}
// N-ary numeric functions.
-absorb_miss boolean function ANY (x != SYSMIS, a[n])
+absorb_miss boolean function ANY (x, a[n])
{
- int sysmis = 0;
- size_t i;
-
- for (i = 0; i < n; i++)
- if (a[i] == x)
- return 1.;
- else if (a[i] == SYSMIS)
- sysmis = 1;
-
- return sysmis ? SYSMIS : 0.;
+ double retval = SYSMIS;
+ if (x != SYSMIS)
+ {
+ for (size_t i = 0; i < n; i++)
+ if (a[i] == x)
+ return 1.;
+ else if (a[i] != SYSMIS)
+ retval = 0.;
+ }
+ return retval;
}
boolean function ANY (string x, string a[n])
absorb_miss boolean function RANGE (x != SYSMIS, a[n*2])
{
- size_t i;
- int sysmis = 0;
-
- for (i = 0; i < n; i++)
+ bool found = false;
+ bool valid = false;
+ for (size_t i = 0; i < n; i++)
{
double w = a[2 * i];
double y = a[2 * i + 1];
if (w != SYSMIS && y != SYSMIS)
{
if (w <= x && x <= y)
- return 1.0;
+ found = true;
+ else if (w <= y)
+ valid = true;
+ else
+ return SYSMIS;
}
- else
- sysmis = 1;
}
- return sysmis ? SYSMIS : 0.;
+ return found ? true : valid ? false : SYSMIS;
}
boolean function RANGE (string x, string a[n*2])
{
- int i;
-
- for (i = 0; i < n; i++)
+ bool found = false;
+ for (size_t i = 0; i < n; i++)
{
struct substring *w = &a[2 * i];
struct substring *y = &a[2 * i + 1];
if (compare_string_3way (w, &x) <= 0 && compare_string_3way (&x, y) <= 0)
- return 1.;
+ found = true;
+ else if (compare_string_3way (w, y) > 0)
+ return SYSMIS;
}
- return 0.;
+ return found;
}
function SD.2 (a[n])
// Time construction & extraction functions.
function TIME.HMS (h, m, s)
+ expression e;
+ expr_node n;
{
if ((h > 0. || m > 0. || s > 0.) && (h < 0. || m < 0. || s < 0.))
{
- msg (SW, _("TIME.HMS cannot mix positive and negative arguments."));
+ msg_at (SW, expr_location (e, n),
+ _("TIME.HMS cannot accept a mix of positive and negative "
+ "arguments."));
+ double args[] = { h, m, s };
+ for (size_t i = 0; i < 3; i++)
+ if (args[i] > 0)
+ msg_at (SN, expr_location (e, n->args[i]),
+ _("This argument has positive value %g."), args[i]);
+ else if (args[i] < 0)
+ msg_at (SN, expr_location (e, n->args[i]),
+ _("This argument has negative value %g."), args[i]);
return SYSMIS;
}
else
function CTIME.SECONDS (time) = time;
// Date construction functions.
-function DATE.DMY (d, m, y) = expr_ymd_to_date (y, m, d);
-function DATE.MDY (m, d, y) = expr_ymd_to_date (y, m, d);
-function DATE.MOYR (m, y) = expr_ymd_to_date (y, m, 1);
-function DATE.QYR (q, y)
+function DATE.DMY (integer d, integer m, integer y)
+ expression e;
+ expr_node n;
+= expr_ymd_to_date (y, m, d, e, n, 3, 2, 1);
+
+function DATE.MDY (integer m, integer d, integer y)
+ expression e;
+ expr_node n;
+= expr_ymd_to_date (y, m, d, e, n, 3, 1, 2);
+
+function DATE.MOYR (integer m, integer y)
+ expression e;
+ expr_node n;
+= expr_ymd_to_date (y, m, 1, e, n, 2, 1, 0);
+
+function DATE.QYR (integer q, integer y)
+ expression e;
+ expr_node n;
+{
+ if (q < 1 || q > 4)
+ {
+ msg_at (SW, expr_location (e, n->args[0]),
+ _("Argument 1 to DATE.QYR must be 1, 2, 3, or 4 (not %d)."), q);
+ return SYSMIS;
+ }
+ return expr_ymd_to_date (y, q * 3 - 2, 1, e, n, 2, 0, 0);
+}
+
+function DATE.WKYR (integer w, integer y)
+ expression e;
+ expr_node n;
+{
+ if (w < 1 || w > 53)
+ {
+ msg_at (SE, expr_location (e, n->args[0]),
+ _("The week argument to DATE.WKYR is outside the acceptable "
+ "range of 1 to 53. The result will be system-missing."));
+ return SYSMIS;
+ }
+ else
+ {
+ double yr_1_1 = expr_ymd_to_ofs (y, 1, 1, e, n, 2, 0, 0);
+ if (yr_1_1 != SYSMIS)
+ return DAY_S * (yr_1_1 + WEEK_DAY * (w - 1));
+ else
+ return SYSMIS;
+ }
+}
+
+function DATE.YRDAY (integer y, integer yd)
+ expression e;
+ expr_node n;
+{
+ if (yd < 1 || yd > 366)
+ {
+ msg_at (SE, expr_location (e, n->args[1]),
+ _("The value %d as day argument to DATE.YRDAY is outside the "
+ "acceptable range of 1 to 366. "
+ "The result will be system-missing."), yd);
+ return SYSMIS;
+ }
+ else
+ {
+ double yr_1_1 = expr_ymd_to_ofs (y, 1, 1, e, n, 1, 0, 0);
+ if (yr_1_1 != SYSMIS)
+ return DAY_S * (yr_1_1 + yd - 1.);
+ else
+ return SYSMIS;
+ }
+}
+
+function YRMODA (integer y, integer m, integer d)
+ expression e;
+ expr_node n;
{
- if (q < 1.0 || q > 4.0 || q != (int) q)
+ if (y >= 0 && y <= 99)
+ y += 1900;
+ else if (y > 47516)
{
- msg (SW, _("The first argument to DATE.QYR must be 1, 2, 3, or 4."));
+ msg_at (SE, expr_location (e, n->args[0]),
+ _("The year argument to YRMODA is greater than 47516. "
+ "The result will be system-missing."));
return SYSMIS;
}
- return expr_ymd_to_date (y, q * 3 - 2, 1);
+
+ return expr_ymd_to_ofs (y, m, d, e, n, 1, 2, 3);
}
-function DATE.WKYR (w, y) = expr_wkyr_to_date (w, y);
-function DATE.YRDAY (y, yday) = expr_yrday_to_date (y, yday);
-function YRMODA (y, m, d) = expr_yrmoda (y, m, d);
// Date extraction functions.
function XDATE.TDAY (date) = floor (date / DAY_S);
// Date arithmetic functions.
no_abbrev function DATEDIFF (date2 >= DAY_S, date1 >= DAY_S, string unit)
- = expr_date_difference (date1, date2, unit);
+ expression e;
+ expr_node n;
+= expr_date_difference (date1, date2, unit, e, n);
+
no_abbrev function DATESUM (date, quantity, string unit)
- = expr_date_sum (date, quantity, unit, ss_cstr ("closest"));
+ expression e;
+ expr_node n;
+= expr_date_sum_closest (date, quantity, unit, e, n);
no_abbrev function DATESUM (date, quantity, string unit, string method)
- = expr_date_sum (date, quantity, unit, method);
+ expression e;
+ expr_node n;
+= expr_date_sum (date, quantity, unit, method, e, n);
// String functions.
function INDEX (string haystack, string needle)
{
- if (needle.length == 0)
- return SYSMIS;
- else
+ if (haystack.length >= needle.length)
{
- int limit = haystack.length - needle.length + 1;
- int i;
- for (i = 1; i <= limit; i++)
+ size_t limit = haystack.length - needle.length + 1;
+ for (size_t i = 1; i <= limit; i++)
if (!memcmp (&haystack.string[i - 1], needle.string, needle.length))
return i;
- return 0;
}
+ return 0;
}
-function INDEX (string haystack, string needles, needle_len_d)
+function INDEX (string haystack, string needles, integer needle_len)
+ expression e;
+ expr_node n;
{
- if (needle_len_d <= INT_MIN || needle_len_d >= INT_MAX
- || (int) needle_len_d != needle_len_d
- || needles.length == 0)
- return SYSMIS;
- else
+ if (needle_len <= 0 || needles.length % needle_len != 0)
{
- int needle_len = needle_len_d;
- if (needle_len < 0 || needle_len > needles.length
- || needles.length % needle_len != 0)
- return SYSMIS;
- else
- {
- int limit = haystack.length - needle_len + 1;
- int i, j;
- for (i = 1; i <= limit; i++)
- for (j = 0; j < needles.length; j += needle_len)
- if (!memcmp (&haystack.string[i - 1], &needles.string[j],
- needle_len))
- return i;
- return 0;
- }
+ msg_at (SE, expr_location (e, n),
+ _("INDEX needle length argument must evenly divide the "
+ "length of the needles argument."));
+ msg_at (SN, expr_location (e, n->args[1]),
+ _("The needles argument has length %zu."), needles.length);
+ msg_at (SN, expr_location (e, n->args[2]),
+ _("The needle length argument has value %d."), needle_len);
+ return SYSMIS;
}
+
+ if (haystack.length >= needle_len)
+ {
+ size_t limit = haystack.length - needle_len + 1;
+ for (size_t i = 1; i <= limit; i++)
+ for (size_t j = 0; j < needles.length; j += needle_len)
+ if (!memcmp (&haystack.string[i - 1], &needles.string[j], needle_len))
+ return i;
+ }
+
+ return 0;
}
function RINDEX (string haystack, string needle)
{
- if (needle.length == 0)
- return SYSMIS;
- else
+ if (haystack.length >= needle.length)
{
- int limit = haystack.length - needle.length + 1;
- int i;
- for (i = limit; i >= 1; i--)
+ size_t limit = haystack.length - needle.length + 1;
+ for (size_t i = limit; i >= 1; i--)
if (!memcmp (&haystack.string[i - 1], needle.string, needle.length))
return i;
- return 0;
}
+
+ return 0;
}
-function RINDEX (string haystack, string needles, needle_len_d)
+function RINDEX (string haystack, string needles, integer needle_len)
+ expression e;
+ expr_node n;
{
- if (needle_len_d <= 0 || needle_len_d >= INT_MAX
- || (int) needle_len_d != needle_len_d
- || needles.length == 0)
- return SYSMIS;
- else
+ if (needle_len <= 0 || needles.length % needle_len != 0)
{
- int needle_len = needle_len_d;
- if (needle_len < 0 || needle_len > needles.length
- || needles.length % needle_len != 0)
- return SYSMIS;
- else
- {
- int limit = haystack.length - needle_len + 1;
- int i, j;
- for (i = limit; i >= 1; i--)
- for (j = 0; j < needles.length; j += needle_len)
- if (!memcmp (&haystack.string[i - 1],
- &needles.string[j], needle_len))
- return i;
- return 0;
- }
+ msg_at (SE, expr_location (e, n),
+ _("RINDEX needle length argument must evenly divide the "
+ "length of the needles argument."));
+ msg_at (SN, expr_location (e, n->args[1]),
+ _("The needles argument has length %zu."), needles.length);
+ msg_at (SN, expr_location (e, n->args[2]),
+ _("The needle length argument has value %d."), needle_len);
+ return SYSMIS;
+ }
+
+ if (haystack.length >= needle_len)
+ {
+ size_t limit = haystack.length - needle_len + 1;
+ for (size_t i = limit; i >= 1; i--)
+ for (size_t j = 0; j < needles.length; j += needle_len)
+ if (!memcmp (&haystack.string[i - 1], &needles.string[j], needle_len))
+ return i;
}
+
+ return 0;
}
function LENGTH (string s)
return s;
}
-absorb_miss string function LPAD (string s, n)
+absorb_miss string function LPAD (string s, integer n)
expression e;
+ expr_node node;
{
- if (n < 0 || n > MAX_STRING || (int) n != n)
- return empty_string;
+ if (n < 0 || n > MAX_STRING)
+ {
+ if (n != INT_MIN)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The length argument to LPAD must be between 0 and %d."),
+ MAX_STRING);
+ msg_at (SN, expr_location (e, node->args[1]),
+ _("The length argument is %d."), n);
+ }
+
+ return s;
+ }
else if (s.length >= n)
return s;
else
{
struct substring t = alloc_string (e, n);
- memset (t.string, ' ', n - s.length);
- memcpy (&t.string[(int) n - s.length], s.string, s.length);
+ size_t pad = n - s.length;
+ memset (t.string, ' ', pad);
+ memcpy (&t.string[pad], s.string, s.length);
return t;
}
}
-absorb_miss string function LPAD (string s, n, string c)
+absorb_miss string function LPAD (string s, integer n, string c)
expression e;
+ expr_node node;
{
- if (n < 0 || n > MAX_STRING || (int) n != n || c.length != 1)
- return empty_string;
+ if (n < 0 || n > MAX_STRING)
+ {
+ if (n != INT_MIN)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The length argument to LPAD must be between 0 and %d."),
+ MAX_STRING);
+ msg_at (SN, expr_location (e, node->args[1]),
+ _("The length argument is %d."), n);
+ }
+
+ return s;
+ }
else if (s.length >= n)
return s;
+ else if (c.length == 0)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The padding argument to LPAD must not be an empty string."));
+ return s;
+ }
else
{
+ size_t n_pad = (n - s.length) / c.length;
+ if (!n_pad)
+ return s;
+
struct substring t = alloc_string (e, n);
- memset (t.string, c.string[0], n - s.length);
- memcpy (&t.string[(int) n - s.length], s.string, s.length);
+ t.length = 0;
+ for (size_t i = 0; i < n_pad; i++)
+ {
+ memcpy (t.string + t.length, c.string, c.length);
+ t.length += c.length;
+ }
+ memcpy (t.string + t.length, s.string, s.length);
+ t.length += s.length;
return t;
}
}
string function REPLACE (string haystack, string needle, string replacement)
expression e;
- = replace_string (e, haystack, needle, replacement, DBL_MAX);
+ = replace_string (e, haystack, needle, replacement, INT_MAX);
absorb_miss string function REPLACE (string haystack, string needle,
- string replacement, n)
+ string replacement, integer n)
expression e;
= replace_string (e, haystack, needle, replacement, n);
-absorb_miss string function RPAD (string s, n)
+absorb_miss string function RPAD (string s, integer n)
expression e;
+ expr_node node;
{
- if (n < 0 || n > MAX_STRING || (int) n != n)
- return empty_string;
+ if (n < 0 || n > MAX_STRING)
+ {
+ if (n != INT_MIN)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The length argument to RPAD must be between 0 and %d."),
+ MAX_STRING);
+ msg_at (SN, expr_location (e, node->args[1]),
+ _("The length argument is %d."), n);
+ }
+
+ return s;
+ }
else if (s.length >= n)
return s;
else
{
struct substring t = alloc_string (e, n);
+ size_t pad = n - s.length;
memcpy (t.string, s.string, s.length);
- memset (&t.string[s.length], ' ', n - s.length);
+ memset (t.string + s.length, ' ', pad);
return t;
}
}
-absorb_miss string function RPAD (string s, n, string c)
+absorb_miss string function RPAD (string s, integer n, string c)
expression e;
+ expr_node node;
{
- if (n < 0 || n > MAX_STRING || (int) n != n || c.length != 1)
- return empty_string;
+ if (n < 0 || n > MAX_STRING)
+ {
+ if (n != INT_MIN)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The length argument to RPAD must be between 0 and %d."),
+ MAX_STRING);
+ msg_at (SN, expr_location (e, node->args[1]),
+ _("The length argument is %d."), n);
+ }
+
+ return s;
+ }
else if (s.length >= n)
return s;
+ else if (c.length == 0)
+ {
+ msg_at (SE, expr_location (e, node),
+ _("The padding argument to RPAD must not be an empty string."));
+ return s;
+ }
else
{
+ size_t n_pad = (n - s.length) / c.length;
+ if (!n_pad)
+ return s;
+
struct substring t = alloc_string (e, n);
memcpy (t.string, s.string, s.length);
- memset (&t.string[s.length], c.string[0], n - s.length);
+ t.length = s.length;
+ for (size_t i = 0; i < n_pad; i++)
+ {
+ memcpy (t.string + t.length, c.string, c.length);
+ t.length += c.length;
+ }
return t;
}
}
string function LTRIM (string s, string c)
{
- if (c.length == 1)
- {
- while (s.length > 0 && s.string[0] == c.string[0])
- {
- s.length--;
- s.string++;
- }
- return s;
- }
- else
- return empty_string;
+ if (c.length > 0)
+ while (s.length >= c.length && !memcmp (s.string, c.string, c.length))
+ {
+ s.length -= c.length;
+ s.string += c.length;
+ }
+ return s;
}
string function RTRIM (string s)
string function RTRIM (string s, string c)
{
- if (c.length == 1)
- {
- while (s.length > 0 && s.string[s.length - 1] == c.string[0])
- s.length--;
- return s;
- }
- else
- return empty_string;
+ if (c.length > 0)
+ while (s.length >= c.length
+ && !memcmp (&s.string[s.length - c.length], c.string, c.length))
+ s.length -= c.length;
+ return s;
}
function NUMBER (string s, ni_format f)
+ expression e;
+ expr_node n;
{
- union value out;
- char *error;
-
if (s.length > f->w)
s.length = f->w;
- error = data_in (s, C_ENCODING, f->type, settings_get_fmt_settings (),
- &out, 0, NULL);
+
+ union value out;
+ char *error = data_in (s, C_ENCODING, f->type, settings_get_fmt_settings (),
+ &out, 0, NULL);
if (error == NULL)
data_in_imply_decimals (s, C_ENCODING, f->type, f->d,
settings_get_fmt_settings (), &out);
else
{
- msg (SE, "Cannot parse `%.*s' as format %s: %s",
- (int) s.length, s.string, fmt_name (f->type), error);
+ msg_at (SE, expr_location (e, n->args[0]),
+ _("Cannot parse \"%.*s\" as format %s: %s"),
+ (int) s.length, s.string, fmt_name (f->type), error);
free (error);
}
return out.f;
return dst;
}
-absorb_miss string function STRUNC (string s, n)
+absorb_miss string function STRUNC (string s, integer n)
{
- if (n < 1 || n == SYSMIS)
- return empty_string;
+ if (n < 1)
+ return n == INT_MIN ? s : empty_string;
if (n < s.length)
s.length = n;
return s;
}
-absorb_miss string function SUBSTR (string s, ofs)
- expression e;
+absorb_miss string function SUBSTR (string s, integer ofs)
{
- if (ofs >= 1 && ofs <= s.length && (int) ofs == ofs)
- return copy_string (e, &s.string[(int) ofs - 1], s.length - ofs + 1);
- else
- return empty_string;
+ return (ofs >= 1 && ofs <= s.length
+ ? ss_substr (s, ofs - 1, SIZE_MAX)
+ : empty_string);
}
-absorb_miss string function SUBSTR (string s, ofs, cnt)
- expression e;
+absorb_miss string function SUBSTR (string s, integer ofs, integer cnt)
{
- if (ofs >= 1 && ofs <= s.length && (int) ofs == ofs
- && cnt >= 1 && cnt <= INT_MAX && (int) cnt == cnt)
- {
- int cnt_max = s.length - (int) ofs + 1;
- return copy_string (e, &s.string[(int) ofs - 1],
- cnt <= cnt_max ? cnt : cnt_max);
- }
- else
- return empty_string;
+ return (ofs >= 1 && cnt >= 1
+ ? ss_substr (s, ofs - 1, cnt)
+ : empty_string);
}
absorb_miss no_opt no_abbrev string function VALUELABEL (var v)
// Artificial.
operator SQUARE (x) = x * x;
-boolean operator NUM_TO_BOOLEAN (x)
+
+absorb_miss boolean operator OPERAND_TO_BOOLEAN (x, expr_node parent)
+ expression e;
+ expr_node n;
+{
+ if (x == 0. || x == 1. || x == SYSMIS)
+ return x;
+
+ switch (parent->n_args)
+ {
+ case 2:
+ msg_at (SE, expr_location (e, parent),
+ /* TRANSLATORS: There are exactly two operands. */
+ _("The operands of %s must have value 0 or 1."),
+ operations[parent->type].name);
+ break;
+
+ case 1:
+ msg_at (SE, expr_location (e, parent),
+ _("The operand of %s must have value 0 or 1."),
+ operations[parent->type].name);
+ break;
+
+ default:
+ NOT_REACHED ();
+ }
+
+ msg_at (SN, expr_location (e, n),
+ _("This operand with unexpected value %g will be treated as 0."), x);
+ return 0.;
+}
+
+absorb_miss boolean operator EXPR_TO_BOOLEAN (x)
+ expression e;
+ expr_node n;
{
if (x == 0. || x == 1. || x == SYSMIS)
return x;
- msg (SE, _("A logical expression was found to have a value other than 0 "
- "(false), 1 (true), or the system-missing value. The result "
- "was forced to 0."));
+ msg_at (SE, expr_location (e, n),
+ _("This expression, which must be 0 or 1, evaluated to %g. "
+ "It will be treated as 0."), x);
return 0.;
}
+operator NUM_TO_INTEGER (x)
+ expression e;
+ expr_node n;
+{
+ if (x == floor (x) && x > INT_MIN && x <= INT_MAX)
+ return x;
+
+ msg_at (SE, expr_location (e, n),
+ _("Treating unexpected non-integer value %g as missing."), x);
+ return SYSMIS;
+}
+
operator BOOLEAN_TO_NUM (boolean x) = x;
// Beta distribution.
{
return case_num (c, v) == SYSMIS;
}
-no_opt boolean function VALUE (num_var v)
+no_opt function VALUE (num_var v)
case c;
{
return case_num (c, v);
}
+no_opt function VALUE (num_vec_elem v)
+{
+ return v;
+}
-no_opt operator VEC_ELEM_NUM (idx)
+// A numeric vector element used in a "normal" context, in which a user-missing
+// value becomes system-missing.
+absorb_miss no_opt operator VEC_ELEM_NUM (idx)
vector v;
case c;
+ expression e;
+ expr_node n;
{
- if (idx >= 1 && idx <= vector_get_n_vars (v))
+ const struct variable *var = expr_index_vector (e, n, v, idx);
+ if (var)
{
- const struct variable *var = vector_get_var (v, (size_t) idx - 1);
- double value = case_num (c, var);
- return !var_is_num_missing (var, value, MV_USER) ? value : SYSMIS;
- }
- else
- {
- if (idx == SYSMIS)
- msg (SE, _("SYSMIS is not a valid index value for vector "
- "%s. The result will be set to SYSMIS."),
- vector_get_name (v));
- else
- msg (SE, _("%g is not a valid index value for vector %s. "
- "The result will be set to SYSMIS."),
- idx, vector_get_name (v));
- return SYSMIS;
+ double d = case_num (c, var);
+ if (!var_is_num_missing (var, d, MV_USER))
+ return d;
}
+ return SYSMIS;
+}
+
+// A numeric vector element used as the argument to the VALUE() function, in
+// which a user-missing value retains its value.
+//
+// All numeric vector elements are initially parsed this way. In most contexts
+// they then get coerced into numbers.
+absorb_miss no_opt num_vec_elem operator VEC_ELEM_NUM_RAW (idx)
+ vector v;
+ case c;
+ expression e;
+ expr_node n;
+{
+ const struct variable *var = expr_index_vector (e, n, v, idx);
+ return var ? case_num (c, var) : SYSMIS;
}
absorb_miss no_opt string operator VEC_ELEM_STR (idx)
expression e;
vector v;
case c;
+ expr_node n;
{
- if (idx >= 1 && idx <= vector_get_n_vars (v))
- {
- struct variable *var = vector_get_var (v, (size_t) idx - 1);
- return copy_string (e, CHAR_CAST_BUG (char *, case_str (c, var)),
- var_get_width (var));
- }
- else
- {
- if (idx == SYSMIS)
- msg (SE, _("SYSMIS is not a valid index value for vector "
- "%s. The result will be set to the empty string."),
- vector_get_name (v));
- else
- msg (SE, _("%g is not a valid index value for vector %s. "
- "The result will be set to the empty string."),
- idx, vector_get_name (v));
- return empty_string;
- }
+ const struct variable *var = expr_index_vector (e, n, v, idx);
+ return (var
+ ? copy_string (e, CHAR_CAST_BUG (char *, case_str (c, var)),
+ var_get_width (var))
+ : empty_string);
}
// Terminals.