/* PSPP - computes sample statistics.
- Copyright (C) 2007, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
+ Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 Free Software Foundation, Inc.
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License as
#include <data/dictionary.h>
#include <data/case.h>
#include <data/casereader.h>
+#include <data/casewriter.h>
#include <data/variable.h>
#include <data/attributes.h>
#include <data/file-handle-def.h>
+#include <data/identifier.h>
+#include <data/settings.h>
#include <data/sys-file-writer.h>
-#include <data/sys-file-reader.h>
#include <data/value.h>
#include <data/vardict.h>
#include <data/value-labels.h>
/* A thin wrapper around sfm_writer */
-struct sysfile_info
+struct syswriter_info
{
bool opened;
struct casewriter *writer;
/* A pointer to the dictionary. Owned externally */
- const struct dictionary *dict;
+ const struct pspp_dict *dict;
/* The scalar containing the dictionary */
SV *dict_sv;
/* A thin wrapper around sfm_reader */
struct sysreader_info
{
- struct sfm_read_info opts;
+ struct any_read_info opts;
/* A pointer to the reader. The reader is owned by the struct */
struct casereader *reader;
/* A pointer to the dictionary. */
- struct dictionary *dict;
+ struct pspp_dict *dict;
};
+struct input_format {
+ struct hmap_node hmap_node; /* In struct pspp_dict's input_formats map. */
+ const struct variable *var;
+ struct fmt_spec input_format;
+};
+
+/* A thin wrapper around struct dictionary.*/
+struct pspp_dict {
+ struct dictionary *dict;
+ struct hmap input_formats; /* Contains struct input_format. */
+};
+
/* A message handler which writes messages to PSPP::errstr */
static void
}
static int
-sysfile_close (struct sysfile_info *sfi)
+sysfile_close (struct syswriter_info *swi)
{
int retval ;
- if ( ! sfi->opened )
+ if ( ! swi->opened )
return 0;
- retval = casewriter_destroy (sfi->writer);
+ retval = casewriter_destroy (swi->writer);
if (retval > 0 )
- sfi->opened = false;
+ swi->opened = false;
return retval;
}
}
-static void
-var_set_input_format (struct variable *v, input_format ip_fmt)
-{
- struct fmt_spec *if_copy = malloc (sizeof (*if_copy));
- memcpy (if_copy, &ip_fmt, sizeof (ip_fmt));
- var_attach_aux (v, if_copy, var_dtor_free);
-}
-
static void
make_value_from_scalar (union value *uv, SV *val, const struct variable *var)
{
scalar_to_value (uv, val, var);
}
+static struct pspp_dict *
+create_pspp_dict (struct dictionary *dict)
+{
+ struct pspp_dict *pspp_dict = xmalloc (sizeof *pspp_dict);
+ pspp_dict->dict = dict;
+ hmap_init (&pspp_dict->input_formats);
+ return pspp_dict;
+}
+
+static const struct fmt_spec *
+find_input_format (const struct pspp_dict *dict, const struct variable *var)
+{
+ struct input_format *input_format;
+
+ HMAP_FOR_EACH_IN_BUCKET (input_format, struct input_format, hmap_node,
+ hash_pointer (var, 0), &dict->input_formats)
+ if (input_format->var == var)
+ return &input_format->input_format;
+
+ return NULL;
+}
+
MODULE = PSPP
i18n_init ();
msg_set_handler (message_handler, NULL);
- settings_init (0, 0);
+ settings_init ();
fh_init ();
SV *
MODULE = PSPP PACKAGE = PSPP::Dict
-struct dictionary *
+struct pspp_dict *
pxs_dict_new()
CODE:
- RETVAL = dict_create ("UTF-8");
+ RETVAL = create_pspp_dict (dict_create ("UTF-8"));
OUTPUT:
RETVAL
void
DESTROY (dict)
- struct dictionary *dict
+ struct pspp_dict *dict
CODE:
- dict_destroy (dict);
+ if (dict != NULL)
+ {
+ struct input_format *input_format, *next_input_format;
+ HMAP_FOR_EACH_SAFE (input_format, next_input_format,
+ struct input_format, hmap_node, &dict->input_formats)
+ {
+ hmap_delete (&dict->input_formats, &input_format->hmap_node);
+ free (input_format);
+ }
+ hmap_destroy (&dict->input_formats);
+ dict_destroy (dict->dict);
+ free (dict);
+ }
int
get_var_cnt (dict)
- struct dictionary *dict
+ struct pspp_dict *dict
CODE:
- RETVAL = dict_get_var_cnt (dict);
+ RETVAL = dict_get_var_cnt (dict->dict);
OUTPUT:
RETVAL
void
set_label (dict, label)
- struct dictionary *dict
+ struct pspp_dict *dict
char *label
CODE:
- dict_set_label (dict, label);
+ dict_set_label (dict->dict, label);
void
set_documents (dict, docs)
- struct dictionary *dict
+ struct pspp_dict *dict
char *docs
CODE:
- dict_set_documents_string (dict, docs);
+ dict_set_documents_string (dict->dict, docs);
void
add_document (dict, doc)
- struct dictionary *dict
+ struct pspp_dict *dict
char *doc
CODE:
- dict_add_document_line (dict, doc, false);
+ dict_add_document_line (dict->dict, doc, false);
void
clear_documents (dict)
- struct dictionary *dict
+ struct pspp_dict *dict
CODE:
- dict_clear_documents (dict);
+ dict_clear_documents (dict->dict);
void
set_weight (dict, var)
- struct dictionary *dict
+ struct pspp_dict *dict
struct variable *var
CODE:
- dict_set_weight (dict, var);
+ dict_set_weight (dict->dict, var);
struct variable *
pxs_get_variable (dict, idx)
- struct dictionary *dict
+ struct pspp_dict *dict
SV *idx
INIT:
SV *errstr = get_sv("PSPP::errstr", TRUE);
sv_setpv (errstr, "");
- if ( SvIV (idx) >= dict_get_var_cnt (dict))
+ if ( SvIV (idx) >= dict_get_var_cnt (dict->dict))
{
sv_setpv (errstr, "The dictionary doesn't have that many variables.");
XSRETURN_UNDEF;
}
CODE:
- RETVAL = dict_get_var (dict, SvIV (idx));
+ RETVAL = dict_get_var (dict->dict, SvIV (idx));
OUTPUT:
RETVAL
struct variable *
pxs_get_var_by_name (dict, name)
- struct dictionary *dict
+ struct pspp_dict *dict
const char *name
INIT:
SV *errstr = get_sv("PSPP::errstr", TRUE);
sv_setpv (errstr, "");
CODE:
- struct variable *var = dict_lookup_var (dict, name);
+ struct variable *var = dict_lookup_var (dict->dict, name);
if ( ! var )
sv_setpv (errstr, "No such variable.");
RETVAL = var;
struct variable *
pxs_dict_create_var (dict, name, ip_fmt)
- struct dictionary * dict
+ struct pspp_dict * dict
char *name
input_format ip_fmt
INIT:
}
CODE:
struct fmt_spec op_fmt;
+ struct input_format *input_format;
struct variable *v;
op_fmt = fmt_for_output_from_input (&ip_fmt);
- v = dict_create_var (dict, name,
+ v = dict_create_var (dict->dict, name,
fmt_is_string (op_fmt.type) ? op_fmt.w : 0);
if ( NULL == v )
{
XSRETURN_UNDEF;
}
var_set_both_formats (v, &op_fmt);
- var_set_input_format (v, ip_fmt);
+
+ input_format = xmalloc (sizeof *input_format);
+ input_format->var = v;
+ input_format->input_format = ip_fmt;
+ hmap_insert (&dict->input_formats, &input_format->hmap_node,
+ hash_pointer (v, 0));
+
RETVAL = v;
OUTPUT:
RETVAL
struct variable *var;
char *label
CODE:
- var_set_label (var, label, false);
+ var_set_label (var, label);
void
MODULE = PSPP PACKAGE = PSPP::Sysfile
-struct sysfile_info *
-pxs_create_sysfile (name, dict_ref, opts_hr)
+struct syswriter_info *
+pxs_create_sysfile (name, dict, opts_hr)
char *name
- SV *dict_ref
+ struct pspp_dict *dict;
SV *opts_hr
INIT:
- SV *dict_sv = SvRV (dict_ref);
- struct dictionary *dict = (void *) SvIV (dict_sv);
+ SV *dict_sv = ST(1);
struct sfm_write_options opts;
if (!SvROK (opts_hr))
{
SV** version = hv_fetch(opt_h, "version", 7, 0);
opts.create_writeable = readonly ? ! SvIV (*readonly) : true;
- opts.compress = compress ? SvIV (*compress) : false;
+ opts.compression = (compress && SvIV (*compress)
+ ? ANY_COMP_SIMPLE
+ : ANY_COMP_NONE);
opts.version = version ? SvIV (*version) : 3 ;
}
CODE:
struct file_handle *fh =
fh_create_file (NULL, name, fh_default_properties () );
- struct sysfile_info *sfi = xmalloc (sizeof (*sfi));
- sfi->writer = sfm_open_writer (fh, dict, opts);
- sfi->dict = dict;
- sfi->opened = true;
- sfi->dict_sv = dict_sv;
- SvREFCNT_inc (sfi->dict_sv);
+ struct syswriter_info *swi = xmalloc (sizeof (*swi));
+ swi->writer = sfm_open_writer (fh, dict->dict, opts);
+ swi->dict = dict;
+ swi->opened = true;
+ swi->dict_sv = dict_sv;
+ SvREFCNT_inc (swi->dict_sv);
- RETVAL = sfi;
+ RETVAL = swi;
OUTPUT:
RETVAL
int
-close (sfi)
- struct sysfile_info *sfi
+close (swi)
+ struct syswriter_info *swi
CODE:
- RETVAL = sysfile_close (sfi);
+ RETVAL = sysfile_close (swi);
OUTPUT:
RETVAL
void
-DESTROY (sfi)
- struct sysfile_info *sfi
+DESTROY (swi)
+ struct syswriter_info *swi
CODE:
- sysfile_close (sfi);
- SvREFCNT_dec (sfi->dict_sv);
- free (sfi);
+ sysfile_close (swi);
+ SvREFCNT_dec (swi->dict_sv);
+ free (swi);
int
-append_case (sfi, ccase)
- struct sysfile_info *sfi
+append_case (swi, ccase)
+ struct syswriter_info *swi
SV *ccase
INIT:
SV *errstr = get_sv("PSPP::errstr", TRUE);
struct ccase *c;
SV *sv;
- if ( av_len (av_case) >= dict_get_var_cnt (sfi->dict))
+ if ( av_len (av_case) >= dict_get_var_cnt (swi->dict->dict))
XSRETURN_UNDEF;
- c = case_create (dict_get_proto (sfi->dict));
+ c = case_create (dict_get_proto (swi->dict->dict));
- dict_get_vars (sfi->dict, &vv, &nv, 1u << DC_ORDINARY | 1u << DC_SYSTEM);
+ dict_get_vars (swi->dict->dict, &vv, &nv,
+ 1u << DC_ORDINARY | 1u << DC_SYSTEM);
for (sv = av_shift (av_case); SvOK (sv); sv = av_shift (av_case))
{
const struct variable *v = vv[i++];
- const struct fmt_spec *ifmt = var_get_aux (v);
+ const struct fmt_spec *ifmt = find_input_format (swi->dict, v);
/* If an input format has been set, then use it.
Otherwise just convert the raw value.
error = data_in (ss, SvUTF8(sv) ? UTF8: "iso-8859-1", ifmt->type,
case_data_rw (c, v), var_get_width (v),
- dict_get_encoding (sfi->dict));
+ dict_get_encoding (swi->dict->dict));
ok = error == NULL;
free (error);
}
/* The remaining variables must be sysmis or blank string */
- while (i < dict_get_var_cnt (sfi->dict))
+ while (i < dict_get_var_cnt (swi->dict->dict))
{
const struct variable *v = vv[i++];
union value *val = case_data_rw (c, v);
value_set_missing (val, var_get_width (v));
}
- casewriter_write (sfi->writer, c);
+ casewriter_write (swi->writer, c);
RETVAL = 1;
finish:
free (vv);
struct sysreader_info *sri = NULL;
struct file_handle *fh =
fh_create_file (NULL, name, fh_default_properties () );
+ struct dictionary *dict;
sri = xmalloc (sizeof (*sri));
- sri->reader = sfm_open_reader (fh, &sri->dict, &sri->opts);
-
- if ( NULL == sri->reader)
- {
- free (sri);
- sri = NULL;
- }
+ sri->reader = any_reader_open_and_decode (fh, NULL, &dict, &sri->opts);
+ if (sri->reader)
+ sri->dict = create_pspp_dict (dict);
+ else
+ {
+ free (sri);
+ sri = NULL;
+ }
RETVAL = sri;
OUTPUT:
RETVAL
-struct dictionary *
+struct pspp_dict *
pxs_get_dict (reader)
struct sysreader_info *reader;
CODE:
OUTPUT:
RETVAL
+SV *
+get_case_cnt (sfr)
+ struct sysreader_info *sfr;
+CODE:
+ SV *ret;
+ casenumber n = casereader_get_case_cnt (sfr->reader);
+ if (n == CASENUMBER_MAX)
+ ret = &PL_sv_undef;
+ else
+ ret = newSViv (n);
+ RETVAL = ret;
+ OUTPUT:
+RETVAL
+
+
void
get_next_case (sfr)
PPCODE:
struct ccase *c;
- if (c = casereader_read (sfr->reader))
+ if ((c = casereader_read (sfr->reader)) != NULL)
{
int v;
- EXTEND (SP, dict_get_var_cnt (sfr->dict));
- for (v = 0; v < dict_get_var_cnt (sfr->dict); ++v )
+ EXTEND (SP, dict_get_var_cnt (sfr->dict->dict));
+ for (v = 0; v < dict_get_var_cnt (sfr->dict->dict); ++v )
{
- const struct variable *var = dict_get_var (sfr->dict, v);
+ const struct variable *var = dict_get_var (sfr->dict->dict, v);
const union value *val = case_data (c, var);
PUSHs (sv_2mortal (value_to_scalar (val, var)));
projects / pspp / blobdiff