1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-9, 2000 Free Software Foundation, Inc.
3 Written by Ben Pfaff <blp@gnu.org>.
5 This program is free software; you can redistribute it and/or
6 modify it under the terms of the GNU General Public License as
7 published by the Free Software Foundation; either version 2 of the
8 License, or (at your option) any later version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
33 #include "debug-print.h"
39 RCD_END, /* sentinel value */
40 RCD_USER, /* user-missing => one */
41 RCD_SINGLE, /* one => one */
42 RCD_HIGH, /* x > a => one */
43 RCD_LOW, /* x < b => one */
44 RCD_RANGE, /* b < x < a => one */
45 RCD_ELSE, /* any but SYSMIS => one */
46 RCD_CONVERT /* "123" => 123 */
49 /* Describes how to recode a single value or range of values into a
54 union value f1, f2; /* Describe value or range as src. Long
55 strings are stored in `c'. */
56 union value t; /* Describes value as dest. Long strings in `c'. */
59 /* Describes how to recode a single variable. */
64 unsigned flags; /* RCD_SRC_* | RCD_DEST_* | RCD_MISC_* */
66 struct variable *src; /* Source variable. */
67 struct variable *dest; /* Destination variable. */
68 char dest_name[9]; /* Name of dest variable if we're creating it. */
70 int has_sysmis; /* Do we recode for SYSMIS? */
71 union value sysmis; /* Coding for SYSMIS (if src is numeric). */
73 struct coding *map; /* Coding for other values. */
74 int nmap, mmap; /* Length of map, max capacity of map. */
77 /* RECODE transformation. */
81 struct rcd_var *codings;
84 /* What we're recoding from (`src'==`source'). */
85 #define RCD_SRC_ERROR 0000u /* Bad value for src. */
86 #define RCD_SRC_NUMERIC 0001u /* Src is numeric. */
87 #define RCD_SRC_STRING 0002u /* Src is short string. */
88 #define RCD_SRC_MASK 0003u /* AND mask to isolate src bits. */
90 /* What we're recoding to (`dest'==`destination'). */
91 #define RCD_DEST_ERROR 0000u /* Bad value for dest. */
92 #define RCD_DEST_NUMERIC 0004u /* Dest is numeric. */
93 #define RCD_DEST_STRING 0010u /* Dest is short string. */
94 #define RCD_DEST_MASK 0014u /* AND mask to isolate dest bits. */
96 /* Miscellaneous bits. */
97 #define RCD_MISC_CREATE 0020u /* We create dest var (numeric only) */
98 #define RCD_MISC_DUPLICATE 0040u /* This var_info has the same MAP
99 value as the previous var_info.
100 Prevents redundant free()ing. */
101 #define RCD_MISC_MISSING 0100u /* Encountered MISSING or SYSMIS in
104 static int parse_dest_spec (struct rcd_var * rcd, union value *v,
105 size_t *max_dst_width);
106 static int parse_src_spec (struct rcd_var * rcd, int type, size_t max_src_width);
107 static int recode_trns_proc (struct trns_header *, struct ccase *);
108 static void recode_trns_free (struct trns_header *);
109 static double convert_to_double (char *, int);
112 static void debug_print (struct rcd_var * head);
117 /* First transformation in the list. rcd is in this list. */
118 static struct rcd_var *head;
120 /* Variables in the current part of the recoding. */
124 /* Parses the RECODE transformation. */
130 /* Transformation that we're constructing. */
133 /* Type of the src variables. */
136 /* Length of longest src string. */
137 size_t max_src_width;
139 /* Length of longest dest string. */
140 size_t max_dst_width;
142 /* For stepping through, constructing the linked list of
144 struct rcd_var *iter;
146 /* The real transformation, just a wrapper for a list of
148 struct recode_trns *trns;
150 lex_match_id ("RECODE");
152 /* Parses each specification between slashes. */
153 head = rcd = xmalloc (sizeof *rcd);
156 /* Whether we've already encountered a specification for SYSMIS. */
159 /* Initialize this rcd_var to ensure proper cleanup. */
162 rcd->nmap = rcd->mmap = 0;
166 /* Parse variable names. */
167 if (!parse_variables (default_dict, &v, &nv, PV_SAME_TYPE))
170 /* Ensure all variables are same type; find length of longest
173 max_src_width = v[0]->width;
176 for (i = 0; i < nv; i++)
177 if (v[i]->width > (int) max_src_width)
178 max_src_width = v[i]->width;
183 rcd->flags |= RCD_SRC_NUMERIC;
185 rcd->flags |= RCD_SRC_STRING;
187 /* Parse each coding in parentheses. */
189 if (!lex_force_match ('('))
193 /* Get the input value (before the `='). */
194 int mark = rcd->nmap;
195 int code = parse_src_spec (rcd, type, max_src_width);
199 /* ELSE is the same as any other input spec except that it
200 precludes later sysmis specifications. */
207 /* If keyword CONVERT was specified, there is no output
213 /* Get the output value (after the `='). */
214 lex_get (); /* Skip `='. */
215 if (!parse_dest_spec (rcd, &output, &max_dst_width))
218 /* Set the value for SYSMIS if requested and if we don't
220 if ((rcd->flags & RCD_MISC_MISSING) && !had_sysmis)
223 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
224 rcd->sysmis.f = output.f;
226 rcd->sysmis.c = xstrdup (output.c);
229 rcd->flags &= ~RCD_MISC_MISSING;
232 /* Since there may be multiple input values for a single
233 output, the output value need to propagated among all
235 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
236 for (i = mark; i < rcd->nmap; i++)
237 rcd->map[i].t.f = output.f;
240 for (i = mark; i < rcd->nmap; i++)
241 rcd->map[i].t.c = xstrdup (output.c);
245 lex_get (); /* Skip `)'. */
246 if (!lex_match ('('))
250 /* Append sentinel value. */
251 rcd->map[rcd->nmap++].type = RCD_END;
253 /* Since multiple variables may use the same recodings, it is
254 necessary to propogate the codings to all of them. */
257 rcd->dest_name[0] = 0;
259 for (i = 1; i < nv; i++)
261 iter = iter->next = xmalloc (sizeof *iter);
263 iter->flags = rcd->flags | RCD_MISC_DUPLICATE;
266 iter->dest_name[0] = 0;
267 iter->has_sysmis = rcd->has_sysmis;
268 iter->sysmis = rcd->sysmis;
269 iter->map = rcd->map;
272 if (lex_match_id ("INTO"))
279 if (!parse_mixed_vars (&names, &nnames, PV_NONE))
284 for (i = 0; i < nnames; i++)
287 msg (SE, _("%d variable(s) cannot be recoded into "
288 "%d variable(s). Specify the same number "
289 "of variables as input and output variables."),
294 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_STRING)
295 for (i = 0, iter = rcd; i < nv; i++, iter = iter->next)
297 struct variable *v = dict_lookup_var (default_dict, names[i]);
301 msg (SE, _("There is no string variable named "
302 "%s. (All string variables specified "
303 "on INTO must already exist. Use the "
304 "STRING command to create a string "
305 "variable.)"), names[i]);
308 if (v->type != ALPHA)
310 msg (SE, _("Type mismatch between input and output "
311 "variables. Output variable %s is not "
312 "a string variable, but all the input "
313 "variables are string variables."), v->name);
316 if (v->width > (int) max_dst_width)
317 max_dst_width = v->width;
321 for (i = 0, iter = rcd; i < nv; i++, iter = iter->next)
323 struct variable *v = dict_lookup_var (default_dict, names[i]);
327 if (v->type != NUMERIC)
329 msg (SE, _("Type mismatch after INTO: %s "
330 "is not a numeric variable."), v->name);
337 strcpy (iter->dest_name, names[i]);
341 /* Note that regardless of whether we succeed or fail,
342 flow-of-control comes here. `success' is the important
343 factor. Ah, if C had garbage collection... */
345 for (i = 0; i < nnames; i++)
353 if (max_src_width > max_dst_width)
354 max_dst_width = max_src_width;
356 if ((rcd->flags & RCD_SRC_MASK) == RCD_SRC_NUMERIC
357 && (rcd->flags & RCD_DEST_MASK) != RCD_DEST_NUMERIC)
359 msg (SE, _("INTO must be used when the input values are "
360 "numeric and output values are string."));
364 if ((rcd->flags & RCD_SRC_MASK) != RCD_SRC_NUMERIC
365 && (rcd->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
367 msg (SE, _("INTO must be used when the input values are "
368 "string and output values are numeric."));
373 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_STRING)
377 for (cp = rcd->map; cp->type != RCD_END; cp++)
380 if (strlen (cp->t.c) < max_dst_width)
382 /* The NULL is only really necessary for the
384 char *repl = xmalloc (max_dst_width + 1);
385 st_pad_copy (repl, cp->t.c, max_dst_width + 1);
390 /* The strings are guaranteed to be in order of
391 nondecreasing length. */
397 if (!lex_match ('/'))
401 rcd = rcd->next = xmalloc (sizeof *rcd);
408 lex_error (_("expecting end of command"));
412 for (rcd = head; rcd; rcd = rcd->next)
413 if (rcd->dest_name[0])
415 rcd->dest = dict_create_var (default_dict, rcd->dest_name, 0);
418 /* FIXME: This can occur if a destname is duplicated.
419 We could give an error at parse time but I don't
421 rcd->dest = dict_lookup_var_assert (default_dict, rcd->dest_name);
425 trns = xmalloc (sizeof *trns);
426 trns->h.proc = recode_trns_proc;
427 trns->h.free = recode_trns_free;
428 trns->codings = head;
429 add_transformation ((struct trns_header *) trns);
439 struct recode_trns t;
442 recode_trns_free ((struct trns_header *) &t);
448 parse_dest_spec (struct rcd_var * rcd, union value * v, size_t *max_dst_width)
458 flags = RCD_DEST_NUMERIC;
460 else if (lex_match_id ("SYSMIS"))
463 flags = RCD_DEST_NUMERIC;
465 else if (token == T_STRING)
467 size_t max = *max_dst_width;
468 size_t toklen = ds_length (&tokstr);
471 v->c = xmalloc (max + 1);
472 st_pad_copy (v->c, ds_value (&tokstr), max + 1);
473 flags = RCD_DEST_STRING;
474 *max_dst_width = max;
477 else if (lex_match_id ("COPY"))
479 if ((rcd->flags & RCD_SRC_MASK) == RCD_SRC_NUMERIC)
481 flags = RCD_DEST_NUMERIC;
486 flags = RCD_DEST_STRING;
491 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_ERROR)
494 else if (((rcd->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC
495 && flags != RCD_DEST_NUMERIC)
496 || ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_STRING
497 && flags != RCD_DEST_STRING))
499 else if ((rcd->flags & RCD_DEST_MASK) ^ flags)
501 msg (SE, _("Inconsistent output types. The output values "
502 "must be all numeric or all string."));
509 /* Reads a set of source specifications and returns one of the
510 following values: 0 on failure; 1 for normal success; 2 for success
511 but with CONVERT as the keyword; 3 for success but with ELSE as the
514 parse_src_spec (struct rcd_var * rcd, int type, size_t max_src_width)
520 if (rcd->nmap >= rcd->mmap - 1)
523 rcd->map = xrealloc (rcd->map, rcd->mmap * sizeof *rcd->map);
526 c = &rcd->map[rcd->nmap];
527 c->f1.c = c->f2.c = NULL;
528 if (lex_match_id ("ELSE"))
534 else if (type == NUMERIC)
538 if (lex_match_id ("LO") || lex_match_id ("LOWEST"))
540 if (!lex_force_match_id ("THRU"))
542 if (lex_match_id ("HI") || lex_match_id ("HIGHEST"))
544 else if (token == T_NUM)
552 lex_error (_("following LO THRU"));
556 else if (lex_match_id ("MISSING"))
559 rcd->flags |= RCD_MISC_MISSING;
561 else if (lex_match_id ("SYSMIS"))
564 rcd->flags |= RCD_MISC_MISSING;
568 lex_error (_("in source value"));
572 else if (token == T_NUM)
576 if (lex_match_id ("THRU"))
578 if (lex_match_id ("HI") || lex_match_id ("HIGHEST"))
580 else if (token == T_NUM)
593 c->type = RCD_SINGLE;
597 lex_error (_("in source value"));
603 assert (type == ALPHA);
604 if (lex_match_id ("CONVERT"))
606 if ((rcd->flags & RCD_DEST_MASK) == RCD_DEST_ERROR)
607 rcd->flags |= RCD_DEST_NUMERIC;
608 else if ((rcd->flags & RCD_DEST_MASK) != RCD_DEST_NUMERIC)
610 msg (SE, _("Keyword CONVERT may only be used with "
611 "string input values and numeric output "
616 c->type = RCD_CONVERT;
622 /* Only the debugging code needs the NULLs at the ends
623 of the strings. However, changing code behavior more
624 than necessary based on the DEBUGGING `#define' is just
626 c->type = RCD_SINGLE;
627 if (!lex_force_string ())
629 c->f1.c = xmalloc (max_src_width + 1);
630 st_pad_copy (c->f1.c, ds_value (&tokstr), max_src_width + 1);
635 if (c->type != RCD_END)
645 /* Data transformation. */
648 recode_trns_free (struct trns_header * t)
651 struct rcd_var *head, *next;
653 head = ((struct recode_trns *) t)->codings;
656 if (head->map && !(head->flags & RCD_MISC_DUPLICATE))
658 if (head->flags & RCD_SRC_STRING)
659 for (i = 0; i < head->nmap; i++)
660 switch (head->map[i].type)
663 free (head->map[i].f2.c);
669 free (head->map[i].f1.c);
678 if (head->flags & RCD_DEST_STRING)
679 for (i = 0; i < head->nmap; i++)
680 if (head->map[i].type != RCD_CONVERT && head->map[i].type != RCD_END)
681 free (head->map[i].t.c);
690 static inline struct coding *
691 find_src_numeric (struct rcd_var * v, struct ccase * c)
693 double cmp = c->data[v->src->fv].f;
698 if (v->sysmis.f != -SYSMIS)
700 if ((v->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
701 c->data[v->dest->fv].f = v->sysmis.f;
703 memcpy (c->data[v->dest->fv].s, v->sysmis.c,
709 for (cp = v->map;; cp++)
715 if (is_num_user_missing (cmp, v->src))
731 if (cmp >= cp->f1.f && cmp <= cp->f2.f)
741 static inline struct coding *
742 find_src_string (struct rcd_var * v, struct ccase * c)
744 char *cmp = c->data[v->src->fv].s;
745 int w = v->src->width;
748 for (cp = v->map;; cp++)
754 if (!memcmp (cp->f1.c, cmp, w))
761 double f = convert_to_double (cmp, w);
764 c->data[v->dest->fv].f = f;
775 recode_trns_proc (struct trns_header * t, struct ccase * c)
780 for (v = ((struct recode_trns *) t)->codings; v; v = v->next)
782 switch (v->flags & RCD_SRC_MASK)
784 case RCD_SRC_NUMERIC:
785 cp = find_src_numeric (v, c);
788 cp = find_src_string (v, c);
794 /* A matching input value was found. */
795 if ((v->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
797 double val = cp->t.f;
799 c->data[v->dest->fv].f = c->data[v->src->fv].f;
801 c->data[v->dest->fv].f = val;
807 st_bare_pad_len_copy (c->data[v->dest->fv].s,
808 c->data[v->src->fv].c,
809 v->dest->width, v->src->width);
811 memcpy (c->data[v->dest->fv].s, cp->t.c, v->dest->width);
822 dump_dest (struct rcd_var * v, union value * c)
824 if ((v->flags & RCD_DEST_MASK) == RCD_DEST_NUMERIC)
827 else if (c->f == -SYSMIS)
830 printf ("=%g", c->f);
832 printf ("=\"%s\"", c->c);
838 debug_print (struct rcd_var * head)
840 struct rcd_var *iter, *start;
844 for (iter = head; iter; iter = iter->next)
847 printf (" %s%s", iter == head ? "" : "/", iter->src->name);
848 while (iter->next && (iter->next->flags & RCD_MISC_DUPLICATE))
851 printf (" %s", iter->src->name);
853 if (iter->has_sysmis)
856 dump_dest (iter, &iter->sysmis);
859 for (c = iter->map; c->type != RCD_END; c++)
862 if ((iter->flags & RCD_SRC_MASK) == RCD_SRC_NUMERIC)
866 printf (_("!!END!!"));
872 printf ("%g", c->f1.f);
875 printf ("%g THRU HIGH", c->f1.f);
878 printf ("LOW THRU %g", c->f1.f);
881 printf ("%g THRU %g", c->f1.f, c->f2.f);
887 printf (_("!!ERROR!!"));
894 printf ("\"%s\"", c->f1.c);
903 printf (_("!!ERROR!!"));
906 if (c->type != RCD_CONVERT)
907 dump_dest (iter, &c->t);
914 start->dest_name[0] ? start->dest_name : start->dest->name);
924 /* Convert NPTR to a `long int' in base 10. Returns the long int on
925 success, NOT_LONG on failure. On success stores a pointer to the
926 first character after the number into *ENDPTR. From the GNU C
929 string_to_long (char *nptr, int width, char **endptr)
932 register unsigned long int cutoff;
933 register unsigned int cutlim;
934 register unsigned long int i;
936 register unsigned char c;
941 /* Check for a sign. */
954 if (s >= nptr + width)
957 /* Save the pointer so we can check later if anything happened. */
960 cutoff = ULONG_MAX / 10ul;
961 cutlim = ULONG_MAX % 10ul;
966 if (isdigit ((unsigned char) c))
970 /* Check for overflow. */
971 if (i > cutoff || (i == cutoff && c > cutlim))
977 if (s >= nptr + width)
982 /* Check if anything actually happened. */
986 /* Check for a value that is within the range of `unsigned long
987 int', but outside the range of `long int'. We limit LONG_MIN and
988 LONG_MAX by one point because we know that NOT_LONG is out there
991 ? -((unsigned long int) LONG_MIN) - 1
992 : ((unsigned long int) LONG_MAX) - 1))
997 /* Return the result of the appropriate sign. */
998 return (negative ? -i : i);
1001 /* Converts S to a double according to format Fx.0. Returns the value
1002 found, or -SYSMIS if there was no valid number in s. WIDTH is the
1003 length of string S. From the GNU C library. */
1005 convert_to_double (char *s, int width)
1007 register const char *end = &s[width];
1011 /* The number so far. */
1014 int got_dot; /* Found a decimal point. */
1015 int got_digit; /* Count of digits. */
1017 /* The exponent of the number. */
1020 /* Eat whitespace. */
1021 while (s < end && isspace ((unsigned char) *s))
1027 sign = *s == '-' ? -1 : 1;
1028 if (*s == '-' || *s == '+')
1039 for (; s < end; ++s)
1041 if (isdigit ((unsigned char) *s))
1045 /* Make sure that multiplication by 10 will not overflow. */
1046 if (num > DBL_MAX * 0.1)
1047 /* The value of the digit doesn't matter, since we have already
1048 gotten as many digits as can be represented in a `double'.
1049 This doesn't necessarily mean the result will overflow.
1050 The exponent may reduce it to within range.
1052 We just need to record that there was another
1053 digit so that we can multiply by 10 later. */
1056 num = (num * 10.0) + (*s - '0');
1058 /* Keep track of the number of digits after the decimal point.
1059 If we just divided by 10 here, we would lose precision. */
1063 else if (!got_dot && *s == '.')
1064 /* Record that we have found the decimal point. */
1073 if (s < end && (tolower ((unsigned char) (*s)) == 'e'
1074 || tolower ((unsigned char) (*s)) == 'd'))
1076 /* Get the exponent specified after the `e' or `E'. */
1083 exp = string_to_long (s, end - s, &s);
1084 if (exp == NOT_LONG || end == s)
1089 while (s < end && isspace ((unsigned char) *s))
1097 /* Multiply NUM by 10 to the EXPONENT power,
1098 checking for overflow and underflow. */
1102 if (-exponent + got_digit > -(DBL_MIN_10_EXP) + 5
1103 || num < DBL_MIN * pow (10.0, (double) -exponent))
1105 num *= pow (10.0, (double) exponent);
1107 else if (exponent > 0)
1109 if (num > DBL_MAX * pow (10.0, (double) -exponent))
1111 num *= pow (10.0, (double) exponent);
1114 return sign > 0 ? num : -num;