basic: util - add base32hexmem() function similar to hexmem()

This implements more of RFC4648.

3 files changed, 530 insertions, 0 deletions

diff --git a/src/basic/util.c b/src/basic/util.c
index bc917ae574..999ef8fd26 100644
--- a/src/basic/util.c
+++ b/src/basic/util.c
@@ -954,6 +954,351 @@ int unhexmem(const char *p, size_t l, void **mem, size_t *len) {
         return 0;
 }
 
+/* https://tools.ietf.org/html/rfc4648#section-6 */
+char base32hexchar(int x) {
+        static const char table[32] = "0123456789"
+                                      "ABCDEFGHIJKLMNOPQRSTUV";
+
+        return table[x & 31];
+}
+
+int unbase32hexchar(char c) {
+        unsigned offset;
+
+        if (c >= '0' && c <= '9')
+                return c - '0';
+
+        offset = '9' - '0' + 1;
+
+        if (c >= 'A' && c <= 'V')
+                return c - 'A' + offset;
+
+        return -EINVAL;
+}
+
+char *base32hexmem(const void *p, size_t l, bool padding) {
+        char *r, *z;
+        const uint8_t *x;
+        size_t len;
+
+        if (padding)
+                /* five input bytes makes eight output bytes, padding is added so we must round up */
+                len = 8 * (l + 4) / 5;
+        else {
+                /* same, but round down as there is no padding */
+                len = 8 * l / 5;
+
+                switch (l % 5) {
+                case 4:
+                        len += 7;
+                        break;
+                case 3:
+                        len += 5;
+                        break;
+                case 2:
+                        len += 4;
+                        break;
+                case 1:
+                        len += 2;
+                        break;
+                }
+        }
+
+        z = r = malloc(len + 1);
+        if (!r)
+                return NULL;
+
+        for (x = p; x < (const uint8_t*) p + (l / 5) * 5; x += 5) {
+                /* x[0] == XXXXXXXX; x[1] == YYYYYYYY; x[2] == ZZZZZZZZ
+                   x[3] == QQQQQQQQ; x[4] == WWWWWWWW */
+                *(z++) = base32hexchar(x[0] >> 3);                    /* 000XXXXX */
+                *(z++) = base32hexchar((x[0] & 7) << 2 | x[1] >> 6);  /* 000XXXYY */
+                *(z++) = base32hexchar((x[1] & 63) >> 1);             /* 000YYYYY */
+                *(z++) = base32hexchar((x[1] & 1) << 4 | x[2] >> 4);  /* 000YZZZZ */
+                *(z++) = base32hexchar((x[2] & 15) << 1 | x[3] >> 7); /* 000ZZZZQ */
+                *(z++) = base32hexchar((x[3] & 127) >> 2);            /* 000QQQQQ */
+                *(z++) = base32hexchar((x[3] & 3) << 3 | x[4] >> 5);  /* 000QQWWW */
+                *(z++) = base32hexchar((x[4] & 31));                  /* 000WWWWW */
+        }
+
+        switch (l % 5) {
+        case 4:
+                *(z++) = base32hexchar(x[0] >> 3);                    /* 000XXXXX */
+                *(z++) = base32hexchar((x[0] & 7) << 2 | x[1] >> 6);  /* 000XXXYY */
+                *(z++) = base32hexchar((x[1] & 63) >> 1);             /* 000YYYYY */
+                *(z++) = base32hexchar((x[1] & 1) << 4 | x[2] >> 4);   /* 000YZZZZ */
+                *(z++) = base32hexchar((x[2] & 15) << 1 | x[3] >> 7); /* 000ZZZZQ */
+                *(z++) = base32hexchar((x[3] & 127) >> 2);            /* 000QQQQQ */
+                *(z++) = base32hexchar((x[3] & 3) << 3);              /* 000QQ000 */
+                if (padding)
+                        *(z++) = '=';
+
+                break;
+
+        case 3:
+                *(z++) = base32hexchar(x[0] >> 3);                   /* 000XXXXX */
+                *(z++) = base32hexchar((x[0] & 7) << 2 | x[1] >> 6); /* 000XXXYY */
+                *(z++) = base32hexchar((x[1] & 63) >> 1);            /* 000YYYYY */
+                *(z++) = base32hexchar((x[1] & 1) << 4 | x[2] >> 4); /* 000YZZZZ */
+                *(z++) = base32hexchar((x[2] & 15) << 1);            /* 000ZZZZ0 */
+                if (padding) {
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                }
+
+                break;
+
+        case 2:
+                *(z++) = base32hexchar(x[0] >> 3);                   /* 000XXXXX */
+                *(z++) = base32hexchar((x[0] & 7) << 2 | x[1] >> 6); /* 000XXXYY */
+                *(z++) = base32hexchar((x[1] & 63) >> 1);            /* 000YYYYY */
+                *(z++) = base32hexchar((x[1] & 1) << 4);             /* 000Y0000 */
+                if (padding) {
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                }
+
+                break;
+
+        case 1:
+                *(z++) = base32hexchar(x[0] >> 3);       /* 000XXXXX */
+                *(z++) = base32hexchar((x[0] & 7) << 2); /* 000XXX00 */
+                if (padding) {
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                        *(z++) = '=';
+                }
+
+                break;
+        }
+
+        *z = 0;
+        return r;
+}
+
+int unbase32hexmem(const char *p, size_t l, bool padding, void **mem, size_t *_len) {
+        _cleanup_free_ uint8_t *r = NULL;
+        int a, b, c, d, e, f, g, h;
+        uint8_t *z;
+        const char *x;
+        size_t len;
+        unsigned pad = 0;
+
+        assert(p);
+
+        /* padding ensures any base32hex input has input divisible by 8 */
+        if (padding && l % 8 != 0)
+                return -EINVAL;
+
+        if (padding) {
+                /* strip the padding */
+                while (l > 0 && p[l - 1] == '=' && pad < 7) {
+                        pad ++;
+                        l --;
+                }
+        }
+
+        /* a group of eight input bytes needs five output bytes, in case of
+           padding we need to add some extra bytes */
+        len = (l / 8) * 5;
+
+        switch (l % 8) {
+        case 7:
+                len += 4;
+                break;
+        case 5:
+                len += 3;
+                break;
+        case 4:
+                len += 2;
+                break;
+        case 2:
+                len += 1;
+                break;
+        case 0:
+                break;
+        default:
+                return -EINVAL;
+        }
+
+        z = r = malloc(len + 1);
+        if (!r)
+                return -ENOMEM;
+
+        for (x = p; x < p + (l / 8) * 8; x += 8) {
+                /* a == 000XXXXX; b == 000YYYYY; c == 000ZZZZZ; d == 000WWWWW
+                   e == 000SSSSS; f == 000QQQQQ; g == 000VVVVV; h == 000RRRRR */
+                a = unbase32hexchar(x[0]);
+                if (a < 0)
+                        return -EINVAL;
+
+                b = unbase32hexchar(x[1]);
+                if (b < 0)
+                        return -EINVAL;
+
+                c = unbase32hexchar(x[2]);
+                if (c < 0)
+                        return -EINVAL;
+
+                d = unbase32hexchar(x[3]);
+                if (d < 0)
+                        return -EINVAL;
+
+                e = unbase32hexchar(x[4]);
+                if (e < 0)
+                        return -EINVAL;
+
+                f = unbase32hexchar(x[5]);
+                if (f < 0)
+                        return -EINVAL;
+
+                g = unbase32hexchar(x[6]);
+                if (g < 0)
+                        return -EINVAL;
+
+                h = unbase32hexchar(x[7]);
+                if (h < 0)
+                        return -EINVAL;
+
+                *(z++) = (uint8_t) a << 3 | (uint8_t) b >> 2;                    /* XXXXXYYY */
+                *(z++) = (uint8_t) b << 6 | (uint8_t) c << 1 | (uint8_t) d >> 4; /* YYZZZZZW */
+                *(z++) = (uint8_t) d << 4 | (uint8_t) e >> 1;                    /* WWWWSSSS */
+                *(z++) = (uint8_t) e << 7 | (uint8_t) f << 2 | (uint8_t) g >> 3; /* SQQQQQVV */
+                *(z++) = (uint8_t) g << 5 | (uint8_t) h;                         /* VVVRRRRR */
+        }
+
+        switch (l % 8) {
+        case 7:
+                a = unbase32hexchar(x[0]);
+                if (a < 0)
+                        return -EINVAL;
+
+                b = unbase32hexchar(x[1]);
+                if (b < 0)
+                        return -EINVAL;
+
+                c = unbase32hexchar(x[2]);
+                if (c < 0)
+                        return -EINVAL;
+
+                d = unbase32hexchar(x[3]);
+                if (d < 0)
+                        return -EINVAL;
+
+                e = unbase32hexchar(x[4]);
+                if (e < 0)
+                        return -EINVAL;
+
+                f = unbase32hexchar(x[5]);
+                if (f < 0)
+                        return -EINVAL;
+
+                g = unbase32hexchar(x[6]);
+                if (g < 0)
+                        return -EINVAL;
+
+                /* g == 000VV000 */
+                if (g & 7)
+                        return -EINVAL;
+
+                *(z++) = (uint8_t) a << 3 | (uint8_t) b >> 2;                    /* XXXXXYYY */
+                *(z++) = (uint8_t) b << 6 | (uint8_t) c << 1 | (uint8_t) d >> 4; /* YYZZZZZW */
+                *(z++) = (uint8_t) d << 4 | (uint8_t) e >> 1;                    /* WWWWSSSS */
+                *(z++) = (uint8_t) e << 7 | (uint8_t) f << 2 | (uint8_t) g >> 3; /* SQQQQQVV */
+
+                break;
+        case 5:
+                a = unbase32hexchar(x[0]);
+                if (a < 0)
+                        return -EINVAL;
+
+                b = unbase32hexchar(x[1]);
+                if (b < 0)
+                        return -EINVAL;
+
+                c = unbase32hexchar(x[2]);
+                if (c < 0)
+                        return -EINVAL;
+
+                d = unbase32hexchar(x[3]);
+                if (d < 0)
+                        return -EINVAL;
+
+                e = unbase32hexchar(x[4]);
+                if (e < 0)
+                        return -EINVAL;
+
+                /* e == 000SSSS0 */
+                if (e & 1)
+                        return -EINVAL;
+
+                *(z++) = (uint8_t) a << 3 | (uint8_t) b >> 2;                    /* XXXXXYYY */
+                *(z++) = (uint8_t) b << 6 | (uint8_t) c << 1 | (uint8_t) d >> 4; /* YYZZZZZW */
+                *(z++) = (uint8_t) d << 4 | (uint8_t) e >> 1;                    /* WWWWSSSS */
+
+                break;
+        case 4:
+                a = unbase32hexchar(x[0]);
+                if (a < 0)
+                        return -EINVAL;
+
+                b = unbase32hexchar(x[1]);
+                if (b < 0)
+                        return -EINVAL;
+
+                c = unbase32hexchar(x[2]);
+                if (c < 0)
+                        return -EINVAL;
+
+                d = unbase32hexchar(x[3]);
+                if (d < 0)
+                        return -EINVAL;
+
+                /* d == 000W0000 */
+                if (d & 15)
+                        return -EINVAL;
+
+                *(z++) = (uint8_t) a << 3 | (uint8_t) b >> 2;                    /* XXXXXYYY */
+                *(z++) = (uint8_t) b << 6 | (uint8_t) c << 1 | (uint8_t) d >> 4; /* YYZZZZZW */
+
+                break;
+        case 2:
+                a = unbase32hexchar(x[0]);
+                if (a < 0)
+                        return -EINVAL;
+
+                b = unbase32hexchar(x[1]);
+                if (b < 0)
+                        return -EINVAL;
+
+                /* b == 000YYY00 */
+                if (b & 3)
+                        return -EINVAL;
+
+                *(z++) = (uint8_t) a << 3 | (uint8_t) b >> 2; /* XXXXXYYY */
+
+                break;
+        case 0:
+                break;
+        default:
+                return -EINVAL;
+        }
+
+        *z = 0;
+
+        *mem = r;
+        r = NULL;
+        *_len = len;
+
+        return 0;
+}
+
 /* https://tools.ietf.org/html/rfc4648#section-4 */
 char base64char(int x) {
         static const char table[64] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
@@ -1117,6 +1462,11 @@ int unbase64mem(const char *p, size_t l, void **mem, size_t *_len) {
                 *(z++) = (uint8_t) a << 2 | (uint8_t) (b >> 4); /* XXXXXXYY */
 
                 break;
+        case 0:
+
+                break;
+        default:
+                return -EINVAL;
         }
 
         *z = 0;
diff --git a/src/basic/util.h b/src/basic/util.h
index dae43006e4..c2e5cc610b 100644
--- a/src/basic/util.h
+++ b/src/basic/util.h
@@ -240,6 +240,8 @@ char octchar(int x) _const_;
 int unoctchar(char c) _const_;
 char decchar(int x) _const_;
 int undecchar(char c) _const_;
+char base32hexchar(int x) _const_;
+int unbase32hexchar(char c) _const_;
 char base64char(int x) _const_;
 int unbase64char(char c) _const_;
 
@@ -618,6 +620,9 @@ static inline void *mempset(void *s, int c, size_t n) {
 char *hexmem(const void *p, size_t l);
 int unhexmem(const char *p, size_t l, void **mem, size_t *len);
 
+char *base32hexmem(const void *p, size_t l, bool padding);
+int unbase32hexmem(const char *p, size_t l, bool padding, void **mem, size_t *len);
+
 char *base64mem(const void *p, size_t l);
 int unbase64mem(const char *p, size_t l, void **mem, size_t *len);
 
diff --git a/src/test/test-util.c b/src/test/test-util.c
index 72fbc345c2..7906c4d7bb 100644
--- a/src/test/test-util.c
+++ b/src/test/test-util.c
@@ -390,6 +390,21 @@ static void test_unhexchar(void) {
         assert_se(unhexchar('0') == 0x0);
 }
 
+static void test_base32hexchar(void) {
+        assert_se(base32hexchar(0) == '0');
+        assert_se(base32hexchar(9) == '9');
+        assert_se(base32hexchar(10) == 'A');
+        assert_se(base32hexchar(31) == 'V');
+}
+
+static void test_unbase32hexchar(void) {
+        assert_se(unbase32hexchar('0') == 0);
+        assert_se(unbase32hexchar('9') == 9);
+        assert_se(unbase32hexchar('A') == 10);
+        assert_se(unbase32hexchar('V') == 31);
+        assert_se(unbase32hexchar('=') == -EINVAL);
+}
+
 static void test_base64char(void) {
         assert_se(base64char(0) == 'A');
         assert_se(base64char(26) == 'a');
@@ -453,6 +468,162 @@ static void test_unhexmem(void) {
 }
 
 /* https://tools.ietf.org/html/rfc4648#section-10 */
+static void test_base32hexmem(void) {
+        char *b32;
+
+        b32 = base32hexmem("", strlen(""), true);
+        assert_se(b32);
+        assert_se(streq(b32, ""));
+        free(b32);
+
+        b32 = base32hexmem("f", strlen("f"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CO======"));
+        free(b32);
+
+        b32 = base32hexmem("fo", strlen("fo"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNG===="));
+        free(b32);
+
+        b32 = base32hexmem("foo", strlen("foo"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMU==="));
+        free(b32);
+
+        b32 = base32hexmem("foob", strlen("foob"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOG="));
+        free(b32);
+
+        b32 = base32hexmem("fooba", strlen("fooba"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOJ1"));
+        free(b32);
+
+        b32 = base32hexmem("foobar", strlen("foobar"), true);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOJ1E8======"));
+        free(b32);
+
+        b32 = base32hexmem("", strlen(""), false);
+        assert_se(b32);
+        assert_se(streq(b32, ""));
+        free(b32);
+
+        b32 = base32hexmem("f", strlen("f"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CO"));
+        free(b32);
+
+        b32 = base32hexmem("fo", strlen("fo"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNG"));
+        free(b32);
+
+        b32 = base32hexmem("foo", strlen("foo"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMU"));
+        free(b32);
+
+        b32 = base32hexmem("foob", strlen("foob"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOG"));
+        free(b32);
+
+        b32 = base32hexmem("fooba", strlen("fooba"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOJ1"));
+        free(b32);
+
+        b32 = base32hexmem("foobar", strlen("foobar"), false);
+        assert_se(b32);
+        assert_se(streq(b32, "CPNMUOJ1E8"));
+        free(b32);
+}
+
+static void test_unbase32hexmem(void) {
+        void *mem;
+        size_t len;
+
+        assert_se(unbase32hexmem("", strlen(""), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), ""));
+        free(mem);
+
+        assert_se(unbase32hexmem("CO======", strlen("CO======"), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "f"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNG====", strlen("CPNG===="), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "fo"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMU===", strlen("CPNMU==="), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foo"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOG=", strlen("CPNMUOG="), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foob"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOJ1", strlen("CPNMUOJ1"), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "fooba"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOJ1E8======", strlen("CPNMUOJ1E8======"), true, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foobar"));
+        free(mem);
+
+        assert_se(unbase32hexmem("A", strlen("A"), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("A=======", strlen("A======="), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAA=====", strlen("AAA====="), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAAA==", strlen("AAAAAA=="), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AB======", strlen("AB======"), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAB====", strlen("AAAB===="), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAB===", strlen("AAAAB==="), true, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAAAB=", strlen("AAAAAAB="), true, &mem, &len) == -EINVAL);
+
+        assert_se(unbase32hexmem("", strlen(""), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), ""));
+        free(mem);
+
+        assert_se(unbase32hexmem("CO", strlen("CO"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "f"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNG", strlen("CPNG"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "fo"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMU", strlen("CPNMU"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foo"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOG", strlen("CPNMUOG"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foob"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOJ1", strlen("CPNMUOJ1"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "fooba"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOJ1E8", strlen("CPNMUOJ1E8"), false, &mem, &len) == 0);
+        assert_se(streq(strndupa(mem, len), "foobar"));
+        free(mem);
+
+        assert_se(unbase32hexmem("CPNMUOG=", strlen("CPNMUOG="), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("CPNMUOJ1E8======", strlen("CPNMUOJ1E8======"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("A", strlen("A"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("A", strlen("A"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAA", strlen("AAA"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAAA", strlen("AAAAAA"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AB", strlen("AB"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAB", strlen("AAAB"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAB", strlen("AAAAB"), false, &mem, &len) == -EINVAL);
+        assert_se(unbase32hexmem("AAAAAAB", strlen("AAAAAAB"), false, &mem, &len) == -EINVAL);
+}
+
+/* https://tools.ietf.org/html/rfc4648#section-10 */
 static void test_base64mem(void) {
         char *b64;
 
@@ -1924,6 +2095,8 @@ int main(int argc, char *argv[]) {
         test_in_charset();
         test_hexchar();
         test_unhexchar();
+        test_base32hexchar();
+        test_unbase32hexchar();
         test_base64char();
         test_unbase64char();
         test_octchar();
@@ -1931,6 +2104,8 @@ int main(int argc, char *argv[]) {
         test_decchar();
         test_undecchar();
         test_unhexmem();
+        test_base32hexmem();
+        test_unbase32hexmem();
         test_base64mem();
         test_unbase64mem();
         test_cescape();