/* List a tar archive, with support routines for reading a tar archive.
- Copyright 1988,92,93,94,96,97,98,1999 Free Software Foundation, Inc.
+
+ Copyright (C) 1988, 1992, 1993, 1994, 1996, 1997, 1998, 1999, 2000,
+ 2001, 2003, 2004, 2005, 2006, 2007, 2010 Free Software Foundation, Inc.
+
Written by John Gilmore, on 1985-08-26.
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
- Free Software Foundation; either version 2, or (at your option) any later
+ Free Software Foundation; either version 3, or (at your option) any later
version.
This program is distributed in the hope that it will be useful, but
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
- 59 Place - Suite 330, Boston, MA 02111-1307, USA. */
+ 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
-/* Define to non-zero for forcing old ctime() instead of isotime(). */
-#undef USE_OLD_CTIME
+#include <system.h>
+#include <inttostr.h>
+#include <quotearg.h>
-#include "system.h"
+#include "common.h"
-#include <ctype.h>
-#include <time.h>
+#define max(a, b) ((a) < (b) ? (b) : (a))
-#define ISODIGIT(Char) \
- ((unsigned char) (Char) >= '0' && (unsigned char) (Char) <= '7')
-#define ISSPACE(Char) (ISASCII (Char) && isspace (Char))
+union block *current_header; /* points to current archive header */
+enum archive_format current_format; /* recognized format */
+union block *recent_long_name; /* recent long name header and contents */
+union block *recent_long_link; /* likewise, for long link */
+size_t recent_long_name_blocks; /* number of blocks in recent_long_name */
+size_t recent_long_link_blocks; /* likewise, for long link */
+union block *recent_global_header; /* Recent global header block */
+
+#define GID_FROM_HEADER(where) gid_from_header (where, sizeof (where))
+#define MAJOR_FROM_HEADER(where) major_from_header (where, sizeof (where))
+#define MINOR_FROM_HEADER(where) minor_from_header (where, sizeof (where))
+#define MODE_FROM_HEADER(where, hbits) \
+ mode_from_header (where, sizeof (where), hbits)
+#define TIME_FROM_HEADER(where) time_from_header (where, sizeof (where))
+#define UID_FROM_HEADER(where) uid_from_header (where, sizeof (where))
+
+static gid_t gid_from_header (const char *buf, size_t size);
+static major_t major_from_header (const char *buf, size_t size);
+static minor_t minor_from_header (const char *buf, size_t size);
+static mode_t mode_from_header (const char *buf, size_t size, unsigned *hbits);
+static time_t time_from_header (const char *buf, size_t size);
+static uid_t uid_from_header (const char *buf, size_t size);
+static uintmax_t from_header (const char *, size_t, const char *,
+ uintmax_t, uintmax_t, bool, bool);
+
+/* Base 64 digits; see Internet RFC 2045 Table 1. */
+static char const base_64_digits[64] =
+{
+ 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
+ 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
+ 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
+ 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
+ '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/'
+};
-#ifndef FNM_LEADING_DIR
-# include <fnmatch.h>
-#endif
+/* Table of base-64 digit values indexed by unsigned chars.
+ The value is 64 for unsigned chars that are not base-64 digits. */
+static char base64_map[UCHAR_MAX + 1];
-#include "common.h"
+static void
+base64_init (void)
+{
+ int i;
+ memset (base64_map, 64, sizeof base64_map);
+ for (i = 0; i < 64; i++)
+ base64_map[(int) base_64_digits[i]] = i;
+}
-union block *current_header; /* points to current archive header */
-struct stat current_stat; /* stat struct corresponding */
-enum archive_format current_format; /* recognized format */
+static char *
+decode_xform (char *file_name, void *data)
+{
+ int type = *(int*)data;
+
+ switch (type)
+ {
+ case XFORM_SYMLINK:
+ /* FIXME: It is not quite clear how and to which extent are the symbolic
+ links subject to filename transformation. In the absence of another
+ solution, symbolic links are exempt from component stripping and
+ name suffix normalization, but subject to filename transformation
+ proper. */
+ return file_name;
+
+ case XFORM_LINK:
+ file_name = safer_name_suffix (file_name, true, absolute_names_option);
+ break;
+
+ case XFORM_REGFILE:
+ file_name = safer_name_suffix (file_name, false, absolute_names_option);
+ break;
+ }
+
+ if (strip_name_components)
+ {
+ size_t prefix_len = stripped_prefix_len (file_name,
+ strip_name_components);
+ if (prefix_len == (size_t) -1)
+ prefix_len = strlen (file_name);
+ file_name += prefix_len;
+ }
+ return file_name;
+}
-/*-----------------------------------.
-| Main loop for reading an archive. |
-`-----------------------------------*/
+static bool
+transform_member_name (char **pinput, int type)
+{
+ return transform_name_fp (pinput, type, decode_xform, &type);
+}
void
-read_and (void (*do_something) ())
+transform_stat_info (int typeflag, struct tar_stat_info *stat_info)
+{
+ if (typeflag == GNUTYPE_VOLHDR)
+ /* Name transformations don't apply to volume headers. */
+ return;
+
+ transform_member_name (&stat_info->file_name, XFORM_REGFILE);
+ switch (typeflag)
+ {
+ case SYMTYPE:
+ transform_member_name (&stat_info->link_name, XFORM_SYMLINK);
+ break;
+
+ case LNKTYPE:
+ transform_member_name (&stat_info->link_name, XFORM_LINK);
+ }
+}
+
+/* Main loop for reading an archive. */
+void
+read_and (void (*do_something) (void))
{
enum read_header status = HEADER_STILL_UNREAD;
enum read_header prev_status;
- char save_typeflag;
+ struct timespec mtime;
+ base64_init ();
name_gather ();
- open_archive (ACCESS_READ);
- while (1)
+ open_archive (ACCESS_READ);
+ do
{
prev_status = status;
- status = read_header ();
+ tar_stat_destroy (¤t_stat_info);
+
+ status = read_header (¤t_header, ¤t_stat_info,
+ read_header_auto);
switch (status)
{
case HEADER_STILL_UNREAD:
+ case HEADER_SUCCESS_EXTENDED:
abort ();
case HEADER_SUCCESS:
/* Valid header. We should decode next field (mode) first.
Ensure incoming names are null terminated. */
-
- if (ending_file_option &&
- fnmatch (ending_file_option, current_file_name,
- FNM_LEADING_DIR) == 0) {
- goto all_done;
- }
-
- /* FIXME: This is a quick kludge before 1.12 goes out. */
- current_stat.st_mtime = TIME_FROM_OCT (current_header->header.mtime);
-
- if (!name_match (current_file_name)
- || current_stat.st_mtime < newer_mtime_option
- || (exclude_option && check_exclude (current_file_name)))
+ decode_header (current_header, ¤t_stat_info,
+ ¤t_format, 1);
+ if (! name_match (current_stat_info.file_name)
+ || (NEWER_OPTION_INITIALIZED (newer_mtime_option)
+ /* FIXME: We get mtime now, and again later; this causes
+ duplicate diagnostics if header.mtime is bogus. */
+ && ((mtime.tv_sec
+ = TIME_FROM_HEADER (current_header->header.mtime)),
+ /* FIXME: Grab fractional time stamps from
+ extended header. */
+ mtime.tv_nsec = 0,
+ current_stat_info.mtime = mtime,
+ OLDER_TAR_STAT_TIME (current_stat_info, m)))
+ || excluded_name (current_stat_info.file_name))
{
- int isextended = 0;
-
- if (current_header->header.typeflag == GNUTYPE_VOLHDR
- || current_header->header.typeflag == GNUTYPE_MULTIVOL
- || current_header->header.typeflag == GNUTYPE_NAMES)
+ switch (current_header->header.typeflag)
{
- (*do_something) ();
- continue;
- }
- if (show_omitted_dirs_option
- && current_header->header.typeflag == DIRTYPE)
- WARN ((0, 0, _("Omitting %s"), current_file_name));
-
- /* Skip past it in the archive. */
-
- if (current_header->oldgnu_header.isextended)
- isextended = 1;
- save_typeflag = current_header->header.typeflag;
- set_next_block_after (current_header);
- if (isextended)
- {
-#if 0
- union block *exhdr;
+ case GNUTYPE_VOLHDR:
+ case GNUTYPE_MULTIVOL:
+ break;
- while (1)
- {
- exhdr = find_next_block ();
- if (!exhdr->sparse_header.isextended)
- {
- set_next_block_after (exhdr);
- break;
- }
- }
- set_next_block_after (exhdr);
-#endif
- skip_extended_headers ();
+ case DIRTYPE:
+ if (show_omitted_dirs_option)
+ WARN ((0, 0, _("%s: Omitting"),
+ quotearg_colon (current_stat_info.file_name)));
+ /* Fall through. */
+ default:
+ skip_member ();
+ continue;
}
-
- /* Skip to the next header on the archive. */
-
- if (save_typeflag != DIRTYPE)
- skip_file (current_stat.st_size);
- continue;
}
-
+ transform_stat_info (current_header->header.typeflag,
+ ¤t_stat_info);
(*do_something) ();
continue;
}
set_next_block_after (current_header);
+
+ if (!ignore_zeros_option)
+ {
+ char buf[UINTMAX_STRSIZE_BOUND];
+
+ status = read_header (¤t_header, ¤t_stat_info,
+ read_header_auto);
+ if (status == HEADER_ZERO_BLOCK)
+ break;
+ WARNOPT (WARN_ALONE_ZERO_BLOCK,
+ (0, 0, _("A lone zero block at %s"),
+ STRINGIFY_BIGINT (current_block_ordinal (), buf)));
+ break;
+ }
status = prev_status;
- if (ignore_zeros_option)
- continue;
- break;
+ continue;
case HEADER_END_OF_FILE:
if (block_number_option)
switch (prev_status)
{
case HEADER_STILL_UNREAD:
- WARN ((0, 0, _("Hmm, this doesn't look like a tar archive")));
+ ERROR ((0, 0, _("This does not look like a tar archive")));
/* Fall through. */
case HEADER_ZERO_BLOCK:
case HEADER_SUCCESS:
- WARN ((0, 0, _("Skipping to next file header")));
+ if (block_number_option)
+ {
+ char buf[UINTMAX_STRSIZE_BOUND];
+ off_t block_ordinal = current_block_ordinal ();
+ block_ordinal -= recent_long_name_blocks;
+ block_ordinal -= recent_long_link_blocks;
+ fprintf (stdlis, _("block %s: "),
+ STRINGIFY_BIGINT (block_ordinal, buf));
+ }
+ ERROR ((0, 0, _("Skipping to next header")));
break;
case HEADER_END_OF_FILE:
case HEADER_FAILURE:
/* We are in the middle of a cascade of errors. */
break;
+
+ case HEADER_SUCCESS_EXTENDED:
+ abort ();
}
continue;
}
break;
}
- all_done: ;
+ while (!all_names_found (¤t_stat_info));
- apply_delayed_set_stat ();
close_archive ();
names_notfound (); /* print names not found */
}
-/*---------------------------------------------.
-| Print a header block, based on tar options. |
-`---------------------------------------------*/
-
+/* Print a header block, based on tar options. */
void
list_archive (void)
{
- int isextended = 0; /* to remember if current_header is extended */
+ off_t block_ordinal = current_block_ordinal ();
/* Print the header block. */
-
if (verbose_option)
- {
- if (verbose_option > 1)
- decode_header (current_header, ¤t_stat, ¤t_format, 0);
- print_header ();
- }
+ print_header (¤t_stat_info, current_header, block_ordinal);
- if (incremental_option && current_header->header.typeflag == GNUTYPE_DUMPDIR)
+ if (incremental_option)
{
- off_t size;
- size_t written, check;
- union block *data_block;
-
- set_next_block_after (current_header);
- if (multi_volume_option)
- {
- assign_string (&save_name, current_file_name);
- save_totsize = current_stat.st_size;
- }
- for (size = current_stat.st_size; size > 0; size -= written)
+ if (verbose_option > 2)
{
- if (multi_volume_option)
- save_sizeleft = size;
- data_block = find_next_block ();
- if (!data_block)
- {
- ERROR ((0, 0, _("EOF in archive file")));
- break; /* FIXME: What happens, then? */
- }
- written = available_space_after (data_block);
- if (written > size)
- written = size;
- errno = 0; /* FIXME: errno should be read-only */
- check = fwrite (data_block->buffer, sizeof (char), written, stdlis);
- set_next_block_after ((union block *)
- (data_block->buffer + written - 1));
- if (check != written)
- {
- ERROR ((0, errno, _("Only wrote %lu of %lu bytes to file %s"),
- (unsigned long) check,
- (unsigned long) written, current_file_name));
- skip_file (size - written);
- break;
- }
+ if (is_dumpdir (¤t_stat_info))
+ list_dumpdir (current_stat_info.dumpdir,
+ dumpdir_size (current_stat_info.dumpdir));
}
- if (multi_volume_option)
- assign_string (&save_name, NULL);
- fputc ('\n', stdlis);
- fflush (stdlis);
- return;
-
}
- /* Check to see if we have an extended header to skip over also. */
-
- if (current_header->oldgnu_header.isextended)
- isextended = 1;
-
- /* Skip past the header in the archive. */
+ skip_member ();
+}
- set_next_block_after (current_header);
+/* Check header checksum */
+/* The standard BSD tar sources create the checksum by adding up the
+ bytes in the header as type char. I think the type char was unsigned
+ on the PDP-11, but it's signed on the Next and Sun. It looks like the
+ sources to BSD tar were never changed to compute the checksum
+ correctly, so both the Sun and Next add the bytes of the header as
+ signed chars. This doesn't cause a problem until you get a file with
+ a name containing characters with the high bit set. So tar_checksum
+ computes two checksums -- signed and unsigned. */
- /* If we needed to skip any extended headers, do so now, by reading
- extended headers and skipping past them in the archive. */
+enum read_header
+tar_checksum (union block *header, bool silent)
+{
+ size_t i;
+ int unsigned_sum = 0; /* the POSIX one :-) */
+ int signed_sum = 0; /* the Sun one :-( */
+ int recorded_sum;
+ uintmax_t parsed_sum;
+ char *p;
- if (isextended)
+ p = header->buffer;
+ for (i = sizeof *header; i-- != 0;)
{
-#if 0
- union block *exhdr;
+ unsigned_sum += (unsigned char) *p;
+ signed_sum += (signed char) (*p++);
+ }
- while (1)
- {
- exhdr = find_next_block ();
+ if (unsigned_sum == 0)
+ return HEADER_ZERO_BLOCK;
- if (!exhdr->sparse_header.isextended)
- {
- set_next_block_after (exhdr);
- break;
- }
- set_next_block_after (exhdr);
- }
-#endif
- skip_extended_headers ();
+ /* Adjust checksum to count the "chksum" field as blanks. */
+
+ for (i = sizeof header->header.chksum; i-- != 0;)
+ {
+ unsigned_sum -= (unsigned char) header->header.chksum[i];
+ signed_sum -= (signed char) (header->header.chksum[i]);
}
+ unsigned_sum += ' ' * sizeof header->header.chksum;
+ signed_sum += ' ' * sizeof header->header.chksum;
- if (multi_volume_option)
- assign_string (&save_name, current_file_name);
+ parsed_sum = from_header (header->header.chksum,
+ sizeof header->header.chksum, 0,
+ (uintmax_t) 0,
+ (uintmax_t) TYPE_MAXIMUM (int), true, silent);
+ if (parsed_sum == (uintmax_t) -1)
+ return HEADER_FAILURE;
- /* Skip to the next header on the archive. */
+ recorded_sum = parsed_sum;
- skip_file (current_stat.st_size);
+ if (unsigned_sum != recorded_sum && signed_sum != recorded_sum)
+ return HEADER_FAILURE;
- if (multi_volume_option)
- assign_string (&save_name, NULL);
+ return HEADER_SUCCESS;
}
-/*-----------------------------------------------------------------------.
-| Read a block that's supposed to be a header block. Return its address |
-| in "current_header", and if it is good, the file's size in |
-| current_stat.st_size. |
-| |
-| Return 1 for success, 0 if the checksum is bad, EOF on eof, 2 for a |
-| block full of zeros (EOF marker). |
-| |
-| You must always set_next_block_after(current_header) to skip past the |
-| header which this routine reads. |
-`-----------------------------------------------------------------------*/
+/* Read a block that's supposed to be a header block. Return its
+ address in *RETURN_BLOCK, and if it is good, the file's size
+ and names (file name, link name) in *INFO.
-/* The standard BSD tar sources create the checksum by adding up the
- bytes in the header as type char. I think the type char was unsigned
- on the PDP-11, but it's signed on the Next and Sun. It looks like the
- sources to BSD tar were never changed to compute the checksum
- currectly, so both the Sun and Next add the bytes of the header as
- signed chars. This doesn't cause a problem until you get a file with
- a name containing characters with the high bit set. So read_header
- computes two checksums -- signed and unsigned. */
+ Return one of enum read_header describing the status of the
+ operation.
+
+ The MODE parameter instructs read_header what to do with special
+ header blocks, i.e.: extended POSIX, GNU long name or long link,
+ etc.:
-/* FIXME: The signed checksum computation is broken on machines where char's
- are unsigned. It's uneasy to handle all cases correctly... */
+ read_header_auto process them automatically,
+ read_header_x_raw when a special header is read, return
+ HEADER_SUCCESS_EXTENDED without actually
+ processing the header,
+ read_header_x_global when a POSIX global header is read,
+ decode it and return HEADER_SUCCESS_EXTENDED.
+
+ You must always set_next_block_after(*return_block) to skip past
+ the header which this routine reads. */
enum read_header
-read_header (void)
+read_header (union block **return_block, struct tar_stat_info *info,
+ enum read_header_mode mode)
{
- size_t i;
- long unsigned_sum; /* the POSIX one :-) */
- long signed_sum; /* the Sun one :-( */
- long recorded_sum;
- char *p;
union block *header;
- char **longp;
+ union block *header_copy;
char *bp;
union block *data_block;
size_t size, written;
- static char *next_long_name, *next_long_link;
+ union block *next_long_name = 0;
+ union block *next_long_link = 0;
+ size_t next_long_name_blocks = 0;
+ size_t next_long_link_blocks = 0;
while (1)
{
+ enum read_header status;
+
header = find_next_block ();
- current_header = header;
+ *return_block = header;
if (!header)
return HEADER_END_OF_FILE;
- recorded_sum = UINTMAX_FROM_OCT (header->header.chksum);
-
- unsigned_sum = 0;
- signed_sum = 0;
- p = header->buffer;
- for (i = sizeof (*header); i-- != 0;)
- {
- /* We can't use unsigned char here because of old compilers,
- e.g. V7. */
-
- unsigned_sum += 0xFF & *p;
- signed_sum += *p++;
- }
-
- /* Adjust checksum to count the "chksum" field as blanks. */
-
- for (i = sizeof (header->header.chksum); i-- != 0;)
- {
- unsigned_sum -= 0xFF & header->header.chksum[i];
- signed_sum -= header->header.chksum[i];
- }
- unsigned_sum += ' ' * sizeof header->header.chksum;
- signed_sum += ' ' * sizeof header->header.chksum;
-
- if (unsigned_sum == sizeof header->header.chksum * ' ')
- {
- /* This is a zeroed block...whole block is 0's except for the
- blanks we faked for the checksum field. */
-
- return HEADER_ZERO_BLOCK;
- }
-
- if (unsigned_sum != recorded_sum && signed_sum != recorded_sum)
- return HEADER_FAILURE;
+ if ((status = tar_checksum (header, false)) != HEADER_SUCCESS)
+ return status;
/* Good block. Decode file size and return. */
if (header->header.typeflag == LNKTYPE)
- current_stat.st_size = 0; /* links 0 size on tape */
+ info->stat.st_size = 0; /* links 0 size on tape */
else
- current_stat.st_size = OFF_FROM_OCT (header->header.size);
+ info->stat.st_size = OFF_FROM_HEADER (header->header.size);
- header->header.name[NAME_FIELD_SIZE - 1] = '\0';
if (header->header.typeflag == GNUTYPE_LONGNAME
- || header->header.typeflag == GNUTYPE_LONGLINK)
+ || header->header.typeflag == GNUTYPE_LONGLINK
+ || header->header.typeflag == XHDTYPE
+ || header->header.typeflag == XGLTYPE
+ || header->header.typeflag == SOLARIS_XHDTYPE)
{
- longp = ((header->header.typeflag == GNUTYPE_LONGNAME)
- ? &next_long_name
- : &next_long_link);
-
- set_next_block_after (header);
- if (*longp)
- free (*longp);
- size = current_stat.st_size;
- if (size != current_stat.st_size)
- FATAL_ERROR ((0, 0, _("Memory exhausted")));
- bp = *longp = (char *) xmalloc (size);
-
- for (; size > 0; size -= written)
+ if (mode == read_header_x_raw)
+ return HEADER_SUCCESS_EXTENDED;
+ else if (header->header.typeflag == GNUTYPE_LONGNAME
+ || header->header.typeflag == GNUTYPE_LONGLINK)
{
- data_block = find_next_block ();
- if (data_block == NULL)
+ size_t name_size = info->stat.st_size;
+ size_t n = name_size % BLOCKSIZE;
+ size = name_size + BLOCKSIZE;
+ if (n)
+ size += BLOCKSIZE - n;
+
+ if (name_size != info->stat.st_size || size < name_size)
+ xalloc_die ();
+
+ header_copy = xmalloc (size + 1);
+
+ if (header->header.typeflag == GNUTYPE_LONGNAME)
{
- ERROR ((0, 0, _("Unexpected EOF on archive file")));
- break;
+ if (next_long_name)
+ free (next_long_name);
+ next_long_name = header_copy;
+ next_long_name_blocks = size / BLOCKSIZE;
}
- written = available_space_after (data_block);
- if (written > size)
- written = size;
-
- memcpy (bp, data_block->buffer, written);
- bp += written;
- set_next_block_after ((union block *)
- (data_block->buffer + written - 1));
+ else
+ {
+ if (next_long_link)
+ free (next_long_link);
+ next_long_link = header_copy;
+ next_long_link_blocks = size / BLOCKSIZE;
+ }
+
+ set_next_block_after (header);
+ *header_copy = *header;
+ bp = header_copy->buffer + BLOCKSIZE;
+
+ for (size -= BLOCKSIZE; size > 0; size -= written)
+ {
+ data_block = find_next_block ();
+ if (! data_block)
+ {
+ ERROR ((0, 0, _("Unexpected EOF in archive")));
+ break;
+ }
+ written = available_space_after (data_block);
+ if (written > size)
+ written = size;
+
+ memcpy (bp, data_block->buffer, written);
+ bp += written;
+ set_next_block_after ((union block *)
+ (data_block->buffer + written - 1));
+ }
+
+ *bp = '\0';
+ }
+ else if (header->header.typeflag == XHDTYPE
+ || header->header.typeflag == SOLARIS_XHDTYPE)
+ xheader_read (&info->xhdr, header,
+ OFF_FROM_HEADER (header->header.size));
+ else if (header->header.typeflag == XGLTYPE)
+ {
+ struct xheader xhdr;
+
+ if (!recent_global_header)
+ recent_global_header = xmalloc (sizeof *recent_global_header);
+ memcpy (recent_global_header, header,
+ sizeof *recent_global_header);
+ memset (&xhdr, 0, sizeof xhdr);
+ xheader_read (&xhdr, header,
+ OFF_FROM_HEADER (header->header.size));
+ xheader_decode_global (&xhdr);
+ xheader_destroy (&xhdr);
+ if (mode == read_header_x_global)
+ return HEADER_SUCCESS_EXTENDED;
}
/* Loop! */
}
else
{
- char *name = next_long_name;
- struct posix_header *h = ¤t_header->header;
- char namebuf[sizeof h->prefix + 1 + sizeof h->name + 1];
+ char const *name;
+ struct posix_header const *h = &header->header;
+ char namebuf[sizeof h->prefix + 1 + NAME_FIELD_SIZE + 1];
+
+ if (recent_long_name)
+ free (recent_long_name);
- if (! name)
+ if (next_long_name)
+ {
+ name = next_long_name->buffer + BLOCKSIZE;
+ recent_long_name = next_long_name;
+ recent_long_name_blocks = next_long_name_blocks;
+ }
+ else
{
/* Accept file names as specified by POSIX.1-1996
section 10.1.1. */
char *np = namebuf;
- if (h->prefix[0])
+
+ if (h->prefix[0] && strcmp (h->magic, TMAGIC) == 0)
{
memcpy (np, h->prefix, sizeof h->prefix);
np[sizeof h->prefix] = '\0';
memcpy (np, h->name, sizeof h->name);
np[sizeof h->name] = '\0';
name = namebuf;
+ recent_long_name = 0;
+ recent_long_name_blocks = 0;
+ }
+ assign_string (&info->orig_file_name, name);
+ assign_string (&info->file_name, name);
+ info->had_trailing_slash = strip_trailing_slashes (info->file_name);
+
+ if (recent_long_link)
+ free (recent_long_link);
+
+ if (next_long_link)
+ {
+ name = next_long_link->buffer + BLOCKSIZE;
+ recent_long_link = next_long_link;
+ recent_long_link_blocks = next_long_link_blocks;
+ }
+ else
+ {
+ memcpy (namebuf, h->linkname, sizeof h->linkname);
+ namebuf[sizeof h->linkname] = '\0';
+ name = namebuf;
+ recent_long_link = 0;
+ recent_long_link_blocks = 0;
}
+ assign_string (&info->link_name, name);
- assign_string (¤t_file_name, name);
- assign_string (¤t_link_name,
- (next_long_link ? next_long_link
- : current_header->header.linkname));
- next_long_link = next_long_name = 0;
return HEADER_SUCCESS;
}
}
}
-/*-------------------------------------------------------------------------.
-| Decode things from a file HEADER block into STAT_INFO, also setting |
-| *FORMAT_POINTER depending on the header block format. If DO_USER_GROUP, |
-| decode the user/group information (this is useful for extraction, but |
-| waste time when merely listing). |
-| |
-| read_header() has already decoded the checksum and length, so we don't. |
-| |
-| This routine should *not* be called twice for the same block, since the |
-| two calls might use different DO_USER_GROUP values and thus might end up |
-| with different uid/gid for the two calls. If anybody wants the uid/gid |
-| they should decode it first, and other callers should decode it without |
-| uid/gid before calling a routine, e.g. print_header, that assumes |
-| decoded data. |
-`-------------------------------------------------------------------------*/
+#define ISOCTAL(c) ((c)>='0'&&(c)<='7')
+
+/* Decode things from a file HEADER block into STAT_INFO, also setting
+ *FORMAT_POINTER depending on the header block format. If
+ DO_USER_GROUP, decode the user/group information (this is useful
+ for extraction, but waste time when merely listing).
+ read_header() has already decoded the checksum and length, so we don't.
+
+ This routine should *not* be called twice for the same block, since
+ the two calls might use different DO_USER_GROUP values and thus
+ might end up with different uid/gid for the two calls. If anybody
+ wants the uid/gid they should decode it first, and other callers
+ should decode it without uid/gid before calling a routine,
+ e.g. print_header, that assumes decoded data. */
void
-decode_header (union block *header, struct stat *stat_info,
+decode_header (union block *header, struct tar_stat_info *stat_info,
enum archive_format *format_pointer, int do_user_group)
{
enum archive_format format;
+ unsigned hbits; /* high bits of the file mode. */
+ mode_t mode = MODE_FROM_HEADER (header->header.mode, &hbits);
if (strcmp (header->header.magic, TMAGIC) == 0)
- format = POSIX_FORMAT;
- else if (strcmp (header->header.magic, OLDGNU_MAGIC) == 0)
- format = OLDGNU_FORMAT;
+ {
+ if (header->star_header.prefix[130] == 0
+ && ISOCTAL (header->star_header.atime[0])
+ && header->star_header.atime[11] == ' '
+ && ISOCTAL (header->star_header.ctime[0])
+ && header->star_header.ctime[11] == ' ')
+ format = STAR_FORMAT;
+ else if (stat_info->xhdr.size)
+ format = POSIX_FORMAT;
+ else
+ format = USTAR_FORMAT;
+ }
+ else if (strcmp (header->buffer + offsetof (struct posix_header, magic),
+ OLDGNU_MAGIC)
+ == 0)
+ format = hbits ? OLDGNU_FORMAT : GNU_FORMAT;
else
format = V7_FORMAT;
*format_pointer = format;
- stat_info->st_mode = MODE_FROM_OCT (header->header.mode);
- stat_info->st_mode &= 07777;
- stat_info->st_mtime = TIME_FROM_OCT (header->header.mtime);
+ stat_info->stat.st_mode = mode;
+ stat_info->mtime.tv_sec = TIME_FROM_HEADER (header->header.mtime);
+ stat_info->mtime.tv_nsec = 0;
+ assign_string (&stat_info->uname,
+ header->header.uname[0] ? header->header.uname : NULL);
+ assign_string (&stat_info->gname,
+ header->header.gname[0] ? header->header.gname : NULL);
if (format == OLDGNU_FORMAT && incremental_option)
{
- stat_info->st_atime = TIME_FROM_OCT (header->oldgnu_header.atime);
- stat_info->st_ctime = TIME_FROM_OCT (header->oldgnu_header.ctime);
+ stat_info->atime.tv_sec = TIME_FROM_HEADER (header->oldgnu_header.atime);
+ stat_info->ctime.tv_sec = TIME_FROM_HEADER (header->oldgnu_header.ctime);
+ stat_info->atime.tv_nsec = stat_info->ctime.tv_nsec = 0;
+ }
+ else if (format == STAR_FORMAT)
+ {
+ stat_info->atime.tv_sec = TIME_FROM_HEADER (header->star_header.atime);
+ stat_info->ctime.tv_sec = TIME_FROM_HEADER (header->star_header.ctime);
+ stat_info->atime.tv_nsec = stat_info->ctime.tv_nsec = 0;
}
+ else
+ stat_info->atime = stat_info->ctime = start_time;
if (format == V7_FORMAT)
{
- stat_info->st_uid = UID_FROM_OCT (header->header.uid);
- stat_info->st_gid = GID_FROM_OCT (header->header.gid);
- stat_info->st_rdev = 0;
+ stat_info->stat.st_uid = UID_FROM_HEADER (header->header.uid);
+ stat_info->stat.st_gid = GID_FROM_HEADER (header->header.gid);
+ stat_info->stat.st_rdev = 0;
}
else
{
if (numeric_owner_option
|| !*header->header.uname
- || !uname_to_uid (header->header.uname, &stat_info->st_uid))
- stat_info->st_uid = UID_FROM_OCT (header->header.uid);
+ || !uname_to_uid (header->header.uname, &stat_info->stat.st_uid))
+ stat_info->stat.st_uid = UID_FROM_HEADER (header->header.uid);
if (numeric_owner_option
|| !*header->header.gname
- || !gname_to_gid (header->header.gname, &stat_info->st_gid))
- stat_info->st_gid = GID_FROM_OCT (header->header.gid);
+ || !gname_to_gid (header->header.gname, &stat_info->stat.st_gid))
+ stat_info->stat.st_gid = GID_FROM_HEADER (header->header.gid);
}
+
switch (header->header.typeflag)
{
-#ifdef S_IFBLK
case BLKTYPE:
- stat_info->st_rdev
- = makedev (MAJOR_FROM_OCT (header->header.devmajor),
- MINOR_FROM_OCT (header->header.devminor));
- break;
-#endif
-
-#ifdef S_IFCHR
case CHRTYPE:
- stat_info->st_rdev
- = makedev (MAJOR_FROM_OCT (header->header.devmajor),
- MINOR_FROM_OCT (header->header.devminor));
+ stat_info->stat.st_rdev =
+ makedev (MAJOR_FROM_HEADER (header->header.devmajor),
+ MINOR_FROM_HEADER (header->header.devminor));
break;
-#endif
default:
- stat_info->st_rdev = 0;
+ stat_info->stat.st_rdev = 0;
}
}
+
+ stat_info->archive_file_size = stat_info->stat.st_size;
+ xheader_decode (stat_info);
+
+ if (sparse_member_p (stat_info))
+ {
+ sparse_fixup_header (stat_info);
+ stat_info->is_sparse = true;
+ }
+ else
+ {
+ stat_info->is_sparse = false;
+ if (((current_format == GNU_FORMAT
+ || current_format == OLDGNU_FORMAT)
+ && current_header->header.typeflag == GNUTYPE_DUMPDIR)
+ || stat_info->dumpdir)
+ stat_info->is_dumpdir = true;
+ }
}
-/*------------------------------------------------------------------------.
-| Quick and dirty octal conversion. Result is -1 if the field is invalid |
-| (all blank, or nonoctal). |
-`------------------------------------------------------------------------*/
+/* Convert buffer at WHERE0 of size DIGS from external format to
+ uintmax_t. DIGS must be positive. If TYPE is nonnull, the data
+ are of type TYPE. The buffer must represent a value in the range
+ -MINUS_MINVAL through MAXVAL. If OCTAL_ONLY, allow only octal
+ numbers instead of the other GNU extensions. Return -1 on error,
+ diagnosing the error if TYPE is nonnull and if !SILENT. */
static uintmax_t
-from_oct (const char *where0, size_t digs0, const char *type, uintmax_t maxval)
+from_header (char const *where0, size_t digs, char const *type,
+ uintmax_t minus_minval, uintmax_t maxval,
+ bool octal_only, bool silent)
{
uintmax_t value;
- const char *where = where0;
- size_t digs = digs0;
+ char const *where = where0;
+ char const *lim = where + digs;
+ int negative = 0;
+
+ /* Accommodate buggy tar of unknown vintage, which outputs leading
+ NUL if the previous field overflows. */
+ where += !*where;
+ /* Accommodate older tars, which output leading spaces. */
for (;;)
{
- if (digs == 0)
+ if (where == lim)
{
- ERROR ((0, 0, _("Blanks in header where octal %s value expected"),
- type));
+ if (type && !silent)
+ ERROR ((0, 0,
+ /* TRANSLATORS: %s is type of the value (gid_t, uid_t,
+ etc.) */
+ _("Blanks in header where numeric %s value expected"),
+ type));
return -1;
}
- if (!ISSPACE (*where))
+ if (!ISSPACE ((unsigned char) *where))
break;
where++;
- digs--;
}
value = 0;
- while (digs != 0 && ISODIGIT (*where))
+ if (ISODIGIT (*where))
{
- /* Scan til nonoctal. */
+ char const *where1 = where;
+ uintmax_t overflow = 0;
+
+ for (;;)
+ {
+ value += *where++ - '0';
+ if (where == lim || ! ISODIGIT (*where))
+ break;
+ overflow |= value ^ (value << LG_8 >> LG_8);
+ value <<= LG_8;
+ }
- if (value << 3 >> 3 != value)
- goto out_of_range;
- value = (value << 3) | (*where++ - '0');
- --digs;
+ /* Parse the output of older, unportable tars, which generate
+ negative values in two's complement octal. If the leading
+ nonzero digit is 1, we can't recover the original value
+ reliably; so do this only if the digit is 2 or more. This
+ catches the common case of 32-bit negative time stamps. */
+ if ((overflow || maxval < value) && '2' <= *where1 && type)
+ {
+ /* Compute the negative of the input value, assuming two's
+ complement. */
+ int digit = (*where1 - '0') | 4;
+ overflow = 0;
+ value = 0;
+ where = where1;
+ for (;;)
+ {
+ value += 7 - digit;
+ where++;
+ if (where == lim || ! ISODIGIT (*where))
+ break;
+ digit = *where - '0';
+ overflow |= value ^ (value << LG_8 >> LG_8);
+ value <<= LG_8;
+ }
+ value++;
+ overflow |= !value;
+
+ if (!overflow && value <= minus_minval)
+ {
+ if (!silent)
+ WARN ((0, 0,
+ /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
+ _("Archive octal value %.*s is out of %s range; assuming two's complement"),
+ (int) (where - where1), where1, type));
+ negative = 1;
+ }
+ }
+
+ if (overflow)
+ {
+ if (type && !silent)
+ ERROR ((0, 0,
+ /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
+ _("Archive octal value %.*s is out of %s range"),
+ (int) (where - where1), where1, type));
+ return -1;
+ }
+ }
+ else if (octal_only)
+ {
+ /* Suppress the following extensions. */
+ }
+ else if (*where == '-' || *where == '+')
+ {
+ /* Parse base-64 output produced only by tar test versions
+ 1.13.6 (1999-08-11) through 1.13.11 (1999-08-23).
+ Support for this will be withdrawn in future releases. */
+ int dig;
+ if (!silent)
+ {
+ static bool warned_once;
+ if (! warned_once)
+ {
+ warned_once = true;
+ WARN ((0, 0, _("Archive contains obsolescent base-64 headers")));
+ }
+ }
+ negative = *where++ == '-';
+ while (where != lim
+ && (dig = base64_map[(unsigned char) *where]) < 64)
+ {
+ if (value << LG_64 >> LG_64 != value)
+ {
+ char *string = alloca (digs + 1);
+ memcpy (string, where0, digs);
+ string[digs] = '\0';
+ if (type && !silent)
+ ERROR ((0, 0,
+ _("Archive signed base-64 string %s is out of %s range"),
+ quote (string), type));
+ return -1;
+ }
+ value = (value << LG_64) | dig;
+ where++;
+ }
+ }
+ else if (*where == '\200' /* positive base-256 */
+ || *where == '\377' /* negative base-256 */)
+ {
+ /* Parse base-256 output. A nonnegative number N is
+ represented as (256**DIGS)/2 + N; a negative number -N is
+ represented as (256**DIGS) - N, i.e. as two's complement.
+ The representation guarantees that the leading bit is
+ always on, so that we don't confuse this format with the
+ others (assuming ASCII bytes of 8 bits or more). */
+ int signbit = *where & (1 << (LG_256 - 2));
+ uintmax_t topbits = (((uintmax_t) - signbit)
+ << (CHAR_BIT * sizeof (uintmax_t)
+ - LG_256 - (LG_256 - 2)));
+ value = (*where++ & ((1 << (LG_256 - 2)) - 1)) - signbit;
+ for (;;)
+ {
+ value = (value << LG_256) + (unsigned char) *where++;
+ if (where == lim)
+ break;
+ if (((value << LG_256 >> LG_256) | topbits) != value)
+ {
+ if (type && !silent)
+ ERROR ((0, 0,
+ _("Archive base-256 value is out of %s range"),
+ type));
+ return -1;
+ }
+ }
+ negative = signbit;
+ if (negative)
+ value = -value;
}
- if (digs != 0 && *where && !ISSPACE (*where))
+ if (where != lim && *where && !ISSPACE ((unsigned char) *where))
{
- ERROR ((0, 0, _("Header contains `%.*s' where octal %s value expected"),
- (int) digs0, where0, type));
+ if (type)
+ {
+ char buf[1000]; /* Big enough to represent any header. */
+ static struct quoting_options *o;
+
+ if (!o)
+ {
+ o = clone_quoting_options (0);
+ set_quoting_style (o, locale_quoting_style);
+ }
+
+ while (where0 != lim && ! lim[-1])
+ lim--;
+ quotearg_buffer (buf, sizeof buf, where0, lim - where, o);
+ if (!silent)
+ ERROR ((0, 0,
+ /* TRANSLATORS: Second %s is a type name (gid_t,uid_t,etc.) */
+ _("Archive contains %.*s where numeric %s value expected"),
+ (int) sizeof buf, buf, type));
+ }
+
return -1;
}
- if (value <= maxval)
- return value;
+ if (value <= (negative ? minus_minval : maxval))
+ return negative ? -value : value;
+
+ if (type && !silent)
+ {
+ char minval_buf[UINTMAX_STRSIZE_BOUND + 1];
+ char maxval_buf[UINTMAX_STRSIZE_BOUND];
+ char value_buf[UINTMAX_STRSIZE_BOUND + 1];
+ char *minval_string = STRINGIFY_BIGINT (minus_minval, minval_buf + 1);
+ char *value_string = STRINGIFY_BIGINT (value, value_buf + 1);
+ if (negative)
+ *--value_string = '-';
+ if (minus_minval)
+ *--minval_string = '-';
+ /* TRANSLATORS: Second %s is type name (gid_t,uid_t,etc.) */
+ ERROR ((0, 0, _("Archive value %s is out of %s range %s..%s"),
+ value_string, type,
+ minval_string, STRINGIFY_BIGINT (maxval, maxval_buf)));
+ }
- out_of_range:
- ERROR ((0, 0, _("Octal value `%.*s' is out of range for %s"),
- (int) digs0, where0, type));
return -1;
}
-gid_t
-gid_from_oct (const char *p, size_t s)
+
+static gid_t
+gid_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "gid_t", (uintmax_t) TYPE_MAXIMUM (gid_t));
+ return from_header (p, s, "gid_t",
+ - (uintmax_t) TYPE_MINIMUM (gid_t),
+ (uintmax_t) TYPE_MAXIMUM (gid_t),
+ false, false);
}
-major_t
-major_from_oct (const char *p, size_t s)
+
+static major_t
+major_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "major_t", (uintmax_t) TYPE_MAXIMUM (major_t));
+ return from_header (p, s, "major_t",
+ - (uintmax_t) TYPE_MINIMUM (major_t),
+ (uintmax_t) TYPE_MAXIMUM (major_t), false, false);
}
-minor_t
-minor_from_oct (const char *p, size_t s)
+
+static minor_t
+minor_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "minor_t", (uintmax_t) TYPE_MAXIMUM (minor_t));
+ return from_header (p, s, "minor_t",
+ - (uintmax_t) TYPE_MINIMUM (minor_t),
+ (uintmax_t) TYPE_MAXIMUM (minor_t), false, false);
}
-mode_t
-mode_from_oct (const char *p, size_t s)
+
+/* Convert P to the file mode, as understood by tar.
+ Store unrecognized mode bits (from 10th up) in HBITS. */
+static mode_t
+mode_from_header (const char *p, size_t s, unsigned *hbits)
{
- return from_oct (p, s, "mode_t", (uintmax_t) TYPE_MAXIMUM (mode_t));
+ unsigned u = from_header (p, s, "mode_t",
+ - (uintmax_t) TYPE_MINIMUM (mode_t),
+ TYPE_MAXIMUM (uintmax_t), false, false);
+ mode_t mode = ((u & TSUID ? S_ISUID : 0)
+ | (u & TSGID ? S_ISGID : 0)
+ | (u & TSVTX ? S_ISVTX : 0)
+ | (u & TUREAD ? S_IRUSR : 0)
+ | (u & TUWRITE ? S_IWUSR : 0)
+ | (u & TUEXEC ? S_IXUSR : 0)
+ | (u & TGREAD ? S_IRGRP : 0)
+ | (u & TGWRITE ? S_IWGRP : 0)
+ | (u & TGEXEC ? S_IXGRP : 0)
+ | (u & TOREAD ? S_IROTH : 0)
+ | (u & TOWRITE ? S_IWOTH : 0)
+ | (u & TOEXEC ? S_IXOTH : 0));
+ *hbits = mode ^ u;
+ return mode;
}
+
off_t
-off_from_oct (const char *p, size_t s)
+off_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "off_t", (uintmax_t) TYPE_MAXIMUM (off_t));
+ /* Negative offsets are not allowed in tar files, so invoke
+ from_header with minimum value 0, not TYPE_MINIMUM (off_t). */
+ return from_header (p, s, "off_t", (uintmax_t) 0,
+ (uintmax_t) TYPE_MAXIMUM (off_t), false, false);
}
-size_t
-size_from_oct (const char *p, size_t s)
-{
- return from_oct (p, s, "size_t", (uintmax_t) TYPE_MAXIMUM (size_t));
-}
-time_t
-time_from_oct (const char *p, size_t s)
+
+static time_t
+time_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "time_t", (uintmax_t) TYPE_MAXIMUM (time_t));
+ return from_header (p, s, "time_t",
+ - (uintmax_t) TYPE_MINIMUM (time_t),
+ (uintmax_t) TYPE_MAXIMUM (time_t), false, false);
}
-uid_t
-uid_from_oct (const char *p, size_t s)
+
+static uid_t
+uid_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "uid_t", (uintmax_t) TYPE_MAXIMUM (uid_t));
+ return from_header (p, s, "uid_t",
+ - (uintmax_t) TYPE_MINIMUM (uid_t),
+ (uintmax_t) TYPE_MAXIMUM (uid_t), false, false);
}
+
uintmax_t
-uintmax_from_oct (const char *p, size_t s)
+uintmax_from_header (const char *p, size_t s)
{
- return from_oct (p, s, "uintmax_t", TYPE_MAXIMUM (uintmax_t));
+ return from_header (p, s, "uintmax_t", (uintmax_t) 0,
+ TYPE_MAXIMUM (uintmax_t), false, false);
}
-
-/*----------------------------------------------------------------------.
-| Format O as a null-terminated decimal string into BUF _backwards_; |
-| return pointer to start of result. |
-`----------------------------------------------------------------------*/
-char *
-stringify_uintmax_t_backwards (uintmax_t o, char *buf)
+/* Return a printable representation of T. The result points to
+ static storage that can be reused in the next call to this
+ function, to ctime, or to asctime. If FULL_TIME, then output the
+ time stamp to its full resolution; otherwise, just output it to
+ 1-minute resolution. */
+char const *
+tartime (struct timespec t, bool full_time)
{
- *--buf = '\0';
- do
- *--buf = '0' + (int) (o % 10);
- while ((o /= 10) != 0);
- return buf;
-}
+ enum { fraclen = sizeof ".FFFFFFFFF" - 1 };
+ static char buffer[max (UINTMAX_STRSIZE_BOUND + 1,
+ INT_STRLEN_BOUND (int) + 16)
+ + fraclen];
+ struct tm *tm;
+ time_t s = t.tv_sec;
+ int ns = t.tv_nsec;
+ bool negative = s < 0;
+ char *p;
-#if !USE_OLD_CTIME
+ if (negative && ns != 0)
+ {
+ s++;
+ ns = 1000000000 - ns;
+ }
-/*-------------------------------------------.
-| Return the time formatted along ISO 8601. |
-`-------------------------------------------*/
+ tm = utc_option ? gmtime (&s) : localtime (&s);
+ if (tm)
+ {
+ if (full_time)
+ {
+ sprintf (buffer, "%04ld-%02d-%02d %02d:%02d:%02d",
+ tm->tm_year + 1900L, tm->tm_mon + 1, tm->tm_mday,
+ tm->tm_hour, tm->tm_min, tm->tm_sec);
+ code_ns_fraction (ns, buffer + strlen (buffer));
+ }
+ else
+ sprintf (buffer, "%04ld-%02d-%02d %02d:%02d",
+ tm->tm_year + 1900L, tm->tm_mon + 1, tm->tm_mday,
+ tm->tm_hour, tm->tm_min);
+ return buffer;
+ }
-/* Also, see http://www.ft.uni-erlangen.de/~mskuhn/iso-time.html. */
+ /* The time stamp cannot be broken down, most likely because it
+ is out of range. Convert it as an integer,
+ right-adjusted in a field with the same width as the usual
+ 4-year ISO time format. */
+ p = umaxtostr (negative ? - (uintmax_t) s : s,
+ buffer + sizeof buffer - UINTMAX_STRSIZE_BOUND - fraclen);
+ if (negative)
+ *--p = '-';
+ while ((buffer + sizeof buffer - sizeof "YYYY-MM-DD HH:MM"
+ + (full_time ? sizeof ":SS.FFFFFFFFF" - 1 : 0))
+ < p)
+ *--p = ' ';
+ if (full_time)
+ code_ns_fraction (ns, buffer + sizeof buffer - 1 - fraclen);
+ return p;
+}
-static char *
-isotime (const time_t *time)
-{
- static char buffer[21];
- struct tm *tm;
+/* Actually print it.
- tm = localtime (time);
- sprintf (buffer, "%4d-%02d-%02d %02d:%02d:%02d\n",
- tm->tm_year + 1900, tm->tm_mon + 1, tm->tm_mday,
- tm->tm_hour, tm->tm_min, tm->tm_sec);
- return buffer;
-}
+ Plain and fancy file header block logging. Non-verbose just prints
+ the name, e.g. for "tar t" or "tar x". This should just contain
+ file names, so it can be fed back into tar with xargs or the "-T"
+ option. The verbose option can give a bunch of info, one line per
+ file. I doubt anybody tries to parse its format, or if they do,
+ they shouldn't. Unix tar is pretty random here anyway. */
-#endif /* not USE_OLD_CTIME */
-/*-------------------------------------------------------------------------.
-| Decode MODE from its binary form in a stat structure, and encode it into |
-| a 9 characters string STRING, terminated with a NUL. |
-`-------------------------------------------------------------------------*/
+/* Width of "user/group size", with initial value chosen
+ heuristically. This grows as needed, though this may cause some
+ stairstepping in the output. Make it too small and the output will
+ almost always look ragged. Make it too large and the output will
+ be spaced out too far. */
+static int ugswidth = 19;
-static void
-decode_mode (mode_t mode, char *string)
-{
- mode_t mask;
- const char *rwx = "rwxrwxrwx";
-
- for (mask = 0400; mask != 0; mask >>= 1)
- if (mode & mask)
- *string++ = *rwx++;
- else
- {
- *string++ = '-';
- rwx++;
- }
-
- if (mode & S_ISUID)
- string[-7] = string[-7] == 'x' ? 's' : 'S';
- if (mode & S_ISGID)
- string[-4] = string[-4] == 'x' ? 's' : 'S';
- if (mode & S_ISVTX)
- string[-1] = string[-1] == 'x' ? 't' : 'T';
-
- *string = '\0';
-}
+/* Width of printed time stamps. It grows if longer time stamps are
+ found (typically, those with nanosecond resolution). Like
+ USGWIDTH, some stairstepping may occur. */
+static int datewidth = sizeof "YYYY-MM-DD HH:MM" - 1;
-/*-------------------------------------------------------------------------.
-| Actually print it. |
-| |
-| Plain and fancy file header block logging. Non-verbose just prints the |
-| name, e.g. for "tar t" or "tar x". This should just contain file names, |
-| so it can be fed back into tar with xargs or the "-T" option. The |
-| verbose option can give a bunch of info, one line per file. I doubt |
-| anybody tries to parse its format, or if they do, they shouldn't. Unix |
-| tar is pretty random here anyway. |
-`-------------------------------------------------------------------------*/
-
-/* FIXME: Note that print_header uses the globals HEAD, HSTAT, and
- HEAD_STANDARD, which must be set up in advance. Not very clean... */
-
-/* UGSWIDTH starts with 18, so with user and group names <= 8 chars, the
- columns never shift during the listing. */
-#define UGSWIDTH 18
-static int ugswidth = UGSWIDTH; /* maximum width encountered so far */
-
-/* DATEWIDTH is the number of columns taken by the date and time fields. */
-#if USE_OLD_CDATE
-# define DATEWIDTH 19
-#else
-# define DATEWIDTH 18
-#endif
+static bool volume_label_printed = false;
-void
-print_header (void)
+static void
+simple_print_header (struct tar_stat_info *st, union block *blk,
+ off_t block_ordinal)
{
char modes[11];
- char *timestamp;
- char uform[11], gform[11]; /* these hold formatted ints */
+ char const *time_stamp;
+ int time_stamp_len;
+ char *temp_name;
+
+ /* These hold formatted ints. */
+ char uform[UINTMAX_STRSIZE_BOUND], gform[UINTMAX_STRSIZE_BOUND];
char *user, *group;
char size[2 * UINTMAX_STRSIZE_BOUND];
/* holds formatted size or major,minor */
char uintbuf[UINTMAX_STRSIZE_BOUND];
- time_t longie; /* to make ctime() call portable */
int pad;
- char *name;
+ int sizelen;
+
+ if (show_transformed_names_option)
+ temp_name = st->file_name ? st->file_name : st->orig_file_name;
+ else
+ temp_name = st->orig_file_name ? st->orig_file_name : st->file_name;
if (block_number_option)
{
char buf[UINTMAX_STRSIZE_BOUND];
+ if (block_ordinal < 0)
+ block_ordinal = current_block_ordinal ();
+ block_ordinal -= recent_long_name_blocks;
+ block_ordinal -= recent_long_link_blocks;
fprintf (stdlis, _("block %s: "),
- STRINGIFY_BIGINT (current_block_ordinal (), buf));
+ STRINGIFY_BIGINT (block_ordinal, buf));
}
if (verbose_option <= 1)
{
/* Just the fax, mam. */
-
- char *quoted_name = quote_copy_string (current_file_name);
-
- if (quoted_name)
- {
- fprintf (stdlis, "%s\n", quoted_name);
- free (quoted_name);
- }
- else
- fprintf (stdlis, "%s\n", current_file_name);
+ fprintf (stdlis, "%s\n", quotearg (temp_name));
}
else
{
/* File type and modes. */
modes[0] = '?';
- switch (current_header->header.typeflag)
+ switch (blk->header.typeflag)
{
case GNUTYPE_VOLHDR:
+ volume_label_printed = true;
modes[0] = 'V';
break;
modes[0] = 'M';
break;
- case GNUTYPE_NAMES:
- modes[0] = 'N';
- break;
-
case GNUTYPE_LONGNAME:
case GNUTYPE_LONGLINK:
- ERROR ((0, 0, _("Visible longname error")));
+ modes[0] = 'L';
+ ERROR ((0, 0, _("Unexpected long name header")));
break;
case GNUTYPE_SPARSE:
case REGTYPE:
case AREGTYPE:
- case LNKTYPE:
modes[0] = '-';
- if (current_file_name[strlen (current_file_name) - 1] == '/')
+ if (temp_name[strlen (temp_name) - 1] == '/')
modes[0] = 'd';
break;
+ case LNKTYPE:
+ modes[0] = 'h';
+ break;
case GNUTYPE_DUMPDIR:
modes[0] = 'd';
break;
break;
}
- decode_mode (current_stat.st_mode, modes + 1);
+ pax_decode_mode (st->stat.st_mode, modes + 1);
- /* Timestamp. */
+ /* Time stamp. */
- longie = current_stat.st_mtime;
-#if USE_OLD_CTIME
- timestamp = ctime (&longie);
- timestamp[16] = '\0';
- timestamp[24] = '\0';
-#else
- timestamp = isotime (&longie);
- timestamp[16] = '\0';
-#endif
+ time_stamp = tartime (st->mtime, full_time_option);
+ time_stamp_len = strlen (time_stamp);
+ if (datewidth < time_stamp_len)
+ datewidth = time_stamp_len;
/* User and group names. */
- if (*current_header->header.uname && current_format != V7_FORMAT)
- user = current_header->header.uname;
+ if (st->uname
+ && st->uname[0]
+ && current_format != V7_FORMAT
+ && !numeric_owner_option)
+ user = st->uname;
else
- user = STRINGIFY_BIGINT (UINTMAX_FROM_OCT (current_header->header.uid),
- uform);
+ {
+ /* Try parsing it as an unsigned integer first, and as a
+ uid_t if that fails. This method can list positive user
+ ids that are too large to fit in a uid_t. */
+ uintmax_t u = from_header (blk->header.uid,
+ sizeof blk->header.uid, 0,
+ (uintmax_t) 0,
+ (uintmax_t) TYPE_MAXIMUM (uintmax_t),
+ false, false);
+ if (u != -1)
+ user = STRINGIFY_BIGINT (u, uform);
+ else
+ {
+ sprintf (uform, "%ld",
+ (long) UID_FROM_HEADER (blk->header.uid));
+ user = uform;
+ }
+ }
- if (*current_header->header.gname && current_format != V7_FORMAT)
- group = current_header->header.gname;
+ if (st->gname
+ && st->gname[0]
+ && current_format != V7_FORMAT
+ && !numeric_owner_option)
+ group = st->gname;
else
- group = STRINGIFY_BIGINT (UINTMAX_FROM_OCT
- (current_header->header.gid),
- gform);
+ {
+ /* Try parsing it as an unsigned integer first, and as a
+ gid_t if that fails. This method can list positive group
+ ids that are too large to fit in a gid_t. */
+ uintmax_t g = from_header (blk->header.gid,
+ sizeof blk->header.gid, 0,
+ (uintmax_t) 0,
+ (uintmax_t) TYPE_MAXIMUM (uintmax_t),
+ false, false);
+ if (g != -1)
+ group = STRINGIFY_BIGINT (g, gform);
+ else
+ {
+ sprintf (gform, "%ld",
+ (long) GID_FROM_HEADER (blk->header.gid));
+ group = gform;
+ }
+ }
/* Format the file size or major/minor device numbers. */
- switch (current_header->header.typeflag)
+ switch (blk->header.typeflag)
{
-#if defined(S_IFBLK) || defined(S_IFCHR)
case CHRTYPE:
case BLKTYPE:
- sprintf (size, "%lu,%lu",
- (unsigned long) major (current_stat.st_rdev),
- (unsigned long) minor (current_stat.st_rdev));
- break;
-#endif
- case GNUTYPE_SPARSE:
strcpy (size,
- STRINGIFY_BIGINT
- (UINTMAX_FROM_OCT (current_header->oldgnu_header.realsize),
- uintbuf));
+ STRINGIFY_BIGINT (major (st->stat.st_rdev), uintbuf));
+ strcat (size, ",");
+ strcat (size,
+ STRINGIFY_BIGINT (minor (st->stat.st_rdev), uintbuf));
break;
+
default:
- strcpy (size, STRINGIFY_BIGINT (current_stat.st_size, uintbuf));
+ /* st->stat.st_size keeps stored file size */
+ strcpy (size, STRINGIFY_BIGINT (st->stat.st_size, uintbuf));
break;
}
/* Figure out padding and print the whole line. */
- pad = strlen (user) + strlen (group) + strlen (size) + 1;
+ sizelen = strlen (size);
+ pad = strlen (user) + 1 + strlen (group) + 1 + sizelen;
if (pad > ugswidth)
ugswidth = pad;
-#if USE_OLD_CTIME
- fprintf (stdlis, "%s %s/%s %*s%s %s %s",
- modes, user, group, ugswidth - pad, "",
- size, timestamp + 4, timestamp + 20);
-#else
- fprintf (stdlis, "%s %s/%s %*s%s %s",
- modes, user, group, ugswidth - pad, "", size, timestamp);
-#endif
-
- name = quote_copy_string (current_file_name);
- if (name)
- {
- fprintf (stdlis, " %s", name);
- free (name);
- }
- else
- fprintf (stdlis, " %s", current_file_name);
+ fprintf (stdlis, "%s %s/%s %*s %-*s",
+ modes, user, group, ugswidth - pad + sizelen, size,
+ datewidth, time_stamp);
+
+ fprintf (stdlis, " %s", quotearg (temp_name));
- switch (current_header->header.typeflag)
+ switch (blk->header.typeflag)
{
case SYMTYPE:
- name = quote_copy_string (current_link_name);
- if (name)
- {
- fprintf (stdlis, " -> %s\n", name);
- free (name);
- }
- else
- fprintf (stdlis, " -> %s\n", current_link_name);
+ fprintf (stdlis, " -> %s\n", quotearg (st->link_name));
break;
case LNKTYPE:
- name = quote_copy_string (current_link_name);
- if (name)
- {
- fprintf (stdlis, _(" link to %s\n"), name);
- free (name);
- }
- else
- fprintf (stdlis, _(" link to %s\n"), current_link_name);
+ fprintf (stdlis, _(" link to %s\n"), quotearg (st->link_name));
break;
default:
- fprintf (stdlis, _(" unknown file type `%c'\n"),
- current_header->header.typeflag);
+ {
+ char type_string[2];
+ type_string[0] = blk->header.typeflag;
+ type_string[1] = '\0';
+ fprintf (stdlis, _(" unknown file type %s\n"),
+ quote (type_string));
+ }
break;
case AREGTYPE:
putc ('\n', stdlis);
break;
+ case GNUTYPE_LONGLINK:
+ fprintf (stdlis, _("--Long Link--\n"));
+ break;
+
+ case GNUTYPE_LONGNAME:
+ fprintf (stdlis, _("--Long Name--\n"));
+ break;
+
case GNUTYPE_VOLHDR:
fprintf (stdlis, _("--Volume Header--\n"));
break;
case GNUTYPE_MULTIVOL:
strcpy (size,
STRINGIFY_BIGINT
- (UINTMAX_FROM_OCT (current_header->oldgnu_header.offset),
+ (UINTMAX_FROM_HEADER (blk->oldgnu_header.offset),
uintbuf));
fprintf (stdlis, _("--Continued at byte %s--\n"), size);
break;
-
- case GNUTYPE_NAMES:
- fprintf (stdlis, _("--Mangled file names--\n"));
- break;
}
}
fflush (stdlis);
}
-/*--------------------------------------------------------------.
-| Print a similar line when we make a directory automatically. |
-`--------------------------------------------------------------*/
+
+static void
+print_volume_label (void)
+{
+ struct tar_stat_info vstat;
+ union block vblk;
+ enum archive_format dummy;
+
+ memset (&vblk, 0, sizeof (vblk));
+ vblk.header.typeflag = GNUTYPE_VOLHDR;
+ if (recent_global_header)
+ memcpy (vblk.header.mtime, recent_global_header->header.mtime,
+ sizeof vblk.header.mtime);
+ tar_stat_init (&vstat);
+ assign_string (&vstat.file_name, ".");
+ decode_header (&vblk, &vstat, &dummy, 0);
+ assign_string (&vstat.file_name, volume_label);
+ simple_print_header (&vstat, &vblk, 0);
+ tar_stat_destroy (&vstat);
+}
void
-print_for_mkdir (char *pathname, int length, mode_t mode)
+print_header (struct tar_stat_info *st, union block *blk,
+ off_t block_ordinal)
+{
+ if (current_format == POSIX_FORMAT && !volume_label_printed && volume_label)
+ {
+ print_volume_label ();
+ volume_label_printed = true;
+ }
+
+ simple_print_header (st, blk, block_ordinal);
+}
+
+/* Print a similar line when we make a directory automatically. */
+void
+print_for_mkdir (char *dirname, int length, mode_t mode)
{
char modes[11];
- char *name;
if (verbose_option > 1)
{
/* File type and modes. */
modes[0] = 'd';
- decode_mode (mode, modes + 1);
+ pax_decode_mode (mode, modes + 1);
if (block_number_option)
{
fprintf (stdlis, _("block %s: "),
STRINGIFY_BIGINT (current_block_ordinal (), buf));
}
- name = quote_copy_string (pathname);
- if (name)
- {
- fprintf (stdlis, "%s %*s %.*s\n", modes, ugswidth + DATEWIDTH,
- _("Creating directory:"), length, name);
- free (name);
- }
- else
- fprintf (stdlis, "%s %*s %.*s\n", modes, ugswidth + DATEWIDTH,
- _("Creating directory:"), length, pathname);
+
+ fprintf (stdlis, "%s %*s %.*s\n", modes, ugswidth + 1 + datewidth,
+ _("Creating directory:"), length, quotearg (dirname));
}
}
-/*--------------------------------------------------------.
-| Skip over SIZE bytes of data in blocks in the archive. |
-`--------------------------------------------------------*/
-
+/* Skip over SIZE bytes of data in blocks in the archive. */
void
skip_file (off_t size)
{
union block *x;
- if (multi_volume_option)
+ /* FIXME: Make sure mv_begin_read is always called before it */
+
+ if (seekable_archive)
{
- save_totsize = size;
- save_sizeleft = size;
+ off_t nblk = seek_archive (size);
+ if (nblk >= 0)
+ size -= nblk * BLOCKSIZE;
+ else
+ seekable_archive = false;
}
+ mv_size_left (size);
+
while (size > 0)
{
x = find_next_block ();
- if (x == NULL)
- FATAL_ERROR ((0, 0, _("Unexpected EOF on archive file")));
+ if (! x)
+ FATAL_ERROR ((0, 0, _("Unexpected EOF in archive")));
set_next_block_after (x);
size -= BLOCKSIZE;
- if (multi_volume_option)
- save_sizeleft -= BLOCKSIZE;
+ mv_size_left (size);
}
}
-/*---.
-| ? |
-`---*/
+/* Skip the current member in the archive.
+ NOTE: Current header must be decoded before calling this function. */
+void
+skip_member (void)
+{
+ if (!current_stat_info.skipped)
+ {
+ char save_typeflag = current_header->header.typeflag;
+ set_next_block_after (current_header);
+
+ mv_begin_read (¤t_stat_info);
+
+ if (current_stat_info.is_sparse)
+ sparse_skip_file (¤t_stat_info);
+ else if (save_typeflag != DIRTYPE)
+ skip_file (current_stat_info.stat.st_size);
+
+ mv_end ();
+ }
+}
void
-skip_extended_headers (void)
+test_archive_label ()
{
- union block *exhdr;
+ base64_init ();
+ name_gather ();
- while (1)
+ open_archive (ACCESS_READ);
+ if (read_header (¤t_header, ¤t_stat_info, read_header_auto)
+ == HEADER_SUCCESS)
{
- exhdr = find_next_block ();
- if (!exhdr->sparse_header.isextended)
+ decode_header (current_header,
+ ¤t_stat_info, ¤t_format, 0);
+ if (current_header->header.typeflag == GNUTYPE_VOLHDR)
+ assign_string (&volume_label, current_header->header.name);
+
+ if (volume_label)
{
- set_next_block_after (exhdr);
- break;
+ if (verbose_option)
+ print_volume_label ();
+ if (!name_match (volume_label) && multi_volume_option)
+ {
+ char *s = drop_volume_label_suffix (volume_label);
+ name_match (s);
+ free (s);
+ }
}
- set_next_block_after (exhdr);
}
+ close_archive ();
+ label_notfound ();
}