![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
diff --git a/src/core/data.c b/src/core/data.c
index 0e40474edaabab45e98b9dc30d31b2b244de33d9..914946fbec6761d3e7b4345742ddbd63668106c1 100644 (file)
--- a/src/core/data.c
+++ b/src/core/data.c
#include <math.h>
+/*
+ * Operator support maxtrix.
+ * <type1> <op> <type2> -> op_matrix[<op>][<type1>][<type2>]
+ */
+
+/* add, sub, mul, div, mod, concat */
+
+/* integer, decimal, string, datetime, binary, regex */
+
+static int op_matrix[6][6][6] = {
+ /* SDB_DATA_ADD */
+ {
+ { SDB_TYPE_INTEGER, -1, -1, -1, -1, -1 },
+ { -1, SDB_TYPE_DECIMAL, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+
+ /* SDB_DATA_SUB */
+ {
+ { SDB_TYPE_INTEGER, -1, -1, -1, -1, -1 },
+ { -1, SDB_TYPE_DECIMAL, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+
+ /* SDB_DATA_MUL */
+ {
+ { SDB_TYPE_INTEGER, -1, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, SDB_TYPE_DECIMAL, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { SDB_TYPE_DATETIME, SDB_TYPE_DATETIME, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+
+ /* SDB_DATA_DIV */
+ {
+ { SDB_TYPE_INTEGER, -1, -1, -1, -1, -1 },
+ { -1, SDB_TYPE_DECIMAL, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { SDB_TYPE_DATETIME, SDB_TYPE_DATETIME, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+
+ /* SDB_DATA_MOD */
+ {
+ { SDB_TYPE_INTEGER, -1, -1, -1, -1, -1 },
+ { -1, SDB_TYPE_DECIMAL, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { SDB_TYPE_DATETIME, SDB_TYPE_DATETIME, -1, SDB_TYPE_DATETIME, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+
+ /* SDB_DATA_CONCAT */
+ {
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, SDB_TYPE_STRING, -1, -1, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ { -1, -1, -1, -1, SDB_TYPE_BINARY, -1 },
+ { -1, -1, -1, -1, -1, -1 },
+ },
+};
+
/*
* private helper functions
*/
v[i] = NULL;
}
}
+ else if (type == SDB_TYPE_BINARY) {
+ struct {
+ size_t length;
+ unsigned char *datum;
+ } *v = datum->data.array.values;
+ size_t i;
+
+ for (i = 0; i < datum->data.array.length; ++i) {
+ if (v[i].datum)
+ free(v[i].datum);
+ v[i].datum = NULL;
+ }
+ }
+ else if (type == SDB_TYPE_REGEX) {
+ struct {
+ char *raw;
+ regex_t regex;
+ } *v = datum->data.array.values;
+ size_t i;
+
+ for (i = 0; i < datum->data.array.length; ++i) {
+ if (v[i].raw) {
+ free(v[i].raw);
+ regfree(&v[i].regex);
+ }
+ v[i].raw = NULL;
+ }
+ }
} /* free_array_values */
+/* compare two arrays element-by-element returning how the first non-equal
+ * elements compare to each other */
+static int
+array_cmp(const sdb_data_t *a1, const sdb_data_t *a2)
+{
+ int type = a1->type & 0xff;
+ size_t len, i;
+
+ assert((a1->type == a2->type) && (a1->type & SDB_TYPE_ARRAY));
+
+ len = SDB_MIN(a1->data.array.length, a2->data.array.length);
+
+ if (type == SDB_TYPE_INTEGER) {
+ int64_t *v1 = a1->data.array.values;
+ int64_t *v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i)
+ if (v1[i] != v2[i])
+ return SDB_CMP(v1[i], v2[i]);
+ }
+ else if (type == SDB_TYPE_DECIMAL) {
+ double *v1 = a1->data.array.values;
+ double *v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i)
+ if (v1[i] != v2[i])
+ return SDB_CMP(v1[i], v2[i]);
+ }
+ else if (type == SDB_TYPE_STRING) {
+ char **v1 = a1->data.array.values;
+ char **v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i) {
+ int diff = strcasecmp(v1[i], v2[i]);
+ if (diff)
+ return diff;
+ }
+ }
+ else if (type == SDB_TYPE_DATETIME) {
+ sdb_time_t *v1 = a1->data.array.values;
+ sdb_time_t *v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i)
+ if (v1[i] != v2[i])
+ return SDB_CMP(v1[i], v2[i]);
+ }
+ else if (type == SDB_TYPE_BINARY) {
+ struct {
+ size_t length;
+ unsigned char *datum;
+ } *v1 = a1->data.array.values;
+ struct {
+ size_t length;
+ unsigned char *datum;
+ } *v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i) {
+ int diff;
+
+ /* on a common prefix, the shorter datum sorts less */
+ if (v1[i].length < v2[i].length) {
+ diff = memcmp(v1[i].datum, v2[i].datum, v1[i].length);
+ diff = diff ? diff : -1;
+ }
+ else if (v1[i].length > v2[i].length) {
+ diff = memcmp(v1[i].datum, v2[i].datum, v2[i].length);
+ diff = diff ? diff : 1;
+ }
+ else
+ diff = memcmp(v1[i].datum, v2[i].datum, v1[i].length);
+
+ if (diff)
+ return diff;
+ }
+ }
+ else if (type == SDB_TYPE_REGEX) {
+ struct {
+ char *raw;
+ regex_t regex;
+ } *v1 = a1->data.array.values;
+ struct {
+ char *raw;
+ regex_t regex;
+ } *v2 = a2->data.array.values;
+
+ for (i = 0; i < len; ++i) {
+ int diff = strcasecmp(v1[i].raw, v2[i].raw);
+ if (diff)
+ return diff;
+ }
+ }
+ else {
+ errno = EINVAL;
+ /* but fall through to ensure stable sorting: */
+ }
+ return SDB_CMP(a1->data.array.length, a2->data.array.length);
+} /* array_cmp */
+
/* Calculate the linear function 'd1 + n * d2'. */
static int
data_lin(const sdb_data_t *d1, int n, const sdb_data_t *d2, sdb_data_t *res)
data_concat(const sdb_data_t *d1, const sdb_data_t *d2, sdb_data_t *res)
{
unsigned char *new;
- unsigned char *s1, *s2;
- size_t len1, len2;
+ const unsigned char *s1, *s2;
+ size_t len1, len2, array1_len = 0, array2_len = 0;
- /* TODO: support array plus element */
- if (d1->type != d2->type)
+ if ((d1->type & 0xff) != (d2->type & 0xff))
return -1;
- if (d1->type == SDB_TYPE_STRING) {
+ if ((d1->type & SDB_TYPE_ARRAY) || (d2->type & SDB_TYPE_ARRAY)) {
+ size_t elem_size = sdb_data_sizeof(d1->type & 0xff);
+ if (d1->type & SDB_TYPE_ARRAY) {
+ s1 = (const unsigned char *)d1->data.array.values;
+ array1_len = d1->data.array.length;
+ }
+ else {
+ /* As per C99, section 6.7.2.1, paragraph 14:
+ * "A pointer to a union object, suitably converted, points to
+ * each of its members" */
+ s1 = (const unsigned char *)&d1->data;
+ array1_len = 1;
+ }
+ if (d2->type & SDB_TYPE_ARRAY) {
+ s2 = (const unsigned char *)d2->data.array.values;
+ array2_len = d2->data.array.length;
+ }
+ else {
+ s2 = (const unsigned char *)&d2->data;
+ array2_len = 1;
+ }
+ len1 = array1_len * elem_size;
+ len2 = array2_len * elem_size;
+ }
+ else if (d1->type == SDB_TYPE_STRING) {
s1 = (unsigned char *)d1->data.string;
s2 = (unsigned char *)d2->data.string;
- len1 = s1 ? strlen((char *)s1) : 0;
- len2 = s2 ? strlen((char *)s2) : 0;
+ len1 = s1 ? strlen((const char *)s1) : 0;
+ len2 = s2 ? strlen((const char *)s2) : 0;
}
else if (d1->type == SDB_TYPE_BINARY) {
s1 = d1->data.binary.datum;
len1 = d1->data.binary.length;
len2 = d2->data.binary.length;
}
- else if (d1->type & SDB_TYPE_ARRAY) {
- size_t elem_size = sdb_data_sizeof(d1->type & 0xff);
- s1 = (unsigned char *)d1->data.array.values;
- s2 = (unsigned char *)d2->data.array.values;
- len1 = d1->data.array.length * elem_size;
- len2 = d2->data.array.length * elem_size;
- }
else
return -1;
- assert(s1 && s2);
-
new = malloc(len1 + len2 + 1);
if (! new)
return -1;
memcpy(new + len1, s2, len2);
new[len1 + len2] = '\0';
- res->type = d1->type;
+ /* element types match and if either datum is an array,
+ * the result is an array as well */
+ res->type = d1->type | d2->type;
if (res->type == SDB_TYPE_STRING) {
res->data.string = (char *)new;
}
res->data.binary.datum = new;
res->data.binary.length = len1 + len2;
}
- else if (d1->type & SDB_TYPE_ARRAY) {
+ else if (res->type & SDB_TYPE_ARRAY) {
res->data.array.values = new;
- res->data.array.length = len1 + len2;
+ res->data.array.length = array1_len + array2_len;
if (copy_array_values(res, res, sdb_data_sizeof(res->type & 0xff))) {
/* this leaks already copied values but there's not much we can
* do and this should only happen if we're in trouble anyway */
return strcmp(d1->data.re.raw, d2->data.re.raw);
}
else if (d1->type & SDB_TYPE_ARRAY) {
- /* TODO */
- errno = ENOTSUP;
- return -1;
+ CMP_NULL(d1->data.array.values, d2->data.array.values);
+ return array_cmp(d1, d2);
}
return -1;
} /* sdb_data_cmp */
CMP_NULL(d1, d2);
- if (sdb_data_format(d1, d1_str, sizeof(d1_str), SDB_UNQUOTED) < 0)
+ if (! sdb_data_format(d1, d1_str, sizeof(d1_str), SDB_UNQUOTED))
return SDB_CMP(sizeof(d1_str), sizeof(d2_str));
- if (sdb_data_format(d2, d2_str, sizeof(d2_str), SDB_UNQUOTED) < 0)
+ if (! sdb_data_format(d2, d2_str, sizeof(d2_str), SDB_UNQUOTED))
return SDB_CMP(sizeof(d1_str), sizeof(d2_str));
return strcasecmp(d1_str, d2_str);
#undef CMP_NULL
} /* sdb_data_strcmp */
-_Bool
+bool
sdb_data_isnull(const sdb_data_t *datum)
{
if (! datum)
return 1;
if ((datum->type == SDB_TYPE_REGEX) && (! datum->data.re.raw))
return 1;
- if ((datum->type & SDB_TYPE_ARRAY) && (! datum->data.array.values))
- return 1;
return 0;
} /* sdb_data_isnull */
+bool
+sdb_data_inarray(const sdb_data_t *value, const sdb_data_t *array)
+{
+ const void *values;
+ size_t length, i;
+ int type = value->type & 0xff;
+
+ if (sdb_data_isnull(value) || sdb_data_isnull(array))
+ return 0;
+ if (! (array->type & SDB_TYPE_ARRAY))
+ return 0;
+ if ((value->type & 0xff) != (array->type & 0xff))
+ return 0;
+
+ if (value->type & SDB_TYPE_ARRAY) {
+ values = value->data.array.values;
+ length = value->data.array.length;
+ }
+ else {
+ values = &value->data;
+ length = 1;
+ }
+
+ for (i = 0; i < length; ++i) {
+ size_t j;
+
+ if (type == SDB_TYPE_INTEGER) {
+ int64_t *v = array->data.array.values;
+ for (j = 0; j < array->data.array.length; ++j)
+ if (((const int64_t *)values)[i] == v[j])
+ break;
+ }
+ else if (type == SDB_TYPE_DECIMAL) {
+ double *v = array->data.array.values;
+ for (j = 0; j < array->data.array.length; ++j)
+ if (((const double *)values)[i] == v[j])
+ break;
+ }
+ else if (type == SDB_TYPE_STRING) {
+ char **v = array->data.array.values;
+ for (j = 0; j < array->data.array.length; ++j)
+ if (!strcasecmp(((const char * const*)values)[i], v[j]))
+ break;
+ }
+ else {
+ /* TODO */
+ errno = ENOTSUP;
+ return 0;
+ }
+
+ if (j >= array->data.array.length)
+ /* value not found */
+ return 0;
+ }
+ return 1;
+} /* sdb_data_inarray */
+
+int
+sdb_data_array_get(const sdb_data_t *array, size_t i, sdb_data_t *value)
+{
+ sdb_data_t tmp = SDB_DATA_INIT;
+ int type;
+
+ if ((! array) || (! (array->type & SDB_TYPE_ARRAY)))
+ return -1;
+ if (i >= array->data.array.length)
+ return -1;
+
+ type = array->type & 0xff;
+ if (type == SDB_TYPE_INTEGER) {
+ int64_t *v = array->data.array.values;
+ tmp.data.integer = v[i];
+ }
+ else if (type == SDB_TYPE_DECIMAL) {
+ double *v = array->data.array.values;
+ tmp.data.decimal = v[i];
+ }
+ else if (type == SDB_TYPE_STRING) {
+ char **v = array->data.array.values;
+ tmp.data.string = v[i];
+ }
+ else {
+ /* TODO */
+ errno = ENOTSUP;
+ return -1;
+ }
+
+ if (value) {
+ *value = tmp;
+ value->type = type;
+ }
+ return 0;
+} /* sdb_data_array_get */
+
int
sdb_data_parse_op(const char *op)
{
return 0;
}
switch (op) {
- case SDB_DATA_CONCAT:
- return data_concat(d1, d2, res);
- case SDB_DATA_ADD:
- return data_lin(d1, 1, d2, res);
- case SDB_DATA_SUB:
- return data_lin(d1, -1, d2, res);
- case SDB_DATA_MUL:
- return data_mul(d1, d2, res);
- case SDB_DATA_DIV:
- return data_div(d1, d2, res, NULL);
- case SDB_DATA_MOD:
- return data_div(d1, d2, NULL, res);
+ case SDB_DATA_CONCAT: return data_concat(d1, d2, res);
+ case SDB_DATA_ADD: return data_lin(d1, 1, d2, res);
+ case SDB_DATA_SUB: return data_lin(d1, -1, d2, res);
+ case SDB_DATA_MUL: return data_mul(d1, d2, res);
+ case SDB_DATA_DIV: return data_div(d1, d2, res, NULL);
+ case SDB_DATA_MOD: return data_div(d1, d2, NULL, res);
}
return -1;
} /* sdb_data_expr_eval */
+int
+sdb_data_expr_type(int op, int type1, int type2)
+{
+ int types_num = (int)SDB_STATIC_ARRAY_LEN(op_matrix[0]);
+
+ assert(SDB_STATIC_ARRAY_LEN(op_matrix[0])
+ == SDB_STATIC_ARRAY_LEN(op_matrix[0][0]));
+
+ if ((op <= 0) || (SDB_STATIC_ARRAY_LEN(op_matrix) < (size_t)op))
+ return -1;
+
+ /* arrays only support concat; element type has to match */
+ if ((type1 & SDB_TYPE_ARRAY) || (type2 & SDB_TYPE_ARRAY)) {
+ if (((type1 & 0xff) != (type2 & 0xff)) || (op != SDB_DATA_CONCAT))
+ return -1;
+ return type1 | SDB_TYPE_ARRAY;
+ }
+ if ((type1 < 0) || (types_num < type1)
+ || (type2 < 0) || (types_num < type2))
+ return -1;
+
+ if ((type1 == SDB_TYPE_NULL) || (type2 == SDB_TYPE_NULL))
+ return SDB_TYPE_NULL;
+ return op_matrix[op - 1][type1 - 1][type2 - 1];
+} /* sdb_data_expr_type */
+
size_t
sdb_data_strlen(const sdb_data_t *datum)
{
if (! datum)
return 0;
- if (datum->type == SDB_TYPE_INTEGER) {
+ if (sdb_data_isnull(datum)) {
+ /* NULL */
+ return 4;
+ }
+ else if (datum->type == SDB_TYPE_INTEGER) {
/* log(64) */
return 20;
}
}
else if (datum->type == SDB_TYPE_STRING) {
if (! datum->data.string)
- return 8; /* "<NULL>" */
+ return 6; /* NULL */
/* in the worst case, each character needs to be escaped */
return 2 * strlen(datum->data.string) + 2;
}
}
else if (datum->type == SDB_TYPE_BINARY) {
if (! datum->data.binary.datum)
- return 8; /* "<NULL>" */
+ return 6; /* NULL */
/* "\xNN" */
return 4 * datum->data.binary.length + 2;
}
else if (datum->type == SDB_TYPE_REGEX) {
if (! datum->data.re.raw)
- return 8; /* "<NULL>" */
+ return 6; /* NULL */
/* "/.../" */
return strlen(datum->data.re.raw) + 4;
}
else if (datum->type & SDB_TYPE_ARRAY) {
- /* TODO */
- errno = ENOTSUP;
- return 0;
+ size_t len = 2; /* [] */
+ size_t i;
+ for (i = 0; i < datum->data.array.length; ++i) {
+ sdb_data_t v = SDB_DATA_INIT;
+ sdb_data_array_get(datum, i, &v);
+ len += sdb_data_strlen(&v) + 1;
+ }
+ return len;
}
return 0;
} /* sdb_data_strlen */
-int
+size_t
sdb_data_format(const sdb_data_t *datum, char *buf, size_t buflen, int quoted)
{
char tmp[sdb_data_strlen(datum) + 1];
char *data = NULL;
- int ret = -1;
+ bool is_null = 0;
+ size_t ret = 0;
size_t i, pos;
- if ((! datum) || (! buf))
- return -1;
+ if (! datum)
+ return 0;
- if (datum->type == SDB_TYPE_INTEGER) {
+ if (datum->type == SDB_TYPE_NULL) {
+ strncpy(buf, "NULL", buflen);
+ ret = 4;
+ }
+ else if (datum->type == SDB_TYPE_INTEGER) {
ret = snprintf(buf, buflen, "%"PRIi64, datum->data.integer);
}
else if (datum->type == SDB_TYPE_DECIMAL) {
- ret = snprintf(buf, buflen, "%g", datum->data.decimal);
+ if (isnan(datum->data.decimal))
+ ret = snprintf(buf, buflen, "nan");
+ else
+ ret = snprintf(buf, buflen, "%g", datum->data.decimal);
}
else if (datum->type == SDB_TYPE_STRING) {
if (! datum->data.string)
- data = "<NULL>";
+ is_null = 1;
else {
pos = 0;
for (i = 0; i < strlen(datum->data.string); ++i) {
@@ -652,23 +993,55 @@ sdb_data_format(const sdb_data_t *datum, char *buf, size_t buflen, int quoted)
data = tmp;
}
else
- data = "<NULL>";
+ is_null = 1;
}
else if (datum->type == SDB_TYPE_REGEX) {
if (! datum->data.re.raw)
- data = "<NULL>";
+ is_null = 1;
else {
snprintf(tmp, sizeof(tmp), "/%s/", datum->data.re.raw);
data = tmp;
}
}
else if (datum->type & SDB_TYPE_ARRAY) {
- /* TODO */
- errno = ENOTSUP;
- return -1;
+ ret = 1;
+ if (buflen > 0)
+ buf[0] = '[';
+ for (i = 0; i < datum->data.array.length; ++i) {
+ sdb_data_t v = SDB_DATA_INIT;
+ size_t n;
+
+ if (ret > 1) {
+ if (buflen > ret + 1) {
+ buf[ret] = ',';
+ buf[ret + 1] = ' ';
+ }
+ ret += 2;
+ }
+
+ sdb_data_array_get(datum, i, &v);
+ if (buflen > ret)
+ n = sdb_data_format(&v, buf + ret, buflen - ret, quoted);
+ else
+ n = sdb_data_format(&v, NULL, 0, quoted);
+ if (n > 0)
+ ret += n;
+ else
+ break;
+ }
+ if (buflen > ret + 1) {
+ buf[ret] = ']';
+ buf[ret + 1] = '\0';
+ }
+ ++ret;
}
- if (data) {
+ if (is_null) {
+ /* never quote NULL */
+ strncpy(buf, "NULL", buflen);
+ ret = 4;
+ }
+ else if (data) {
if (quoted == SDB_UNQUOTED)
ret = snprintf(buf, buflen, "%s", data);
else if (quoted == SDB_SINGLE_QUOTED)
@@ -676,17 +1049,23 @@ sdb_data_format(const sdb_data_t *datum, char *buf, size_t buflen, int quoted)
else
ret = snprintf(buf, buflen, "\"%s\"", data);
}
- buf[buflen - 1] = '\0';
+ if (buflen > 0)
+ buf[buflen - 1] = '\0';
return ret;
} /* sdb_data_format */
int
-sdb_data_parse(char *str, int type, sdb_data_t *data)
+sdb_data_parse(const char *str, int type, sdb_data_t *data)
{
sdb_data_t tmp;
char *endptr = NULL;
+ if (! str) {
+ errno = EINVAL;
+ return -1;
+ }
+
errno = 0;
if (type == SDB_TYPE_INTEGER) {
tmp.data.integer = strtoll(str, &endptr, 0);
tmp.data.decimal = strtod(str, &endptr);
}
else if (type == SDB_TYPE_STRING) {
- tmp.data.string = str;
+ tmp.data.string = strdup(str);
+ if (! tmp.data.string)
+ return -1;
}
else if (type == SDB_TYPE_DATETIME) {
double datetime = strtod(str, &endptr);
tmp.data.datetime = DOUBLE_TO_SDB_TIME(datetime);
}
else if (type == SDB_TYPE_BINARY) {
- /* we don't support any binary information containing 0-bytes */
+ /* we don't support any binary information containing 0-bytes here */
+ tmp.data.binary.datum = (unsigned char *)strdup(str);
+ if (! tmp.data.binary.datum)
+ return -1;
tmp.data.binary.length = strlen(str);
- tmp.data.binary.datum = (unsigned char *)str;
}
else if (type == SDB_TYPE_REGEX) {
tmp.data.re.raw = strdup(str);