#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"
/* Fields from the top-level header record. */
struct sfm_header_record
{
+ char magic[5]; /* First 4 bytes of file, then null. */
int weight_idx; /* 0 if unweighted, otherwise a var index. */
int nominal_case_size; /* Number of var positions. */
static const char *choose_encoding (
struct sfm_reader *,
+ const struct sfm_header_record *,
const struct sfm_extension_record *ext_integer,
const struct sfm_extension_record *ext_encoding);
/* 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, 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;
bool
sfm_detect (FILE *file)
{
- char rec_type[5];
+ char magic[5];
- if (fread (rec_type, 4, 1, file) != 1)
+ if (fread (magic, 4, 1, file) != 1)
return false;
- rec_type[4] = '\0';
+ magic[4] = '\0';
- return !strcmp ("$FL2", rec_type);
+ return !strcmp (ASCII_MAGIC, magic) || !strcmp (EBCDIC_MAGIC, magic);
}
\f
/* Reads the global header of the system file. Initializes *HEADER and *INFO,
read_header (struct sfm_reader *r, struct sfm_read_info *info,
struct sfm_header_record *header)
{
- char rec_type[5];
uint8_t raw_layout_code[4];
uint8_t raw_bias[8];
- read_string (r, rec_type, sizeof rec_type);
+ read_string (r, header->magic, sizeof header->magic);
read_string (r, header->eye_catcher, sizeof header->eye_catcher);
- if (strcmp ("$FL2", rec_type) != 0)
+ if (strcmp (ASCII_MAGIC, header->magic)
+ && strcmp (EBCDIC_MAGIC, header->magic))
sys_error (r, 0, _("This is not an SPSS system file."));
/* Identify integer format. */
static const char *
choose_encoding (struct sfm_reader *r,
+ const struct sfm_header_record *header,
const struct sfm_extension_record *ext_integer,
const struct sfm_extension_record *ext_encoding)
{
}
}
+ /* If the file magic number is EBCDIC then its character data is too. */
+ if (!strcmp (header->magic, EBCDIC_MAGIC))
+ return "EBCDIC-US";
+
return locale_charset ();
}
_("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, ' ');
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