CTVL_BOTH = SETTINGS_VALUE_SHOW_BOTH,
};
+enum ctables_weighting
+ {
+ CTW_EFFECTIVE,
+ CTW_DICTIONARY,
+ CTW_UNWEIGHTED
+#define N_CTWS 3
+ };
+
enum ctables_function_type
{
- /* A function that operates on data in a single cell. The function does
- not have an unweighted version. */
+ /* A function that operates on data in a single cell. It operates on
+ effective weights. It does not have an unweighted version. */
CTFT_CELL,
- /* A function that operates on data in a single cell. The function has an
- unweighted version. */
+ /* A function that operates on data in a single cell. The function
+ operates on effective weights and has a U-prefixed unweighted
+ version. */
CTFT_UCELL,
- /* A function that operates on an area of cells. The function has an
- unweighted version. */
+ /* A function that operates on data in a single cell. It operates on
+ dictionary weights, and has U-prefixed unweighted version and an
+ E-prefixed effective weight version. */
+ CTFT_UECELL,
+
+ /* A function that operates on an area of cells. It operates on effective
+ weights and has a U-prefixed unweighted version. */
CTFT_AREA,
};
enum ctables_format format;
enum ctables_function_availability availability;
- bool may_be_unweighted;
- bool is_area;
+ bool u_prefix; /* Accepts a 'U' prefix (for unweighted)? */
+ bool e_prefix; /* Accepts an 'E' prefix (for effective)? */
+ bool is_area; /* Needs an area prefix. */
};
static const struct ctables_function_info ctables_function_info[N_CTSF_FUNCTIONS] = {
#define S(ENUM, NAME, TYPE, FORMAT, AVAILABILITY) \
.type = TYPE, \
.format = FORMAT, \
.availability = AVAILABILITY, \
- .may_be_unweighted = (TYPE) == CTFT_UCELL || (TYPE) == CTFT_AREA, \
+ .u_prefix = (TYPE) == CTFT_UCELL || (TYPE) == CTFT_UECELL || (TYPE) == CTFT_AREA, \
+ .e_prefix = (TYPE) == CTFT_UECELL, \
.is_area = (TYPE) == CTFT_AREA \
},
#include "ctables.inc"
#undef S
};
-static bool ctables_summary_function_is_count (enum ctables_summary_function);
-
enum ctables_area_type
{
/* Within a section, where stacked variables divide one section from
const struct ctables_cell *example;
size_t sequence;
- double d_valid; /* Dictionary weight. */
- double d_count;
- double d_total;
- double e_valid; /* Effective weight */
- double e_count;
- double e_total;
- double u_valid; /* Unweighted. */
- double u_count;
- double u_total;
+ double count[N_CTWS];
+ double valid[N_CTWS];
+ double total[N_CTWS];
struct ctables_sum *sums;
};
struct ctables_sum
{
- double e_sum;
- double u_sum;
+ double sum[N_CTWS];
};
enum ctables_summary_variant
struct variable **vars;
size_t n;
size_t scale_idx;
+ size_t summary_idx;
size_t *areas[N_CTATS];
size_t n_areas[N_CTATS];
size_t group_head;
/* CCT_FUNCTION. */
enum ctables_summary_function sort_function;
- bool weighted;
+ enum ctables_weighting weighting;
enum ctables_area_type area;
struct variable *sort_var;
double percentile;
cell, it must be 0). For CTSF_PTILE only, 'percentile' is the
percentile between 0 and 100 (for other functions it must be 0). */
enum ctables_summary_function function;
- bool weighted;
+ enum ctables_weighting weighting;
enum ctables_area_type calc_area;
double percentile; /* CTSF_PTILE only. */
return availability[f];
}
-static bool
-ctables_summary_function_is_count (enum ctables_summary_function f)
-{
- return f == CTSF_COUNT || f == CTSF_ECOUNT;
-}
-
static bool
parse_ctables_summary_function (struct lexer *lexer,
enum ctables_summary_function *function,
- bool *weighted,
+ enum ctables_weighting *weighting,
enum ctables_area_type *area)
{
if (!lex_force_id (lexer))
return false;
struct substring name = lex_tokss (lexer);
- *weighted = !(ss_match_byte (&name, 'U') || ss_match_byte (&name, 'u'));
+ bool u = ss_match_byte (&name, 'U') || ss_match_byte (&name, 'u');
+ bool e = !u && (ss_match_byte (&name, 'E') || ss_match_byte (&name, 'e'));
bool has_area = false;
*area = 0;
{
/* Special case where .COUNT suffix is omitted. */
*function = CTSF_areaPCT_COUNT;
+ *weighting = CTW_EFFECTIVE;
lex_get (lexer);
return true;
}
if (ss_equals_case (cfi->basename, name))
{
*function = f;
- if (!*weighted && !cfi->may_be_unweighted)
- break;
- if (has_area != cfi->is_area)
+ if ((u && !cfi->u_prefix) || (e && !cfi->e_prefix) || (has_area != cfi->is_area))
break;
+ *weighting = (e ? CTW_EFFECTIVE
+ : u ? CTW_UNWEIGHTED
+ : cfi->e_prefix ? CTW_DICTIONARY
+ : CTW_EFFECTIVE);
lex_get (lexer);
return true;
}
static const char *
ctables_summary_label__ (const struct ctables_summary_spec *spec)
{
- bool w = spec->weighted;
+ bool w = spec->weighting != CTW_UNWEIGHTED;
+ bool d = spec->weighting == CTW_DICTIONARY;
enum ctables_area_type a = spec->user_area;
switch (spec->function)
{
case CTSF_COUNT:
- return w ? N_("Count") : N_("Unweighted Count");
-
- case CTSF_ECOUNT:
- return N_("Adjusted Count");
+ return (d ? N_("Count")
+ : w ? N_("Adjusted Count")
+ : N_("Unweighted Count"));
case CTSF_areaPCT_COUNT:
switch (a)
case CTSF_MAXIMUM: return N_("Maximum");
case CTSF_MEAN: return w ? N_("Mean") : N_("Unweighted Mean");
- case CTSF_MEDIAN: return N_("Median");
+ case CTSF_MEDIAN: return w ? N_("Median") : N_("Unweighted Median");
case CTSF_MINIMUM: return N_("Minimum");
- case CTSF_MISSING: return N_("Missing");
- case CTSF_MODE: return N_("Mode");
+ case CTSF_MISSING: return w ? N_("Missing") : N_("Unweighted Missing");
+ case CTSF_MODE: return w ? N_("Mode") : N_("Unweighted Mode");
case CTSF_PTILE: NOT_REACHED ();
case CTSF_RANGE: return N_("Range");
- case CTSF_SEMEAN: return N_("Std Error of Mean");
- case CTSF_STDDEV: return N_("Std Deviation");
- case CTSF_SUM: return N_("Sum");
- case CTSF_TOTALN: return N_("Total N");
- case CTSF_ETOTALN: return N_("Adjusted Total N");
- case CTSF_VALIDN: return N_("Valid N");
- case CTSF_EVALIDN: return N_("Adjusted Valid N");
- case CTSF_VARIANCE: return N_("Variance");
+ case CTSF_SEMEAN: return w ? N_("Std Error of Mean") : N_("Unweighted Std Error of Mean");
+ case CTSF_STDDEV: return w ? N_("Std Deviation") : N_("Unweighted Std Deviation");
+ case CTSF_SUM: return w ? N_("Sum") : N_("Unweighted Sum");
+ case CTSF_TOTALN: return (d ? N_("Total N")
+ : w ? N_("Adjusted Total N")
+ : N_("Unweighted Total N"));
+ case CTSF_VALIDN: return (d ? N_("Valid N")
+ : w ? N_("Adjusted Valid N")
+ : N_("Unweighted Valid N"));
+ case CTSF_VARIANCE: return w ? N_("Variance") : N_("Unweighted Variance");
case CTSF_areaPCT_SUM:
switch (a)
{
if (spec->function == CTSF_PTILE)
{
double p = spec->percentile;
- char *s = (spec->weighted
+ char *s = (spec->weighting != CTW_UNWEIGHTED
? xasprintf (_("Percentile %.2f"), p)
: xasprintf (_("Unweighted Percentile %.2f"), p));
return pivot_value_new_user_text_nocopy (s);
static const char *
ctables_summary_function_name (enum ctables_summary_function function,
- bool weighted,
+ enum ctables_weighting weighting,
enum ctables_area_type area,
char *buffer, size_t bufsize)
{
const struct ctables_function_info *cfi = &ctables_function_info[function];
snprintf (buffer, bufsize, "%s%s%s",
- weighted ? "" : "U",
+ (weighting == CTW_UNWEIGHTED ? "U"
+ : weighting == CTW_DICTIONARY ? ""
+ : cfi->e_prefix ? "E"
+ : ""),
cfi->is_area ? ctables_area_type_name[area] : "",
cfi->basename.string);
return buffer;
static bool
add_summary_spec (struct ctables_axis *axis,
- enum ctables_summary_function function, bool weighted,
+ enum ctables_summary_function function,
+ enum ctables_weighting weighting,
enum ctables_area_type area, double percentile,
const char *label, const struct fmt_spec *format,
bool is_ctables_format, const struct msg_location *loc,
if (axis->op == CTAO_VAR)
{
char function_name[128];
- ctables_summary_function_name (function, weighted, area,
+ ctables_summary_function_name (function, weighting, area,
function_name, sizeof function_name);
const char *var_name = var_get_name (axis->var);
switch (ctables_function_availability (function))
struct ctables_summary_spec *dst = &set->specs[set->n++];
*dst = (struct ctables_summary_spec) {
.function = function,
- .weighted = weighted,
+ .weighting = weighting,
.calc_area = area,
.user_area = area,
.percentile = percentile,
else
{
for (size_t i = 0; i < 2; i++)
- if (!add_summary_spec (axis->subs[i], function, weighted, area,
+ if (!add_summary_spec (axis->subs[i], function, weighting, area,
percentile, label, format, is_ctables_format,
loc, sv))
return false;
/* Parse function. */
enum ctables_summary_function function;
- bool weighted;
+ enum ctables_weighting weighting;
enum ctables_area_type area;
- if (!parse_ctables_summary_function (ctx->lexer, &function, &weighted,
+ if (!parse_ctables_summary_function (ctx->lexer, &function, &weighting,
&area))
goto error;
struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
lex_ofs (ctx->lexer) - 1);
- add_summary_spec (sub, function, weighted, area, percentile, label,
+ add_summary_spec (sub, function, weighting, area, percentile, label,
formatp, is_ctables_format, loc, sv);
free (label);
msg_location_destroy (loc);
{
cat.type = CCT_FUNCTION;
if (!parse_ctables_summary_function (lexer, &cat.sort_function,
- &cat.weighted, &cat.area))
+ &cat.weighting, &cat.area))
goto error;
if (lex_match (lexer, T_LPAREN))
.scale_idx = (a->scale_idx != SIZE_MAX ? a->scale_idx
: b->scale_idx != SIZE_MAX ? a->n + b->scale_idx
: SIZE_MAX),
+ .summary_idx = (a->summary_idx != SIZE_MAX ? a->summary_idx
+ : b->summary_idx != SIZE_MAX ? a->n + b->summary_idx
+ : SIZE_MAX),
.n = n,
};
for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
struct variable **vars = xmalloc (sizeof *vars);
*vars = a->var;
+ bool is_summary = a->specs[CSV_CELL].n || a->scale;
struct ctables_nest *nest = xmalloc (sizeof *nest);
*nest = (struct ctables_nest) {
.vars = vars,
.n = 1,
.scale_idx = a->scale ? 0 : SIZE_MAX,
+ .summary_idx = is_summary ? 0 : SIZE_MAX,
};
- if (a->specs[CSV_CELL].n || a->scale)
+ if (is_summary)
for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
{
ctables_summary_spec_set_clone (&nest->specs[sv], &a->specs[sv]);
double ovalid;
double ovalue;
};
-
- /* XXX multiple response */
};
static void
switch (ss->function)
{
case CTSF_COUNT:
- case CTSF_ECOUNT:
case CTSF_areaPCT_COUNT:
case CTSF_areaPCT_VALIDN:
case CTSF_areaPCT_TOTALN:
case CTSF_MISSING:
case CTSF_TOTALN:
- case CTSF_ETOTALN:
case CTSF_VALIDN:
- case CTSF_EVALIDN:
s->count = 0;
break;
break;
case CTSF_MEAN:
+ case CTSF_SUM:
+ case CTSF_areaPCT_SUM:
+ s->moments = moments1_create (MOMENT_MEAN);
+ break;
+
case CTSF_SEMEAN:
case CTSF_STDDEV:
- case CTSF_SUM:
case CTSF_VARIANCE:
- case CTSF_areaPCT_SUM:
s->moments = moments1_create (MOMENT_VARIANCE);
break;
switch (ss->function)
{
case CTSF_COUNT:
- case CTSF_ECOUNT:
case CTSF_areaPCT_COUNT:
case CTSF_areaPCT_VALIDN:
case CTSF_areaPCT_TOTALN:
case CTSF_MISSING:
case CTSF_TOTALN:
- case CTSF_ETOTALN:
case CTSF_VALIDN:
- case CTSF_EVALIDN:
break;
case CTSF_areaID:
static void
ctables_summary_add (union ctables_summary *s,
const struct ctables_summary_spec *ss,
- const struct variable *var, const union value *value,
+ const union value *value,
bool is_scale, bool is_scale_missing,
- bool is_missing, bool excluded_missing,
- double d_weight, double e_weight)
+ bool is_missing, bool is_included,
+ double weight)
{
/* To determine whether a case is included in a given table for a particular
kind of summary, consider the following charts for each variable in the
switch (ss->function)
{
case CTSF_TOTALN:
+ s->count += weight;
+ break;
+
case CTSF_areaPCT_TOTALN:
- s->count += ss->weighted ? d_weight : 1.0;
+ s->count += weight;
break;
case CTSF_COUNT:
+ if (is_scale || is_included)
+ s->count += weight;
+ break;
+
case CTSF_areaPCT_COUNT:
- if (is_scale || !excluded_missing)
- s->count += ss->weighted ? d_weight : 1.0;
+ if (is_scale || is_included)
+ s->count += weight;
break;
case CTSF_VALIDN:
+ if (is_scale
+ ? !is_scale_missing
+ : !is_missing)
+ s->count += weight;
+ break;
+
case CTSF_areaPCT_VALIDN:
if (is_scale
? !is_scale_missing
: !is_missing)
- s->count += ss->weighted ? d_weight : 1.0;
+ s->count += weight;
break;
case CTSF_areaID:
if (is_scale
? is_scale_missing
: is_missing)
- s->count += ss->weighted ? d_weight : 1.0;
- break;
-
- case CTSF_ECOUNT:
- if (is_scale || !excluded_missing)
- s->count += e_weight;
- break;
-
- case CTSF_EVALIDN:
- if (is_scale
- ? !is_scale_missing
- : !is_missing)
- s->count += e_weight;
- break;
-
- case CTSF_ETOTALN:
- s->count += e_weight;
+ s->count += weight;
break;
case CTSF_MAXIMUM:
case CTSF_RANGE:
if (!is_scale_missing)
{
- assert (!var_is_alpha (var)); /* XXX? */
if (s->min == SYSMIS || value->f < s->min)
s->min = value->f;
if (s->max == SYSMIS || value->f > s->max)
case CTSF_STDDEV:
case CTSF_SUM:
case CTSF_VARIANCE:
- case CTSF_areaPCT_SUM:
if (!is_scale_missing)
- moments1_add (s->moments, value->f, ss->weighted ? e_weight : 1.0);
+ moments1_add (s->moments, value->f, weight);
+ break;
+
+ case CTSF_areaPCT_SUM:
+ if (!is_missing && !is_scale_missing)
+ moments1_add (s->moments, value->f, weight);
break;
case CTSF_MEDIAN:
case CTSF_PTILE:
if (!is_scale_missing)
{
- double w = ss->weighted ? e_weight : 1.0;
- s->ovalid += w;
+ s->ovalid += weight;
struct ccase *c = case_create (casewriter_get_proto (s->writer));
*case_num_rw_idx (c, 0) = value->f;
- *case_num_rw_idx (c, 1) = w;
+ *case_num_rw_idx (c, 1) = weight;
casewriter_write (s->writer, c);
}
break;
switch (ss->function)
{
case CTSF_COUNT:
- case CTSF_ECOUNT:
return s->count;
case CTSF_areaID:
case CTSF_areaPCT_COUNT:
{
const struct ctables_area *a = cell->areas[ss->calc_area];
- double a_count = ss->weighted ? a->e_count : a->u_count;
+ double a_count = a->count[ss->weighting];
return a_count ? s->count / a_count * 100 : SYSMIS;
}
case CTSF_areaPCT_VALIDN:
{
const struct ctables_area *a = cell->areas[ss->calc_area];
- double a_valid = ss->weighted ? a->e_valid : a->u_valid;
+ double a_valid = a->valid[ss->weighting];
return a_valid ? s->count / a_valid * 100 : SYSMIS;
}
case CTSF_areaPCT_TOTALN:
{
const struct ctables_area *a = cell->areas[ss->calc_area];
- double a_total = ss->weighted ? a->e_total : a->u_total;
+ double a_total = a->total[ss->weighting];
return a_total ? s->count / a_total * 100 : SYSMIS;
}
case CTSF_MISSING:
case CTSF_TOTALN:
- case CTSF_ETOTALN:
case CTSF_VALIDN:
- case CTSF_EVALIDN:
return s->count;
case CTSF_MAXIMUM:
const struct ctables_area *a = cell->areas[ss->calc_area];
const struct ctables_sum *sum = &a->sums[ss->sum_var_idx];
- double denom = ss->weighted ? sum->e_sum : sum->u_sum;
+ double denom = sum->sum[ss->weighting];
return denom != 0 ? weight * mean / denom * 100 : SYSMIS;
}
return 0;
}
-/* Algorithm:
-
- For each row:
- For each ctables_table:
- For each combination of row vars:
- For each combination of column vars:
- For each combination of layer vars:
- Add entry
- Make a table of row values:
- Sort entries by row values
- Assign a 0-based index to each actual value
- Construct a dimension
- Make a table of column values
- Make a table of layer values
- For each entry:
- Fill the table entry using the indexes from before.
- */
-
static struct ctables_area *
ctables_area_insert (struct ctables_section *s, struct ctables_cell *cell,
enum ctables_area_type area)
static struct ctables_cell *
ctables_cell_insert__ (struct ctables_section *s, const struct ccase *c,
- const struct ctables_category *cats[PIVOT_N_AXES][10])
+ const struct ctables_category **cats[PIVOT_N_AXES])
{
size_t hash = 0;
enum ctables_summary_variant sv = CSV_CELL;
|| cat->type == CCT_SUBTOTAL
|| cat->type == CCT_POSTCOMPUTE)
{
- /* XXX these should be more encompassing I think.*/
-
switch (a)
{
case PIVOT_AXIS_COLUMN:
}
static bool
-is_scale_missing (const struct ctables_summary_spec_set *specs,
- const struct ccase *c)
+is_listwise_missing (const struct ctables_summary_spec_set *specs,
+ const struct ccase *c)
{
- if (!specs->is_scale)
- return false;
-
- if (var_is_num_missing (specs->var, case_num (c, specs->var)))
- return true;
-
for (size_t i = 0; i < specs->n_listwise_vars; i++)
{
const struct variable *var = specs->listwise_vars[i];
return false;
}
+static void
+add_weight (double dst[N_CTWS], const double src[N_CTWS])
+{
+ for (enum ctables_weighting wt = 0; wt < N_CTWS; wt++)
+ dst[wt] += src[wt];
+}
+
static void
ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
- const struct ctables_category *cats[PIVOT_N_AXES][10],
- bool is_missing, bool excluded_missing,
- double d_weight, double e_weight)
+ const struct ctables_category **cats[PIVOT_N_AXES],
+ bool is_included, double weight[N_CTWS])
{
struct ctables_cell *cell = ctables_cell_insert__ (s, c, cats);
const struct ctables_nest *ss = s->nests[s->table->summary_axis];
const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
+ const union value *value = case_data (c, specs->var);
+ bool is_missing = var_is_value_missing (specs->var, value);
+ bool scale_missing = specs->is_scale && (is_missing || is_listwise_missing (specs, c));
- bool scale_missing = is_scale_missing (specs, c);
for (size_t i = 0; i < specs->n; i++)
- ctables_summary_add (&cell->summaries[i], &specs->specs[i],
- specs->var, case_data (c, specs->var), specs->is_scale,
- scale_missing, is_missing, excluded_missing,
- d_weight, e_weight);
+ ctables_summary_add (&cell->summaries[i], &specs->specs[i], value,
+ specs->is_scale, scale_missing, is_missing,
+ is_included, weight[specs->specs[i].weighting]);
for (enum ctables_area_type at = 0; at < N_CTATS; at++)
if (!(cell->omit_areas && (1u << at)))
{
struct ctables_area *a = cell->areas[at];
- a->d_total += d_weight;
- a->e_total += e_weight;
- a->u_total += 1.0;
- if (!excluded_missing)
- {
- a->d_count += d_weight;
- a->e_count += e_weight;
- a->u_count += 1.0;
- }
+
+ add_weight (a->total, weight);
+ if (is_included)
+ add_weight (a->count, weight);
if (!is_missing)
{
- a->d_valid += d_weight;
- a->e_valid += e_weight;
- a->u_count += 1.0;
+ add_weight (a->valid, weight);
- for (size_t i = 0; i < s->table->n_sum_vars; i++)
- {
- /* XXX listwise_missing??? */
- const struct variable *var = s->table->sum_vars[i];
- double addend = case_num (c, var);
- if (!var_is_num_missing (var, addend))
- {
- struct ctables_sum *sum = &a->sums[i];
- sum->e_sum += addend * e_weight;
- sum->u_sum += addend;
- }
- }
+ if (!scale_missing)
+ for (size_t i = 0; i < s->table->n_sum_vars; i++)
+ {
+ const struct variable *var = s->table->sum_vars[i];
+ double addend = case_num (c, var);
+ if (!var_is_num_missing (var, addend))
+ for (enum ctables_weighting wt = 0; wt < N_CTWS; wt++)
+ a->sums[i].sum[wt] += addend * weight[wt];
+ }
}
}
}
static void
recurse_totals (struct ctables_section *s, const struct ccase *c,
- const struct ctables_category *cats[PIVOT_N_AXES][10],
- bool is_missing, bool excluded_missing,
- double d_weight, double e_weight,
+ const struct ctables_category **cats[PIVOT_N_AXES],
+ bool is_included, double weight[N_CTWS],
enum pivot_axis_type start_axis, size_t start_nest)
{
for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
{
const struct ctables_category *save = cats[a][i];
cats[a][i] = total;
- ctables_cell_add__ (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight);
- recurse_totals (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight, a, i + 1);
+ ctables_cell_add__ (s, c, cats, is_included, weight);
+ recurse_totals (s, c, cats, is_included, weight, a, i + 1);
cats[a][i] = save;
}
}
static void
recurse_subtotals (struct ctables_section *s, const struct ccase *c,
- const struct ctables_category *cats[PIVOT_N_AXES][10],
- bool is_missing, bool excluded_missing,
- double d_weight, double e_weight,
+ const struct ctables_category **cats[PIVOT_N_AXES],
+ bool is_included, double weight[N_CTWS],
enum pivot_axis_type start_axis, size_t start_nest)
{
for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
if (save->subtotal)
{
cats[a][i] = save->subtotal;
- ctables_cell_add__ (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight);
- recurse_subtotals (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight, a, i + 1);
+ ctables_cell_add__ (s, c, cats, is_included, weight);
+ recurse_subtotals (s, c, cats, is_included, weight, a, i + 1);
cats[a][i] = save;
}
}
}
static void
-ctables_cell_insert (struct ctables_section *s,
- const struct ccase *c,
- double d_weight, double e_weight)
+ctables_cell_insert (struct ctables_section *s, const struct ccase *c,
+ double weight[N_CTWS])
{
- const struct ctables_category *cats[PIVOT_N_AXES][10]; /* XXX */
-
- /* Does at least one categorical variable have a missing value in an included
- or excluded category? */
- bool is_missing = false;
+ const struct ctables_category *layer_cats[s->nests[PIVOT_AXIS_LAYER]->n];
+ const struct ctables_category *row_cats[s->nests[PIVOT_AXIS_ROW]->n];
+ const struct ctables_category *column_cats[s->nests[PIVOT_AXIS_COLUMN]->n];
+ const struct ctables_category **cats[PIVOT_N_AXES] =
+ {
+ [PIVOT_AXIS_LAYER] = layer_cats,
+ [PIVOT_AXIS_ROW] = row_cats,
+ [PIVOT_AXIS_COLUMN] = column_cats,
+ };
- /* Does at least one categorical variable have a missing value in an excluded
- category? */
- bool excluded_missing = false;
+ bool is_included = true;
for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
{
const struct ctables_nest *nest = s->nests[a];
for (size_t i = 0; i < nest->n; i++)
- {
- if (i == nest->scale_idx)
- continue;
-
- const struct variable *var = nest->vars[i];
- const union value *value = case_data (c, var);
-
- bool var_missing = var_is_value_missing (var, value) != 0;
- if (var_missing)
- is_missing = true;
-
- cats[a][i] = ctables_categories_match (
- s->table->categories[var_get_dict_index (var)], value, var);
- if (!cats[a][i])
- {
- if (!var_missing)
- return;
+ if (i != nest->scale_idx)
+ {
+ const struct variable *var = nest->vars[i];
+ const union value *value = case_data (c, var);
- static const struct ctables_category cct_excluded_missing = {
- .type = CCT_EXCLUDED_MISSING,
- .hide = true,
- };
- cats[a][i] = &cct_excluded_missing;
- excluded_missing = true;
- }
+ cats[a][i] = ctables_categories_match (
+ s->table->categories[var_get_dict_index (var)], value, var);
+ if (!cats[a][i])
+ {
+ if (i != nest->summary_idx)
+ return;
+
+ if (!var_is_value_missing (var, value))
+ return;
+
+ static const struct ctables_category cct_excluded_missing = {
+ .type = CCT_EXCLUDED_MISSING,
+ .hide = true,
+ };
+ cats[a][i] = &cct_excluded_missing;
+ is_included = false;
+ }
}
}
- if (!excluded_missing)
+ if (is_included)
for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
{
const struct ctables_nest *nest = s->nests[a];
}
}
- ctables_cell_add__ (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight);
-
- //if (!excluded_missing)
- {
- recurse_totals (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight, 0, 0);
- recurse_subtotals (s, c, cats, is_missing, excluded_missing,
- d_weight, e_weight, 0, 0);
- }
+ ctables_cell_add__ (s, c, cats, is_included, weight);
+ recurse_totals (s, c, cats, is_included, weight, 0, 0);
+ recurse_subtotals (s, c, cats, is_included, weight, 0, 0);
}
struct merge_item
const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
if (as->function != bs->function)
return as->function > bs->function ? 1 : -1;
- else if (as->weighted != bs->weighted)
- return as->weighted > bs->weighted ? 1 : -1;
+ else if (as->weighting != bs->weighting)
+ return as->weighting > bs->weighting ? 1 : -1;
else if (as->calc_area != bs->calc_area)
return as->calc_area > bs->calc_area ? 1 : -1;
else if (as->percentile != bs->percentile)
for (size_t i = 0; i < nest->n; i++)
hmap_init (&s->occurrences[a][i]);
}
- for (size_t i = 0; i < N_CTATS; i++)
- hmap_init (&s->areas[i]);
+ for (enum ctables_area_type at = 0; at < N_CTATS; at++)
+ hmap_init (&s->areas[at]);
}
}
{
const struct ctables_summary_spec *ss2 = &pc->specs->specs[i];
if (ss->function == ss2->function
- && ss->weighted == ss2->weighted
+ && ss->weighting == ss2->weighting
&& ss->calc_area == ss2->calc_area
&& ss->percentile == ss2->percentile)
{
struct pivot_value *value;
if (ct->hide_threshold != 0
&& d < ct->hide_threshold
- && ctables_summary_function_is_count (ss->function))
+ && ss->function == CTSF_COUNT)
{
value = pivot_value_new_user_text_nocopy (
xasprintf ("<%d", ct->hide_threshold));
nest->areas[at] = xmalloc (nest->n * sizeof *nest->areas[at]);
nest->n_areas[at] = 0;
- bool add_vars = (at == CTAT_LAYER ? a == PIVOT_AXIS_LAYER
- : at == CTAT_LAYERROW ? a != PIVOT_AXIS_COLUMN
- : at == CTAT_LAYERCOL ? a != PIVOT_AXIS_ROW
- : at == CTAT_TABLE ? false
- : true);
- if (add_vars)
- for (size_t k = 0; k < nest->n; k++)
- {
- if (k == nest->scale_idx)
- continue;
- nest->areas[at][nest->n_areas[at]++] = k;
- }
- else if ((at == CTAT_LAYERCOL && a == PIVOT_AXIS_ROW && t->label_axis[a] != a)
- || (at == CTAT_LAYERROW && a == PIVOT_AXIS_COLUMN && t->label_axis[a] != a)
- || (at == CTAT_LAYER && t->label_axis[a] == PIVOT_AXIS_LAYER))
- {
- for (size_t k = nest->n - 1; k < nest->n; k--)
- {
- if (k == nest->scale_idx)
- continue;
- nest->areas[at][nest->n_areas[at]++] = k;
- break;
- }
- continue;
- }
-
enum pivot_axis_type ata, atb;
if (at == CTAT_ROW || at == CTAT_LAYERROW)
{
atb = PIVOT_AXIS_ROW;
}
- size_t n_drop = 0;
- bool drop_inner = false;
+ if (at == CTAT_LAYER
+ ? a != PIVOT_AXIS_LAYER && t->label_axis[a] == PIVOT_AXIS_LAYER
+ : at == CTAT_LAYERCOL || at == CTAT_LAYERROW
+ ? a == atb && t->label_axis[a] != a
+ : false)
+ {
+ for (size_t k = nest->n - 1; k < nest->n; k--)
+ if (k != nest->scale_idx)
+ {
+ nest->areas[at][nest->n_areas[at]++] = k;
+ break;
+ }
+ continue;
+ }
+
+ if (at == CTAT_LAYER ? a != PIVOT_AXIS_LAYER
+ : at == CTAT_LAYERROW || at == CTAT_LAYERCOL ? a == atb
+ : at == CTAT_TABLE ? true
+ : false)
+ continue;
+
+ for (size_t k = 0; k < nest->n; k++)
+ if (k != nest->scale_idx)
+ nest->areas[at][nest->n_areas[at]++] = k;
+
+ int n_drop;
switch (at)
{
case CTAT_SUBTABLE:
- if (t->clabels_from_axis == PIVOT_AXIS_LAYER)
- n_drop = a != PIVOT_AXIS_LAYER;
- else if (t->clabels_to_axis != PIVOT_AXIS_LAYER)
- n_drop = a == t->clabels_from_axis ? 2 : 0;
- else
- {
- n_drop = a != PIVOT_AXIS_LAYER && a != t->clabels_from_axis;
- drop_inner = a == t->clabels_from_axis;
- }
+#define L PIVOT_AXIS_LAYER
+ n_drop = (t->clabels_from_axis == L ? a != L
+ : t->clabels_to_axis == L ? (t->clabels_from_axis == a ? -1 : a != L)
+ : t->clabels_from_axis == a ? 2
+ : 0);
+#undef L
break;
case CTAT_LAYERROW:
case CTAT_ROW:
case CTAT_COL:
- if (a == atb)
- {
- if (t->label_axis[ata] == atb)
- ;
- else if (t->label_axis[atb] == ata)
- drop_inner = true;
- else if (t->label_axis[atb] == PIVOT_AXIS_LAYER)
- drop_inner = true;
- else
- n_drop = 1;
- }
- if (a == ata && t->label_axis[ata] == atb)
- n_drop++;
+ n_drop = (a == ata ? t->label_axis[ata] == atb
+ : a != atb ? 0
+ : t->clabels_from_axis == atb ? -1
+ : t->clabels_to_axis != atb ? 1
+ : 0);
break;
case CTAT_LAYER:
break;
}
- if (drop_inner)
+ if (n_drop < 0)
{
size_t n = nest->n_areas[at];
if (n > 1)
nest->n_areas[at]--;
}
}
- for (size_t i = 0; i < n_drop; i++)
- if (nest->n_areas[at] > 0)
- nest->n_areas[at]--;
+ else
+ {
+ for (int i = 0; i < n_drop; i++)
+ if (nest->n_areas[at] > 0)
+ nest->n_areas[at]--;
+ }
}
}
}
else
{
struct ctables_nest *nest = xmalloc (sizeof *nest);
- *nest = (struct ctables_nest) { .n = 0 };
+ *nest = (struct ctables_nest) {
+ .n = 0,
+ .scale_idx = SIZE_MAX,
+ .summary_idx = SIZE_MAX
+ };
t->stacks[a] = (struct ctables_stack) { .nests = nest, .n = 1 };
/* There's no point in moving labels away from an axis that has no
struct ctables_nest *nest = &stack->nests[i];
if (!nest->specs[CSV_CELL].n)
{
- struct ctables_summary_spec_set *specs = &nest->specs[CSV_CELL];
- specs->specs = xmalloc (sizeof *specs->specs);
- specs->n = 1;
+ struct ctables_summary_spec_set *ss = &nest->specs[CSV_CELL];
+ ss->specs = xmalloc (sizeof *ss->specs);
+ ss->n = 1;
enum ctables_summary_function function
- = specs->is_scale ? CTSF_MEAN : CTSF_COUNT;
+ = ss->is_scale ? CTSF_MEAN : CTSF_COUNT;
- *specs->specs = (struct ctables_summary_spec) {
+ if (!ss->var)
+ {
+ nest->summary_idx = nest->n - 1;
+ ss->var = nest->vars[nest->summary_idx];
+ }
+ *ss->specs = (struct ctables_summary_spec) {
.function = function,
- .weighted = true,
- .format = ctables_summary_default_format (function, specs->var),
+ .weighting = ss->is_scale ? CTW_EFFECTIVE : CTW_DICTIONARY,
+ .format = ctables_summary_default_format (function, ss->var),
};
- if (!specs->var)
- specs->var = nest->vars[0];
ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
&nest->specs[CSV_CELL]);
static void
ctables_section_recurse_add_empty_categories (
struct ctables_section *s,
- const struct ctables_category *cats[PIVOT_N_AXES][10], struct ccase *c,
+ const struct ctables_category **cats[PIVOT_N_AXES], struct ccase *c,
enum pivot_axis_type a, size_t a_idx)
{
if (a >= PIVOT_N_AXES)
if (!show_empty)
return;
- const struct ctables_category *cats[PIVOT_N_AXES][10]; /* XXX */
+ const struct ctables_category *layer_cats[s->nests[PIVOT_AXIS_LAYER]->n];
+ const struct ctables_category *row_cats[s->nests[PIVOT_AXIS_ROW]->n];
+ const struct ctables_category *column_cats[s->nests[PIVOT_AXIS_COLUMN]->n];
+ const struct ctables_category **cats[PIVOT_N_AXES] =
+ {
+ [PIVOT_AXIS_LAYER] = layer_cats,
+ [PIVOT_AXIS_ROW] = row_cats,
+ [PIVOT_AXIS_COLUMN] = column_cats,
+ };
struct ccase *c = case_create (dict_get_proto (s->table->ctables->dict));
ctables_section_recurse_add_empty_categories (s, cats, c, 0, 0);
case_unref (c);
}
hmap_destroy (&s->cells);
- for (size_t i = 0; i < N_CTATS; i++)
- hmap_destroy (&s->areas[i]);
+ for (enum ctables_area_type at = 0; at < N_CTATS; at++)
+ hmap_destroy (&s->areas[at]);
}
static void
for (struct ccase *c = casereader_read (group); c;
case_unref (c), c = casereader_read (group))
{
- double d_weight = dict_get_case_weight (dict, c, &warn_on_invalid);
+ double d_weight = dict_get_rounded_case_weight (dict, c, &warn_on_invalid);
double e_weight = (ct->e_weight
? var_force_valid_weight (ct->e_weight,
case_num (c, ct->e_weight),
&warn_on_invalid)
: d_weight);
+ double weight[] = {
+ [CTW_DICTIONARY] = d_weight,
+ [CTW_EFFECTIVE] = e_weight,
+ [CTW_UNWEIGHTED] = 1.0,
+ };
for (size_t i = 0; i < ct->n_tables; i++)
{
struct ctables_table *t = ct->tables[i];
for (size_t j = 0; j < t->n_sections; j++)
- ctables_cell_insert (&t->sections[j], c, d_weight, e_weight);
+ ctables_cell_insert (&t->sections[j], c, weight);
for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
if (t->label_axis[a] != a)
{
/* Parse function. */
enum ctables_summary_function function;
- bool weighted;
+ enum ctables_weighting weighting;
enum ctables_area_type area;
- if (!parse_ctables_summary_function (lexer, &function, &weighted, &area))
+ if (!parse_ctables_summary_function (lexer, &function, &weighting, &area))
goto error;
/* Parse percentile. */
sizeof *sss->specs);
sss->specs[sss->n++] = (struct ctables_summary_spec) {
.function = function,
- .weighted = weighted,
+ .weighting = weighting,
.calc_area = area,
.user_area = area,
.percentile = percentile,