/* PSPP - a program for statistical analysis.
- Copyright (C) 1997-9, 2000, 2009, 2010 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 "data/case.h"
#include "data/data-in.h"
+#include "data/dataset.h"
#include "data/dictionary.h"
#include "data/format.h"
-#include "data/procedure.h"
#include "data/transformations.h"
#include "data/variable.h"
#include "language/command.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"
{
struct pool *pool;
-
-
/* Variable types, for convenience. */
enum val_type src_type; /* src_vars[*] type. */
enum val_type dst_type; /* dst_vars[*] type. */
};
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;
{
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);
}
- 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
into TRNS->mappings and TRNS->map_cnt. 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 map_allocated;
bool have_dst_type;
trns->map_cnt = 0;
map_allocated = 0;
have_dst_type = false;
- if (!lex_force_match (lexer, '('))
+ if (!lex_force_match (lexer, T_LPAREN))
return false;
do
{
struct map_in in;
if (!parse_map_in (lexer, &in, trns->pool,
- trns->src_type, trns->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;
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_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;
}
}
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;
value_init_pool (pool, &in->x, width);
value_copy_buf_rpad (&in->x, width,
- CHAR_CAST_BUG (uint8_t *, ds_data (string)),
- ds_length (string), ' ');
+ CHAR_CAST (uint8_t *, s), strlen (s), ' ');
+ free (s);
}
/* Parses a mapping output value into OUT, allocating memory from
set_map_out_num (out, SYSMIS);
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)
{
out->copy_input = false;
value_init_pool (pool, &out->value, length);
- memcpy (value_str_rw (&out->value, length), string, length);
+ memcpy (out->value.s, string, length);
out->width = length;
}
/* 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 i;
-
+ const struct variable *narrow_var = NULL;
+ int min_dst_width = INT_MAX;
trns->max_dst_width = 0;
+
for (i = 0; i < trns->var_cnt; i++)
{
const struct variable *v = trns->dst_vars[i];
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++)
{
struct map_out *out = &trns->mappings[i].out;
if (!out->copy_input)
- value_resize_pool (trns->pool, &out->value,
- out->width, trns->max_dst_width);
+ {
+ 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;
- trns->dst_dict = dict;
-
for (i = 0; i < trns->var_cnt; i++)
{
const struct variable **var = &trns->dst_vars[i];
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++)
{
switch (in->type)
{
case MAP_SINGLE:
- match = !memcmp (value, value_str (&in->x, trns->max_src_width),
- 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, &uv, 0, encoding);
+ match = error == NULL;
+ free (error);
- msg_disable ();
- match = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
- LEGACY_NATIVE, FMT_F, 0, 0, trns->dst_dict,
- &uv, 0);
- msg_enable ();
out->value.f = uv.f;
break;
}
if (out != NULL)
{
if (!out->copy_input)
- memcpy (dst, value_str (&out->value, trns->max_dst_width),
- var_get_width (dst_var));
+ memcpy (dst, out->value.s, var_get_width (dst_var));
else if (trns->src_vars != trns->dst_vars)
{
union value *dst_data = case_data_rw (*c, dst_var);