hexdecoct: ignore whitespace within the input hexadecimal text of unhexmem()

2 files changed, 81 insertions, 25 deletions

diff --git a/src/basic/hexdecoct.c b/src/basic/hexdecoct.c
index 7ab5c5f70..1c1949f1f 100644
--- a/src/basic/hexdecoct.c
+++ b/src/basic/hexdecoct.c
@@ -77,33 +77,69 @@ char *hexmem(const void *p, size_t l) {
         return r;
 }
 
-int unhexmem(const char *p, size_t l, void **mem, size_t *len) {
-        _cleanup_free_ uint8_t *r = NULL;
-        uint8_t *z;
+static int unhex_next(const char **p, size_t *l) {
+        int r;
+
+        assert(p);
+        assert(l);
+
+        /* Find the next non-whitespace character, and decode it. We
+         * greedily skip all preceeding and all following whitespace. */
+
+        for (;;) {
+                if (*l == 0)
+                        return -EPIPE;
+
+                if (!strchr(WHITESPACE, **p))
+                        break;
+
+                /* Skip leading whitespace */
+                (*p)++, (*l)--;
+        }
+
+        r = unhexchar(**p);
+        if (r < 0)
+                return r;
+
+        for (;;) {
+                (*p)++, (*l)--;
+
+                if (*l == 0 || !strchr(WHITESPACE, **p))
+                        break;
+
+                /* Skip following whitespace */
+        }
+
+        return r;
+}
+
+int unhexmem(const char *p, size_t l, void **ret, size_t *ret_len) {
+        _cleanup_free_ uint8_t *buf = NULL;
         const char *x;
+        uint8_t *z;
 
-        assert(mem);
-        assert(len);
+        assert(ret);
+        assert(ret_len);
         assert(p || l == 0);
 
         if (l == (size_t) -1)
                 l = strlen(p);
 
-        if (l % 2 != 0)
-                return -EINVAL;
-
-        z = r = malloc((l + 1) / 2 + 1);
-        if (!r)
+        /* Note that the calculation of memory size is an upper boundary, as we ignore whitespace while decoding */
+        buf = malloc((l + 1) / 2 + 1);
+        if (!buf)
                 return -ENOMEM;
 
-        for (x = p; x < p + l; x += 2) {
+        for (x = p, z = buf;;) {
                 int a, b;
 
-                a = unhexchar(x[0]);
+                a = unhex_next(&x, &l);
+                if (a == -EPIPE) /* End of string */
+                        break;
                 if (a < 0)
                         return a;
 
-                b = unhexchar(x[1]);
+                b = unhex_next(&x, &l);
                 if (b < 0)
                         return b;
 
@@ -112,8 +148,8 @@ int unhexmem(const char *p, size_t l, void **mem, size_t *len) {
 
         *z = 0;
 
-        *mem = TAKE_PTR(r);
-        *len = (l + 1) / 2;
+        *ret_len = (size_t) (z - buf);
+        *ret = TAKE_PTR(buf);
 
         return 0;
 }
diff --git a/src/test/test-hexdecoct.c b/src/test/test-hexdecoct.c
index 74b8e3b68..048e8cb82 100644
--- a/src/test/test-hexdecoct.c
+++ b/src/test/test-hexdecoct.c
@@ -79,20 +79,40 @@ static void test_undecchar(void) {
         assert_se(undecchar('9') == 9);
 }
 
-static void test_unhexmem(void) {
-        const char *hex = "efa2149213";
-        const char *hex_invalid = "efa214921o";
-        _cleanup_free_ char *hex2 = NULL;
+static void test_unhexmem_one(const char *s, size_t l, int retval) {
+        _cleanup_free_ char *hex = NULL;
         _cleanup_free_ void *mem = NULL;
         size_t len;
 
-        assert_se(unhexmem(hex_invalid, strlen(hex_invalid), &mem, &len) == -EINVAL);
-        assert_se(unhexmem(hex, strlen(hex) + 1, &mem, &len) == -EINVAL);
-        assert_se(unhexmem(hex, strlen(hex) - 1, &mem, &len) == -EINVAL);
-        assert_se(unhexmem(hex, strlen(hex), &mem, &len) == 0);
+        assert_se(unhexmem(s, l, &mem, &len) == retval);
+        if (retval == 0) {
+                char *answer;
+
+                if (l == (size_t) - 1)
+                        l = strlen(s);
+
+                assert_se((hex = hexmem(mem, len)));
+                answer = strndupa(s, l);
+                assert_se(streq(delete_chars(answer, WHITESPACE), hex));
+        }
+}
+
+static void test_unhexmem(void) {
+        const char *hex = "efa2149213";
+        const char *hex_space = "  e f   a\n 2\r  14\n\r\t9\t2 \n1\r3 \r\r\t";
+        const char *hex_invalid = "efa214921o";
 
-        assert_se((hex2 = hexmem(mem, len)));
-        assert_se(streq(hex, hex2));
+        test_unhexmem_one(NULL, 0, 0);
+        test_unhexmem_one("", 0, 0);
+        test_unhexmem_one("", (size_t) -1, 0);
+        test_unhexmem_one("   \n \t\r   \t\t \n\n\n", (size_t) -1, 0);
+        test_unhexmem_one(hex_invalid, strlen(hex_invalid), -EINVAL);
+        test_unhexmem_one(hex_invalid, (size_t) - 1, -EINVAL);
+        test_unhexmem_one(hex, strlen(hex) - 1, -EPIPE);
+        test_unhexmem_one(hex, strlen(hex), 0);
+        test_unhexmem_one(hex, (size_t) -1, 0);
+        test_unhexmem_one(hex_space, strlen(hex_space), 0);
+        test_unhexmem_one(hex_space, (size_t) -1, 0);
 }
 
 /* https://tools.ietf.org/html/rfc4648#section-10 */