Better support for reading/writing non-ASCII xattr values from/to tar

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

9 files changed, 50 insertions, 23 deletions

diff --git a/include/tar.h b/include/tar.h
index c496095..22563c4 100644
--- a/include/tar.h
+++ b/include/tar.h
@@ -71,7 +71,8 @@ typedef struct {
 typedef struct tar_xattr_t {
 	struct tar_xattr_t *next;
 	char *key;
-	char *value;
+	sqfs_u8 *value;
+	size_t value_len;
 	char data[];
 } tar_xattr_t;
 
diff --git a/lib/tar/base64.c b/lib/tar/base64.c
index 0ccf3c4..f88444a 100644
--- a/lib/tar/base64.c
+++ b/lib/tar/base64.c
@@ -23,18 +23,37 @@ static sqfs_u8 convert(char in)
 	return 0;
 }
 
-void base64_decode(sqfs_u8 *out, const char *in)
+size_t base64_decode(sqfs_u8 *out, const char *in, size_t len)
 {
-	char temp[4];
+	sqfs_u8 *start = out;
 
-	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++));
+	while (len > 0) {
+		unsigned int diff = 0, value = 0;
 
-		*(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);
+		while (diff < 4 && len > 0) {
+			if (*in == '=' || *in == '_' || *in == '\0') {
+				len = 0;
+			} else {
+				value = (value << 6) | convert(*(in++));
+				--len;
+				++diff;
+			}
+		}
+
+		if (diff < 2)
+			break;
+
+		value <<= 6 * (4 - diff);
+
+		switch (diff) {
+		case 4:  out[2] = value & 0xff; /* fall-through */
+		case 3:  out[1] = (value >> 8) & 0xff; /* fall-through */
+		default: out[0] = (value >> 16) & 0xff;
+		}
+
+		out += (diff * 3) / 4;
 	}
+
+	*out = '\0';
+	return out - start;
 }
diff --git a/lib/tar/internal.h b/lib/tar/internal.h
index 055637a..0220f05 100644
--- a/lib/tar/internal.h
+++ b/lib/tar/internal.h
@@ -61,7 +61,7 @@ void free_sparse_list(sparse_map_t *sparse);
 
 void free_xattr_list(tar_xattr_t *list);
 
-void base64_decode(sqfs_u8 *out, const char *in);
+size_t base64_decode(sqfs_u8 *out, const char *in, size_t len);
 
 void urldecode(char *str);
 
diff --git a/lib/tar/read_header.c b/lib/tar/read_header.c
index 845afc3..20e7a9f 100644
--- a/lib/tar/read_header.c
+++ b/lib/tar/read_header.c
@@ -68,7 +68,8 @@ static tar_xattr_t *mkxattr(const char *key, size_t keylen,
 		return NULL;
 
 	xattr->key = xattr->data;
-	xattr->value = xattr->data + keylen + 1;
+	xattr->value = (sqfs_u8 *)xattr->data + keylen + 1;
+	xattr->value_len = valuelen;
 	memcpy(xattr->key, key, keylen);
 	memcpy(xattr->value, value, valuelen);
 	return xattr;
@@ -201,7 +202,8 @@ static int read_pax_header(FILE *fp, sqfs_u64 entsize, unsigned int *set_by_pax,
 
 			value = ptr + 1;
 
-			xattr = mkxattr(key, ptr - key, value, strlen(value));
+			xattr = mkxattr(key, ptr - key,
+					value, len - (value - line) - 1);
 			if (xattr == NULL)
 				goto fail_errno;
 
@@ -221,7 +223,8 @@ static int read_pax_header(FILE *fp, sqfs_u64 entsize, unsigned int *set_by_pax,
 				goto fail_errno;
 
 			urldecode(xattr->key);
-			base64_decode((sqfs_u8 *)xattr->value, value);
+			xattr->value_len = base64_decode(xattr->value, value,
+							 xattr->value_len);
 
 			xattr->next = out->xattr;
 			out->xattr = xattr;
diff --git a/lib/tar/write_header.c b/lib/tar/write_header.c
index c235014..acce280 100644
--- a/lib/tar/write_header.c
+++ b/lib/tar/write_header.c
@@ -133,7 +133,7 @@ static int write_schily_xattr(FILE *fp, const struct stat *orig,
 	struct stat sb;
 
 	for (it = xattr; it != NULL; it = it->next) {
-		len = strlen(prefix) + strlen(it->key) + strlen(it->value) + 2;
+		len = strlen(prefix) + strlen(it->key) + it->value_len + 2;
 
 		total_size += num_digits(len) + 1 + len;
 	}
@@ -146,11 +146,12 @@ static int write_schily_xattr(FILE *fp, const struct stat *orig,
 		return -1;
 
 	for (it = xattr; it != NULL; it = it->next) {
-		len = strlen(prefix) + strlen(it->key) + strlen(it->value) + 2;
+		len = strlen(prefix) + strlen(it->key) + it->value_len + 2;
 		len += num_digits(len) + 1;
 
-		fprintf(fp, PRI_SZ " %s%s=%s\n", len,
-			prefix, it->key, it->value);
+		fprintf(fp, PRI_SZ " %s%s=", len, prefix, it->key);
+		fwrite(it->value, 1, it->value_len, fp);
+		fputc('\n', fp);
 	}
 
 	return padd_file(fp, total_size);
diff --git a/tar/sqfs2tar.c b/tar/sqfs2tar.c
index 6767b95..58fb611 100644
--- a/tar/sqfs2tar.c
+++ b/tar/sqfs2tar.c
@@ -267,9 +267,10 @@ static int get_xattrs(const char *name, const sqfs_inode_generic_t *inode,
 		ent->key = ent->data;
 		strcpy(ent->key, (const char *)key->key);
 
-		ent->value = ent->key + strlen(ent->key) + 1;
+		ent->value = (sqfs_u8 *)ent->key + strlen(ent->key) + 1;
 		memcpy(ent->value, value->value, value->size + 1);
 
+		ent->value_len = value->size;
 		ent->next = list;
 		list = ent;
 
diff --git a/tar/tar2sqfs.c b/tar/tar2sqfs.c
index 5d40fc5..6be682f 100644
--- a/tar/tar2sqfs.c
+++ b/tar/tar2sqfs.c
@@ -334,7 +334,7 @@ static int copy_xattr(tree_node_t *node, const tar_header_decoded_t *hdr)
 		}
 
 		ret = sqfs_xattr_writer_add(sqfs.xwr, xattr->key, xattr->value,
-					    strlen(xattr->value));
+					    xattr->value_len);
 		if (ret) {
 			sqfs_perror(hdr->name, "storing xattr key-value pair",
 				    ret);
diff --git a/tests/tar_xattr_bsd.c b/tests/tar_xattr_bsd.c
index dcaba1e..0741511 100644
--- a/tests/tar_xattr_bsd.c
+++ b/tests/tar_xattr_bsd.c
@@ -54,7 +54,8 @@ int main(void)
 
 	assert(hdr.xattr != NULL);
 	assert(strcmp(hdr.xattr->key, "user.mime_type") == 0);
-	assert(strcmp(hdr.xattr->value, "text/plain") == 0);
+	assert(strcmp((const char *)hdr.xattr->value, "text/plain") == 0);
+	assert(hdr.xattr->value_len == 10);
 	assert(hdr.xattr->next == NULL);
 
 	clear_header(&hdr);
diff --git a/tests/tar_xattr_schily.c b/tests/tar_xattr_schily.c
index 6cc31f5..f562b43 100644
--- a/tests/tar_xattr_schily.c
+++ b/tests/tar_xattr_schily.c
@@ -54,7 +54,8 @@ int main(void)
 
 	assert(hdr.xattr != NULL);
 	assert(strcmp(hdr.xattr->key, "user.mime_type") == 0);
-	assert(strcmp(hdr.xattr->value, "text/plain") == 0);
+	assert(strcmp((const char *)hdr.xattr->value, "text/plain") == 0);
+	assert(hdr.xattr->value_len == 10);
 	assert(hdr.xattr->next == NULL);
 
 	clear_header(&hdr);