/* PSPP - a program for statistical analysis.
- Copyright (C) 1997-9, 2000, 2009 Free Software Foundation, Inc.
+ Copyright (C) 1997-9, 2000, 2009, 2010, 2011 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 published by
#include <math.h>
#include <stdlib.h>
-#include <data/case.h>
-#include <data/data-in.h>
-#include <data/format.h>
-#include <data/dictionary.h>
-#include <data/procedure.h>
-#include <data/transformations.h>
-#include <data/variable.h>
-#include <language/command.h>
-#include <language/lexer/lexer.h>
-#include <language/lexer/variable-parser.h>
-#include <language/lexer/range-parser.h>
-#include <libpspp/assertion.h>
-#include <libpspp/compiler.h>
-#include <libpspp/message.h>
-#include <libpspp/pool.h>
-#include <libpspp/str.h>
-
-#include "xalloc.h"
+#include "data/case.h"
+#include "data/data-in.h"
+#include "data/dataset.h"
+#include "data/dictionary.h"
+#include "data/format.h"
+#include "data/transformations.h"
+#include "data/variable.h"
+#include "language/command.h"
+#include "language/lexer/lexer.h"
+#include "language/lexer/value-parser.h"
+#include "language/lexer/variable-parser.h"
+#include "libpspp/assertion.h"
+#include "libpspp/cast.h"
+#include "libpspp/compiler.h"
+#include "libpspp/i18n.h"
+#include "libpspp/message.h"
+#include "libpspp/pool.h"
+#include "libpspp/str.h"
+
+#include "gl/xalloc.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
MAP_CONVERT /* "123" => 123. */
};
-/* A value involved in a RECODE mapping. */
-union recode_value
- {
- double f; /* Numeric. */
- char *c; /* Short or long string. */
- };
-
/* Describes input values to be mapped. */
struct map_in
{
enum map_in_type type; /* One of MAP_*. */
- union recode_value x, y; /* Source values. */
+ union value x, y; /* Source values. */
};
/* Describes the value used as output from a mapping. */
struct map_out
{
bool copy_input; /* If true, copy input to output. */
- union recode_value value; /* If copy_input false, recoded value. */
+ union value value; /* If copy_input false, recoded value. */
int width; /* If copy_input false, output value width. */
};
/* Variables. */
const struct variable **src_vars; /* Source variables. */
const struct variable **dst_vars; /* Destination variables. */
+ const struct dictionary *dst_dict; /* Dictionary of dst_vars */
char **dst_names; /* Name of dest variables, if they're new. */
- size_t var_cnt; /* Number of variables. */
+ size_t n_vars; /* Number of variables. */
/* Mappings. */
struct mapping *mappings; /* Value mappings. */
- size_t map_cnt; /* Number of mappings. */
+ size_t n_maps; /* Number of mappings. */
+ int max_src_width; /* Maximum width of src_vars[*]. */
+ int max_dst_width; /* Maximum width of any map_out in mappings. */
};
static bool parse_src_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
-static bool parse_mappings (struct lexer *, struct recode_trns *);
+static bool parse_mappings (struct lexer *, struct recode_trns *,
+ const char *dict_encoding);
static bool parse_dst_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
static void add_mapping (struct recode_trns *,
size_t *map_allocated, const struct map_in *);
static bool parse_map_in (struct lexer *lexer, struct map_in *, struct pool *,
- enum val_type src_type, size_t max_src_width);
+ enum val_type src_type, size_t max_src_width,
+ const char *dict_encoding);
static void set_map_in_generic (struct map_in *, enum map_in_type);
static void set_map_in_num (struct map_in *, enum map_in_type, double, double);
static void set_map_in_str (struct map_in *, struct pool *,
- const struct string *, size_t width);
+ struct substring, size_t width,
+ const char *dict_encoding);
static bool parse_map_out (struct lexer *lexer, struct pool *, struct map_out *);
static void set_map_out_num (struct map_out *, double);
static void set_map_out_str (struct map_out *, struct pool *,
- const struct string *);
+ struct substring);
-static void enlarge_dst_widths (struct recode_trns *);
+static bool enlarge_dst_widths (struct recode_trns *);
static void create_dst_vars (struct recode_trns *, struct dictionary *);
-static trns_proc_func recode_trns_proc;
-static trns_free_func recode_trns_free;
+static bool recode_trns_free (void *trns_);
+
+static const struct trns_class recode_trns_class;
\f
/* Parser. */
{
do
{
+ struct dictionary *dict = dataset_dict (ds);
struct recode_trns *trns
= pool_create_container (struct recode_trns, pool);
/* Parse source variable names,
then input to output mappings,
then destintation variable names. */
- if (!parse_src_vars (lexer, trns, dataset_dict (ds) )
- || !parse_mappings (lexer, trns)
- || !parse_dst_vars (lexer, trns, dataset_dict (ds)))
+ if (!parse_src_vars (lexer, trns, dict)
+ || !parse_mappings (lexer, trns, dict_get_encoding (dict))
+ || !parse_dst_vars (lexer, trns, dict))
{
recode_trns_free (trns);
return CMD_FAILURE;
/* Ensure that all the output strings are at least as wide
as the widest destination variable. */
if (trns->dst_type == VAL_STRING)
- enlarge_dst_widths (trns);
+ {
+ if (! enlarge_dst_widths (trns))
+ {
+ recode_trns_free (trns);
+ return CMD_FAILURE;
+ }
+ }
/* Create destination variables, if needed.
This must be the final step; otherwise we'd have to
delete destination variables on failure. */
+ trns->dst_dict = dict;
if (trns->src_vars != trns->dst_vars)
- create_dst_vars (trns, dataset_dict (ds));
+ create_dst_vars (trns, dict);
/* Done. */
- add_transformation (ds,
- recode_trns_proc, recode_trns_free, trns);
+ add_transformation (ds, &recode_trns_class, trns);
}
- while (lex_match (lexer, '/'));
+ while (lex_match (lexer, T_SLASH));
- return lex_end_of_command (lexer);
+ return CMD_SUCCESS;
}
/* Parses a set of variables to recode into TRNS->src_vars and
- TRNS->var_cnt. Sets TRNS->src_type. Returns true if
+ TRNS->n_vars. Sets TRNS->src_type. Returns true if
successful, false on parse error. */
static bool
parse_src_vars (struct lexer *lexer,
struct recode_trns *trns, const struct dictionary *dict)
{
- if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->var_cnt,
+ if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->n_vars,
PV_SAME_TYPE))
return false;
pool_register (trns->pool, free, trns->src_vars);
}
/* Parses a set of mappings, which take the form (input=output),
- into TRNS->mappings and TRNS->map_cnt. Sets TRNS->dst_type.
+ into TRNS->mappings and TRNS->n_maps. Sets TRNS->dst_type.
Returns true if successful, false on parse error. */
static bool
-parse_mappings (struct lexer *lexer, struct recode_trns *trns)
+parse_mappings (struct lexer *lexer, struct recode_trns *trns,
+ const char *dict_encoding)
{
- size_t max_src_width;
size_t map_allocated;
bool have_dst_type;
size_t i;
/* Find length of longest source variable. */
- max_src_width = var_get_width (trns->src_vars[0]);
- for (i = 1; i < trns->var_cnt; i++)
+ trns->max_src_width = var_get_width (trns->src_vars[0]);
+ for (i = 1; i < trns->n_vars; i++)
{
size_t var_width = var_get_width (trns->src_vars[i]);
- if (var_width > max_src_width)
- max_src_width = var_width;
+ if (var_width > trns->max_src_width)
+ trns->max_src_width = var_width;
}
/* Parse the mappings in parentheses. */
trns->mappings = NULL;
- trns->map_cnt = 0;
+ trns->n_maps = 0;
map_allocated = 0;
have_dst_type = false;
- if (!lex_force_match (lexer, '('))
+ if (!lex_force_match (lexer, T_LPAREN))
return false;
do
{
size_t first_map_idx;
size_t i;
- first_map_idx = trns->map_cnt;
+ first_map_idx = trns->n_maps;
/* Parse source specifications. */
do
struct map_in in;
if (!parse_map_in (lexer, &in, trns->pool,
- trns->src_type, max_src_width))
+ trns->src_type, trns->max_src_width,
+ dict_encoding))
return false;
add_mapping (trns, &map_allocated, &in);
- lex_match (lexer, ',');
+ lex_match (lexer, T_COMMA);
}
- while (!lex_match (lexer, '='));
+ while (!lex_match (lexer, T_EQUALS));
if (!parse_map_out (lexer, trns->pool, &out))
return false;
return false;
}
- for (i = first_map_idx; i < trns->map_cnt; i++)
+ for (i = first_map_idx; i < trns->n_maps; i++)
trns->mappings[i].out = out;
}
else
struct map_in in;
set_map_in_generic (&in, MAP_CONVERT);
add_mapping (trns, &map_allocated, &in);
- set_map_out_num (&trns->mappings[trns->map_cnt - 1].out, 0.0);
+ set_map_out_num (&trns->mappings[trns->n_maps - 1].out, 0.0);
dst_type = VAL_NUMERIC;
if (trns->src_type != VAL_STRING
trns->dst_type = dst_type;
have_dst_type = true;
- if (!lex_force_match (lexer, ')'))
+ if (!lex_force_match (lexer, T_RPAREN))
return false;
}
- while (lex_match (lexer, '('));
+ while (lex_match (lexer, T_LPAREN));
return true;
}
false on parse error. */
static bool
parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
- enum val_type src_type, size_t max_src_width)
+ enum val_type src_type, size_t max_src_width,
+ const char *dict_encoding)
{
if (lex_match_id (lexer, "ELSE"))
- {
set_map_in_generic (in, MAP_ELSE);
- }
else if (src_type == VAL_NUMERIC)
{
if (lex_match_id (lexer, "MISSING"))
set_map_in_generic (in, MAP_MISSING);
else if (!lex_force_string (lexer))
return false;
- else
+ else
{
- set_map_in_str (in, pool, lex_tokstr (lexer), max_src_width);
+ set_map_in_str (in, pool, lex_tokss (lexer), max_src_width,
+ dict_encoding);
lex_get (lexer);
if (lex_token (lexer) == T_ID
- && lex_id_match (ss_cstr ("THRU"), ss_cstr (lex_tokid (lexer))))
+ && lex_id_match (ss_cstr ("THRU"), lex_tokss (lexer)))
{
- msg (SE, _("THRU is not allowed with string variables."));
+ msg (SE, _("%s is not allowed with string variables."), "THRU");
return false;
}
}
size_t *map_allocated, const struct map_in *in)
{
struct mapping *m;
- if (trns->map_cnt >= *map_allocated)
+ if (trns->n_maps >= *map_allocated)
trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
map_allocated,
sizeof *trns->mappings);
- m = &trns->mappings[trns->map_cnt++];
+ m = &trns->mappings[trns->n_maps++];
m->in = *in;
}
right to WIDTH characters long. */
static void
set_map_in_str (struct map_in *in, struct pool *pool,
- const struct string *string, size_t width)
+ struct substring string, size_t width,
+ const char *dict_encoding)
{
+ char *s = recode_string (dict_encoding, "UTF-8",
+ ss_data (string), ss_length (string));
in->type = MAP_SINGLE;
- in->x.c = pool_alloc_unaligned (pool, width);
- buf_copy_rpad (in->x.c, width, ds_data (string), ds_length (string));
+ value_init_pool (pool, &in->x, width);
+ value_copy_buf_rpad (&in->x, width,
+ CHAR_CAST (uint8_t *, s), strlen (s), ' ');
+ free (s);
}
/* Parses a mapping output value into OUT, allocating memory from
}
else if (lex_match_id (lexer, "SYSMIS"))
set_map_out_num (out, SYSMIS);
- else if (lex_token (lexer) == T_STRING)
+ else if (lex_is_string (lexer))
{
- set_map_out_str (out, pool, lex_tokstr (lexer));
+ set_map_out_str (out, pool, lex_tokss (lexer));
lex_get (lexer);
}
- else if (lex_match_id (lexer, "COPY"))
+ else if (lex_match_id (lexer, "COPY"))
{
out->copy_input = true;
- out->width = 0;
+ out->width = 0;
}
else
{
/* Sets OUT as a string mapping output with the given VALUE. */
static void
set_map_out_str (struct map_out *out, struct pool *pool,
- const struct string *value)
+ const struct substring value)
{
- const char *string = ds_data (value);
- size_t length = ds_length (value);
+ const char *string = ss_data (value);
+ size_t length = ss_length (value);
+
+ if (length == 0)
+ {
+ /* A length of 0 will yield a numeric value, which is not
+ what we want. */
+ string = " ";
+ length = 1;
+ }
out->copy_input = false;
- out->value.c = pool_alloc_unaligned (pool, length);
- memcpy (out->value.c, string, length);
+ value_init_pool (pool, &out->value, length);
+ memcpy (out->value.s, string, length);
out->width = length;
}
if (lex_match_id (lexer, "INTO"))
{
- size_t name_cnt;
+ size_t n_names;
size_t i;
if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
- &trns->dst_names, &name_cnt,
+ &trns->dst_names, &n_names,
PV_NONE))
return false;
- if (name_cnt != trns->var_cnt)
+ if (n_names != trns->n_vars)
{
msg (SE, _("%zu variable(s) cannot be recoded into "
"%zu variable(s). Specify the same number "
"of variables as source and target variables."),
- trns->var_cnt, name_cnt);
+ trns->n_vars, n_names);
return false;
}
trns->dst_vars = pool_nalloc (trns->pool,
- trns->var_cnt, sizeof *trns->dst_vars);
- for (i = 0; i < trns->var_cnt; i++)
+ trns->n_vars, sizeof *trns->dst_vars);
+ for (i = 0; i < trns->n_vars; i++)
{
const struct variable *v;
v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
}
}
- for (i = 0; i < trns->var_cnt; i++)
+ for (i = 0; i < trns->n_vars; i++)
{
const struct variable *v = trns->dst_vars[i];
if (v != NULL && var_get_type (v) != trns->dst_type)
/* Ensures that all the output values in TRNS are as wide as the
widest destination variable. */
-static void
+static bool
enlarge_dst_widths (struct recode_trns *trns)
{
- size_t max_dst_width;
size_t i;
+ const struct variable *narrow_var = NULL;
+ int min_dst_width = INT_MAX;
+ trns->max_dst_width = 0;
- max_dst_width = 0;
- for (i = 0; i < trns->var_cnt; i++)
+ for (i = 0; i < trns->n_vars; i++)
{
const struct variable *v = trns->dst_vars[i];
- if (var_get_width (v) > max_dst_width)
- max_dst_width = var_get_width (v);
+ if (var_get_width (v) > trns->max_dst_width)
+ trns->max_dst_width = var_get_width (v);
+
+ if (var_get_width (v) < min_dst_width)
+ {
+ min_dst_width = var_get_width (v);
+ narrow_var = v;
+ }
}
- for (i = 0; i < trns->map_cnt; i++)
+ for (i = 0; i < trns->n_maps; i++)
{
struct map_out *out = &trns->mappings[i].out;
- if (!out->copy_input && out->width < max_dst_width)
- {
- char *s = pool_alloc_unaligned (trns->pool, max_dst_width + 1);
- buf_copy_rpad (s, max_dst_width + 1, out->value.c, out->width);
- out->value.c = s;
- }
+ if (!out->copy_input)
+ {
+ if (out->width > min_dst_width)
+ {
+ msg (ME,
+ _("Cannot recode because the variable %s would require a width of %d bytes or greater, but it has a width of only %d bytes."),
+ var_get_name (narrow_var), out->width, min_dst_width);
+ return false;
+ }
+
+ value_resize_pool (trns->pool, &out->value,
+ out->width, trns->max_dst_width);
+ }
}
+
+ return true;
}
/* Creates destination variables that don't already exist. */
{
size_t i;
- for (i = 0; i < trns->var_cnt; i++)
+ for (i = 0; i < trns->n_vars; i++)
{
const struct variable **var = &trns->dst_vars[i];
const char *name = trns->dst_names[i];
{
struct mapping *m;
- for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
+ for (m = trns->mappings; m < trns->mappings + trns->n_maps; m++)
{
const struct map_in *in = &m->in;
const struct map_out *out = &m->out;
/* Returns the output mapping in TRNS for an input of VALUE with
the given WIDTH, or a null pointer if there is no mapping. */
static const struct map_out *
-find_src_string (struct recode_trns *trns, const char *value, const struct variable *src_var)
+find_src_string (struct recode_trns *trns, const uint8_t *value,
+ const struct variable *src_var)
{
- struct mapping *m;
+ const char *encoding = dict_get_encoding (trns->dst_dict);
int width = var_get_width (src_var);
+ struct mapping *m;
- for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
+ for (m = trns->mappings; m < trns->mappings + trns->n_maps; m++)
{
const struct map_in *in = &m->in;
struct map_out *out = &m->out;
switch (in->type)
{
case MAP_SINGLE:
- match = !memcmp (value, in->x.c, width);
+ match = !memcmp (value, in->x.s, width);
break;
case MAP_ELSE:
match = true;
case MAP_CONVERT:
{
union value uv;
+ char *error;
+
+ error = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
+ C_ENCODING, FMT_F, settings_get_fmt_settings (),
+ &uv, 0, encoding);
+ match = error == NULL;
+ free (error);
- msg_disable ();
- match = data_in (ss_buffer (value, width), LEGACY_NATIVE,
- FMT_F, 0, 0, 0, &uv, 0);
- msg_enable ();
out->value.f = uv.f;
break;
}
}
/* Performs RECODE transformation. */
-static int
+static enum trns_result
recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
{
struct recode_trns *trns = trns_;
size_t i;
*c = case_unshare (*c);
- for (i = 0; i < trns->var_cnt; i++)
+ for (i = 0; i < trns->n_vars; i++)
{
const struct variable *src_var = trns->src_vars[i];
const struct variable *dst_var = trns->dst_vars[i];
-
- const union value *src_data = case_data (*c, src_var);
- union value *dst_data = case_data_rw (*c, dst_var);
-
const struct map_out *out;
if (trns->src_type == VAL_NUMERIC)
- out = find_src_numeric (trns, src_data->f, src_var);
+ out = find_src_numeric (trns, case_num (*c, src_var), src_var);
else
- out = find_src_string (trns, src_data->s, src_var);
+ out = find_src_string (trns, case_str (*c, src_var), src_var);
if (trns->dst_type == VAL_NUMERIC)
{
+ double *dst = case_num_rw (*c, dst_var);
if (out != NULL)
- dst_data->f = !out->copy_input ? out->value.f : src_data->f;
+ *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
else if (trns->src_vars != trns->dst_vars)
- dst_data->f = SYSMIS;
+ *dst = SYSMIS;
}
else
{
+ char *dst = CHAR_CAST_BUG (char *, case_str_rw (*c, dst_var));
if (out != NULL)
{
if (!out->copy_input)
- memcpy (dst_data->s, out->value.c, var_get_width (dst_var));
+ memcpy (dst, out->value.s, var_get_width (dst_var));
else if (trns->src_vars != trns->dst_vars)
- buf_copy_rpad (dst_data->s, var_get_width (dst_var),
- src_data->s, var_get_width (src_var));
+ {
+ union value *dst_data = case_data_rw (*c, dst_var);
+ const union value *src_data = case_data (*c, src_var);
+ value_copy_rpad (dst_data, var_get_width (dst_var),
+ src_data, var_get_width (src_var), ' ');
+ }
}
else if (trns->src_vars != trns->dst_vars)
- memset (dst_data->s, ' ', var_get_width (dst_var));
+ memset (dst, ' ', var_get_width (dst_var));
}
}
pool_destroy (trns->pool);
return true;
}
+
+static const struct trns_class recode_trns_class = {
+ .name = "RECODE",
+ .execute = recode_trns_proc,
+ .destroy = recode_trns_free,
+};
projects / pspp / blobdiff