/**
* Gets all meta data for a file and returns a filled meta_data_t *
* structure.
* @param file_event is the structure that contains directory and fileinfo
* structures. directory is the directory we are iterating over. It
* is used here to build the filename name. fileinfo is a glib
* structure that contains all meta data and more for a file.
* @param filter is a structure containing all structures needed to filter
* files as database, regex_exclude_list and contains a boolean
* that is set to TRUE if the file has been filtered out and FALSE
* otherwise.
* @param opt are the selected options for the program.
* @returns a newly allocated and filled meta_data_t * structure.
*/
static meta_data_t *get_meta_data_from_fileinfo(file_event_t *file_event, filter_file_t *filter, options_t *opt)
{
meta_data_t *meta = NULL;
gchar *directory = NULL;
GFileInfo *fileinfo = NULL;
if (file_event != NULL)
{
directory = file_event->directory;
fileinfo = file_event->fileinfo;
}
if (directory != NULL && fileinfo != NULL && filter != NULL && filter->database != NULL)
{
/* filling meta data for the file represented by fileinfo */
meta = new_meta_data_t();
meta->file_type = g_file_info_get_file_type(fileinfo);
meta->name = g_build_path(G_DIR_SEPARATOR_S, directory, g_file_info_get_name(fileinfo), NULL);
if (exclude_file(filter->regex_exclude_list, meta->name) == FALSE)
{
/* fills meta_data_t *meta structure */
get_file_attributes(meta, fileinfo);
meta->blocksize = calculate_file_blocksize(opt, meta->size);
/* We need to determine if the file has already been saved by looking into the local database
* This is usefull only when carving directories at the begining of the process as when called
* by m_fanotify we already know that the file was written and that something changed.
*/
meta->in_cache = is_file_in_cache(filter->database, meta);
filter->excluded = FALSE;
}
else
{
filter->excluded = TRUE;
}
}
return meta;
}
/* RMBIN -- Remove all binaries in a directory or in a directory tree.
* We chdir to each directory to minimize path searches.
*/
static void
rmbin (
char *dir,
int recurse,
char *path /* pathname of current directory */
)
{
char newpath[SZ_PATHNAME+1];
char fname[SZ_PATHNAME+1];
int dp, ftype;
if ((dp = os_diropen (dir)) == ERR) {
fprintf (stderr, "cannot open directory `%s'\n", dir);
fflush (stderr);
return;
}
sprintf (newpath, "%s%s/", path, dir);
/* Descend into the subdirectory.
*/
if (strcmp (dir, ".") != 0)
if (os_chdir (dir) == ERR) {
os_dirclose (dp);
fprintf (stderr, "cannot change directory to `%s'\n", newpath);
fflush (stderr);
return;
}
/* Scan through the directory.
*/
while (os_gfdir (dp, fname, SZ_PATHNAME) > 0) {
if (os_symlink (fname, 0, 0))
continue;
if ((ftype = os_filetype (fname)) == DIRECTORY_FILE)
rmbin (fname, recurse, newpath);
else {
if (only[0] != NULL) {
if (exclude_file (fname))
continue;
} else if (ftype != BINARY_FILE || exclude_file (fname))
continue;
/* We have a binary file which is not excluded from deletion
* by its extension, so delete it.
*/
if (interactive && (verify_delete (fname, newpath) == NO))
continue;
if (verbose) {
printf ("%s%s\n", newpath, fname);
fflush (stdout);
}
if (execute)
if (os_delete (fname) == ERR) {
fprintf (stderr, "cannot delete `%s'\n", fname);
fflush (stderr);
}
}
}
/* Return from the subdirectory.
*/
if (strcmp (dir, ".") != 0)
if (os_chdir ("..") == ERR) {
fprintf (stderr, "cannot return from subdirectory `%s'\n",
newpath);
fflush (stderr);
}
os_dirclose (dp);
}
static int FAST_FUNC writeFileToTarball(const char *fileName, struct stat *statbuf,
void *userData, int depth UNUSED_PARAM)
{
struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData;
const char *header_name;
int inputFileFd = -1;
/* Strip leading '/' (must be before memorizing hardlink's name) */
header_name = fileName;
while (header_name[0] == '/') {
static smallint warned;
if (!warned) {
bb_error_msg("removing leading '/' from member names");
warned = 1;
}
header_name++;
}
if (header_name[0] == '\0')
return TRUE;
/* It is against the rules to archive a socket */
if (S_ISSOCK(statbuf->st_mode)) {
bb_error_msg("%s: socket ignored", fileName);
return TRUE;
}
/*
* Check to see if we are dealing with a hard link.
* If so -
* Treat the first occurance of a given dev/inode as a file while
* treating any additional occurances as hard links. This is done
* by adding the file information to the HardLinkInfo linked list.
*/
tbInfo->hlInfo = NULL;
if (statbuf->st_nlink > 1) {
tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf);
if (tbInfo->hlInfo == NULL)
addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, header_name);
}
/* It is a bad idea to store the archive we are in the process of creating,
* so check the device and inode to be sure that this particular file isn't
* the new tarball */
if (tbInfo->statBuf.st_dev == statbuf->st_dev
&& tbInfo->statBuf.st_ino == statbuf->st_ino
) {
bb_error_msg("%s: file is the archive; skipping", fileName);
return TRUE;
}
if (exclude_file(tbInfo->excludeList, header_name))
return SKIP;
#if !ENABLE_FEATURE_TAR_GNU_EXTENSIONS
if (strlen(header_name) >= NAME_SIZE) {
bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported");
return TRUE;
}
#endif
/* Is this a regular file? */
if (tbInfo->hlInfo == NULL && S_ISREG(statbuf->st_mode)) {
/* open the file we want to archive, and make sure all is well */
inputFileFd = open_or_warn(fileName, O_RDONLY);
if (inputFileFd < 0) {
return FALSE;
}
}
/* Add an entry to the tarball */
if (writeTarHeader(tbInfo, header_name, fileName, statbuf) == FALSE) {
return FALSE;
}
/* If it was a regular file, write out the body */
if (inputFileFd >= 0) {
size_t readSize;
/* Write the file to the archive. */
/* We record size into header first, */
/* and then write out file. If file shrinks in between, */
/* tar will be corrupted. So we don't allow for that. */
/* NB: GNU tar 1.16 warns and pads with zeroes */
/* or even seeks back and updates header */
bb_copyfd_exact_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
////off_t readSize;
////readSize = bb_copyfd_size(inputFileFd, tbInfo->tarFd, statbuf->st_size);
////if (readSize != statbuf->st_size && readSize >= 0) {
//// bb_error_msg_and_die("short read from %s, aborting", fileName);
////}
/* Check that file did not grow in between? */
/* if (safe_read(inputFileFd, 1) == 1) warn but continue? */
close(inputFileFd);
/* Pad the file up to the tar block size */
/* (a few tricks here in the name of code size) */
readSize = (-(int)statbuf->st_size) & (TAR_BLOCK_SIZE-1);
memset(block_buf, 0, readSize);
xwrite(tbInfo->tarFd, block_buf, readSize);
}
return TRUE;
}
static int writeFileToTarball(const char *fileName, struct stat *statbuf,
void *userData)
{
struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData;
const char *header_name;
/*
** Check to see if we are dealing with a hard link.
** If so -
** Treat the first occurance of a given dev/inode as a file while
** treating any additional occurances as hard links. This is done
** by adding the file information to the HardLinkInfo linked list.
*/
tbInfo->hlInfo = NULL;
if (statbuf->st_nlink > 1) {
tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf);
if (tbInfo->hlInfo == NULL)
addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, fileName);
}
/* It is against the rules to archive a socket */
if (S_ISSOCK(statbuf->st_mode)) {
bb_error_msg("%s: socket ignored", fileName);
return (TRUE);
}
/* It is a bad idea to store the archive we are in the process of creating,
* so check the device and inode to be sure that this particular file isn't
* the new tarball */
if (tbInfo->statBuf.st_dev == statbuf->st_dev &&
tbInfo->statBuf.st_ino == statbuf->st_ino) {
bb_error_msg("%s: file is the archive; skipping", fileName);
return (TRUE);
}
header_name = fileName;
while (header_name[0] == '/') {
static int alreadyWarned = FALSE;
if (alreadyWarned == FALSE) {
bb_error_msg("Removing leading '/' from member names");
alreadyWarned = TRUE;
}
header_name++;
}
if (strlen(fileName) >= NAME_SIZE) {
bb_error_msg(bb_msg_name_longer_than_foo, NAME_SIZE);
return (TRUE);
}
if (header_name[0] == '\0')
return TRUE;
# ifdef CONFIG_FEATURE_TAR_FROM
if (exclude_file(tbInfo->excludeList, header_name)) {
return SKIP;
}
# endif /* CONFIG_FEATURE_TAR_FROM */
if (writeTarHeader(tbInfo, header_name, fileName, statbuf) == FALSE) {
return (FALSE);
}
/* Now, if the file is a regular file, copy it out to the tarball */
if ((tbInfo->hlInfo == NULL)
&& (S_ISREG(statbuf->st_mode))) {
int inputFileFd;
ssize_t readSize = 0;
/* open the file we want to archive, and make sure all is well */
if ((inputFileFd = open(fileName, O_RDONLY)) < 0) {
bb_perror_msg("%s: Cannot open", fileName);
return (FALSE);
}
/* write the file to the archive */
readSize = bb_copyfd_eof(inputFileFd, tbInfo->tarFd);
/* Pad the file up to the tar block size */
for (; (readSize % TAR_BLOCK_SIZE) != 0; readSize++) {
write(tbInfo->tarFd, "\0", 1);
}
close(inputFileFd);
}
return (TRUE);
}
/*
* Read a tar file and extract or list the specified files within it.
* If the list is empty than all files are extracted or listed.
*/
static int readTarFile(int tarFd, int extractFlag, int listFlag,
int tostdoutFlag, int verboseFlag, char** extractList,
char** excludeList)
{
int status;
int errorFlag=FALSE;
int skipNextHeaderFlag=FALSE;
TarHeader rawHeader;
TarInfo header;
/* Read the tar file, and iterate over it one file at a time */
while ( (status = full_read(tarFd, (char*)&rawHeader, TAR_BLOCK_SIZE)) == TAR_BLOCK_SIZE ) {
/* Try to read the header */
if ( readTarHeader(&rawHeader, &header) == FALSE ) {
if ( *(header.name) == '\0' ) {
goto endgame;
} else {
errorFlag=TRUE;
error_msg("Bad tar header, skipping");
continue;
}
}
if ( *(header.name) == '\0' )
continue;
header.tarFd = tarFd;
/* Skip funky extra GNU headers that precede long files */
if ( (header.type == GNULONGNAME) || (header.type == GNULONGLINK) ) {
skipNextHeaderFlag=TRUE;
if (tarExtractRegularFile(&header, FALSE, FALSE) == FALSE)
errorFlag = TRUE;
continue;
}
if ( skipNextHeaderFlag == TRUE ) {
skipNextHeaderFlag=FALSE;
error_msg(name_longer_than_foo, NAME_SIZE);
if (tarExtractRegularFile(&header, FALSE, FALSE) == FALSE)
errorFlag = TRUE;
continue;
}
#if defined BB_FEATURE_TAR_EXCLUDE
if (exclude_file(excludeList, header.name)) {
/* There are not the droids you're looking for, move along */
/* If it is a regular file, pretend to extract it with
* the extractFlag set to FALSE, so the junk in the tarball
* is properly skipped over */
if ( header.type==REGTYPE || header.type==REGTYPE0 ) {
if (tarExtractRegularFile(&header, FALSE, FALSE) == FALSE)
errorFlag = TRUE;
}
continue;
}
#endif
if (!extract_file(extractList, header.name)) {
/* There are not the droids you're looking for, move along */
/* If it is a regular file, pretend to extract it with
* the extractFlag set to FALSE, so the junk in the tarball
* is properly skipped over */
if ( header.type==REGTYPE || header.type==REGTYPE0 ) {
if (tarExtractRegularFile(&header, FALSE, FALSE) == FALSE)
errorFlag = TRUE;
}
continue;
}
if (listFlag == TRUE) {
/* Special treatment if the list (-t) flag is on */
if (verboseFlag == TRUE) {
int len, len1;
char buf[35];
struct tm *tm = localtime (&(header.mtime));
len=printf("%s ", mode_string(header.mode));
my_getpwuid(buf, header.uid);
if (! *buf)
len+=printf("%d", header.uid);
else
len+=printf("%s", buf);
my_getgrgid(buf, header.gid);
if (! *buf)
len+=printf("/%-d ", header.gid);
else
len+=printf("/%-s ", buf);
if (header.type==CHRTYPE || header.type==BLKTYPE) {
len1=snprintf(buf, sizeof(buf), "%ld,%-ld ",
header.devmajor, header.devminor);
} else {
len1=snprintf(buf, sizeof(buf), "%lu ", (long)header.size);
}
/* Jump through some hoops to make the columns match up */
for(;(len+len1)<31;len++)
printf(" ");
printf(buf);
/* Use ISO 8610 time format */
if (tm) {
printf ("%04d-%02d-%02d %02d:%02d:%02d ",
tm->tm_year + 1900, tm->tm_mon + 1, tm->tm_mday,
tm->tm_hour, tm->tm_min, tm->tm_sec);
}
}
printf("%s", header.name);
if (verboseFlag == TRUE) {
if (header.type==LNKTYPE) /* If this is a link, say so */
printf(" link to %s", header.linkname);
else if (header.type==SYMTYPE)
printf(" -> %s", header.linkname);
}
printf("\n");
}
/* List contents if we are supposed to do that */
if (verboseFlag == TRUE && extractFlag == TRUE) {
/* Now the normal listing */
FILE *vbFd = stdout;
if (tostdoutFlag == TRUE) // If the archive goes to stdout, verbose to stderr
vbFd = stderr;
fprintf(vbFd, "%s\n", header.name);
}
/* Remove files if we would overwrite them */
if (extractFlag == TRUE && tostdoutFlag == FALSE)
unlink(header.name);
/* If we got here, we can be certain we have a legitimate
* header to work with. So work with it. */
switch ( header.type ) {
case REGTYPE:
case REGTYPE0:
/* If the name ends in a '/' then assume it is
* supposed to be a directory, and fall through */
if (!last_char_is(header.name,'/')) {
if (tarExtractRegularFile(&header, extractFlag, tostdoutFlag)==FALSE)
errorFlag=TRUE;
break;
}
case DIRTYPE:
if (tarExtractDirectory( &header, extractFlag, tostdoutFlag)==FALSE)
errorFlag=TRUE;
break;
case LNKTYPE:
if (tarExtractHardLink( &header, extractFlag, tostdoutFlag)==FALSE)
errorFlag=TRUE;
break;
case SYMTYPE:
if (tarExtractSymLink( &header, extractFlag, tostdoutFlag)==FALSE)
errorFlag=TRUE;
break;
case CHRTYPE:
case BLKTYPE:
case FIFOTYPE:
if (tarExtractSpecial( &header, extractFlag, tostdoutFlag)==FALSE)
errorFlag=TRUE;
break;
#if 0
/* Handled earlier */
case GNULONGNAME:
case GNULONGLINK:
skipNextHeaderFlag=TRUE;
break;
#endif
default:
error_msg("Unknown file type '%c' in tar file", header.type);
close( tarFd);
return( FALSE);
}
}
close(tarFd);
if (status > 0) {
/* Bummer - we read a partial header */
perror_msg("Error reading tar file");
return ( FALSE);
}
else if (errorFlag==TRUE) {
error_msg( "Error exit delayed from previous errors");
return( FALSE);
} else
return( status);
/* Stuff to do when we are done */
endgame:
close( tarFd);
if ( *(header.name) == '\0' ) {
if (errorFlag==TRUE)
error_msg( "Error exit delayed from previous errors");
else
return( TRUE);
}
return( FALSE);
}
