#include <config.h>
+#include "data/casereader.h"
#include "data/dataset.h"
#include "data/dictionary.h"
+#include "data/mrset.h"
#include "language/command.h"
#include "language/lexer/format-parser.h"
#include "language/lexer/lexer.h"
#include "language/lexer/variable-parser.h"
+#include "language/stats/freq.h"
+#include "libpspp/array.h"
+#include "libpspp/assertion.h"
#include "libpspp/hmap.h"
#include "libpspp/message.h"
+#include "libpspp/string-array.h"
#include "output/pivot-table.h"
#include "gl/minmax.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
+#define N_(msgid) (msgid)
enum ctables_vlabel
{
}
}
+/* XXX:
+ - unweighted summaries (U*)
+ - lower confidence limits (*.LCL)
+ - upper confidence limits (*.UCL)
+ - standard error (*.SE)
+ */
+#define SUMMARIES \
+ /* All variables. */ \
+ S(CTSF_COUNT, "COUNT", N_("Count"), CTF_COUNT, CTFA_ALL) \
+ S(CTSF_ECOUNT, "ECOUNT", N_("Adjusted Count"), CTF_COUNT, CTFA_ALL) \
+ S(CTSF_ROWPCT_COUNT, "ROWPCT.COUNT", N_("Row %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_COLPCT_COUNT, "COLPCT.COUNT", N_("Column %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_TABLEPCT_COUNT, "TABLEPCT.COUNT", N_("Table %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT.COUNT", N_("Subtable %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERPCT_COUNT, "LAYERPCT.COUNT", N_("Layer %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT.COUNT", N_("Layer Row %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT.COUNT", N_("Layer Column %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_ROWPCT_VALIDN, "ROWPCT.VALIDN", N_("Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_COLPCT_VALIDN, "COLPCT.VALIDN", N_("Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_TABLEPCT_VALIDN, "TABLEPCT.VALIDN", N_("Table Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_SUBTABLEPCT_VALIDN, "SUBTABLEPCT.VALIDN", N_("Subtable Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERPCT_VALIDN, "LAYERPCT.VALIDN", N_("Layer Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERROWPCT_VALIDN, "LAYERROWPCT.VALIDN", N_("Layer Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERCOLPCT_VALIDN, "LAYERCOLPCT.VALIDN", N_("Layer Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_ROWPCT_TOTALN, "ROWPCT.TOTALN", N_("Row Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_COLPCT_TOTALN, "COLPCT.TOTALN", N_("Column Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_TABLEPCT_TOTALN, "TABLEPCT.TOTALN", N_("Table Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_SUBTABLEPCT_TOTALN, "SUBTABLEPCT.TOTALN", N_("Subtable Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERPCT_TOTALN, "LAYERPCT.TOTALN", N_("Layer Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERROWPCT_TOTALN, "LAYERROWPCT.TOTALN", N_("Layer Row Total N %"), CTF_PERCENT, CTFA_ALL) \
+ S(CTSF_LAYERCOLPCT_TOTALN, "LAYERCOLPCT.TOTALN", N_("Layer Column Total N %"), CTF_PERCENT, CTFA_ALL) \
+ \
+ /* Scale variables, totals, and subtotals. */ \
+ S(CTSF_MAXIMUM, "MAXIMUM", N_("Maximum"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_MEAN, "MEAN", N_("Mean"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_MEDIAN, "MEDIAN", N_("Median"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_MINIMUM, "MINIMUM", N_("Minimum"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_MISSING, "MISSING", N_("Missing"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_MODE, "MODE", N_("Mode"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_PTILE, "PTILE", N_("Percentile"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_RANGE, "RANGE", N_("Range"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_SEMEAN, "SEMEAN", N_("Std Error of Mean"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_STDDEV, "STDDEV", N_("Std Deviation"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_SUM, "SUM", N_("Sum"), CTF_GENERAL, CTFA_SCALE) \
+ S(CSTF_TOTALN, "TOTALN", N_("Total N"), CTF_COUNT, CTFA_SCALE) \
+ S(CTSF_ETOTALN, "ETOTALN", N_("Adjusted Total N"), CTF_COUNT, CTFA_SCALE) \
+ S(CTSF_VALIDN, "VALIDN", N_("Valid N"), CTF_COUNT, CTFA_SCALE) \
+ S(CTSF_EVALIDN, "EVALIDN", N_("Adjusted Valid N"), CTF_COUNT, CTFA_SCALE) \
+ S(CTSF_VARIANCE, "VARIANCE", N_("Variance"), CTF_GENERAL, CTFA_SCALE) \
+ S(CTSF_ROWPCT_SUM, "ROWPCT.SUM", N_("Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_COLPCT_SUM, "COLPCT.SUM", N_("Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_TABLEPCT_SUM, "TABLEPCT.SUM", N_("Table Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_SUBTABLEPCT_SUM, "SUBTABLEPCT.SUM", N_("Subtable Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_LAYERPCT_SUM, "LAYERPCT.SUM", N_("Layer Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_LAYERROWPCT_SUM, "LAYERROWPCT.SUM", N_("Layer Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ S(CTSF_LAYERCOLPCT_SUM, "LAYERCOLPCT.SUM", N_("Layer Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
+ \
+ /* Multiple response sets. */ \
+ S(CTSF_RESPONSES, "RESPONSES", N_("Responses"), CTF_COUNT, CTFA_MRSETS) \
+ S(CTSF_ROWPCT_RESPONSES, "ROWPCT.RESPONSES", N_("Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_COLPCT_RESPONSES, "COLPCT.RESPONSES", N_("Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_TABLEPCT_RESPONSES, "TABLEPCT.RESPONSES", N_("Table Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_SUBTABLEPCT_RESPONSES, "SUBTABLEPCT.RESPONSES", N_("Subtable Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERPCT_RESPONSES, "LAYERPCT.RESPONSES", N_("Layer Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERROWPCT_RESPONSES, "LAYERROWPCT.RESPONSES", N_("Layer Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERCOLPCT_RESPONSES, "LAYERCOLPCT.RESPONSES", N_("Layer Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_ROWPCT_RESPONSES_COUNT, "ROWPCT.RESPONSES.COUNT", N_("Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_COLPCT_RESPONSES_COUNT, "COLPCT.RESPONSES.COUNT", N_("Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_TABLEPCT_RESPONSES_COUNT, "TABLEPCT.RESPONSES.COUNT", N_("Table Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_SUBTABLEPCT_RESPONSES_COUNT, "SUBTABLEPCT.RESPONSES.COUNT", N_("Subtable Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERPCT_RESPONSES_COUNT, "LAYERPCT.RESPONSES.COUNT", N_("Layer Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERROWPCT_RESPONSES_COUNT, "LAYERROWPCT.RESPONSES.COUNT", N_("Layer Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERCOLPCT_RESPONSES_COUNT, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_ROWPCT_COUNT_RESPONSES, "ROWPCT.COUNT.RESPONSES", N_("Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_COLPCT_COUNT_RESPONSES, "COLPCT.COUNT.RESPONSES", N_("Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_TABLEPCT_COUNT_RESPONSES, "TABLEPCT.COUNT.RESPONSES", N_("Table Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_SUBTABLEPCT_COUNT_RESPONSES, "SUBTABLEPCT.COUNT.RESPONSES", N_("Subtable Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERPCT_COUNT_RESPONSES, "LAYERPCT.COUNT.RESPONSES", N_("Layer Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERROWPCT_COUNT_RESPONSES, "LAYERROWPCT.COUNT.RESPONSES", N_("Layer Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
+ S(CTSF_LAYERCOLPCT_COUNT_RESPONSES, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS)
+
+enum ctables_summary_function
+ {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) ENUM,
+ SUMMARIES
+#undef S
+ };
+
+enum {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) +1
+ N_CTSF_FUNCTIONS = SUMMARIES
+#undef S
+};
+
struct ctables
{
struct pivot_table_look *look;
format. Otherwise, this string is displayed. */
char *missing;
+ /* Indexed by variable dictionary index. */
enum ctables_vlabel *vlabels;
bool mrsets_count_duplicates; /* MRSETS. */
/* CTPO_CAT_RANGE.
XXX what about string ranges? */
- struct
- {
- double low; /* -DBL_MAX for LO. */
- double high; /* DBL_MAX for HIGH. */
- }
- range;
+ double range[2];
/* CTPO_ADD, CTPO_SUB, CTPO_MUL, CTPO_DIV, CTPO_POW. */
struct ctables_postcompute_expr *subs[2];
enum ctables_label_position row_labels;
enum ctables_label_position col_labels;
- /* XXX CATEGORIES */
+ /* Indexed by variable dictionary index. */
+ struct ctables_categories **categories;
+ size_t n_categories;
double cilevel;
struct ctables_pairwise *pairwise;
};
+struct ctables_var
+ {
+ bool is_mrset;
+ union
+ {
+ struct variable *var;
+ const struct mrset *mrset;
+ };
+ };
+
+static const struct fmt_spec *
+ctables_var_get_print_format (const struct ctables_var *var)
+{
+ return (var->is_mrset
+ ? var_get_print_format (var->mrset->vars[0])
+ : var_get_print_format (var->var));
+}
+
+static const char *
+ctables_var_name (const struct ctables_var *var)
+{
+ return var->is_mrset ? var->mrset->name : var_get_name (var->var);
+}
+
+struct ctables_categories
+ {
+ size_t n_refs;
+
+ /* Explicit categories. */
+ struct ctables_cat_value *values;
+ size_t n_values;
+
+ /* Implicit categories. */
+ bool sort_ascending;
+ bool include_missing;
+ enum { CTCS_VALUE, CTCS_LABEL, CTCS_FUNCTION } key;
+ enum ctables_summary_function sort_func;
+ struct variable *sort_func_var;
+ double percentile;
+
+ /* Totals. */
+ bool show_totals;
+ bool totals_before;
+ char *total_label;
+
+ /* Empty categories. */
+ bool show_empty;
+ };
+
+struct ctables_cat_value
+ {
+ enum ctables_cat_value_type
+ {
+ CCVT_NUMBER,
+ CCVT_STRING,
+ CCVT_RANGE,
+ CCVT_MISSING,
+ CCVT_OTHERNM,
+ CCVT_SUBTOTAL,
+ CCVT_HSUBTOTAL,
+ }
+ type;
+
+ union
+ {
+ double number; /* CCVT_NUMBER. */
+ char *string; /* CCVT_STRING. */
+ double range[2]; /* CCVT_RANGE. */
+ char *subtotal_label; /* CCVT_SUBTOTAL, CCVT_HSUBTOTAL. */
+ };
+ };
+
+static void
+ctables_cat_value_uninit (struct ctables_cat_value *cv)
+{
+ if (!cv)
+ return;
+
+ switch (cv->type)
+ {
+ case CCVT_NUMBER:
+ case CCVT_RANGE:
+ case CCVT_MISSING:
+ case CCVT_OTHERNM:
+ break;
+
+ case CCVT_STRING:
+ free (cv->string);
+ break;
+
+ case CCVT_SUBTOTAL:
+ case CCVT_HSUBTOTAL:
+ free (cv->subtotal_label);
+ }
+}
+
+static void
+ctables_categories_unref (struct ctables_categories *c)
+{
+ if (!c)
+ return;
+
+ assert (c->n_refs > 0);
+ if (--c->n_refs)
+ return;
+
+ for (size_t i = 0; i < c->n_values; i++)
+ ctables_cat_value_uninit (&c->values[i]);
+ free (c->values);
+ free (c->total_label);
+ free (c);
+}
+
/* Chi-square test (SIGTEST). */
struct ctables_chisq
{
{
/* Terminals. */
CTAO_VAR,
- CTAO_MRSET,
/* Nonterminals. */
CTAO_STACK, /* + */
/* Terminals. */
struct
{
- union
- {
- struct variable *var;
- const struct mrset *mrset;
- };
-
+ struct ctables_var var;
bool scale;
struct ctables_summary *summaries;
size_t n_summaries;
+ size_t allocated_summaries;
};
/* Nonterminals. */
struct ctables_axis *subs[2];
};
+
+ struct msg_location *loc;
};
static void ctables_axis_destroy (struct ctables_axis *);
-#define SUMMARIES \
- /* All variables. */ \
- S(CTSF_COUNT, "COUNT") \
- S(CTSF_ECOUNT, "ECOUNT") \
- S(CTSF_ROWPCT_COUNT, "ROWPCT.COUNT") \
- S(CTSF_COLPCT_COUNT, "COLPCT.COUNT") \
- S(CTSF_TABLEPCT_COUNT, "TABLEPCT.COUNT") \
- S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT.COUNT") \
- S(CTSF_LAYERPCT_COUNT, "LAYERPCT.COUNT") \
- S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT.COUNT") \
- S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT.COUNT") \
- S(CTSF_ROWPCT_VALIDN, "ROWPCT.VALIDN") \
- S(CTSF_COLPCT_VALIDN, "COLPCT.VALIDN") \
- S(CTSF_TABLEPCT_VALIDN, "TABLEPCT.VALIDN") \
- S(CTSF_SUBTABLEPCT_VALIDN, "SUBTABLEPCT.VALIDN") \
- S(CTSF_LAYERPCT_VALIDN, "LAYERPCT.VALIDN") \
- S(CTSF_LAYERROWPCT_VALIDN, "LAYERROWPCT.VALIDN") \
- S(CTSF_LAYERCOLPCT_VALIDN, "LAYERCOLPCT.VALIDN") \
- S(CTSF_ROWPCT_TOTALN, "ROWPCT.TOTALN") \
- S(CTSF_COLPCT_TOTALN, "COLPCT.TOTALN") \
- S(CTSF_TABLEPCT_TOTALN, "TABLEPCT.TOTALN") \
- S(CTSF_SUBTABLEPCT_TOTALN, "SUBTABLEPCT.TOTALN") \
- S(CTSF_LAYERPCT_TOTALN, "LAYERPCT.TOTALN") \
- S(CTSF_LAYERROWPCT_TOTALN, "LAYERROWPCT.TOTALN") \
- S(CTSF_LAYERCOLPCT_TOTALN, "LAYERCOLPCT.TOTALN") \
- \
- /* Scale variables, totals, and subtotals. */ \
- S(CTSF_MAXIMUM, "!MAXIMUM") \
- S(CTSF_MEAN, "!MEAN") \
- S(CTSF_MEDIAN, "!MEDIAN") \
- S(CTSF_MINIMUM, "!MINIMUM") \
- S(CTSF_MISSING, "!MISSING") \
- S(CTSF_MODE, "!MODE") \
- S(CTSF_PTILE, "!PTILE") \
- S(CTSF_RANGE, "!RANGE") \
- S(CTSF_SEMAN, "!SEMAN") \
- S(CTSF_STDDEV, "!STDDEV") \
- S(CTSF_SUM, "!SUM") \
- S(CSTF_TOTALN, "!TOTALN") \
- S(CTSF_ETOTALN, "!ETOTALN") \
- S(CTSF_VALIDN, "!VALIDN") \
- S(CTSF_EVALIDN, "!EVALIDN") \
- S(CTSF_VARIANCE, "!VARIANCE") \
- S(CTSF_ROWPCT_SUM, "ROWPCT.SUM") \
- S(CTSF_COLPCT_SUM, "COLPCT.SUM") \
- S(CTSF_TABLEPCT_SUM, "TABLEPCT.SUM") \
- S(CTSF_SUBTABLEPCT_SUM, "SUBTABLEPCT.SUM") \
- S(CTSF_LAYERPCT_SUM, "LAYERPCT.SUM") \
- S(CTSF_LAYERROWPCT_SUM, "LAYERROWPCT.SUM") \
- S(CTSF_LAYERCOLPCT_SUM, "LAYERCOLPCT.SUM") \
- \
- /* Multiple response sets. */ \
- S(CTSF_ROWPCT_RESPONSES, "ROWPCT.RESPONSES") \
- S(CTSF_COLPCT_RESPONSES, "COLPCT.RESPONSES") \
- S(CTSF_TABLEPCT_RESPONSES, "TABLEPCT.RESPONSES") \
- S(CTSF_SUBTABLEPCT_RESPONSES, "SUBTABLEPCT.RESPONSES") \
- S(CTSF_LAYERPCT_RESPONSES, "LAYERPCT.RESPONSES") \
- S(CTSF_LAYERROWPCT_RESPONSES, "LAYERROWPCT.RESPONSES") \
- S(CTSF_LAYERCOLPCT_RESPONSES, "LAYERCOLPCT.RESPONSES") \
- S(CTSF_ROWPCT_RESPONSES_COUNT, "ROWPCT.RESPONSES.COUNT") \
- S(CTSF_COLPCT_RESPONSES_COUNT, "COLPCT.RESPONSES.COUNT") \
- S(CTSF_TABLEPCT_RESPONSES_COUNT, "TABLEPCT.RESPONSES.COUNT") \
- S(CTSF_SUBTABLEPCT_RESPONSES_COUNT, "SUBTABLEPCT.RESPONSES.COUNT") \
- S(CTSF_LAYERPCT_RESPONSES_COUNT, "LAYERPCT.RESPONSES.COUNT") \
- S(CTSF_LAYERROWPCT_RESPONSES_COUNT, "LAYERROWPCT.RESPONSES.COUNT") \
- S(CTSF_LAYERCOLPCT_RESPONSES_COUNT, "LAYERCOLPCT.RESPONSES.COUNT") \
- S(CTSF_ROWPCT_COUNT_RESPONSES, "ROWPCT.COUNT.RESPONSES") \
- S(CTSF_COLPCT_COUNT_RESPONSES, "COLPCT.COUNT.RESPONSES") \
- S(CTSF_TABLEPCT_COUNT_RESPONSES, "TABLEPCT.COUNT.RESPONSES") \
- S(CTSF_SUBTABLEPCT_COUNT_RESPONSES, "SUBTABLEPCT.COUNT.RESPONSES") \
- S(CTSF_LAYERPCT_COUNT_RESPONSES, "LAYERPCT.COUNT.RESPONSES") \
- S(CTSF_LAYERROWPCT_COUNT_RESPONSES, "LAYERROWPCT.COUNT.RESPONSES") \
- S(CTSF_LAYERCOLPCT_COUNT_RESPONSES, "LAYERCOLPCT.COUNT.RESPONSES")
-
-enum ctables_summary_function
+enum ctables_format
{
-#define S(ENUM, NAME) ENUM,
- SUMMARIES
-#undef S
+ CTF_COUNT,
+ CTF_PERCENT,
+ CTF_GENERAL
};
-enum {
-#define S(ENUM, NAME) +1
- N_CTSF_FUNCTIONS = SUMMARIES
-#undef S
-};
+enum ctables_function_availability
+ {
+ CTFA_ALL, /* Any variables. */
+ CTFA_SCALE, /* Only scale variables, totals, and subtotals. */
+ CTFA_MRSETS, /* Only multiple-response sets */
+ };
struct ctables_summary
{
enum ctables_summary_function function;
+ double percentile; /* CTSF_PTILE only. */
char *label;
struct fmt_spec format; /* XXX extra CTABLES formats */
};
return true;
}
+static enum ctables_function_availability
+ctables_function_availability (enum ctables_summary_function f)
+{
+ static enum ctables_function_availability availability[] = {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = AVAILABILITY,
+ SUMMARIES
+#undef S
+ };
+
+ return availability[f];
+}
+
static bool
parse_ctables_summary_function (struct lexer *lexer,
enum ctables_summary_function *f)
struct substring name;
};
static struct pair names[] = {
-#define S(ENUM, NAME) { ENUM, SS_LITERAL_INITIALIZER (NAME) },
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) \
+ { ENUM, SS_LITERAL_INITIALIZER (NAME) },
SUMMARIES
/* The .COUNT suffix may be omitted. */
- S(CTSF_ROWPCT_COUNT, "ROWPCT")
- S(CTSF_COLPCT_COUNT, "COLPCT")
- S(CTSF_TABLEPCT_COUNT, "TABLEPCT")
- S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT")
- S(CTSF_LAYERPCT_COUNT, "LAYERPCT")
- S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT")
- S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT")
+ S(CTSF_ROWPCT_COUNT, "ROWPCT", _, _, _)
+ S(CTSF_COLPCT_COUNT, "COLPCT", _, _, _)
+ S(CTSF_TABLEPCT_COUNT, "TABLEPCT", _, _, _)
+ S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT", _, _, _)
+ S(CTSF_LAYERPCT_COUNT, "LAYERPCT", _, _, _)
+ S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT", _, _, _)
+ S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT", _, _, _)
#undef S
};
if (ss_equals_case (names[i].name, lex_tokss (lexer)))
{
*f = names[i].function;
+ lex_get (lexer);
return true;
}
switch (axis->op)
{
case CTAO_VAR:
- case CTAO_MRSET:
for (size_t i = 0; i < axis->n_summaries; i++)
ctables_summary_uninit (&axis->summaries[i]);
free (axis->summaries);
ctables_axis_destroy (axis->subs[1]);
break;
}
+ msg_location_destroy (axis->loc);
free (axis);
}
static struct ctables_axis *
ctables_axis_new_nonterminal (enum ctables_axis_op op,
struct ctables_axis *sub0,
- struct ctables_axis *sub1)
+ struct ctables_axis *sub1,
+ struct lexer *lexer, int start_ofs)
{
struct ctables_axis *axis = xmalloc (sizeof *axis);
- *axis = (struct ctables_axis) { .op = op, .subs = { sub0, sub1 } };
+ *axis = (struct ctables_axis) {
+ .op = op,
+ .subs = { sub0, sub1 },
+ .loc = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1),
+ };
return axis;
}
struct ctables_table *t;
};
-static struct ctables_summary *
-add_summary (struct ctables_axis *axis, size_t *allocated_summaries)
+static struct fmt_spec
+ctables_summary_default_format (enum ctables_summary_function function,
+ const struct ctables_var *var)
+{
+ static const enum ctables_format default_formats[] = {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = FORMAT,
+ SUMMARIES
+#undef S
+ };
+ switch (default_formats[function])
+ {
+ case CTF_COUNT:
+ return (struct fmt_spec) { .type = FMT_F, .w = 40 };
+
+ case CTF_PERCENT:
+ return (struct fmt_spec) { .type = FMT_PCT, .w = 40, .d = 1 };
+
+ case CTF_GENERAL:
+ return *ctables_var_get_print_format (var);
+
+ default:
+ NOT_REACHED ();
+ }
+}
+
+static const char *
+ctables_summary_function_name (enum ctables_summary_function function)
+{
+ static const char *names[] = {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = NAME,
+ SUMMARIES
+#undef S
+ };
+ return names[function];
+}
+
+static bool
+add_summary (struct ctables_axis *axis,
+ enum ctables_summary_function function, double percentile,
+ const char *label, const struct fmt_spec *format,
+ const struct msg_location *loc)
{
- if (axis->n_summaries >= *allocated_summaries)
- axis->summaries = x2nrealloc (axis->summaries, allocated_summaries,
- sizeof *axis->summaries);
+ if (axis->op == CTAO_VAR)
+ {
+ if (axis->n_summaries >= axis->allocated_summaries)
+ axis->summaries = x2nrealloc (axis->summaries,
+ &axis->allocated_summaries,
+ sizeof *axis->summaries);
+
+ const char *function_name = ctables_summary_function_name (function);
+ const char *var_name = ctables_var_name (&axis->var);
+ switch (ctables_function_availability (function))
+ {
+ case CTFA_MRSETS:
+ if (!axis->var.is_mrset)
+ {
+ msg_at (SE, loc, _("Summary function %s applies only to multiple "
+ "response sets."), function_name);
+ msg_at (SN, axis->loc, _("'%s' is not a multiple response set."),
+ var_name);
+ return false;
+ }
+ break;
+
+ case CTFA_SCALE:
+ if (!axis->scale)
+ {
+ msg_at (SE, loc,
+ _("Summary function %s applies only to scale variables."),
+ function_name);
+ msg_at (SN, axis->loc, _("'%s' is not a scale variable."),
+ var_name);
+ return false;
+ }
+ break;
- struct ctables_summary *s = &axis->summaries[axis->n_summaries++];
- *s = (struct ctables_summary) { .function = CTSF_COUNT };
- return s;
+ case CTFA_ALL:
+ break;
+ }
+
+ struct ctables_summary *dst = &axis->summaries[axis->n_summaries++];
+ *dst = (struct ctables_summary) {
+ .function = function,
+ .percentile = percentile,
+ .label = xstrdup (label),
+ .format = (format ? *format
+ : ctables_summary_default_format (function, &axis->var)),
+ };
+ return true;
+ }
+ else
+ {
+ for (size_t i = 0; i < 2; i++)
+ if (!add_summary (axis->subs[i], function, percentile, label, format,
+ loc))
+ return false;
+ return true;
+ }
}
static struct ctables_axis *ctables_axis_parse_stack (
struct ctables_axis_parse_ctx *);
+static bool
+ctables_var_parse (struct lexer *lexer, struct dictionary *dict,
+ struct ctables_var *var)
+{
+ if (ss_starts_with (lex_tokss (lexer), ss_cstr ("$")))
+ {
+ *var = (struct ctables_var) {
+ .is_mrset = true,
+ .mrset = dict_lookup_mrset (dict, lex_tokcstr (lexer))
+ };
+ if (!var->mrset)
+ {
+ lex_error (lexer, _("'%s' does not name a multiple-response set "
+ "in the active file dictionary."),
+ lex_tokcstr (lexer));
+ return false;
+ }
+ lex_get (lexer);
+ return true;
+ }
+ else
+ {
+ *var = (struct ctables_var) {
+ .is_mrset = false,
+ .var = parse_variable (lexer, dict),
+ };
+ return var->var != NULL;
+ }
+}
+
static struct ctables_axis *
ctables_axis_parse_primary (struct ctables_axis_parse_ctx *ctx)
{
if (!lex_force_id (ctx->lexer))
return NULL;
- const struct mrset *mrset = NULL;
- struct variable *var = NULL;
- if (ss_starts_with (lex_tokss (ctx->lexer), ss_cstr ("$")))
- {
- mrset = dict_lookup_mrset (ctx->dict, lex_tokcstr (ctx->lexer));
- if (!mrset)
- {
- lex_error (ctx->lexer, _("'%s' is not the name of a "
- "multiple-response set in the active file "
- "dictionary."),
- lex_tokcstr (ctx->lexer));
- return NULL;
- }
- lex_get (ctx->lexer);
- }
- else
- {
- var = parse_variable (ctx->lexer, ctx->dict);
- if (!var)
- return NULL;
- }
+ int start_ofs = lex_ofs (ctx->lexer);
+ struct ctables_var var;
+ if (!ctables_var_parse (ctx->lexer, ctx->dict, &var))
+ return NULL;
struct ctables_axis *axis = xmalloc (sizeof *axis);
- if (mrset)
- *axis = (struct ctables_axis) { .op = CTAO_MRSET, .mrset = mrset };
- else
- *axis = (struct ctables_axis) { .op = CTAO_VAR, .var = var };
+ *axis = (struct ctables_axis) { .op = CTAO_VAR, .var = var };
/* XXX should figure out default measures by reading data */
- axis->scale = (mrset ? false
+ axis->scale = (var.is_mrset ? false
: lex_match_phrase (ctx->lexer, "[S]") ? true
: lex_match_phrase (ctx->lexer, "[C]") ? false
- : var_get_measure (var) == MEASURE_SCALE);
+ : var_get_measure (var.var) == MEASURE_SCALE);
+ axis->loc = lex_ofs_location (ctx->lexer, start_ofs,
+ lex_ofs (ctx->lexer) - 1);
+ return axis;
+}
- size_t allocated_summaries = 0;
- if (lex_match (ctx->lexer, T_LBRACK))
+static struct ctables_axis *
+ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
+{
+ struct ctables_axis *sub = ctables_axis_parse_primary (ctx);
+ if (!sub || !lex_match (ctx->lexer, T_LBRACK))
+ return sub;
+
+ do
{
- do
+ int start_ofs = lex_ofs (ctx->lexer);
+
+ /* Parse function. */
+ enum ctables_summary_function function;
+ if (!parse_ctables_summary_function (ctx->lexer, &function))
+ goto error;
+
+ /* Parse percentile. */
+ double percentile = 0;
+ if (function == CTSF_PTILE)
{
- struct ctables_summary *s = add_summary (axis, &allocated_summaries);
- if (!parse_ctables_summary_function (ctx->lexer, &s->function))
+ if (!lex_force_num_range_closed (ctx->lexer, "PTILE", 0, 100))
goto error;
- if (lex_is_string (ctx->lexer))
- {
- s->label = ss_xstrdup (lex_tokss (ctx->lexer));
- lex_get (ctx->lexer);
- }
- if (lex_token (ctx->lexer) == T_ID)
+ percentile = lex_number (ctx->lexer);
+ lex_get (ctx->lexer);
+ }
+
+ /* Parse label. */
+ char *label;
+ if (lex_is_string (ctx->lexer))
+ {
+ label = ss_xstrdup (lex_tokss (ctx->lexer));
+ lex_get (ctx->lexer);
+ }
+ else if (function == CTSF_PTILE)
+ label = xasprintf (_("Percentile %.2f"), percentile);
+ else
+ {
+ static const char *default_labels[] = {
+#define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = LABEL,
+ SUMMARIES
+#undef S
+ };
+ label = xstrdup (gettext (default_labels[function]));
+ }
+
+ /* Parse format. */
+ struct fmt_spec format;
+ const struct fmt_spec *formatp;
+ if (lex_token (ctx->lexer) == T_ID)
+ {
+ if (!parse_format_specifier (ctx->lexer, &format)
+ || !fmt_check_output (&format)
+ || !fmt_check_type_compat (&format, VAL_NUMERIC))
{
- if (!parse_format_specifier (ctx->lexer, &s->format)
- || !fmt_check_output (&s->format)
- || !fmt_check_type_compat (&s->format, VAL_NUMERIC))
- goto error;
+ free (label);
+ goto error;
}
- lex_match (ctx->lexer, T_COMMA);
+ formatp = &format;
}
- while (!lex_match (ctx->lexer, T_RBRACK));
+ else
+ formatp = NULL;
+
+ struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
+ lex_ofs (ctx->lexer) - 1);
+ add_summary (sub, function, percentile, label, formatp, loc);
+ free (label);
+ msg_location_destroy (loc);
+
+ lex_match (ctx->lexer, T_COMMA);
+ }
+ while (!lex_match (ctx->lexer, T_RBRACK));
+
+ return sub;
+
+error:
+ ctables_axis_destroy (sub);
+ return NULL;
+}
+
+static const struct ctables_axis *
+find_scale (const struct ctables_axis *axis)
+{
+ if (!axis)
+ return NULL;
+ else if (axis->op == CTAO_VAR)
+ {
+ if (axis->scale)
+ {
+ assert (!axis->var.is_mrset);
+ return axis;
+ }
+ else
+ return NULL;
}
else
{
- struct ctables_summary *s = add_summary (axis, &allocated_summaries);
- s->function = axis->scale ? CTSF_MEAN : CTSF_COUNT;
- s->label = xstrdup (axis->scale ? _("Mean") : _("Count"));
- s->format = (struct fmt_spec) { .type = FMT_F, .w = 40 };
+ for (size_t i = 0; i < 2; i++)
+ {
+ const struct ctables_axis *scale = find_scale (axis->subs[i]);
+ if (scale)
+ return scale;
+ }
+ return NULL;
}
- return axis;
+}
-error:
- ctables_axis_destroy (axis);
- return NULL;
+static const struct ctables_axis *
+find_categorical_summary (const struct ctables_axis *axis)
+{
+ if (!axis)
+ return NULL;
+ else if (axis->op == CTAO_VAR)
+ return !axis->scale && axis->n_summaries ? axis : NULL;
+ else
+ {
+ for (size_t i = 0; i < 2; i++)
+ {
+ const struct ctables_axis *sum
+ = find_categorical_summary (axis->subs[i]);
+ if (sum)
+ return sum;
+ }
+ return NULL;
+ }
}
static struct ctables_axis *
ctables_axis_parse_nest (struct ctables_axis_parse_ctx *ctx)
{
- struct ctables_axis *lhs = ctables_axis_parse_primary (ctx);
+ int start_ofs = lex_ofs (ctx->lexer);
+ struct ctables_axis *lhs = ctables_axis_parse_postfix (ctx);
if (!lhs)
return NULL;
- while (lex_match (ctx->lexer, T_PLUS))
+ while (lex_match (ctx->lexer, T_GT))
{
- struct ctables_axis *rhs = ctables_axis_parse_primary (ctx);
+ struct ctables_axis *rhs = ctables_axis_parse_postfix (ctx);
if (!rhs)
return NULL;
- lhs = ctables_axis_new_nonterminal (CTAO_NEST, lhs, rhs);
+ struct ctables_axis *nest = ctables_axis_new_nonterminal (
+ CTAO_NEST, lhs, rhs, ctx->lexer, start_ofs);
+
+ const struct ctables_axis *outer_scale = find_scale (lhs);
+ const struct ctables_axis *inner_scale = find_scale (rhs);
+ if (outer_scale && inner_scale)
+ {
+ msg_at (SE, nest->loc, _("Cannot nest scale variables."));
+ msg_at (SN, outer_scale->loc, _("This is an outer scale variable."));
+ msg_at (SN, inner_scale->loc, _("This is an inner scale variable."));
+ ctables_axis_destroy (nest);
+ return NULL;
+ }
+
+ const struct ctables_axis *outer_sum = find_categorical_summary (lhs);
+ if (outer_sum)
+ {
+ msg_at (SE, nest->loc,
+ _("Summaries may only be requested for categorical variables "
+ "at the innermost nesting level."));
+ msg_at (SN, outer_sum->loc,
+ _("This outer categorical variable has a summary."));
+ ctables_axis_destroy (nest);
+ return NULL;
+ }
+
+ lhs = nest;
}
return lhs;
static struct ctables_axis *
ctables_axis_parse_stack (struct ctables_axis_parse_ctx *ctx)
{
+ int start_ofs = lex_ofs (ctx->lexer);
struct ctables_axis *lhs = ctables_axis_parse_nest (ctx);
if (!lhs)
return NULL;
if (!rhs)
return NULL;
- lhs = ctables_axis_new_nonterminal (CTAO_STACK, lhs, rhs);
+ lhs = ctables_axis_new_nonterminal (CTAO_STACK, lhs, rhs,
+ ctx->lexer, start_ofs);
}
return lhs;
return t->axes[a] != NULL;
}
+static void
+ctables_chisq_destroy (struct ctables_chisq *chisq)
+{
+ free (chisq);
+}
+
+static void
+ctables_pairwise_destroy (struct ctables_pairwise *pairwise)
+{
+ free (pairwise);
+}
+
+static void
+ctables_table_uninit (struct ctables_table *t)
+{
+ if (!t)
+ return;
+
+ for (size_t i = 0; i < t->n_categories; i++)
+ ctables_categories_unref (t->categories[i]);
+ free (t->categories);
+
+ ctables_axis_destroy (t->axes[PIVOT_AXIS_COLUMN]);
+ ctables_axis_destroy (t->axes[PIVOT_AXIS_ROW]);
+ ctables_axis_destroy (t->axes[PIVOT_AXIS_LAYER]);
+ free (t->caption);
+ free (t->corner);
+ free (t->title);
+ ctables_chisq_destroy (t->chisq);
+ ctables_pairwise_destroy (t->pairwise);
+}
+
+static void
+ctables_destroy (struct ctables *ct)
+{
+ if (!ct)
+ return;
+
+ pivot_table_look_unref (ct->look);
+ free (ct->zero);
+ free (ct->missing);
+ free (ct->vlabels);
+ for (size_t i = 0; i < ct->n_tables; i++)
+ ctables_table_uninit (&ct->tables[i]);
+ free (ct->tables);
+ free (ct);
+}
+
+static struct ctables_cat_value
+ccvt_range (double low, double high)
+{
+ return (struct ctables_cat_value) {
+ .type = CCVT_RANGE,
+ .range = { low, high }
+ };
+}
+
+static bool
+ctables_table_parse_categories (struct lexer *lexer, struct dictionary *dict,
+ struct ctables_table *t)
+{
+ if (!lex_match_id (lexer, "VARIABLES"))
+ return false;
+ lex_match (lexer, T_EQUALS);
+
+ struct variable **vars;
+ size_t n_vars;
+ if (!parse_variables (lexer, dict, &vars, &n_vars, PV_NO_SCRATCH))
+ return false;
+
+ struct ctables_categories *c = xmalloc (sizeof *c);
+ *c = (struct ctables_categories) { .n_refs = n_vars };
+ for (size_t i = 0; i < n_vars; i++)
+ {
+ struct ctables_categories **cp
+ = &t->categories[var_get_dict_index (vars[i])];
+ ctables_categories_unref (*cp);
+ *cp = c;
+ }
+ free (vars);
+
+ if (lex_match (lexer, T_LBRACK))
+ {
+ size_t allocated_values = 0;
+ do
+ {
+ if (c->n_values >= allocated_values)
+ c->values = x2nrealloc (c->values, &allocated_values,
+ sizeof *c->values);
+
+ struct ctables_cat_value *v = &c->values[c->n_values];
+ if (lex_match_id (lexer, "OTHERNM"))
+ v->type = CCVT_OTHERNM;
+ else if (lex_match_id (lexer, "MISSING"))
+ v->type = CCVT_MISSING;
+ else if (lex_match_id (lexer, "SUBTOTAL"))
+ *v = (struct ctables_cat_value)
+ { .type = CCVT_SUBTOTAL, .subtotal_label = NULL };
+ else if (lex_match_id (lexer, "HSUBTOTAL"))
+ *v = (struct ctables_cat_value)
+ { .type = CCVT_HSUBTOTAL, .subtotal_label = NULL };
+ else if (lex_match_id (lexer, "LO"))
+ {
+ if (!lex_force_match_id (lexer, "THRU") || lex_force_num (lexer))
+ return false;
+ *v = ccvt_range (-DBL_MAX, lex_number (lexer));
+ lex_get (lexer);
+ }
+ else if (lex_is_number (lexer))
+ {
+ double number = lex_number (lexer);
+ lex_get (lexer);
+ if (lex_match_id (lexer, "THRU"))
+ {
+ v->type = CCVT_RANGE;
+ v->range[0] = number;
+ if (lex_match_id (lexer, "HI"))
+ *v = ccvt_range (number, DBL_MAX);
+ else
+ {
+ if (!lex_force_num (lexer))
+ return false;
+ *v = ccvt_range (number, lex_number (lexer));
+ lex_get (lexer);
+ }
+ }
+ else
+ *v = (struct ctables_cat_value) {
+ .type = CCVT_NUMBER,
+ .number = number
+ };
+ }
+ else if (lex_is_string (lexer))
+ {
+ *v = (struct ctables_cat_value) {
+ .type = CCVT_STRING,
+ .string = ss_xstrdup (lex_tokss (lexer)),
+ };
+ lex_get (lexer);
+ }
+ else
+ {
+ lex_error (lexer, NULL);
+ return false;
+ }
+
+ if ((v->type == CCVT_SUBTOTAL || v->type == CCVT_HSUBTOTAL)
+ && lex_match (lexer, T_EQUALS))
+ {
+ if (!lex_force_string (lexer))
+ return false;
+
+ v->subtotal_label = ss_xstrdup (lex_tokss (lexer));
+ lex_get (lexer);
+ }
+
+ c->n_values++;
+ lex_match (lexer, T_COMMA);
+ }
+ while (!lex_match (lexer, T_RBRACK));
+ }
+
+ while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
+ {
+ if (!c->n_values && lex_match_id (lexer, "ORDER"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (lex_match_id (lexer, "A"))
+ c->sort_ascending = true;
+ else if (lex_match_id (lexer, "D"))
+ c->sort_ascending = false;
+ else
+ {
+ lex_error_expecting (lexer, "A", "D");
+ return false;
+ }
+ }
+ else if (!c->n_values && lex_match_id (lexer, "KEY"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (lex_match_id (lexer, "VALUE"))
+ c->key = CTCS_VALUE;
+ else if (lex_match_id (lexer, "LABEL"))
+ c->key = CTCS_LABEL;
+ else
+ {
+ c->key = CTCS_FUNCTION;
+ if (!parse_ctables_summary_function (lexer, &c->sort_func))
+ return false;
+
+ if (lex_match (lexer, T_LPAREN))
+ {
+ c->sort_func_var = parse_variable (lexer, dict);
+ if (!c->sort_func_var)
+ return false;
+
+ if (c->sort_func == CTSF_PTILE)
+ {
+ lex_match (lexer, T_COMMA);
+ if (!lex_force_num_range_closed (lexer, "PTILE", 0, 100))
+ return false;
+ c->percentile = lex_number (lexer);
+ lex_get (lexer);
+ }
+
+ if (!lex_force_match (lexer, T_RPAREN))
+ return false;
+ }
+ else if (ctables_function_availability (c->sort_func)
+ == CTFA_SCALE)
+ {
+ bool UNUSED b = lex_force_match (lexer, T_LPAREN);
+ return false;
+ }
+ }
+ }
+ else if (!c->n_values && lex_match_id (lexer, "MISSING"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (lex_match_id (lexer, "INCLUDE"))
+ c->include_missing = true;
+ else if (lex_match_id (lexer, "EXCLUDE"))
+ c->include_missing = false;
+ else
+ {
+ lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
+ return false;
+ }
+ }
+ else if (lex_match_id (lexer, "TOTAL"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (!parse_bool (lexer, &c->show_totals))
+ return false;
+ }
+ else if (lex_match_id (lexer, "LABEL"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (!lex_force_string (lexer))
+ return false;
+ free (c->total_label);
+ c->total_label = ss_xstrdup (lex_tokss (lexer));
+ lex_get (lexer);
+ }
+ else if (lex_match_id (lexer, "POSITION"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (lex_match_id (lexer, "BEFORE"))
+ c->totals_before = true;
+ else if (lex_match_id (lexer, "AFTER"))
+ c->totals_before = false;
+ else
+ {
+ lex_error_expecting (lexer, "BEFORE", "AFTER");
+ return false;
+ }
+ }
+ else if (lex_match_id (lexer, "EMPTY"))
+ {
+ lex_match (lexer, T_EQUALS);
+ if (lex_match_id (lexer, "INCLUDE"))
+ c->show_empty = true;
+ else if (lex_match_id (lexer, "EXCLUDE"))
+ c->show_empty = false;
+ else
+ {
+ lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
+ return false;
+ }
+ }
+ else
+ {
+ if (!c->n_values)
+ lex_error_expecting (lexer, "ORDER", "KEY", "MISSING",
+ "TOTAL", "LABEL", "POSITION", "EMPTY");
+ else
+ lex_error_expecting (lexer, "TOTAL", "LABEL", "POSITION", "EMPTY");
+ return false;
+ }
+ }
+ return true;
+}
+
+struct var_array
+ {
+ struct variable **vars;
+ size_t n;
+ };
+
+static void
+var_array_uninit (struct var_array *va)
+{
+ if (va)
+ free (va->vars);
+}
+
+struct var_array2
+ {
+ struct var_array *vas;
+ size_t n;
+ };
+
+static void
+var_array2_uninit (struct var_array2 *vaa)
+{
+ if (vaa)
+ {
+ for (size_t i = 0; i < vaa->n; i++)
+ var_array_uninit (&vaa->vas[i]);
+ free (vaa->vas);
+ }
+}
+
+static struct var_array2
+nest_fts (struct var_array2 va0, struct var_array2 va1)
+{
+ if (!va0.n)
+ return va1;
+ else if (!va1.n)
+ return va0;
+
+ struct var_array2 vaa = { .vas = xnmalloc (va0.n, va1.n * sizeof *vaa.vas) };
+ for (size_t i = 0; i < va0.n; i++)
+ for (size_t j = 0; j < va1.n; j++)
+ {
+ size_t allocate = va0.vas[i].n + va1.vas[j].n;
+ struct variable **vars = xnmalloc (allocate, sizeof *vars);
+ size_t n = 0;
+ for (size_t k = 0; k < va0.vas[i].n; k++)
+ vars[n++] = va0.vas[i].vars[k];
+ for (size_t k = 0; k < va1.vas[j].n; k++)
+ vars[n++] = va1.vas[j].vars[k];
+ assert (n == allocate);
+
+ vaa.vas[vaa.n++] = (struct var_array) { .vars = vars, n = n };
+ }
+ var_array2_uninit (&va0);
+ var_array2_uninit (&va1);
+ return vaa;
+}
+
+static struct var_array2
+stack_fts (struct var_array2 va0, struct var_array2 va1)
+{
+ struct var_array2 vaa = { .vas = xnmalloc (va0.n + va1.n, sizeof *vaa.vas) };
+ for (size_t i = 0; i < va0.n; i++)
+ vaa.vas[vaa.n++] = va0.vas[i];
+ for (size_t i = 0; i < va1.n; i++)
+ vaa.vas[vaa.n++] = va1.vas[i];
+ assert (vaa.n == va0.n + va1.n);
+ free (va0.vas);
+ free (va1.vas);
+ return vaa;
+}
+
+static struct var_array2
+enumerate_fts (const struct ctables_axis *a)
+{
+ if (!a)
+ return (struct var_array2) { .n = 0 };
+
+ switch (a->op)
+ {
+ case CTAO_VAR:
+ assert (!a->var.is_mrset);
+ struct variable **v = xmalloc (sizeof *v);
+ *v = a->var.var;
+ struct var_array *va = xmalloc (sizeof *va);
+ *va = (struct var_array) { .vars = v, .n = 1 };
+ return (struct var_array2) { .vas = va, .n = 1 };
+
+ case CTAO_STACK:
+ return stack_fts (enumerate_fts (a->subs[0]),
+ enumerate_fts (a->subs[1]));
+
+ case CTAO_NEST:
+ return nest_fts (enumerate_fts (a->subs[0]),
+ enumerate_fts (a->subs[1]));
+ }
+
+ NOT_REACHED ();
+}
+
+struct ctables_freqtab
+ {
+ struct var_array vars;
+ struct hmap data; /* Contains "struct freq"s. */
+ struct freq **sorted;
+ };
+
+static int
+compare_freq_3way (const void *a_, const void *b_, const void *vars_)
+{
+ const struct var_array *vars = vars_;
+ struct freq *const *a = a_;
+ struct freq *const *b = b_;
+
+ for (size_t i = 0; i < vars->n; i++)
+ {
+ int cmp = value_compare_3way (&(*a)->values[i], &(*b)->values[i],
+ var_get_width (vars->vars[i]));
+ if (cmp)
+ return cmp;
+ }
+
+ return 0;
+}
+
+static bool
+ctables_execute (struct dataset *ds, struct ctables *ct)
+{
+ struct ctables_freqtab **fts = NULL;
+ size_t n_fts = 0;
+ size_t allocated_fts = 0;
+ for (size_t i = 0; i < ct->n_tables; i++)
+ {
+ struct ctables_table *t = &ct->tables[i];
+ struct var_array2 vaa = enumerate_fts (t->axes[PIVOT_AXIS_ROW]);
+ vaa = nest_fts (vaa, enumerate_fts (t->axes[PIVOT_AXIS_COLUMN]));
+ vaa = nest_fts (vaa, enumerate_fts (t->axes[PIVOT_AXIS_LAYER]));
+ for (size_t i = 0; i < vaa.n; i++)
+ {
+ for (size_t j = 0; j < vaa.vas[i].n; j++)
+ {
+ if (j)
+ fputs (", ", stdout);
+ fputs (var_get_name (vaa.vas[i].vars[j]), stdout);
+ }
+ putchar ('\n');
+ }
+
+ for (size_t j = 0; j < vaa.n; j++)
+ {
+ if (n_fts >= allocated_fts)
+ fts = x2nrealloc (fts, &allocated_fts, sizeof *fts);
+
+ struct ctables_freqtab *ft = xmalloc (sizeof *ft);
+ *ft = (struct ctables_freqtab) {
+ .vars = vaa.vas[j],
+ .data = HMAP_INITIALIZER (ft->data),
+ };
+ fts[n_fts++] = ft;
+ }
+
+ free (vaa.vas);
+ }
+
+ struct casereader *input = casereader_create_filter_weight (proc_open (ds),
+ dataset_dict (ds),
+ NULL, NULL);
+ bool warn_on_invalid = true;
+ for (struct ccase *c = casereader_read (input); c;
+ case_unref (c), c = casereader_read (input))
+ {
+ double weight = dict_get_case_weight (dataset_dict (ds), c,
+ &warn_on_invalid);
+ for (size_t i = 0; i < n_fts; i++)
+ {
+ struct ctables_freqtab *ft = fts[i];
+
+ size_t hash = 0;
+
+ for (size_t j = 0; j < ft->vars.n; j++)
+ {
+ const struct variable *var = ft->vars.vars[j];
+ hash = value_hash (case_data (c, var), var_get_width (var), hash);
+ }
+
+ struct freq *f;
+ HMAP_FOR_EACH_WITH_HASH (f, struct freq, node, hash, &ft->data)
+ {
+ for (size_t j = 0; j < ft->vars.n; j++)
+ {
+ const struct variable *var = ft->vars.vars[j];
+ if (!value_equal (case_data (c, var), &f->values[j],
+ var_get_width (var)))
+ goto next_hash_node;
+ }
+
+ f->count += weight;
+ goto next_ft;
+
+ next_hash_node: ;
+ }
+
+ f = xmalloc (table_entry_size (ft->vars.n));
+ f->count = weight;
+ for (size_t j = 0; j < ft->vars.n; j++)
+ {
+ const struct variable *var = ft->vars.vars[j];
+ value_clone (&f->values[j], case_data (c, var),
+ var_get_width (var));
+ }
+ hmap_insert (&ft->data, &f->node, hash);
+
+ next_ft: ;
+ }
+ }
+ casereader_destroy (input);
+
+ for (size_t i = 0; i < n_fts; i++)
+ {
+ struct ctables_freqtab *ft = fts[i];
+ ft->sorted = xnmalloc (ft->data.count, sizeof *ft->sorted);
+
+ struct freq *f;
+ size_t n = 0;
+ HMAP_FOR_EACH (f, struct freq, node, &ft->data)
+ ft->sorted[n++] = f;
+ sort (ft->sorted, ft->data.count, sizeof *ft->sorted,
+ compare_freq_3way, &ft->vars);
+ }
+
+ for (size_t i = 0; i < n_fts; i++)
+ {
+ struct ctables_freqtab *ft = fts[i];
+ struct freq *f, *next;
+ HMAP_FOR_EACH_SAFE (f, next, struct freq, node, &ft->data)
+ {
+ hmap_delete (&ft->data, &f->node);
+ for (size_t j = 0; j < ft->vars.n; j++)
+ {
+ const struct variable *var = ft->vars.vars[j];
+ value_destroy (&f->values[j], var_get_width (var));
+ }
+ free (f);
+ }
+ hmap_destroy (&ft->data);
+ free (ft->sorted);
+ var_array_uninit (&ft->vars);
+ free (ft);
+ }
+ free (fts);
+
+ return proc_commit (ds);
+}
+
int
cmd_ctables (struct lexer *lexer, struct dataset *ds)
{
size_t n_vars = dict_get_n_vars (dataset_dict (ds));
enum ctables_vlabel *vlabels = xnmalloc (n_vars, sizeof *vlabels);
- for (size_t i = 0; n_vars; i++)
+ for (size_t i = 0; i < n_vars; i++)
vlabels[i] = CTVL_DEFAULT;
struct ctables *ct = xmalloc (sizeof *ct);
.slabels_visible = true,
.row_labels = CTLP_NORMAL,
.col_labels = CTLP_NORMAL,
+ .categories = xcalloc (dict_get_n_vars (dataset_dict (ds)),
+ sizeof *t->categories),
+ .n_categories = dict_get_n_vars (dataset_dict (ds)),
.cilevel = 95,
};
lex_match (lexer, T_EQUALS);
if (!ctables_axis_parse (lexer, dataset_dict (ds), ct, t, PIVOT_AXIS_ROW))
goto error;
-
if (lex_match (lexer, T_BY))
{
if (!ctables_axis_parse (lexer, dataset_dict (ds),
goto error;
}
}
+
+ if (!t->axes[PIVOT_AXIS_ROW] && !t->axes[PIVOT_AXIS_COLUMN]
+ && !t->axes[PIVOT_AXIS_LAYER])
+ {
+ lex_error (lexer, _("At least one variable must be specified."));
+ goto error;
+ }
+
+ const struct ctables_axis *scales[PIVOT_N_AXES];
+ size_t n_scales = 0;
+ for (size_t i = 0; i < 3; i++)
+ {
+ scales[i] = find_scale (t->axes[i]);
+ if (scales[i])
+ n_scales++;
+ }
+ if (n_scales > 1)
+ {
+ msg (SE, _("Scale variables may appear only on one dimension."));
+ if (scales[PIVOT_AXIS_ROW])
+ msg_at (SN, scales[PIVOT_AXIS_ROW]->loc,
+ _("This scale variable appears in the rows dimension."));
+ if (scales[PIVOT_AXIS_COLUMN])
+ msg_at (SN, scales[PIVOT_AXIS_COLUMN]->loc,
+ _("This scale variable appears in the columns dimension."));
+ if (scales[PIVOT_AXIS_LAYER])
+ msg_at (SN, scales[PIVOT_AXIS_LAYER]->loc,
+ _("This scale variable appears in the layer dimension."));
+ goto error;
+ }
+
+ if (lex_token (lexer) == T_ENDCMD)
+ break;
if (!lex_force_match (lexer, T_SLASH))
- goto error;
+ break;
- /* XXX Validate axes. */
while (!lex_match_id (lexer, "TABLE") && lex_token (lexer) != T_ENDCMD)
{
if (lex_match_id (lexer, "SLABELS"))
t->slabels_position = PIVOT_AXIS_LAYER;
else
{
- lex_error_expecting (lexer, "COLUMN", "ROW",
- "LAYER");
+ lex_error_expecting (lexer, "COLUMN", "ROW", "LAYER");
goto error;
}
}
t->cilevel = lex_number (lexer);
lex_get (lexer);
}
+ else if (lex_match_id (lexer, "CATEGORIES"))
+ {
+ if (!ctables_table_parse_categories (lexer, dataset_dict (ds), t))
+ goto error;
+ }
else if (lex_match_id (lexer, "TITLES"))
{
do
goto error;
}
}
+
+ if (t->row_labels != CTLP_NORMAL && t->col_labels != CTLP_NORMAL)
+ {
+ msg (SE, _("ROWLABELS and COLLABELS may not both be specified."));
+ goto error;
+ }
+
}
while (lex_token (lexer) != T_ENDCMD);
- return CMD_SUCCESS;
+ bool ok = ctables_execute (ds, ct);
+ ctables_destroy (ct);
+ return ok ? CMD_SUCCESS : CMD_FAILURE;
error:
- /* XXX free */
+ ctables_destroy (ct);
return CMD_FAILURE;
}
projects / pspp / blobdiff