWC
* #define mbtowc(wcp, p, n) ((*wcp = *p), 1)
* #endif
*/
/*
* Print a string, taking care with any non-printable characters.
*
* Note that we use a stack-allocated buffer to receive the formatted
* string if we can. This is partly performance (avoiding a call to
* malloc()), partly out of expedience (we have to call vsnprintf()
* before malloc() anyway to find out how big a buffer we need; we may
* as well point that first call at a small local buffer in case it
* works), but mostly for safety (so we can use this to print messages
* about out-of-memory conditions).
*/
void
safe_fprintf(FILE *f, const char *fmt, ...)
{
char fmtbuff_stack[256]; /* Place to format the printf() string. */
char outbuff[256]; /* Buffer for outgoing characters. */
char *fmtbuff_heap; /* If fmtbuff_stack is too small, we use malloc */
char *fmtbuff; /* Pointer to fmtbuff_stack or fmtbuff_heap. */
int fmtbuff_length;
int length;
va_list ap;
const char *p;
unsigned i;
wchar_t wc;
char try_wc;
/* Use a stack-allocated buffer if we can, for speed and safety. */
fmtbuff_heap = NULL;
fmtbuff_length = sizeof(fmtbuff_stack);
fmtbuff = fmtbuff_stack;
/* Try formatting into the stack buffer. */
va_start(ap, fmt);
length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap);
va_end(ap);
/* If the result was too large, allocate a buffer on the heap. */
if (length >= fmtbuff_length) {
fmtbuff_length = length+1;
fmtbuff_heap = malloc(fmtbuff_length);
/* Reformat the result into the heap buffer if we can. */
if (fmtbuff_heap != NULL) {
fmtbuff = fmtbuff_heap;
va_start(ap, fmt);
length = vsnprintf(fmtbuff, fmtbuff_length, fmt, ap);
va_end(ap);
} else {
/* Leave fmtbuff pointing to the truncated
* string in fmtbuff_stack. */
length = sizeof(fmtbuff_stack) - 1;
}
}
/* Note: mbrtowc() has a cleaner API, but mbtowc() seems a bit
* more portable, so we use that here instead. */
mbtowc(NULL, NULL, 0); /* Reset the shift state. */
/* Write data, expanding unprintable characters. */
p = fmtbuff;
i = 0;
try_wc = 1;
while (*p != '\0') {
int n;
/* Convert to wide char, test if the wide
* char is printable in the current locale. */
if (try_wc && (n = mbtowc(&wc, p, length)) != -1) {
length -= n;
if (iswprint(wc) && wc != L'\\') {
/* Printable, copy the bytes through. */
while (n-- > 0)
outbuff[i++] = *p++;
} else {
/* Not printable, format the bytes. */
while (n-- > 0)
i += bsdtar_expand_char(
outbuff, i, *p++);
}
} else {
/* After any conversion failure, don't bother
* trying to convert the rest. */
i += bsdtar_expand_char(outbuff, i, *p++);
try_wc = 0;
}
/* If our output buffer is full, dump it and keep going. */
if (i > (sizeof(outbuff) - 20)) {
outbuff[i++] = '\0';
fprintf(f, "%s", outbuff);
i = 0;
}
}
outbuff[i++] = '\0';
fprintf(f, "%s", outbuff);
/* If we allocated a heap-based formatting buffer, free it now. */
if (fmtbuff_heap != NULL)
free(fmtbuff_heap);
}
/*
* Render an arbitrary sequence of bytes into printable ASCII characters.
*/
static size_t
bsdtar_expand_char(char *buff, size_t offset, char c)
{
size_t i = offset;
if (isprint(c) && c != '\\')
buff[i++] = c;
else {
buff[i++] = '\\';
switch (c) {
case '\a': buff[i++] = 'a'; break;
case '\b': buff[i++] = 'b'; break;
case '\f': buff[i++] = 'f'; break;
case '\n': buff[i++] = 'n'; break;
#if '\r' != '\n'
/* On some platforms, \n and \r are the same. */
case '\r': buff[i++] = 'r'; break;
#endif
case '\t': buff[i++] = 't'; break;
case '\v': buff[i++] = 'v'; break;
case '\\': buff[i++] = '\\'; break;
default:
sprintf(buff + i, "%03o", 0xFF & (int)c);
i += 3;
}
}
return (i - offset);
}
static void
bsdtar_vwarnc(struct bsdtar *bsdtar, int code, const char *fmt, va_list ap)
{
fprintf(stderr, "%s: ", bsdtar->progname);
vfprintf(stderr, fmt, ap);
if (code != 0)
fprintf(stderr, ": %s", strerror(code));
fprintf(stderr, "\n");
}
void
bsdtar_warnc(struct bsdtar *bsdtar, int code, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
bsdtar_vwarnc(bsdtar, code, fmt, ap);
va_end(ap);
}
void
bsdtar_errc(struct bsdtar *bsdtar, int eval, int code, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
bsdtar_vwarnc(bsdtar, code, fmt, ap);
va_end(ap);
exit(eval);
}
int
yes(const char *fmt, ...)
{
char buff[32];
char *p;
ssize_t l;
va_list ap;
va_start(ap, fmt);
vfprintf(stderr, fmt, ap);
va_end(ap);
fprintf(stderr, " (y/N)? ");
fflush(stderr);
l = read(2, buff, sizeof(buff) - 1);
if (l <= 0)
return (0);
buff[l] = 0;
for (p = buff; *p != '\0'; p++) {
if (isspace(0xff & (int)*p))
continue;
switch(*p) {
case 'y': case 'Y':
return (1);
case 'n': case 'N':
return (0);
default:
return (0);
}
}
return (0);
}
/*
* Read lines from file and do something with each one. If option_null
* is set, lines are terminated with zero bytes; otherwise, they're
* terminated with newlines.
*
* This uses a self-sizing buffer to handle arbitrarily-long lines.
* If the "process" function returns non-zero for any line, this
* function will return non-zero after attempting to process all
* remaining lines.
*/
int
process_lines(struct bsdtar *bsdtar, const char *pathname,
int (*process)(struct bsdtar *, const char *))
{
FILE *f;
char *buff, *buff_end, *line_start, *line_end, *p;
size_t buff_length, new_buff_length, bytes_read, bytes_wanted;
int separator;
int ret;
separator = bsdtar->option_null ? '\0' : '\n';
ret = 0;
if (strcmp(pathname, "-") == 0)
f = stdin;
else
f = fopen(pathname, "r");
if (f == NULL)
bsdtar_errc(bsdtar, 1, errno, "Couldn't open %s", pathname);
buff_length = 8192;
buff = malloc(buff_length);
if (buff == NULL)
bsdtar_errc(bsdtar, 1, ENOMEM, "Can't read %s", pathname);
line_start = line_end = buff_end = buff;
for (;;) {
/* Get some more data into the buffer. */
bytes_wanted = buff + buff_length - buff_end;
bytes_read = fread(buff_end, 1, bytes_wanted, f);
buff_end += bytes_read;
/* Process all complete lines in the buffer. */
while (line_end < buff_end) {
if (*line_end == separator) {
*line_end = '\0';
if ((*process)(bsdtar, line_start) != 0)
ret = -1;
line_start = line_end + 1;
line_end = line_start;
} else
line_end++;
}
if (feof(f))
break;
if (ferror(f))
bsdtar_errc(bsdtar, 1, errno,
"Can't read %s", pathname);
if (line_start > buff) {
/* Move a leftover fractional line to the beginning. */
memmove(buff, line_start, buff_end - line_start);
buff_end -= line_start - buff;
line_end -= line_start - buff;
line_start = buff;
} else {
/* Line is too big; enlarge the buffer. */
new_buff_length = buff_length * 2;
if (new_buff_length <= buff_length)
bsdtar_errc(bsdtar, 1, ENOMEM,
"Line too long in %s", pathname);
buff_length = new_buff_length;
p = realloc(buff, buff_length);
if (p == NULL)
bsdtar_errc(bsdtar, 1, ENOMEM,
"Line too long in %s", pathname);
buff_end = p + (buff_end - buff);
line_end = p + (line_end - buff);
line_start = buff = p;
}
}
/* At end-of-file, handle the final line. */
if (line_end > line_start) {
*line_end = '\0';
if ((*process)(bsdtar, line_start) != 0)
ret = -1;
}
free(buff);
if (f != stdin)
fclose(f);
return (ret);
}
/*-
* The logic here for -C <dir> attempts to avoid
* chdir() as long as possible. For example:
* "-C /foo -C /bar file" needs chdir("/bar") but not chdir("/foo")
* "-C /foo -C bar file" needs chdir("/foo/bar")
* "-C /foo -C bar /file1" does not need chdir()
* "-C /foo -C bar /file1 file2" needs chdir("/foo/bar") before file2
*
* The only correct way to handle this is to record a "pending" chdir
* request and combine multiple requests intelligently until we
* need to process a non-absolute file. set_chdir() adds the new dir
* to the pending list; do_chdir() actually executes any pending chdir.
*
* This way, programs that build tar command lines don't have to worry
* about -C with non-existent directories; such requests will only
* fail if the directory must be accessed.
*/
void
set_chdir(struct bsdtar *bsdtar, const char *newdir)
{
if (newdir[0] == '/') {
/* The -C /foo -C /bar case; dump first one. */
free(bsdtar->pending_chdir);
bsdtar->pending_chdir = NULL;
}
if (bsdtar->pending_chdir == NULL)
/* Easy case: no previously-saved dir. */
bsdtar->pending_chdir = strdup(newdir);
else {
/* The -C /foo -C bar case; concatenate */
char *old_pending = bsdtar->pending_chdir;
size_t old_len = strlen(old_pending);
bsdtar->pending_chdir = malloc(old_len + strlen(newdir) + 2);
if (old_pending[old_len - 1] == '/')
old_pending[old_len - 1] = '\0';
if (bsdtar->pending_chdir != NULL)
sprintf(bsdtar->pending_chdir, "%s/%s",
old_pending, newdir);
free(old_pending);
}
if (bsdtar->pending_chdir == NULL)
bsdtar_errc(bsdtar, 1, errno, "No memory");
}
void
do_chdir(struct bsdtar *bsdtar)
{
if (bsdtar->pending_chdir == NULL)
return;
if (chdir(bsdtar->pending_chdir) != 0) {
bsdtar_errc(bsdtar, 1, 0, "could not chdir to '%s'\n",
bsdtar->pending_chdir);
}
free(bsdtar->pending_chdir);
bsdtar->pending_chdir = NULL;
}
const char *
strip_components(const char *path, int elements)
{
const char *p = path;
while (elements > 0) {
switch (*p++) {
case '/':
elements--;
path = p;
break;
case '\0':
/* Path is too short, skip it. */
return (NULL);
}
}
while (*path == '/')
++path;
if (*path == '\0')
return (NULL);
return (path);
}
/*
* Handle --strip-components and any future path-rewriting options.
* Returns non-zero if the pathname should not be extracted.
*
* TODO: Support pax-style regex path rewrites.
*/
int
edit_pathname(struct bsdtar *bsdtar, struct archive_entry *entry)
{
const char *name = archive_entry_pathname(entry);
#if HAVE_REGEX_H
char *subst_name;
#endif
int r;
#if HAVE_REGEX_H
r = apply_substitution(bsdtar, name, &subst_name, 0);
if (r == -1) {
bsdtar_warnc(bsdtar, 0, "Invalid substituion, skipping entry");
return 1;
}
if (r == 1) {
archive_entry_copy_pathname(entry, subst_name);
if (*subst_name == '\0') {
free(subst_name);
return -1;
} else
free(subst_name);
name = archive_entry_pathname(entry);
}
if (archive_entry_hardlink(entry)) {
r = apply_substitution(bsdtar, archive_entry_hardlink(entry), &subst_name, 1);
if (r == -1) {
bsdtar_warnc(bsdtar, 0, "Invalid substituion, skipping entry");
return 1;
}
if (r == 1) {
archive_entry_copy_hardlink(entry, subst_name);
free(subst_name);
}
}
if (archive_entry_symlink(entry) != NULL) {
r = apply_substitution(bsdtar, archive_entry_symlink(entry), &subst_name, 1);
if (r == -1) {
bsdtar_warnc(bsdtar, 0, "Invalid substituion, skipping entry");
return 1;
}
if (r == 1) {
archive_entry_copy_symlink(entry, subst_name);
free(subst_name);
}
}
#endif
/* Strip leading dir names as per --strip-components option. */
if (bsdtar->strip_components > 0) {
const char *linkname = archive_entry_hardlink(entry);
name = strip_components(name, bsdtar->strip_components);
if (name == NULL)
return (1);
if (linkname != NULL) {
linkname = strip_components(linkname,
bsdtar->strip_components);
if (linkname == NULL)
return (1);
archive_entry_copy_hardlink(entry, linkname);
}
}
/* Strip redundant leading '/' characters. */
while (name[0] == '/' && name[1] == '/')
name++;
/* Strip leading '/' unless user has asked us not to. */
if (name[0] == '/' && !bsdtar->option_absolute_paths) {
/* Generate a warning the first time this happens. */
if (!bsdtar->warned_lead_slash) {
bsdtar_warnc(bsdtar, 0,
"Removing leading '/' from member names");
bsdtar->warned_lead_slash = 1;
}
name++;
/* Special case: Stripping leading '/' from "/" yields ".". */
if (*name == '\0')
name = ".";
}
/* Safely replace name in archive_entry. */
if (name != archive_entry_pathname(entry)) {
char *q = strdup(name);
archive_entry_copy_pathname(entry, q);
free(q);
}
return (0);
}
/*
* Like strcmp(), but try to be a little more aware of the fact that
* we're comparing two paths. Right now, it just handles leading
* "./" and trailing '/' specially, so that "a/b/" == "./a/b"
*
* TODO: Make this better, so that "./a//b/./c/" == "a/b/c"
* TODO: After this works, push it down into libarchive.
* TODO: Publish the path normalization routines in libarchive so
* that bsdtar can normalize paths and use fast strcmp() instead
* of this.
*/
int
pathcmp(const char *a, const char *b)
{
/* Skip leading './' */
if (a[0] == '.' && a[1] == '/' && a[2] != '\0')
a += 2;
if (b[0] == '.' && b[1] == '/' && b[2] != '\0')
b += 2;
/* Find the first difference, or return (0) if none. */
while (*a == *b) {
if (*a == '\0')
return (0);
a++;
b++;
}
/*
* If one ends in '/' and the other one doesn't,
* they're the same.
*/
if (a[0] == '/' && a[1] == '\0' && b[0] == '\0')
return (0);
if (a[0] == '\0' && b[0] == '/' && b[1] == '\0')
return (0);
/* They're really different, return the correct sign. */
return (*(const unsigned char *)a - *(const unsigned char *)b);
}