Add a single, central base64 decoder

Similar to the hex blob decoder, we need this once for tar and
once for the filemap xattr parser.

Simply add a single, central implementation to libutil, with a
simple unit test, and then use it in both libtar and gensquashfs.

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

7 files changed, 202 insertions, 132 deletions

diff --git a/bin/gensquashfs/filemap_xattr.c b/bin/gensquashfs/filemap_xattr.c
index 059ce81..0b73d36 100644
--- a/bin/gensquashfs/filemap_xattr.c
+++ b/bin/gensquashfs/filemap_xattr.c
@@ -11,25 +11,6 @@
 #define NEW_FILE_START "# file: "
 
 // Taken from attr-2.5.1/tools/setfattr.c
-static int
-base64_digit(char c) {
-	if (c >= 'A' && c <= 'Z')
-		return c - 'A';
-	else if (c >= 'a' && c <= 'z')
-		return 26 + c - 'a';
-	else if (c >= '0' && c <= '9')
-		return 52 + c - '0';
-	else if (c == '+')
-		return 62;
-	else if (c == '/')
-		return 63;
-	else if (c == '=')
-		return -2;
-	else
-		return -1;
-}
-
-// Taken from attr-2.5.1/tools/setfattr.c
 static char *
 decode(const char *value, size_t *size) {
 	char *decoded = NULL;
@@ -50,71 +31,20 @@ decode(const char *value, size_t *size) {
 			return NULL;
 		}
 	} else if (value[0] == '0' && (value[1] == 's' || value[1] == 'S')) {
-		const char *v = value + 2, *end = value + *size;
-		int d0, d1, d2, d3;
-		char *d;
+		size_t input_len = *size - 2;
 
-		decoded = realloc(decoded, *size / 4 * 3);
+		*size = (input_len / 4) * 3;
+
+		decoded = realloc(decoded, *size);
 		if (decoded == NULL) {
 			return NULL;
 		}
-		d = decoded;
-		for (;;) {
-			while (v < end && isspace(*v))
-				v++;
-			if (v == end) {
-				d0 = d1 = d2 = d3 = -2;
-				break;
-			}
-			if (v + 4 > end) {
-			bad_base64_encoding:
-				free(decoded);
-				fprintf(stderr, "bad input encoding\n");
-				return NULL;
-			}
-			d0 = base64_digit(*v++);
-			d1 = base64_digit(*v++);
-			d2 = base64_digit(*v++);
-			d3 = base64_digit(*v++);
-			if (d0 < 0 || d1 < 0 || d2 < 0 || d3 < 0)
-				break;
 
-			*d++ = (char)((d0 << 2) | (d1 >> 4));
-			*d++ = (char)((d1 << 4) | (d2 >> 2));
-			*d++ = (char)((d2 << 6) | d3);
-		}
-		if (d0 == -2) {
-			if (d1 != -2 || d2 != -2 || d3 != -2)
-				goto bad_base64_encoding;
-			goto base64_end;
-		}
-		if (d0 == -1 || d1 < 0 || d2 == -1 || d3 == -1)
-			goto bad_base64_encoding;
-		*d++ = (char)((d0 << 2) | (d1 >> 4));
-		if (d2 != -2)
-			*d++ = (char)((d1 << 4) | (d2 >> 2));
-		else {
-			if (d1 & 0x0F || d3 != -2)
-				goto bad_base64_encoding;
-			goto base64_end;
-		}
-		if (d3 != -2)
-			*d++ = (char)((d2 << 6) | d3);
-		else if (d2 & 0x03)
-			goto bad_base64_encoding;
-	base64_end:
-		while (v < end && isspace(*v))
-			v++;
-		if (v + 4 <= end && *v == '=') {
-			if (*++v != '=' || *++v != '=' || *++v != '=')
-				goto bad_base64_encoding;
-			v++;
+		if (base64_decode(value + 2, input_len, decoded, size)) {
+			free(decoded);
+			fprintf(stderr, "bad input encoding\n");
+			return NULL;
 		}
-		while (v < end && isspace(*v))
-			v++;
-		if (v < end)
-			goto bad_base64_encoding;
-		*size = d - decoded;
 	} else {
 		const char *v = value, *end = value + *size;
 		char *d;
diff --git a/include/util/util.h b/include/util/util.h
index 787580c..cd32887 100644
--- a/include/util/util.h
+++ b/include/util/util.h
@@ -80,4 +80,7 @@ SQFS_INTERNAL int check_file_range_equal(sqfs_file_t *file, void *scratch,
 SQFS_INTERNAL int hex_decode(const char *in, size_t in_sz,
 			     sqfs_u8 *out, size_t out_sz);
 
+SQFS_INTERNAL int base64_decode(const char *in, size_t in_len,
+				sqfs_u8 *out, size_t *out_len);
+
 #endif /* SQFS_UTIL_H */
diff --git a/lib/tar/pax_header.c b/lib/tar/pax_header.c
index cc2f2b0..b61aab6 100644
--- a/lib/tar/pax_header.c
+++ b/lib/tar/pax_header.c
@@ -11,56 +11,6 @@
 #include <string.h>
 #include <stdlib.h>
 
-static sqfs_u8 base64_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;
-}
-
-static size_t base64_decode(sqfs_u8 *out, const char *in, size_t len)
-{
-	sqfs_u8 *start = out;
-
-	while (len > 0) {
-		unsigned int diff = 0, value = 0;
-
-		while (diff < 4 && len > 0) {
-			if (*in == '=' || *in == '_' || *in == '\0') {
-				len = 0;
-			} else {
-				value = (value << 6) | base64_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;
-}
-
 static int pax_read_decimal(const char *str, sqfs_u64 *out)
 {
 	sqfs_u64 result = 0;
@@ -201,10 +151,16 @@ static int pax_xattr_schily(tar_header_decoded_t *out,
 static int pax_xattr_libarchive(tar_header_decoded_t *out,
 				tar_xattr_t *xattr)
 {
+	int ret;
+
+	ret = base64_decode((const char *)xattr->value, xattr->value_len,
+			    xattr->value, &xattr->value_len);
+	if (ret)
+		return -1;
+
 	urldecode(xattr->key);
-	xattr->value_len = base64_decode(xattr->value,
-					 (const char *)xattr->value,
-					 xattr->value_len);
+
+	xattr->value[xattr->value_len] = '\0';
 	xattr->next = out->xattr;
 	out->xattr = xattr;
 	return 0;
diff --git a/lib/util/Makemodule.am b/lib/util/Makemodule.am
index 4f1f0da..ec38b7a 100644
--- a/lib/util/Makemodule.am
+++ b/lib/util/Makemodule.am
@@ -15,6 +15,7 @@ libutil_a_SOURCES += lib/util/filename_sane.c
 libutil_a_SOURCES += lib/util/source_date_epoch.c
 libutil_a_SOURCES += lib/util/file_cmp.c
 libutil_a_SOURCES += lib/util/hex_decode.c
+libutil_a_SOURCES += lib/util/base64_decode.c
 libutil_a_CFLAGS = $(AM_CFLAGS)
 libutil_a_CPPFLAGS = $(AM_CPPFLAGS)
 
diff --git a/lib/util/base64_decode.c b/lib/util/base64_decode.c
new file mode 100644
index 0000000..b1cf5b6
--- /dev/null
+++ b/lib/util/base64_decode.c
@@ -0,0 +1,103 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * base64_decode.c
+ *
+ * Copyright (C) 2022 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "util/util.h"
+#include "util/test.h"
+
+#include <ctype.h>
+
+static int base64_digit(int c)
+{
+	if (isupper(c))
+		return c - 'A';
+	if (islower(c))
+		return c - 'a' + 26;
+	if (isdigit(c))
+		return c - '0' + 52;
+	if (c == '+')
+		return 62;
+	if (c == '/' || c == '-')
+		return 63;
+	return -1;
+}
+
+int base64_decode(const char *in, size_t in_len, sqfs_u8 *out, size_t *out_len)
+{
+	int i1, i2, i3, i4;
+	size_t count = 0;
+
+	while (in_len >= 4) {
+		i1 = base64_digit(*(in++));
+		i2 = base64_digit(*(in++));
+		i3 = *(in++);
+		i4 = *(in++);
+		in_len -= 4;
+
+		if (i1 < 0 || i2 < 0 || count >= *out_len)
+			goto fail;
+
+		out[count++] = (i1 << 2) | (i2 >> 4);
+
+		if (i3 == '=' || i3 == '_') {
+			if ((i4 != '=' && i4 != '_') || in_len > 0)
+				goto fail;
+			break;
+		}
+
+		i3 = base64_digit(i3);
+		if (i3 < 0 || count >= *out_len)
+			goto fail;
+
+		out[count++] = ((i2 & 0x0F) << 4) | (i3 >> 2);
+
+		if (i4 == '=' || i4 == '_') {
+			if (in_len > 0)
+				goto fail;
+			break;
+		}
+
+		i4 = base64_digit(i4);
+		if (i4 < 0 || count >= *out_len)
+			goto fail;
+
+		out[count++] = ((i3 & 0x3) << 6) | i4;
+	}
+
+	/* libarchive has this bizarre bastardization of truncated base64 */
+	if (in_len > 0) {
+		if (in_len == 1)
+			goto fail;
+
+		i1 = base64_digit(*(in++));
+		i2 = base64_digit(*(in++));
+		in_len -= 2;
+
+		if (i1 < 0 || i2 < 0 || count >= *out_len)
+			goto fail;
+
+		out[count++] = (i1 << 2) | (i2 >> 4);
+
+		if (in_len > 0) {
+			i3 = *(in++);
+			--in_len;
+
+			if (i3 != '=' && i3 != '_') {
+				i3 = base64_digit(i3);
+				if (i3 < 0 || count >= *out_len)
+					goto fail;
+
+				out[count++] = ((i2 & 0x0F) << 4) | (i3 >> 2);
+			}
+		}
+	}
+
+	*out_len = count;
+	return 0;
+fail:
+	*out_len = 0;
+	return -1;
+}
diff --git a/tests/libutil/Makemodule.am b/tests/libutil/Makemodule.am
index 5a69de4..724af50 100644
--- a/tests/libutil/Makemodule.am
+++ b/tests/libutil/Makemodule.am
@@ -33,10 +33,13 @@ test_sdate_epoch_LDADD = libutil.a libcompat.a
 test_hex_decode_SOURCES = tests/libutil/hex_decode.c
 test_hex_decode_LDADD = libutil.a libcompat.a
 
+test_base64_decode_SOURCES = tests/libutil/base64_decode.c
+test_base64_decode_LDADD = libutil.a libcompat.a
+
 LIBUTIL_TESTS = \
 	test_str_table test_rbtree test_xxhash test_threadpool test_ismemzero \
 	test_canonicalize_name test_filename_sane test_filename_sane_w32 \
-	test_sdate_epoch test_hex_decode
+	test_sdate_epoch test_hex_decode test_base64_decode
 
 check_PROGRAMS += $(LIBUTIL_TESTS)
 TESTS += $(LIBUTIL_TESTS)
diff --git a/tests/libutil/base64_decode.c b/tests/libutil/base64_decode.c
new file mode 100644
index 0000000..8f22a86
--- /dev/null
+++ b/tests/libutil/base64_decode.c
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * base64_decode.c
+ *
+ * Copyright (C) 2022 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "util/util.h"
+#include "util/test.h"
+
+static const struct {
+	int result;
+	const char *in;
+	const char *out;
+} test_vec[] = {
+	{ 0, "", "" },
+	{ 0, "Zg", "f" },
+	{ 0, "Zg==", "f" },
+	{ 0, "Zm8=", "fo" },
+	{ 0, "Zm9v", "foo" },
+	{ 0, "Zm9vYg==", "foob" },
+	{ 0, "Zm9vYmE=", "fooba" },
+	{ 0, "Zm9vYmFy", "foobar" },
+	{ 0, "TGV0J3MgYWxsIGxvdmUgTGFpbiEK", "Let's all love Lain!\n" },
+	{ -1, "Zg==X",     "XX" },
+};
+
+int main(int argc, char **argv)
+{
+	sqfs_u8 buffer[256];
+	size_t i, j;
+	(void)argc; (void)argv;
+
+	for (i = 0; i < sizeof(test_vec) / sizeof(test_vec[0]); ++i) {
+		const size_t in_len = strlen(test_vec[i].in);
+		const size_t out_len = strlen(test_vec[i].out);
+		size_t real_out;
+		int ret;
+
+		/* initialize the buffer */
+		for (j = 0; j < sizeof(buffer); ++j) {
+			buffer[j] = (j % 2) ? 0xAA : 0x55;
+		}
+
+		/* convert */
+		real_out = sizeof(buffer);
+		ret = base64_decode(test_vec[i].in, in_len, buffer, &real_out);
+
+		/* make sure pattern is un-touched after expected offset */
+		j = (in_len / 4) * 3;
+		if (in_len % 4)
+			j += 3;
+
+		for (; j < sizeof(buffer); ++j) {
+			TEST_ASSERT(buffer[j] == ((j % 2) ? 0xAA : 0x55));
+		}
+
+		/* check result */
+		if (test_vec[i].result == 0) {
+			TEST_ASSERT(ret == 0);
+			TEST_EQUAL_UI(real_out, out_len);
+			ret = memcmp(buffer, test_vec[i].out, out_len);
+			TEST_ASSERT(ret == 0);
+		} else {
+			TEST_ASSERT(ret != 0);
+			TEST_EQUAL_UI(real_out, 0);
+		}
+
+		fprintf(stderr, "CASE %lu OK\n", (unsigned long)i);
+	}
+
+	return EXIT_SUCCESS;
+}
+