/* PSPP - a program for statistical analysis.
- Copyright (C) 1997-2000, 2006-2007, 2009-2012 Free Software Foundation, Inc.
+ Copyright (C) 1997-2000, 2006-2007, 2009-2013 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 "libpspp/str.h"
#include "libpspp/stringi-set.h"
+#include "gl/c-strtod.h"
#include "gl/c-ctype.h"
#include "gl/inttostr.h"
#include "gl/localcharset.h"
EXT_VAR_ATTRS = 18, /* Variable attributes. */
EXT_MRSETS2 = 19, /* Multiple response sets (extended). */
EXT_ENCODING = 20, /* Character encoding. */
- EXT_LONG_LABELS = 21 /* Value labels for long strings. */
+ EXT_LONG_LABELS = 21, /* Value labels for long strings. */
+ EXT_DATAVIEW = 24 /* "Format properties in dataview table". */
};
/* Fields from the top-level header record. */
/* Opens the system file designated by file handle FH for reading. Reads the
system file's dictionary into *DICT.
+ Ordinarily the reader attempts to automatically detect the character
+ encoding based on the file's contents. This isn't always possible,
+ especially for files written by old versions of SPSS or PSPP, so specifying
+ a nonnull ENCODING overrides the choice of character encoding.
+
If INFO is non-null, then it receives additional info about the system file,
which the caller must eventually free with sfm_read_info_destroy() when it
is no longer needed. */
struct casereader *
-sfm_open_reader (struct file_handle *fh, struct dictionary **dictp,
- struct sfm_read_info *infop)
+sfm_open_reader (struct file_handle *fh, const char *volatile encoding,
+ struct dictionary **dictp, struct sfm_read_info *infop)
{
struct sfm_reader *volatile r = NULL;
- struct sfm_read_info info;
+ struct sfm_read_info *volatile info;
struct sfm_header_record header;
struct sfm_extension_record *extensions[32];
- struct dictionary *dict = NULL;
+ struct dictionary *volatile dict = NULL;
size_t i;
/* Create and initialize reader. */
r->opcode_idx = sizeof r->opcodes;
r->corruption_warning = false;
- memset (&info, 0, sizeof info);
+ info = infop ? infop : xmalloc (sizeof *info);
+ memset (info, 0, sizeof *info);
/* TRANSLATORS: this fragment will be interpolated into
messages in fh_lock() that identify types of files. */
goto error;
/* Read header. */
- read_header (r, &info, &header);
+ read_header (r, info, &header);
vars = NULL;
n_vars = allocated_vars = 0;
First, figure out the correct character encoding, because this determines
how the rest of the header data is to be interpreted. */
- dict = dict_create (choose_encoding (r, &header, extensions[EXT_INTEGER],
- extensions[EXT_ENCODING]));
+ dict = dict_create (encoding
+ ? encoding
+ : choose_encoding (r, &header, extensions[EXT_INTEGER],
+ extensions[EXT_ENCODING]));
r->encoding = dict_get_encoding (dict);
/* These records don't use variables at all. */
parse_document (dict, document);
if (extensions[EXT_INTEGER] != NULL)
- parse_machine_integer_info (r, extensions[EXT_INTEGER], &info);
+ parse_machine_integer_info (r, extensions[EXT_INTEGER], info);
if (extensions[EXT_FLOAT] != NULL)
parse_machine_float_info (r, extensions[EXT_FLOAT]);
if (extensions[EXT_FILE_ATTRS] != NULL)
parse_data_file_attributes (r, extensions[EXT_FILE_ATTRS], dict);
- parse_header (r, &header, &info, dict);
+ parse_header (r, &header, info, dict);
/* Parse the variable records, the basis of almost everything else. */
parse_variable_records (r, dict, vars, n_vars);
wrong when very long strings are involved, so don't warn in
that case. */
if (header.nominal_case_size != -1 && header.nominal_case_size != n_vars
- && info.version_major != 13)
+ && info->version_major != 13)
sys_warn (r, -1, _("File header claims %d variable positions but "
"%zu were read from file."),
header.nominal_case_size, n_vars);
r->proto = caseproto_ref_pool (dict_get_proto (dict), r->pool);
*dictp = dict;
- if (infop)
- *infop = info;
- else
- sfm_read_info_destroy (&info);
+ if (infop != info)
+ {
+ sfm_read_info_destroy (info);
+ free (info);
+ }
return casereader_create_sequential
(NULL, r->proto,
&sys_file_casereader_class, r);
error:
- sfm_read_info_destroy (&info);
+ if (infop != info)
+ {
+ sfm_read_info_destroy (info);
+ free (info);
+ }
+
close_reader (r);
dict_destroy (dict);
*dictp = NULL;
{ EXT_VAR_SETS, 0, 0 },
{ EXT_DATE, 0, 0 },
{ EXT_DATA_ENTRY, 0, 0 },
+ { EXT_DATAVIEW, 0, 0 },
};
const struct extension_record_type *type;
{
double low = parse_float (r, rec->missing, 0);
double high = parse_float (r, rec->missing, 8);
+
+ /* Deal with SPSS 21 change in representation. */
+ if (low == SYSMIS)
+ low = LOWEST;
+
mv_add_range (&mv, low, high);
ofs += 16;
}
double lowest = parse_float (r, record->data, 16);
if (sysmis != SYSMIS)
- sys_warn (r, record->pos, _("File specifies unexpected value %g as %s."),
- sysmis, "SYSMIS");
+ sys_warn (r, record->pos,
+ _("File specifies unexpected value %g (%a) as %s, "
+ "instead of %g (%a)."),
+ sysmis, sysmis, "SYSMIS", SYSMIS, SYSMIS);
if (highest != HIGHEST)
- sys_warn (r, record->pos, _("File specifies unexpected value %g as %s."),
- highest, "HIGHEST");
-
- if (lowest != LOWEST)
- sys_warn (r, record->pos, _("File specifies unexpected value %g as %s."),
- lowest, "LOWEST");
+ sys_warn (r, record->pos,
+ _("File specifies unexpected value %g (%a) as %s, "
+ "instead of %g (%a)."),
+ highest, highest, "HIGHEST", HIGHEST, HIGHEST);
+
+ /* SPSS before version 21 used a unique value just bigger than SYSMIS as
+ LOWEST. SPSS 21 uses SYSMIS for LOWEST, which is OK because LOWEST only
+ appears in a context (missing values) where SYSMIS cannot. */
+ if (lowest != LOWEST && lowest != SYSMIS)
+ sys_warn (r, record->pos,
+ _("File specifies unexpected value %g (%a) as %s, "
+ "instead of %g (%a) or %g (%a)."),
+ lowest, lowest, "LOWEST", LOWEST, LOWEST, SYSMIS, SYSMIS);
}
/* Parses record type 7, subtype 7 or 19. */
_("MRSET %s has only %zu variables."), mrset->name,
mrset->n_vars);
mrset_destroy (mrset);
+ stringi_set_destroy (&var_names);
continue;
}
mrset->width = width;
value_init (&mrset->counted, width);
if (width == 0)
- mrset->counted.f = strtod (counted, NULL);
+ mrset->counted.f = c_strtod (counted, NULL);
else
value_copy_str_rpad (&mrset->counted, width,
(const uint8_t *) counted, ' ');
align = parse_int (r, record->data, ofs);
ofs += 4;
- /* SPSS 14 sometimes seems to set string variables' measure
- to zero. */
- if (0 == measure && var_is_alpha (v))
+ /* SPSS sometimes seems to set variables' measure to zero. */
+ if (0 == measure)
measure = 1;
if (measure < 1 || measure > 3 || align < 0 || align > 2)
if (record == NULL)
{
- /* Convert variable names to lowercase. */
+ /* There are no long variable names. Use the short variable names,
+ converted to lowercase, as the long variable names. */
size_t i;
for (i = 0; i < dict_get_var_cnt (dict); i++)
struct variable *var = dict_get_var (dict, i);
char *new_name;
- new_name = xstrdup (var_get_name (var));
- str_lowercase (new_name);
-
+ new_name = utf8_to_lower (var_get_name (var));
rename_var_and_save_short_names (dict, var, new_name);
-
free (new_name);
}
while (read_variable_to_value_pair (r, dict, text, &var, &long_name))
{
/* Validate long name. */
- /* XXX need to reencode name to UTF-8 */
if (!dict_id_is_valid (dict, long_name, false))
{
sys_warn (r, record->pos,
}
/* Identify any duplicates. */
- if (strcasecmp (var_get_short_name (var, 0), long_name)
+ if (utf8_strcasecmp (var_get_short_name (var, 0), long_name)
&& dict_lookup_var (dict, long_name) != NULL)
{
sys_warn (r, record->pos,
start = text->pos;
n = 0;
- for (;;)
+ while (text->pos < text->buffer.length)
{
int c = text->buffer.string[text->pos];
if (c < '0' || c > '9')
n = (n * 10) + (c - '0');
text->pos++;
}
- if (start == text->pos)
+ if (text->pos >= text->buffer.length || start == text->pos)
{
sys_warn (r, text->start,
_("Expecting digit at offset %zu in MRSETS record."),
projects / pspp / blobdiff