libtar: add support for xattr extensions

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

10 files changed, 274 insertions, 3 deletions

diff --git a/include/tar.h b/include/tar.h
index fd701d9..424be80 100644
--- a/include/tar.h
+++ b/include/tar.h
@@ -56,6 +56,13 @@ typedef struct {
 	char padding[7];
 } gnu_sparse_t;
 
+typedef struct tar_xattr_t {
+	struct tar_xattr_t *next;
+	char *key;
+	char *value;
+	char data[];
+} tar_xattr_t;
+
 typedef struct {
 	struct stat sb;
 	char *name;
@@ -64,6 +71,7 @@ typedef struct {
 	uint64_t actual_size;
 	uint64_t record_size;
 	bool unknown_record;
+	tar_xattr_t *xattr;
 } tar_header_decoded_t;
 
 #define TAR_TYPE_FILE '0'
diff --git a/lib/Makemodule.am b/lib/Makemodule.am
index 23ce70d..865d99a 100644
--- a/lib/Makemodule.am
+++ b/lib/Makemodule.am
@@ -10,7 +10,8 @@ libfstree_a_CPPFLAGS = $(AM_CPPFLAGS)
 libtar_a_SOURCES = lib/tar/read_header.c lib/tar/write_header.c lib/tar/skip.c
 libtar_a_SOURCES += lib/tar/number.c lib/tar/checksum.c lib/tar/cleanup.c
 libtar_a_SOURCES += lib/tar/read_sparse_map.c lib/tar/read_sparse_map_old.c
-libtar_a_SOURCES += lib/tar/internal.h include/tar.h
+libtar_a_SOURCES += lib/tar/base64.c lib/tar/urldecode.c lib/tar/internal.h
+libtar_a_SOURCES += include/tar.h
 libtar_a_CFLAGS = $(AM_CFLAGS)
 libtar_a_CPPFLAGS = $(AM_CPPFLAGS)
 
diff --git a/lib/tar/base64.c b/lib/tar/base64.c
new file mode 100644
index 0000000..313e9f4
--- /dev/null
+++ b/lib/tar/base64.c
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "internal.h"
+
+static uint8_t convert(char in)
+{
+	if (isupper(in))
+		return in - 'A';
+	if (islower(in))
+		return in - 'a' + 26;
+	if (isdigit(in))
+		return in - '0' + 52;
+	if (in == '+')
+		return 62;
+	if (in == '/' || in == '-')
+		return 63;
+	return 0;
+}
+
+void base64_decode(uint8_t *out, const char *in)
+{
+	char temp[4];
+
+	while (*in != '\0' && *in != '=') {
+		temp[0] = *in == '\0' ? 0 : convert(*(in++));
+		temp[1] = *in == '\0' ? 0 : convert(*(in++));
+		temp[2] = *in == '\0' ? 0 : convert(*(in++));
+		temp[3] = *in == '\0' ? 0 : convert(*(in++));
+
+		*(out++) = ((temp[0] << 2) & 0xFC) | ((temp[1] >> 4) & 0x03);
+		*(out++) = ((temp[1] << 4) & 0xF0) | ((temp[2] >> 2) & 0x0F);
+		*(out++) = ((temp[2] << 6) & 0xC0) | ( temp[3]       & 0x3F);
+	}
+}
diff --git a/lib/tar/cleanup.c b/lib/tar/cleanup.c
index c4d1734..a34c28b 100644
--- a/lib/tar/cleanup.c
+++ b/lib/tar/cleanup.c
@@ -12,8 +12,20 @@ void free_sparse_list(sparse_map_t *sparse)
 	}
 }
 
+void free_xattr_list(tar_xattr_t *list)
+{
+	tar_xattr_t *old;
+
+	while (list != NULL) {
+		old = list;
+		list = list->next;
+		free(old);
+	}
+}
+
 void clear_header(tar_header_decoded_t *hdr)
 {
+	free_xattr_list(hdr->xattr);
 	free_sparse_list(hdr->sparse);
 	free(hdr->name);
 	free(hdr->link_target);
diff --git a/lib/tar/internal.h b/lib/tar/internal.h
index 3c0e27f..04a97d3 100644
--- a/lib/tar/internal.h
+++ b/lib/tar/internal.h
@@ -50,4 +50,10 @@ sparse_map_t *read_gnu_old_sparse(int fd, tar_header_t *hdr);
 
 void free_sparse_list(sparse_map_t *sparse);
 
+void free_xattr_list(tar_xattr_t *list);
+
+void base64_decode(uint8_t *out, const char *in);
+
+void urldecode(char *str);
+
 #endif /* INTERNAL_H */
diff --git a/lib/tar/read_header.c b/lib/tar/read_header.c
index db40d7d..94c4a69 100644
--- a/lib/tar/read_header.c
+++ b/lib/tar/read_header.c
@@ -60,12 +60,29 @@ fail:
 	return NULL;
 }
 
+static tar_xattr_t *mkxattr(const char *key, size_t keylen,
+			    const char *value, size_t valuelen)
+{
+	tar_xattr_t *xattr;
+
+	xattr = calloc(1, sizeof(*xattr) + keylen + 1 + valuelen + 1);
+	if (xattr == NULL)
+		return NULL;
+
+	xattr->key = xattr->data;
+	xattr->value = xattr->data + keylen + 1;
+	memcpy(xattr->key, key, keylen);
+	memcpy(xattr->value, value, valuelen);
+	return xattr;
+}
+
 static int read_pax_header(int fd, uint64_t entsize, unsigned int *set_by_pax,
 			   tar_header_decoded_t *out)
 {
 	sparse_map_t *sparse_last = NULL, *sparse;
 	uint64_t field, offset = 0, num_bytes = 0;
-	char *buffer, *line;
+	char *buffer, *line, *key, *ptr, *value;
+	tar_xattr_t *xattr;
 	uint64_t i;
 
 	buffer = record_to_memory(fd, entsize);
@@ -183,6 +200,39 @@ static int read_pax_header(int fd, uint64_t entsize, unsigned int *set_by_pax,
 				sparse_last->next = sparse;
 				sparse_last = sparse;
 			}
+		} else if (!strncmp(line, "SCHILY.xattr.", 13)) {
+			key = line + 13;
+
+			ptr = strrchr(key, '=');
+			if (ptr == NULL || ptr == key)
+				continue;
+
+			value = ptr + 1;
+
+			xattr = mkxattr(key, ptr - key, value, strlen(value));
+			if (xattr == NULL)
+				goto fail_errno;
+
+			xattr->next = out->xattr;
+			out->xattr = xattr;
+		} else if (!strncmp(line, "LIBARCHIVE.xattr.", 17)) {
+			key = line + 17;
+
+			ptr = strrchr(key, '=');
+			if (ptr == NULL || ptr == key)
+				continue;
+
+			value = ptr + 1;
+
+			xattr = mkxattr(key, ptr - key, value, strlen(value));
+			if (xattr == NULL)
+				goto fail_errno;
+
+			urldecode(xattr->key);
+			base64_decode((uint8_t *)xattr->value, value);
+
+			xattr->next = out->xattr;
+			out->xattr = xattr;
 		}
 	}
 
diff --git a/lib/tar/urldecode.c b/lib/tar/urldecode.c
new file mode 100644
index 0000000..ac03f10
--- /dev/null
+++ b/lib/tar/urldecode.c
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "internal.h"
+
+static int xdigit(int x)
+{
+	if (isupper(x))
+		return x - 'A' + 0x0A;
+	if (islower(x))
+		return x - 'a' + 0x0A;
+	return x - '0';
+}
+
+void urldecode(char *str)
+{
+	unsigned char *out = (unsigned char *)str;
+	char *in = str;
+	int x;
+
+	while (*in != '\0') {
+		x = *(in++);
+
+		if (x == '%' && isxdigit(in[0]) && isxdigit(in[1])) {
+			x = xdigit(*(in++)) << 4;
+			x |= xdigit(*(in++));
+		}
+
+		*(out++) = x;
+	}
+
+	*out = '\0';
+}
diff --git a/tests/Makemodule.am b/tests/Makemodule.am
index 2c8344a..ad6532c 100644
--- a/tests/Makemodule.am
+++ b/tests/Makemodule.am
@@ -60,18 +60,28 @@ test_tar_sparse_gnu2_LDADD = libtar.a libutil.a
 test_tar_sparse_gnu2_CPPFLAGS = $(AM_CPPFLAGS)
 test_tar_sparse_gnu2_CPPFLAGS += -DTESTPATH=$(top_srcdir)/tests/tar
 
+test_tar_xattr_bsd_SOURCES = tests/tar_xattr_bsd.c
+test_tar_xattr_bsd_LDADD = libtar.a libutil.a
+test_tar_xattr_bsd_CPPFLAGS = $(AM_CPPFLAGS) -DTESTPATH=$(top_srcdir)/tests/tar
+
+test_tar_xattr_schily_SOURCES = tests/tar_xattr_schily.c
+test_tar_xattr_schily_LDADD = libtar.a libutil.a
+test_tar_xattr_schily_CPPFLAGS = $(AM_CPPFLAGS)
+test_tar_xattr_schily_CPPFLAGS += -DTESTPATH=$(top_srcdir)/tests/tar
+
 check_PROGRAMS += test_canonicalize_name test_mknode_simple test_mknode_slink
 check_PROGRAMS += test_mknode_reg test_mknode_dir test_gen_inode_table
 check_PROGRAMS += test_add_by_path test_get_path test_fstree_sort
 check_PROGRAMS += test_fstree_from_file test_fstree_init test_fstree_xattr
 check_PROGRAMS += test_tar_ustar test_tar_pax test_tar_gnu test_tar_sparse_gnu
 check_PROGRAMS += test_tar_sparse_gnu1 test_tar_sparse_gnu2
+check_PROGRAMS += test_tar_xattr_bsd test_tar_xattr_schily
 
 TESTS += test_canonicalize_name test_mknode_simple test_mknode_slink
 TESTS += test_mknode_reg test_mknode_dir test_gen_inode_table
 TESTS += test_add_by_path test_get_path test_fstree_sort test_fstree_from_file
 TESTS += test_fstree_init test_fstree_xattr test_tar_ustar test_tar_pax
 TESTS += test_tar_gnu test_tar_sparse_gnu test_tar_sparse_gnu1
-TESTS += test_tar_sparse_gnu2
+TESTS += test_tar_sparse_gnu2 test_tar_xattr_bsd test_tar_xattr_schily
 
 EXTRA_DIST += $(top_srcdir)/tests/tar
diff --git a/tests/tar_xattr_bsd.c b/tests/tar_xattr_bsd.c
new file mode 100644
index 0000000..cfd7ad7
--- /dev/null
+++ b/tests/tar_xattr_bsd.c
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <unistd.h>
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <stdio.h>
+
+#define STR(x) #x
+#define STRVALUE(x) STR(x)
+
+#define TEST_PATH STRVALUE(TESTPATH)
+
+static int open_read(const char *path)
+{
+	int fd = open(path, O_RDONLY);
+
+	if (fd < 0) {
+		perror(path);
+		exit(EXIT_FAILURE);
+	}
+
+	return fd;
+}
+
+int main(void)
+{
+	tar_header_decoded_t hdr;
+	char buffer[6];
+	int fd;
+
+	assert(chdir(TEST_PATH) == 0);
+
+	fd = open_read("xattr/xattr-libarchive.tar");
+	assert(read_header(fd, &hdr) == 0);
+	assert(hdr.sb.st_mode == (S_IFREG | 0644));
+	assert(hdr.sb.st_uid == 01750);
+	assert(hdr.sb.st_gid == 01750);
+	assert(hdr.sb.st_size == 5);
+	assert(hdr.sb.st_mtime == 1543094477);
+	assert(hdr.sb.st_atime == 1543094642);
+	assert(hdr.sb.st_ctime == 1543094606);
+	assert(strcmp(hdr.name, "input.txt") == 0);
+	assert(!hdr.unknown_record);
+	assert(read_retry(fd, buffer, 5) == 5);
+	buffer[5] = '\0';
+	assert(strcmp(buffer, "test\n") == 0);
+
+	assert(hdr.xattr != NULL);
+	assert(strcmp(hdr.xattr->key, "user.mime_type") == 0);
+	assert(strcmp(hdr.xattr->value, "text/plain") == 0);
+	assert(hdr.xattr->next == NULL);
+
+	clear_header(&hdr);
+	close(fd);
+	return EXIT_SUCCESS;
+}
diff --git a/tests/tar_xattr_schily.c b/tests/tar_xattr_schily.c
new file mode 100644
index 0000000..481e6ea
--- /dev/null
+++ b/tests/tar_xattr_schily.c
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "util.h"
+#include "tar.h"
+
+#include <unistd.h>
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <stdio.h>
+
+#define STR(x) #x
+#define STRVALUE(x) STR(x)
+
+#define TEST_PATH STRVALUE(TESTPATH)
+
+static int open_read(const char *path)
+{
+	int fd = open(path, O_RDONLY);
+
+	if (fd < 0) {
+		perror(path);
+		exit(EXIT_FAILURE);
+	}
+
+	return fd;
+}
+
+int main(void)
+{
+	tar_header_decoded_t hdr;
+	char buffer[6];
+	int fd;
+
+	assert(chdir(TEST_PATH) == 0);
+
+	fd = open_read("xattr/xattr-schily.tar");
+	assert(read_header(fd, &hdr) == 0);
+	assert(hdr.sb.st_mode == (S_IFREG | 0644));
+	assert(hdr.sb.st_uid == 01750);
+	assert(hdr.sb.st_gid == 01750);
+	assert(hdr.sb.st_size == 5);
+	assert(hdr.sb.st_mtime == 1543094477);
+	assert(hdr.sb.st_atime == 1543094642);
+	assert(hdr.sb.st_ctime == 1543094606);
+	assert(strcmp(hdr.name, "input.txt") == 0);
+	assert(!hdr.unknown_record);
+	assert(read_retry(fd, buffer, 5) == 5);
+	buffer[5] = '\0';
+	assert(strcmp(buffer, "test\n") == 0);
+
+	assert(hdr.xattr != NULL);
+	assert(strcmp(hdr.xattr->key, "user.mime_type") == 0);
+	assert(strcmp(hdr.xattr->value, "text/plain") == 0);
+	assert(hdr.xattr->next == NULL);
+
+	clear_header(&hdr);
+	close(fd);
+	return EXIT_SUCCESS;
+}