1 /* PSPP - computes sample statistics.
2 Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014,
3 2020 Free Software Foundation, Inc.
5 This program is free software: you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation, either version 3 of the License, or
8 (at your option) any later version.
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU 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, see <http://www.gnu.org/licenses/>. */
21 /* The Gnulib "strftime" module defines my_strftime in <config.h> for use by
22 gl/strftime.c. Perl also defines my_strftime in embed.h for some other
23 purpose. The former definition doesn't matter in this file, so suppress it
24 to avoid a compiler warning. */
34 #include <libpspp/hmap.h>
35 #include <libpspp/hash-functions.h>
36 #include <libpspp/message.h>
37 #include <libpspp/version.h>
38 #include <libpspp/i18n.h>
39 #include <gl/xalloc.h>
40 #include <data/dictionary.h>
41 #include <data/case.h>
42 #include <data/casereader.h>
43 #include <data/casewriter.h>
44 #include <data/variable.h>
45 #include <data/attributes.h>
46 #include <data/file-handle-def.h>
47 #include <data/identifier.h>
48 #include <data/settings.h>
49 #include <data/sys-file-writer.h>
50 #include <data/value.h>
51 #include <data/vardict.h>
52 #include <data/value-labels.h>
53 #include <data/format.h>
54 #include <data/data-in.h>
55 #include <data/data-out.h>
58 typedef struct fmt_spec input_format ;
59 typedef struct fmt_spec output_format ;
62 /* A thin wrapper around sfm_writer */
67 /* A pointer to the writer. The writer is owned by the struct */
68 struct casewriter *writer;
70 /* A pointer to the dictionary. Owned externally */
71 const struct pspp_dict *dict;
73 /* The scalar containing the dictionary */
78 /* A thin wrapper around sfm_reader */
81 struct any_read_info opts;
83 /* A pointer to the reader. The reader is owned by the struct */
84 struct casereader *reader;
86 /* A pointer to the dictionary. */
87 struct pspp_dict *dict;
92 struct hmap_node hmap_node; /* In struct pspp_dict's input_formats map. */
93 const struct variable *var;
94 struct fmt_spec input_format;
97 /* A thin wrapper around struct dictionary.*/
99 struct dictionary *dict;
100 struct hmap input_formats; /* Contains struct input_format. */
104 /* A message handler which writes messages to PSPP::errstr */
106 message_handler (const struct msg *m, void *aux)
108 SV *errstr = get_sv("PSPP::errstr", TRUE);
109 sv_setpv (errstr, m->text);
113 sysfile_close (struct syswriter_info *swi)
119 retval = casewriter_destroy (swi->writer);
127 scalar_to_value (union value *val, SV *scalar, const struct variable *var)
129 if ( var_is_numeric (var))
131 if ( SvNOK (scalar) || SvIOK (scalar) )
132 val->f = SvNV (scalar);
139 const char *p = SvPV (scalar, len);
140 int width = var_get_width (var);
141 value_set_missing (val, width);
142 memcpy (val->s, p, len);
148 value_to_scalar (const union value *val, const struct variable *var)
150 if ( var_is_numeric (var))
152 if (var_is_value_missing (var, val) == MV_SYSTEM)
153 return newSVpvn ("", 0);
155 return newSVnv (val->f);
159 int width = var_get_width (var);
160 return newSVpvn ((char *) val->s, width);
166 make_value_from_scalar (union value *uv, SV *val, const struct variable *var)
168 value_init (uv, var_get_width (var));
169 scalar_to_value (uv, val, var);
172 static struct pspp_dict *
173 create_pspp_dict (struct dictionary *dict)
175 struct pspp_dict *pspp_dict = xmalloc (sizeof *pspp_dict);
176 pspp_dict->dict = dict;
177 hmap_init (&pspp_dict->input_formats);
181 static const struct fmt_spec *
182 find_input_format (const struct pspp_dict *dict, const struct variable *var)
184 struct input_format *input_format;
186 HMAP_FOR_EACH_IN_BUCKET (input_format, struct input_format, hmap_node,
187 hash_pointer (var, 0), &dict->input_formats)
188 if (input_format->var == var)
189 return &input_format->input_format;
197 MODULE = PSPP PACKAGE = PSPP
205 /* Check that the version is correct up to the length of 'ver'.
206 This allows PSPP autobuilders to add a "-build#" suffix to the
207 PSPP version without causing failures here. */
208 assert (0 == strncmp (ver, bare_version, strlen (ver)));
211 const struct msg_handler mh = { .output_msg = message_handler };
212 msg_set_handler (&mh);
217 format_value (val, var)
222 const struct fmt_spec *fmt = var_get_print_format (var);
225 make_value_from_scalar (&uv, val, var);
226 s = data_out (&uv, var_get_encoding (var), fmt, settings_get_fmt_settings ());
227 value_destroy (&uv, var_get_width (var));
228 ret = newSVpv (s, fmt->w);
236 value_is_missing (val, var)
242 make_value_from_scalar (&uv, val, var);
243 ret = var_is_value_missing (var, &uv) != 0;
244 value_destroy (&uv, var_get_width (var));
251 MODULE = PSPP PACKAGE = PSPP::Dict
256 RETVAL = create_pspp_dict (dict_create ("UTF-8"));
263 struct pspp_dict *dict
267 struct input_format *input_format, *next_input_format;
269 HMAP_FOR_EACH_SAFE (input_format, next_input_format,
270 struct input_format, hmap_node, &dict->input_formats)
272 hmap_delete (&dict->input_formats, &input_format->hmap_node);
275 hmap_destroy (&dict->input_formats);
276 dict_unref (dict->dict);
282 struct pspp_dict *dict
284 RETVAL = dict_get_n_vars (dict->dict);
289 set_label (dict, label)
290 struct pspp_dict *dict
293 dict_set_label (dict->dict, label);
296 set_documents (dict, docs)
297 struct pspp_dict *dict
300 dict_set_documents_string (dict->dict, docs);
304 add_document (dict, doc)
305 struct pspp_dict *dict
308 dict_add_document_line (dict->dict, doc, false);
312 clear_documents (dict)
313 struct pspp_dict *dict
315 dict_clear_documents (dict->dict);
319 set_weight (dict, var)
320 struct pspp_dict *dict
323 dict_set_weight (dict->dict, var);
327 pxs_get_variable (dict, idx)
328 struct pspp_dict *dict
331 SV *errstr = get_sv("PSPP::errstr", TRUE);
332 sv_setpv (errstr, "");
333 if ( SvIV (idx) >= dict_get_n_vars (dict->dict))
335 sv_setpv (errstr, "The dictionary doesn't have that many variables.");
339 RETVAL = dict_get_var (dict->dict, SvIV (idx));
345 pxs_get_var_by_name (dict, name)
346 struct pspp_dict *dict
349 SV *errstr = get_sv("PSPP::errstr", TRUE);
350 sv_setpv (errstr, "");
352 struct variable *var = dict_lookup_var (dict->dict, name);
354 sv_setpv (errstr, "No such variable.");
360 MODULE = PSPP PACKAGE = PSPP::Var
364 pxs_dict_create_var (dict, name, ip_fmt)
365 struct pspp_dict * dict
369 SV *errstr = get_sv("PSPP::errstr", TRUE);
370 sv_setpv (errstr, "");
371 if ( ! id_is_plausible (name, false))
373 sv_setpv (errstr, "The variable name is not valid.");
377 struct fmt_spec op_fmt;
378 struct input_format *input_format;
381 op_fmt = fmt_for_output_from_input (&ip_fmt, settings_get_fmt_settings ());
382 v = dict_create_var (dict->dict, name,
383 fmt_is_string (op_fmt.type) ? op_fmt.w : 0);
386 sv_setpv (errstr, "The variable could not be created (probably already exists).");
389 var_set_both_formats (v, &op_fmt);
391 input_format = xmalloc (sizeof *input_format);
392 input_format->var = v;
393 input_format->input_format = ip_fmt;
394 hmap_insert (&dict->input_formats, &input_format->hmap_node,
395 hash_pointer (v, 0));
403 set_missing_values (var, v1, ...)
404 struct variable *var;
411 croak ("No more than 3 missing values are permitted");
413 for (i = 0; i < items - 1; ++i)
414 make_value_from_scalar (&val[i], ST(i+1), var);
416 struct missing_values mv;
417 mv_init (&mv, var_get_width (var));
418 for (i = 0 ; i < items - 1; ++i )
420 mv_add_value (&mv, &val[i]);
421 value_destroy (&val[i], var_get_width (var));
423 var_set_missing_values (var, &mv);
427 set_label (var, label)
428 struct variable *var;
431 var_set_label (var, label);
435 clear_value_labels (var)
436 struct variable *var;
438 var_clear_value_labels (var);
441 get_write_format (var)
444 HV *fmthash = (HV *) sv_2mortal ((SV *) newHV());
445 const struct fmt_spec *fmt = var_get_write_format (var);
447 hv_store (fmthash, "fmt", 3, newSVnv (fmt->type), 0);
448 hv_store (fmthash, "decimals", 8, newSVnv (fmt->d), 0);
449 hv_store (fmthash, "width", 5, newSVnv (fmt->w), 0);
451 RETVAL = newRV ((SV *) fmthash);
456 get_print_format (var)
459 HV *fmthash = (HV *) sv_2mortal ((SV *) newHV());
460 const struct fmt_spec *fmt = var_get_print_format (var);
462 hv_store (fmthash, "fmt", 3, newSVnv (fmt->type), 0);
463 hv_store (fmthash, "decimals", 8, newSVnv (fmt->d), 0);
464 hv_store (fmthash, "width", 5, newSVnv (fmt->w), 0);
466 RETVAL = newRV ((SV *) fmthash);
472 pxs_set_write_format (var, fmt)
476 var_set_write_format (var, &fmt);
480 pxs_set_print_format (var, fmt)
484 var_set_print_format (var, &fmt);
487 pxs_set_output_format (var, fmt)
491 var_set_both_formats (var, &fmt);
495 add_value_label (var, key, label)
500 SV *errstr = get_sv("PSPP::errstr", TRUE);
501 sv_setpv (errstr, "");
503 union value the_value;
504 int width = var_get_width (var);
507 value_init (&the_value, width);
508 if ( var_is_numeric (var))
510 if ( ! looks_like_number (key))
512 sv_setpv (errstr, "Cannot add label with string key to a numeric variable");
513 value_destroy (&the_value, width);
516 the_value.f = SvNV (key);
520 value_copy_str_rpad (&the_value, width, SvPV_nolen(key), ' ');
522 ok = var_add_value_label (var, &the_value, label);
523 value_destroy (&the_value, width);
526 sv_setpv (errstr, "Something went wrong");
536 HV *attrhash = (HV *) sv_2mortal ((SV *) newHV());
538 struct attrset *as = var_get_attributes (var);
542 struct attrset_iterator iter;
543 struct attribute *attr;
545 for (attr = attrset_first (as, &iter);
547 attr = attrset_next (as, &iter))
550 const char *name = attribute_get_name (attr);
552 AV *values = newAV ();
554 for (i = 0 ; i < attribute_get_n_values (attr); ++i )
556 const char *value = attribute_get_value (attr, i);
557 av_push (values, newSVpv (value, 0));
560 hv_store (attrhash, name, strlen (name),
561 newRV_noinc ((SV*) values), 0);
565 RETVAL = newRV ((SV *) attrhash);
572 struct variable * var
574 RETVAL = var_get_name (var);
581 struct variable * var
583 RETVAL = var_get_label (var);
589 get_value_labels (var)
592 HV *labelhash = (HV *) sv_2mortal ((SV *) newHV());
593 const struct val_lab *vl;
594 const struct val_labs *labels = var_get_value_labels (var);
598 for (vl = val_labs_first (labels);
600 vl = val_labs_next (labels, vl))
602 SV *sv = value_to_scalar (&vl->value, var);
604 const char *s = SvPV (sv, len);
605 hv_store (labelhash, s, len, newSVpv (val_lab_get_label (vl), 0), 0);
609 RETVAL = newRV ((SV *) labelhash);
615 MODULE = PSPP PACKAGE = PSPP::Sysfile
618 struct syswriter_info *
619 pxs_create_sysfile (name, dict, opts_hr)
621 struct pspp_dict *dict;
625 struct sfm_write_options opts;
626 if (!SvROK (opts_hr))
628 opts = sfm_writer_default_options ();
632 HV *opt_h = (HV *) SvRV (opts_hr);
633 SV** readonly = hv_fetch(opt_h, "readonly", 8, 0);
634 SV** compress = hv_fetch(opt_h, "compress", 8, 0);
635 SV** version = hv_fetch(opt_h, "version", 7, 0);
637 opts.create_writeable = readonly ? ! SvIV (*readonly) : true;
638 opts.compression = (compress && SvIV (*compress)
641 opts.version = version ? SvIV (*version) : 3 ;
644 struct file_handle *fh =
645 fh_create_file (NULL, name, "UTF-8", fh_default_properties () );
646 struct syswriter_info *swi = xmalloc (sizeof (*swi));
647 swi->writer = sfm_open_writer (fh, dict->dict, opts);
650 swi->dict_sv = dict_sv;
651 SvREFCNT_inc (swi->dict_sv);
659 struct syswriter_info *swi
661 RETVAL = sysfile_close (swi);
667 struct syswriter_info *swi
670 SvREFCNT_dec (swi->dict_sv);
674 append_case (swi, ccase)
675 struct syswriter_info *swi
678 SV *errstr = get_sv("PSPP::errstr", TRUE);
679 sv_setpv (errstr, "");
680 if ( (!SvROK(ccase)))
686 AV *av_case = (AV*) SvRV (ccase);
688 const struct variable **vv;
692 if ( av_len (av_case) >= dict_get_n_vars (swi->dict->dict))
695 c = case_create (dict_get_proto (swi->dict->dict));
697 dict_get_vars (swi->dict->dict, &vv, &nv,
698 1u << DC_ORDINARY | 1u << DC_SYSTEM);
700 for (SV *sv = av_shift (av_case); SvOK (sv); sv = av_shift (av_case))
702 const struct variable *v = vv[i++];
703 const struct fmt_spec *ifmt = find_input_format (swi->dict, v);
705 /* If an input format has been set, then use it.
706 Otherwise just convert the raw value. */
709 struct substring ss = ss_cstr (SvPV_nolen (sv));
710 char *error = data_in (ss,
711 SvUTF8(sv) ? UTF8: "iso-8859-1",
713 settings_get_fmt_settings (),
716 dict_get_encoding (swi->dict->dict));
721 SvREFCNT_dec_NN (sv);
728 scalar_to_value (case_data_rw (c, v), sv, v);
730 SvREFCNT_dec_NN (sv);
733 /* The remaining variables must be sysmis or blank string */
734 while (i < dict_get_n_vars (swi->dict->dict))
736 const struct variable *v = vv[i++];
737 union value *val = case_data_rw (c, v);
738 value_set_missing (val, var_get_width (v));
740 casewriter_write (swi->writer, c);
750 MODULE = PSPP PACKAGE = PSPP::Reader
752 struct sysreader_info *
753 pxs_open_sysfile (name)
756 struct sysreader_info *sri = NULL;
757 struct file_handle *fh =
758 fh_create_file (NULL, name, "UTF-8", fh_default_properties () );
759 struct dictionary *dict;
761 sri = xmalloc (sizeof (*sri));
762 sri->reader = any_reader_open_and_decode (fh, NULL, &dict, &sri->opts);
764 sri->dict = create_pspp_dict (dict);
777 pxs_get_dict (reader)
778 struct sysreader_info *reader;
780 RETVAL = reader->dict;
786 struct sysreader_info *sfr;
789 casenumber n = casereader_get_n_cases (sfr->reader);
790 if (n == CASENUMBER_MAX)
802 struct sysreader_info *sfr;
806 if ((c = casereader_read (sfr->reader)) != NULL)
810 EXTEND (SP, dict_get_n_vars (sfr->dict->dict));
811 for (v = 0; v < dict_get_n_vars (sfr->dict->dict); ++v )
813 const struct variable *var = dict_get_var (sfr->dict->dict, v);
814 const union value *val = case_data (c, var);
816 PUSHs (sv_2mortal (value_to_scalar (val, var)));