wrong behaviour of 'strverscmp'

[黄建忠 <jianzhong.huang@i-soft.com.cn>]

Hi,

When I try to run some testcases, I found 'strverscmp' do not work like
expected.

According to man page, 'Thus, the ordering is 000, 00, 01, 010, 09, 0,
1, 9, 10.'.

So the return value of strverscmp('000', '00) should less than zero, but
musl return '0'-0(*l - *r), that is 48.

some test cases here:
ASSERT (strverscmp ("", "") == 0);
ASSERT (strverscmp ("a", "a") == 0);
ASSERT (strverscmp ("a", "b") < 0);
ASSERT (strverscmp ("b", "a") > 0);
ASSERT (strverscmp ("000", "00") < 0);
ASSERT (strverscmp ("00", "000") > 0);
ASSERT (strverscmp ("a0", "a") > 0);
ASSERT (strverscmp ("00", "01") < 0);
ASSERT (strverscmp ("01", "010") < 0);
ASSERT (strverscmp ("010", "09") < 0);
ASSERT (strverscmp ("09", "0") < 0);
ASSERT (strverscmp ("9", "10") < 0);
ASSERT (strverscmp ("0a", "0") > 0);

Here is a correct version from gnulib and uclibc use this one. Maybe it
can not be used directly. (License issue.)

/* states: S_N: normal, S_I: comparing integral part, S_F: comparing
fractionnal parts, S_Z: idem but with leading Zeroes only */
#define S_N 0x0
#define S_I 0x4
#define S_F 0x8
#define S_Z 0xC

/* result_type: CMP: return diff; LEN: compare using len_diff/diff */
#define CMP 2
#define LEN 3

/* Compare S1 and S2 as strings holding indices/version numbers,
returning less than, equal to or greater than zero if S1 is less than,
equal to or greater than S2 (for more info, see the texinfo doc).
*/
int strverscmp (const char *s1, const char *s2)
{
const unsigned char *p1 = (const unsigned char *) s1;
const unsigned char *p2 = (const unsigned char *) s2;
unsigned char c1, c2;
int state;
int diff;

/* Symbol(s) 0 [1-9] others (padding)
Transition (10) 0 (01) d (00) x (11) - */
static const unsigned int next_state[] =
{
/* state x d 0 - */
/* S_N */ S_N, S_I, S_Z, S_N,
/* S_I */ S_N, S_I, S_I, S_I,
/* S_F */ S_N, S_F, S_F, S_F,
/* S_Z */ S_N, S_F, S_Z, S_Z
};
static const int result_type[] =
{
/* state x/x x/d x/0 x/- d/x d/d d/0 d/-
0/x 0/d 0/0 0/- -/x -/d -/0 -/- */

/* S_N */ CMP, CMP, CMP, CMP, CMP, LEN, CMP, CMP,
CMP, CMP, CMP, CMP, CMP, CMP, CMP, CMP,
/* S_I */ CMP, -1, -1, CMP, +1, LEN, LEN, CMP,
+1, LEN, LEN, CMP, CMP, CMP, CMP, CMP,
/* S_F */ CMP, CMP, CMP, CMP, CMP, LEN, CMP, CMP,
CMP, CMP, CMP, CMP, CMP, CMP, CMP, CMP,
/* S_Z */ CMP, +1, +1, CMP, -1, CMP, CMP, CMP,
-1, CMP, CMP, CMP
};

if (p1 == p2)
return 0;

c1 = *p1++;
c2 = *p2++;
/* Hint: '0' is a digit too. */
state = S_N | ((c1 == '0') + (isdigit (c1) != 0));
while ((diff = c1 - c2) == 0 && c1 != '\0')
{
state = next_state[state];
c1 = *p1++;
c2 = *p2++;
state |= (c1 == '0') + (isdigit (c1) != 0);
}

state = result_type[state << 2 | (((c2 == '0') + (isdigit (c2) != 0)))];

switch (state)
{
case CMP:
return diff;

case LEN:
while (isdigit (*p1++))
if (!isdigit (*p2++))
return 1;

return isdigit (*p2) ? -1 : diff;

default:
return state;
}
}

-- 
Huang JianZhong