-/* PSPP - computes sample statistics.
- Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc.
- Written by Ben Pfaff <blp@gnu.org>.
+/* PSPP - a program for statistical analysis.
+ Copyright (C) 1997-9, 2000, 2006, 2007 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 the Free Software Foundation; either version 2 of the
- License, or (at your option) any later version.
+ 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
+ the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
- This program is distributed in the hope that it will be useful, but
- WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- General Public License for more details.
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
You should have received a copy of the GNU General Public License
- along with this program; if not, write to the Free Software
- Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
- 02110-1301, USA. */
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
#include <config.h>
-#include <stdlib.h>
+#include <data/sys-file-reader.h>
+#include <data/sys-file-private.h>
+
#include <errno.h>
#include <float.h>
-#include <c-ctype.h>
+#include <inttypes.h>
+#include <setjmp.h>
+#include <stdlib.h>
-#include <libpspp/alloc.h>
+#include <libpspp/assertion.h>
#include <libpspp/message.h>
#include <libpspp/compiler.h>
-#include <libpspp/magic.h>
#include <libpspp/misc.h>
+#include <libpspp/pool.h>
#include <libpspp/str.h>
-
-#include "sys-file-reader.h"
-#include "sfm-private.h"
-#include "case.h"
-#include "dictionary.h"
-#include "file-handle-def.h"
-#include "file-name.h"
-#include "format.h"
-#include "value-labels.h"
-#include "variable.h"
+#include <libpspp/hash.h>
+#include <libpspp/array.h>
+
+#include <data/case.h>
+#include <data/casereader-provider.h>
+#include <data/casereader.h>
+#include <data/dictionary.h>
+#include <data/file-handle-def.h>
+#include <data/file-name.h>
+#include <data/format.h>
+#include <data/missing-values.h>
+#include <data/value-labels.h>
+#include <data/variable.h>
+#include <data/value.h>
+
+#include "c-ctype.h"
+#include "inttostr.h"
+#include "minmax.h"
+#include "unlocked-io.h"
+#include "xalloc.h"
+#include "xsize.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
+#define N_(msgid) (msgid)
/* System file reader. */
struct sfm_reader
{
- struct file_handle *fh; /* File handle. */
- FILE *file; /* File stream. */
+ /* Resource tracking. */
+ struct pool *pool; /* All system file state. */
+ jmp_buf bail_out; /* longjmp() target for error handling. */
- int reverse_endian; /* 1=file has endianness opposite us. */
- int fix_specials; /* 1=SYSMIS/HIGHEST/LOWEST differs from us. */
- int value_cnt; /* Number of `union values's per case. */
- long case_cnt; /* Number of cases, -1 if unknown. */
- int compressed; /* 1=compressed, 0=not compressed. */
+ /* File state. */
+ struct file_handle *fh; /* File handle. */
+ FILE *file; /* File stream. */
+ bool error; /* I/O or corruption error? */
+ size_t value_cnt; /* Number of "union value"s in struct case. */
+
+ /* File format. */
+ enum integer_format integer_format; /* On-disk integer format. */
+ enum float_format float_format; /* On-disk floating point format. */
+ int oct_cnt; /* Number of 8-byte units per case. */
+ struct sfm_var *sfm_vars; /* Variables. */
+ size_t sfm_var_cnt; /* Number of variables. */
+ casenumber case_cnt; /* Number of cases */
+ bool has_long_var_names; /* File has a long variable name map */
+
+ /* Decompression. */
+ bool compressed; /* File is compressed? */
double bias; /* Compression bias, usually 100.0. */
- int weight_idx; /* 0-based index of weighting variable, or -1. */
- bool ok; /* False after an I/O error or corrupt data. */
-
- /* Variables. */
- struct sfm_var *vars; /* Variables. */
-
- /* File's special constants. */
- flt64 sysmis;
- flt64 highest;
- flt64 lowest;
-
- /* Decompression buffer. */
- flt64 *buf; /* Buffer data. */
- flt64 *ptr; /* Current location in buffer. */
- flt64 *end; /* End of buffer data. */
-
- /* Compression instruction octet. */
- unsigned char x[8]; /* Current instruction octet. */
- unsigned char *y; /* Location in current instruction octet. */
+ uint8_t opcodes[8]; /* Current block of opcodes. */
+ size_t opcode_idx; /* Next opcode to interpret, 8 if none left. */
};
-/* A variable in a system file. */
-struct sfm_var
- {
- int width; /* 0=numeric, otherwise string width. */
- int fv; /* Index into case. */
- };
-\f
-/* Utilities. */
+static struct casereader_class sys_file_casereader_class;
-/* Swap bytes *A and *B. */
-static inline void
-bswap (char *a, char *b)
-{
- char t = *a;
- *a = *b;
- *b = t;
-}
+static bool close_reader (struct sfm_reader *);
-/* Reverse the byte order of 32-bit integer *X. */
-static inline void
-bswap_int32 (int32_t *x_)
-{
- char *x = (char *) x_;
- bswap (x + 0, x + 3);
- bswap (x + 1, x + 2);
-}
+static struct variable **make_var_by_value_idx (struct sfm_reader *,
+ struct dictionary *);
+static struct variable *lookup_var_by_value_idx (struct sfm_reader *,
+ struct variable **,
+ int value_idx);
-/* Reverse the byte order of 64-bit floating point *X. */
-static inline void
-bswap_flt64 (flt64 *x_)
-{
- char *x = (char *) x_;
- bswap (x + 0, x + 7);
- bswap (x + 1, x + 6);
- bswap (x + 2, x + 5);
- bswap (x + 3, x + 4);
-}
-
-static void
-corrupt_msg (int class, const char *format,...)
+static void sys_warn (struct sfm_reader *, const char *, ...)
PRINTF_FORMAT (2, 3);
-/* Displays a corrupt sysfile error. */
-static void
-corrupt_msg (int class, const char *format,...)
-{
- struct error e;
- va_list args;
- struct string text;
-
- ds_create (&text, _("corrupt system file: "));
- va_start (args, format);
- ds_vprintf (&text, format, args);
- va_end (args);
-
- e.category = msg_class_to_category (class);
- e.severity = msg_class_to_severity (class);
- e.where.file_name = NULL;
- e.where.line_number = 0;
- e.text = ds_c_str (&text);
-
- err_msg (&e);
-}
-
-/* Closes a system file after we're done with it. */
-void
-sfm_close_reader (struct sfm_reader *r)
-{
- if (r == NULL)
- return;
-
- if (r->file)
- {
- if (fn_close (fh_get_file_name (r->fh), r->file) == EOF)
- msg (ME, _("%s: Closing system file: %s."),
- fh_get_file_name (r->fh), strerror (errno));
- r->file = NULL;
- }
-
- if (r->fh != NULL)
- fh_close (r->fh, "system file", "rs");
-
- free (r->vars);
- free (r->buf);
- free (r);
-}
+static void sys_error (struct sfm_reader *, const char *, ...)
+ PRINTF_FORMAT (2, 3)
+ NO_RETURN;
+
+static void read_bytes (struct sfm_reader *, void *, size_t);
+static bool try_read_bytes (struct sfm_reader *, void *, size_t);
+static int read_int (struct sfm_reader *);
+static double read_float (struct sfm_reader *);
+static void read_string (struct sfm_reader *, char *, size_t);
+static void skip_bytes (struct sfm_reader *, size_t);
+
+static struct variable_to_value_map *open_variable_to_value_map (
+ struct sfm_reader *, size_t size);
+static void close_variable_to_value_map (struct sfm_reader *r,
+ struct variable_to_value_map *);
+static bool read_variable_to_value_map (struct sfm_reader *,
+ struct dictionary *,
+ struct variable_to_value_map *,
+ struct variable **var, char **value,
+ int *warning_cnt);
+
+static bool close_reader (struct sfm_reader *r);
\f
/* Dictionary reader. */
-static void buf_unread(struct sfm_reader *r, size_t byte_cnt);
-
-static void *buf_read (struct sfm_reader *, void *buf, size_t byte_cnt,
- size_t min_alloc);
-
-static int read_header (struct sfm_reader *,
- struct dictionary *, struct sfm_read_info *);
-static int parse_format_spec (struct sfm_reader *, int32_t,
- struct fmt_spec *, const struct variable *);
-static int read_value_labels (struct sfm_reader *, struct dictionary *,
- struct variable **var_by_idx);
-static int read_variables (struct sfm_reader *,
- struct dictionary *, struct variable ***var_by_idx);
-static int read_machine_int32_info (struct sfm_reader *, int size, int count);
-static int read_machine_flt64_info (struct sfm_reader *, int size, int count);
-static int read_documents (struct sfm_reader *, struct dictionary *);
-
-static int fread_ok (struct sfm_reader *, void *, size_t);
-
-/* Displays the message X with corrupt_msg, then jumps to the error
- label. */
-#define lose(X) \
- do { \
- corrupt_msg X; \
- goto error; \
- } while (0)
-
-/* Calls buf_read with the specified arguments, and jumps to
- error if the read fails. */
-#define assertive_buf_read(a,b,c,d) \
- do { \
- if (!buf_read (a,b,c,d)) \
- goto error; \
- } while (0)
+enum which_format
+ {
+ PRINT_FORMAT,
+ WRITE_FORMAT
+ };
+
+static void read_header (struct sfm_reader *, struct dictionary *,
+ int *weight_idx, int *claimed_oct_cnt,
+ struct sfm_read_info *);
+static void read_variable_record (struct sfm_reader *, struct dictionary *,
+ int *format_warning_cnt);
+static void parse_format_spec (struct sfm_reader *, unsigned int,
+ enum which_format, struct variable *,
+ int *format_warning_cnt);
+static void setup_weight (struct sfm_reader *, int weight_idx,
+ struct variable **var_by_value_idx,
+ struct dictionary *);
+static void read_documents (struct sfm_reader *, struct dictionary *);
+static void read_value_labels (struct sfm_reader *, struct dictionary *,
+ struct variable **var_by_value_idx);
+
+static void read_extension_record (struct sfm_reader *, struct dictionary *,
+ struct sfm_read_info *);
+static void read_machine_integer_info (struct sfm_reader *,
+ size_t size, size_t count,
+ struct sfm_read_info *);
+static void read_machine_float_info (struct sfm_reader *,
+ size_t size, size_t count);
+static void read_display_parameters (struct sfm_reader *,
+ size_t size, size_t count,
+ struct dictionary *);
+static void read_long_var_name_map (struct sfm_reader *,
+ size_t size, size_t count,
+ struct dictionary *);
+static void read_long_string_map (struct sfm_reader *,
+ size_t size, size_t count,
+ struct dictionary *);
+
/* Opens the system file designated by file handle FH for
reading. Reads the system file's dictionary into *DICT.
If INFO is non-null, then it receives additional info about the
system file. */
-struct sfm_reader *
+struct casereader *
sfm_open_reader (struct file_handle *fh, struct dictionary **dict,
- struct sfm_read_info *info)
+ struct sfm_read_info *volatile info)
{
- struct sfm_reader *r = NULL;
- struct variable **var_by_idx = NULL;
+ struct sfm_reader *volatile r = NULL;
+ struct variable **var_by_value_idx;
+ struct sfm_read_info local_info;
+ int format_warning_cnt = 0;
+ int weight_idx;
+ int claimed_oct_cnt;
+ int rec_type;
- *dict = dict_create ();
if (!fh_open (fh, FH_REF_FILE, "system file", "rs"))
- goto error;
+ return NULL;
+
+ *dict = dict_create ();
/* Create and initialize reader. */
- r = xmalloc (sizeof *r);
+ r = pool_create_container (struct sfm_reader, pool);
r->fh = fh;
r->file = fn_open (fh_get_file_name (fh), "rb");
+ r->error = false;
+ r->oct_cnt = 0;
+ r->has_long_var_names = false;
+ r->opcode_idx = sizeof r->opcodes;
- r->reverse_endian = 0;
- r->fix_specials = 0;
- r->value_cnt = 0;
- r->case_cnt = 0;
- r->compressed = 0;
- r->bias = 100.0;
- r->weight_idx = -1;
- r->ok = true;
+ /* Initialize info. */
+ if (info == NULL)
+ info = &local_info;
+ memset (info, 0, sizeof *info);
- r->vars = NULL;
-
- r->sysmis = -FLT64_MAX;
- r->highest = FLT64_MAX;
- r->lowest = second_lowest_flt64;
-
- r->buf = r->ptr = r->end = NULL;
- r->y = r->x + sizeof r->x;
+ if (setjmp (r->bail_out))
+ {
+ close_reader (r);
+ dict_destroy (*dict);
+ *dict = NULL;
+ return NULL;
+ }
- /* Check that file open succeeded. */
if (r->file == NULL)
{
- msg (ME, _("An error occurred while opening \"%s\" for reading "
- "as a system file: %s."),
+ msg (ME, _("Error opening \"%s\" for reading as a system file: %s."),
fh_get_file_name (r->fh), strerror (errno));
- goto error;
+ longjmp (r->bail_out, 1);
}
- /* Read header and variables. */
- if (!read_header (r, *dict, info) || !read_variables (r, *dict, &var_by_idx))
- goto error;
-
+ /* Read header. */
+ read_header (r, *dict, &weight_idx, &claimed_oct_cnt, info);
- /* Handle weighting. */
- if (r->weight_idx != -1)
+ /* Read all the variable definition records. */
+ rec_type = read_int (r);
+ while (rec_type == 2)
{
- struct variable *weight_var;
+ read_variable_record (r, *dict, &format_warning_cnt);
+ rec_type = read_int (r);
+ }
+
+ /* Figure out the case format. */
+ var_by_value_idx = make_var_by_value_idx (r, *dict);
+ setup_weight (r, weight_idx, var_by_value_idx, *dict);
- if (r->weight_idx < 0 || r->weight_idx >= r->value_cnt)
- lose ((ME, _("%s: Index of weighting variable (%d) is not between 0 "
- "and number of elements per case (%d)."),
- fh_get_file_name (r->fh), r->weight_idx, r->value_cnt));
+ /* Read all the rest of the dictionary records. */
+ while (rec_type != 999)
+ {
+ switch (rec_type)
+ {
+ case 3:
+ read_value_labels (r, *dict, var_by_value_idx);
+ break;
+ case 4:
+ sys_error (r, _("Misplaced type 4 record."));
- weight_var = var_by_idx[r->weight_idx];
+ case 6:
+ read_documents (r, *dict);
+ break;
- if (weight_var == NULL)
- lose ((ME,
- _("%s: Weighting variable may not be a continuation of "
- "a long string variable."), fh_get_file_name (fh)));
- else if (weight_var->type == ALPHA)
- lose ((ME, _("%s: Weighting variable may not be a string variable."),
- fh_get_file_name (fh)));
+ case 7:
+ read_extension_record (r, *dict, info);
+ break;
- dict_set_weight (*dict, weight_var);
+ default:
+ sys_error (r, _("Unrecognized record type %d."), rec_type);
+ }
+ rec_type = read_int (r);
}
- else
- dict_set_weight (*dict, NULL);
-
- /* Read records of types 3, 4, 6, and 7. */
- for (;;)
- {
- int32_t rec_type;
- assertive_buf_read (r, &rec_type, sizeof rec_type, 0);
- if (r->reverse_endian)
- bswap_int32 (&rec_type);
- switch (rec_type)
+ if ( ! r->has_long_var_names )
+ {
+ int i;
+ for (i = 0; i < dict_get_var_cnt (*dict); i++)
{
- case 3:
- if (!read_value_labels (r, *dict, var_by_idx))
- goto error;
- break;
-
- case 4:
- lose ((ME, _("%s: Orphaned variable index record (type 4). Type 4 "
- "records must always immediately follow type 3 "
- "records."),
- fh_get_file_name (r->fh)));
-
- case 6:
- if (!read_documents (r, *dict))
- goto error;
- break;
-
- case 7:
- {
- struct
- {
- int32_t subtype P;
- int32_t size P;
- int32_t count P;
- }
- data;
- unsigned long bytes;
-
- int skip = 0;
-
- assertive_buf_read (r, &data, sizeof data, 0);
- if (r->reverse_endian)
- {
- bswap_int32 (&data.subtype);
- bswap_int32 (&data.size);
- bswap_int32 (&data.count);
- }
- bytes = data.size * data.count;
- if (bytes < data.size || bytes < data.count)
- lose ((ME, "%s: Record type %d subtype %d too large.",
- fh_get_file_name (r->fh), rec_type, data.subtype));
-
- switch (data.subtype)
- {
- case 3:
- if (!read_machine_int32_info (r, data.size, data.count))
- goto error;
- break;
-
- case 4:
- if (!read_machine_flt64_info (r, data.size, data.count))
- goto error;
- break;
-
- case 5:
- case 6: /* ?? Used by SPSS 8.0. */
- skip = 1;
- break;
-
- case 11: /* Variable display parameters */
- {
- const int n_vars = data.count / 3 ;
- int i;
- if ( data.count % 3 || n_vars > dict_get_var_cnt(*dict) )
- {
- msg (MW, _("%s: Invalid subrecord length. "
- "Record: 7; Subrecord: 11"),
- fh_get_file_name (r->fh));
- skip = 1;
- }
-
- for ( i = 0 ; i < min(n_vars, dict_get_var_cnt(*dict)) ; ++i )
- {
- struct
- {
- int32_t measure P;
- int32_t width P;
- int32_t align P;
- }
- params;
-
- struct variable *v;
-
- assertive_buf_read (r, ¶ms, sizeof(params), 0);
-
- v = dict_get_var(*dict, i);
-
- v->measure = params.measure;
- v->display_width = params.width;
- v->alignment = params.align;
- }
- }
- break;
-
- case 13: /* SPSS 12.0 Long variable name map */
- {
- char *buf, *short_name, *save_ptr;
- int idx;
-
- /* Read data. */
- buf = xmalloc (bytes + 1);
- if (!buf_read (r, buf, bytes, 0))
- {
- free (buf);
- goto error;
- }
- buf[bytes] = '\0';
-
- /* Parse data. */
- for (short_name = strtok_r (buf, "=", &save_ptr), idx = 0;
- short_name != NULL;
- short_name = strtok_r (NULL, "=", &save_ptr), idx++)
- {
- char *long_name = strtok_r (NULL, "\t", &save_ptr);
- struct variable *v;
-
- /* Validate long name. */
- if (long_name == NULL)
- {
- msg (MW, _("%s: Trailing garbage in long variable "
- "name map."),
- fh_get_file_name (r->fh));
- break;
- }
- if (!var_is_valid_name (long_name, false))
- {
- msg (MW, _("%s: Long variable mapping to invalid "
- "variable name `%s'."),
- fh_get_file_name (r->fh), long_name);
- break;
- }
-
- /* Find variable using short name. */
- v = dict_lookup_var (*dict, short_name);
- if (v == NULL)
- {
- msg (MW, _("%s: Long variable mapping for "
- "nonexistent variable %s."),
- fh_get_file_name (r->fh), short_name);
- break;
- }
-
- /* Identify any duplicates. */
- if ( compare_var_names(short_name, long_name, 0) &&
- NULL != dict_lookup_var (*dict, long_name))
- lose ((ME, _("%s: Duplicate long variable name `%s' "
- "within system file."),
- fh_get_file_name (r->fh), long_name));
-
-
- /* Set long name.
- Renaming a variable may clear the short
- name, but we want to retain it, so
- re-set it explicitly. */
- dict_rename_var (*dict, v, long_name);
- var_set_short_name (v, short_name);
-
- /* For compatability, make sure dictionary
- is in long variable name map order. In
- the common case, this has no effect,
- because the dictionary and the long
- variable name map are already in the
- same order. */
- dict_reorder_var (*dict, v, idx);
- }
-
- /* Free data. */
- free (buf);
- }
- break;
-
- default:
- msg (MW, _("%s: Unrecognized record type 7, subtype %d "
- "encountered in system file."),
- fh_get_file_name (r->fh), data.subtype);
- skip = 1;
- }
-
- if (skip)
- {
- void *x = buf_read (r, NULL, data.size * data.count, 0);
- if (x == NULL)
- goto error;
- free (x);
- }
- }
- break;
-
- case 999:
- {
- int32_t filler;
-
- assertive_buf_read (r, &filler, sizeof filler, 0);
- goto success;
- }
-
- default:
- corrupt_msg(MW, _("%s: Unrecognized record type %d."),
- fh_get_file_name (r->fh), rec_type);
- }
- }
+ struct variable *var = dict_get_var (*dict, i);
+ char short_name [SHORT_NAME_LEN + 1];
+ char long_name [SHORT_NAME_LEN + 1];
-success:
- /* Come here on successful completion. */
- free (var_by_idx);
- return r;
+ strcpy (short_name, var_get_name (var));
-error:
- /* Come here on unsuccessful completion. */
- sfm_close_reader (r);
- free (var_by_idx);
- if (*dict != NULL)
- {
- dict_destroy (*dict);
- *dict = NULL;
- }
- return NULL;
-}
+ strcpy (long_name, short_name);
+ str_lowercase (long_name);
-/* Read record type 7, subtype 3. */
-static int
-read_machine_int32_info (struct sfm_reader *r, int size, int count)
-{
- int32_t data[8];
- int file_bigendian;
+ /* Set long name. Renaming a variable may clear the short
+ name, but we want to retain it, so re-set it
+ explicitly. */
+ dict_rename_var (*dict, var, long_name);
+ var_set_short_name (var, 0, short_name);
+ }
- int i;
+ r->has_long_var_names = true;
+ }
- if (size != sizeof (int32_t) || count != 8)
- lose ((ME, _("%s: Bad size (%d) or count (%d) field on record type 7, "
- "subtype 3. Expected size %d, count 8."),
- fh_get_file_name (r->fh), size, count, sizeof (int32_t)));
-
- assertive_buf_read (r, data, sizeof data, 0);
- if (r->reverse_endian)
- for (i = 0; i < 8; i++)
- bswap_int32 (&data[i]);
-
-#ifdef FPREP_IEEE754
- if (data[4] != 1)
- lose ((ME, _("%s: Floating-point representation in system file is not "
- "IEEE-754. PSPP cannot convert between floating-point "
- "formats."),
- fh_get_file_name (r->fh)));
-#else
-#error Add support for your floating-point format.
-#endif
-
-#ifdef WORDS_BIGENDIAN
- file_bigendian = 1;
-#else
- file_bigendian = 0;
-#endif
- if (r->reverse_endian)
- file_bigendian ^= 1;
- if (file_bigendian ^ (data[6] == 1))
- lose ((ME, _("%s: File-indicated endianness (%s) does not match "
- "endianness intuited from file header (%s)."),
- fh_get_file_name (r->fh),
- file_bigendian ? _("big-endian") : _("little-endian"),
- data[6] == 1 ? _("big-endian") : (data[6] == 2 ? _("little-endian")
- : _("unknown"))));
-
- /* PORTME: Character representation code. */
- if (data[7] != 2 && data[7] != 3)
- lose ((ME, _("%s: File-indicated character representation code (%s) is "
- "not ASCII."),
- fh_get_file_name (r->fh),
- (data[7] == 1 ? "EBCDIC"
- : (data[7] == 4 ? _("DEC Kanji") : _("Unknown")))));
-
- return 1;
-
-error:
- return 0;
+ /* Read record 999 data, which is just filler. */
+ read_int (r);
+
+ /* Warn if the actual amount of data per case differs from the
+ amount that the header claims. SPSS version 13 gets this
+ wrong when very long strings are involved, so don't warn in
+ that case. */
+ if (claimed_oct_cnt != -1 && claimed_oct_cnt != r->oct_cnt
+ && info->version_major != 13)
+ sys_warn (r, _("File header claims %d variable positions but "
+ "%d were read from file."),
+ claimed_oct_cnt, r->oct_cnt);
+
+ /* Create an index of dictionary variable widths for
+ sfm_read_case to use. We cannot use the `struct variable's
+ from the dictionary we created, because the caller owns the
+ dictionary and may destroy or modify its variables. */
+ sfm_dictionary_to_sfm_vars (*dict, &r->sfm_vars, &r->sfm_var_cnt);
+ pool_register (r->pool, free, r->sfm_vars);
+
+ pool_free (r->pool, var_by_value_idx);
+ r->value_cnt = dict_get_next_value_idx (*dict);
+ return casereader_create_sequential
+ (NULL, r->value_cnt,
+ r->case_cnt == -1 ? CASENUMBER_MAX: r->case_cnt,
+ &sys_file_casereader_class, r);
}
-/* Read record type 7, subtype 4. */
-static int
-read_machine_flt64_info (struct sfm_reader *r, int size, int count)
+/* Closes a system file after we're done with it.
+ Returns true if an I/O error has occurred on READER, false
+ otherwise. */
+static bool
+close_reader (struct sfm_reader *r)
{
- flt64 data[3];
- int i;
+ bool error;
- if (size != sizeof (flt64) || count != 3)
- lose ((ME, _("%s: Bad size (%d) or count (%d) field on record type 7, "
- "subtype 4. Expected size %d, count 8."),
- fh_get_file_name (r->fh), size, count, sizeof (flt64)));
-
- assertive_buf_read (r, data, sizeof data, 0);
- if (r->reverse_endian)
- for (i = 0; i < 3; i++)
- bswap_flt64 (&data[i]);
+ if (r == NULL)
+ return true;
- if (data[0] != SYSMIS || data[1] != FLT64_MAX
- || data[2] != second_lowest_flt64)
+ if (r->file)
{
- r->sysmis = data[0];
- r->highest = data[1];
- r->lowest = data[2];
- msg (MW, _("%s: File-indicated value is different from internal value "
- "for at least one of the three system values. SYSMIS: "
- "indicated %g, expected %g; HIGHEST: %g, %g; LOWEST: "
- "%g, %g."),
- fh_get_file_name (r->fh), (double) data[0], (double) SYSMIS,
- (double) data[1], (double) FLT64_MAX,
- (double) data[2], (double) second_lowest_flt64);
+ if (fn_close (fh_get_file_name (r->fh), r->file) == EOF)
+ {
+ msg (ME, _("Error closing system file \"%s\": %s."),
+ fh_get_file_name (r->fh), strerror (errno));
+ r->error = true;
+ }
+ r->file = NULL;
}
-
- return 1;
-error:
- return 0;
+ if (r->fh != NULL)
+ fh_close (r->fh, "system file", "rs");
+
+ error = r->error;
+ pool_destroy (r->pool);
+
+ return !error;
}
-static int
-read_header (struct sfm_reader *r,
- struct dictionary *dict, struct sfm_read_info *info)
+/* Destroys READER. */
+static void
+sys_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
{
- struct sysfile_header hdr; /* Disk buffer. */
- char prod_name[sizeof hdr.prod_name + 1]; /* Buffer for product name. */
- int skip_amt = 0; /* Amount of product name to omit. */
- int i;
+ struct sfm_reader *r = r_;
+ close_reader (r);
+}
- /* Read header, check magic. */
- assertive_buf_read (r, &hdr, sizeof hdr, 0);
- if (strncmp ("$FL2", hdr.rec_type, 4) != 0)
- lose ((ME, _("%s: Bad magic. Proper system files begin with "
- "the four characters `$FL2'. This file will not be read."),
- fh_get_file_name (r->fh)));
-
- /* Check eye-category.her string. */
- memcpy (prod_name, hdr.prod_name, sizeof hdr.prod_name);
- for (i = 0; i < 60; i++)
- if (!c_isprint ((unsigned char) prod_name[i]))
- prod_name[i] = ' ';
- for (i = 59; i >= 0; i--)
- if (!c_isgraph ((unsigned char) prod_name[i]))
- {
- prod_name[i] = '\0';
- break;
- }
- prod_name[60] = '\0';
-
- {
-#define N_PREFIXES 2
- static const char *prefix[N_PREFIXES] =
- {
- "@(#) SPSS DATA FILE",
- "SPSS SYSTEM FILE.",
- };
+/* Returns true if FILE is an SPSS system file,
+ false otherwise. */
+bool
+sfm_detect (FILE *file)
+{
+ char rec_type[5];
- int i;
+ if (fread (rec_type, 4, 1, file) != 1)
+ return false;
+ rec_type[4] = '\0';
- for (i = 0; i < N_PREFIXES; i++)
- if (!strncmp (prefix[i], hdr.prod_name, strlen (prefix[i])))
- {
- skip_amt = strlen (prefix[i]);
- break;
- }
- }
-
- /* Check endianness. */
- if (hdr.layout_code == 2)
- r->reverse_endian = 0;
- else
+ return !strcmp ("$FL2", rec_type);
+}
+\f
+/* Reads the global header of the system file.
+ Sets DICT's file label to the system file's label.
+ Sets *WEIGHT_IDX to 0 if the system file is unweighted,
+ or to the value index of the weight variable otherwise.
+ Sets *CLAIMED_OCT_CNT to the number of "octs" (8-byte units)
+ per case that the file claims to have (although it is not
+ always correct).
+ Initializes INFO with header information. */
+static void
+read_header (struct sfm_reader *r, struct dictionary *dict,
+ int *weight_idx, int *claimed_oct_cnt,
+ struct sfm_read_info *info)
+{
+ char rec_type[5];
+ char eye_catcher[61];
+ uint8_t raw_layout_code[4];
+ uint8_t raw_bias[8];
+ char creation_date[10];
+ char creation_time[9];
+ char file_label[65];
+ struct substring file_label_ss;
+ struct substring product;
+
+ read_string (r, rec_type, sizeof rec_type);
+ read_string (r, eye_catcher, sizeof eye_catcher);
+
+ if (strcmp ("$FL2", rec_type) != 0)
+ sys_error (r, _("This is not an SPSS system file."));
+
+ /* Identify integer format. */
+ read_bytes (r, raw_layout_code, sizeof raw_layout_code);
+ if ((!integer_identify (2, raw_layout_code, sizeof raw_layout_code,
+ &r->integer_format)
+ && !integer_identify (3, raw_layout_code, sizeof raw_layout_code,
+ &r->integer_format))
+ || (r->integer_format != INTEGER_MSB_FIRST
+ && r->integer_format != INTEGER_LSB_FIRST))
+ sys_error (r, _("This is not an SPSS system file."));
+
+ *claimed_oct_cnt = read_int (r);
+ if (*claimed_oct_cnt < 0 || *claimed_oct_cnt > INT_MAX / 16)
+ *claimed_oct_cnt = -1;
+
+ r->compressed = read_int (r) != 0;
+
+ *weight_idx = read_int (r);
+
+ r->case_cnt = read_int (r);
+ if ( r->case_cnt > INT_MAX / 2)
+ r->case_cnt = -1;
+
+
+ /* Identify floating-point format and obtain compression bias. */
+ read_bytes (r, raw_bias, sizeof raw_bias);
+ if (float_identify (100.0, raw_bias, sizeof raw_bias, &r->float_format) == 0)
{
- bswap_int32 (&hdr.layout_code);
- if (hdr.layout_code != 2)
- lose ((ME, _("%s: File layout code has unexpected value %d. Value "
- "should be 2, in big-endian or little-endian format."),
- fh_get_file_name (r->fh), hdr.layout_code));
-
- r->reverse_endian = 1;
- bswap_int32 (&hdr.case_size);
- bswap_int32 (&hdr.compress);
- bswap_int32 (&hdr.weight_idx);
- bswap_int32 (&hdr.case_cnt);
- bswap_flt64 (&hdr.bias);
+ sys_warn (r, _("Compression bias (%g) is not the usual "
+ "value of 100, or system file uses unrecognized "
+ "floating-point format."),
+ r->bias);
+ if (r->integer_format == INTEGER_MSB_FIRST)
+ r->float_format = FLOAT_IEEE_DOUBLE_BE;
+ else
+ r->float_format = FLOAT_IEEE_DOUBLE_LE;
}
+ float_convert (r->float_format, raw_bias, FLOAT_NATIVE_DOUBLE, &r->bias);
+ read_string (r, creation_date, sizeof creation_date);
+ read_string (r, creation_time, sizeof creation_time);
+ read_string (r, file_label, sizeof file_label);
+ skip_bytes (r, 3);
- /* Copy basic info and verify correctness. */
- r->value_cnt = hdr.case_size;
-
- /* If value count is rediculous, then force it to -1 (a sentinel value) */
- if ( r->value_cnt < 0 ||
- r->value_cnt > (INT_MAX / (int) sizeof (union value) / 2))
- r->value_cnt = -1;
+ file_label_ss = ss_cstr (file_label);
+ ss_trim (&file_label_ss, ss_cstr (" "));
+ if (!ss_is_empty (file_label_ss))
+ {
+ ss_data (file_label_ss)[ss_length (file_label_ss)] = '\0';
+ dict_set_label (dict, ss_data (file_label_ss));
+ }
- r->compressed = hdr.compress;
+ strcpy (info->creation_date, creation_date);
+ strcpy (info->creation_time, creation_time);
+ info->integer_format = r->integer_format;
+ info->float_format = r->float_format;
+ info->compressed = r->compressed;
+ info->case_cnt = r->case_cnt;
+
+ product = ss_cstr (eye_catcher);
+ ss_match_string (&product, ss_cstr ("@(#) SPSS DATA FILE"));
+ ss_trim (&product, ss_cstr (" "));
+ str_copy_buf_trunc (info->product, sizeof info->product,
+ ss_data (product), ss_length (product));
+}
- r->weight_idx = hdr.weight_idx - 1;
+/* Reads a variable (type 2) record from R and adds the
+ corresponding variable to DICT.
+ Also skips past additional variable records for long string
+ variables. */
+static void
+read_variable_record (struct sfm_reader *r, struct dictionary *dict,
+ int *format_warning_cnt)
+{
+ int width;
+ int has_variable_label;
+ int missing_value_code;
+ int print_format;
+ int write_format;
+ char name[9];
+
+ struct variable *var;
+ int nv;
+
+ width = read_int (r);
+ has_variable_label = read_int (r);
+ missing_value_code = read_int (r);
+ print_format = read_int (r);
+ write_format = read_int (r);
+ read_string (r, name, sizeof name);
+ name[strcspn (name, " ")] = '\0';
+
+ /* Check variable name. */
+ if (name[0] == '$' || name[0] == '#')
+ sys_error (r, "Variable name begins with invalid character `%c'.",
+ name[0]);
+ if (!var_is_plausible_name (name, false))
+ sys_error (r, _("Invalid variable name `%s'."), name);
+
+ /* Create variable. */
+ if (width < 0 || width > 255)
+ sys_error (r, _("Bad variable width %d."), width);
+ var = dict_create_var (dict, name, width);
+ if (var == NULL)
+ sys_error (r,
+ _("Duplicate variable name `%s' within system file."),
+ name);
+
+ /* Set the short name the same as the long name. */
+ var_set_short_name (var, 0, var_get_name (var));
+
+ /* Get variable label, if any. */
+ if (has_variable_label != 0 && has_variable_label != 1)
+ sys_error (r, _("Variable label indicator field is not 0 or 1."));
+ if (has_variable_label == 1)
+ {
+ size_t len;
+ char label[255 + 1];
- r->case_cnt = hdr.case_cnt;
- if (r->case_cnt < -1 || r->case_cnt > INT_MAX / 2)
- lose ((ME,
- _("%s: Number of cases in file (%ld) is not between -1 and %d."),
- fh_get_file_name (r->fh), (long) r->case_cnt, INT_MAX / 2));
+ len = read_int (r);
+ if (len >= sizeof label)
+ sys_error (r, _("Variable %s has label of invalid length %zu."),
+ name, len);
+ read_string (r, label, len + 1);
+ var_set_label (var, label);
- r->bias = hdr.bias;
- if (r->bias != 100.0)
- corrupt_msg (MW, _("%s: Compression bias (%g) is not the usual "
- "value of 100."),
- fh_get_file_name (r->fh), r->bias);
+ skip_bytes (r, ROUND_UP (len, 4) - len);
+ }
- /* Make a file label only on the condition that the given label is
- not all spaces or nulls. */
- {
- int i;
-
- for (i = sizeof hdr.file_label - 1; i >= 0; i--)
- {
- if (!c_isspace ((unsigned char) hdr.file_label[i])
- && hdr.file_label[i] != 0)
- {
- char *label = xmalloc (i + 2);
- memcpy (label, hdr.file_label, i + 1);
- label[i + 1] = 0;
- dict_set_label (dict, label);
- free (label);
- break;
- }
- }
- }
-
- if (info)
+ /* Set missing values. */
+ if (missing_value_code != 0)
{
- char *cp;
-
- memcpy (info->creation_date, hdr.creation_date, 9);
- info->creation_date[9] = 0;
-
- memcpy (info->creation_time, hdr.creation_time, 8);
- info->creation_time[8] = 0;
+ struct missing_values mv;
+ int i;
-#ifdef WORDS_BIGENDIAN
- info->big_endian = !r->reverse_endian;
-#else
- info->big_endian = r->reverse_endian;
-#endif
+ mv_init (&mv, var_get_width (var));
+ if (var_is_numeric (var))
+ {
+ if (missing_value_code < -3 || missing_value_code > 3
+ || missing_value_code == -1)
+ sys_error (r, _("Numeric missing value indicator field is not "
+ "-3, -2, 0, 1, 2, or 3."));
+ if (missing_value_code < 0)
+ {
+ double low = read_float (r);
+ double high = read_float (r);
+ mv_add_num_range (&mv, low, high);
+ missing_value_code = -missing_value_code - 2;
+ }
+ for (i = 0; i < missing_value_code; i++)
+ mv_add_num (&mv, read_float (r));
+ }
+ else if (var_get_width (var) <= MAX_SHORT_STRING)
+ {
+ if (missing_value_code < 1 || missing_value_code > 3)
+ sys_error (r, _("String missing value indicator field is not "
+ "0, 1, 2, or 3."));
+ for (i = 0; i < missing_value_code; i++)
+ {
+ char string[9];
+ read_string (r, string, sizeof string);
+ mv_add_str (&mv, string);
+ }
+ }
+ else
+ sys_error (r, _("Long string variable %s may not have missing "
+ "values."), name);
+ var_set_missing_values (var, &mv);
+ }
- info->compressed = hdr.compress;
+ /* Set formats. */
+ parse_format_spec (r, print_format, PRINT_FORMAT, var, format_warning_cnt);
+ parse_format_spec (r, write_format, WRITE_FORMAT, var, format_warning_cnt);
- info->case_cnt = hdr.case_cnt;
+ /* Account for values.
+ Skip long string continuation records, if any. */
+ nv = width == 0 ? 1 : DIV_RND_UP (width, 8);
+ r->oct_cnt += nv;
+ if (width > 8)
+ {
+ int i;
- for (cp = &prod_name[skip_amt]; cp < &prod_name[60]; cp++)
- if (c_isgraph ((unsigned char) *cp))
- break;
- strcpy (info->product, cp);
+ for (i = 1; i < nv; i++)
+ {
+ /* Check for record type 2 and width -1. */
+ if (read_int (r) != 2 || read_int (r) != -1)
+ sys_error (r, _("Missing string continuation record."));
+
+ /* Skip and ignore remaining continuation data. */
+ has_variable_label = read_int (r);
+ missing_value_code = read_int (r);
+ print_format = read_int (r);
+ write_format = read_int (r);
+ read_string (r, name, sizeof name);
+
+ /* Variable label fields on continuation records have
+ been spotted in system files created by "SPSS Power
+ Macintosh Release 6.1". */
+ if (has_variable_label)
+ skip_bytes (r, ROUND_UP (read_int (r), 4));
+ }
}
-
- return 1;
-
-error:
- return 0;
}
-/* Reads most of the dictionary from file H; also fills in the
- associated VAR_BY_IDX array. */
-static int
-read_variables (struct sfm_reader *r,
- struct dictionary *dict, struct variable ***var_by_idx)
+/* Translates the format spec from sysfile format to internal
+ format. */
+static void
+parse_format_spec (struct sfm_reader *r, unsigned int s,
+ enum which_format which, struct variable *v,
+ int *format_warning_cnt)
{
- int i;
+ const int max_format_warnings = 8;
+ struct fmt_spec f;
+ uint8_t raw_type = s >> 16;
+ uint8_t w = s >> 8;
+ uint8_t d = s;
- struct sysfile_variable sv; /* Disk buffer. */
- int long_string_count = 0; /* # of long string continuation
- records still expected. */
- int next_value = 0; /* Index to next `value' structure. */
+ bool ok;
- assert(r);
+ if (!fmt_from_io (raw_type, &f.type))
+ sys_error (r, _("Unknown variable format %"PRIu8"."), raw_type);
+ f.w = w;
+ f.d = d;
- *var_by_idx = 0;
+ msg_disable ();
+ ok = fmt_check_output (&f) && fmt_check_width_compat (&f, var_get_width (v));
+ msg_enable ();
- /* Pre-allocate variables. */
- if (r->value_cnt != -1)
+ if (ok)
{
- *var_by_idx = xnmalloc (r->value_cnt, sizeof **var_by_idx);
- r->vars = xnmalloc (r->value_cnt, sizeof *r->vars);
+ if (which == PRINT_FORMAT)
+ var_set_print_format (v, &f);
+ else
+ var_set_write_format (v, &f);
}
+ else if (*++format_warning_cnt <= max_format_warnings)
+ {
+ char fmt_string[FMT_STRING_LEN_MAX + 1];
+ sys_warn (r, _("%s variable %s has invalid %s format %s."),
+ var_is_numeric (v) ? _("Numeric") : _("String"),
+ var_get_name (v),
+ which == PRINT_FORMAT ? _("print") : _("write"),
+ fmt_to_string (&f, fmt_string));
+
+ if (*format_warning_cnt == max_format_warnings)
+ sys_warn (r, _("Suppressing further invalid format warnings."));
+ }
+}
-
- /* Read in the entry for each variable and use the info to
- initialize the dictionary. */
- for (i = 0; ; ++i)
+/* Sets the weighting variable in DICT to the variable
+ corresponding to the given 1-based VALUE_IDX, if VALUE_IDX is
+ nonzero. */
+static void
+setup_weight (struct sfm_reader *r, int weight_idx,
+ struct variable **var_by_value_idx, struct dictionary *dict)
+{
+ if (weight_idx != 0)
{
- struct variable *vv;
- char name[SHORT_NAME_LEN + 1];
- int nv;
- int j;
+ struct variable *weight_var
+ = lookup_var_by_value_idx (r, var_by_value_idx, weight_idx);
+ if (var_is_numeric (weight_var))
+ dict_set_weight (dict, weight_var);
+ else
+ sys_error (r, _("Weighting variable must be numeric."));
+ }
+}
- if ( r->value_cnt != -1 && i >= r->value_cnt )
- break;
+/* Reads a document record, type 6, from system file R, and sets up
+ the documents and n_documents fields in the associated
+ dictionary. */
+static void
+read_documents (struct sfm_reader *r, struct dictionary *dict)
+{
+ int line_cnt;
+ char *documents;
- assertive_buf_read (r, &sv, sizeof sv, 0);
+ if (dict_get_documents (dict) != NULL)
+ sys_error (r, _("Multiple type 6 (document) records."));
- if (r->reverse_endian)
- {
- bswap_int32 (&sv.rec_type);
- bswap_int32 (&sv.type);
- bswap_int32 (&sv.has_var_label);
- bswap_int32 (&sv.n_missing_values);
- bswap_int32 (&sv.print);
- bswap_int32 (&sv.write);
- }
+ line_cnt = read_int (r);
+ if (line_cnt <= 0)
+ sys_error (r, _("Number of document lines (%d) "
+ "must be greater than 0."), line_cnt);
- /* We've come to the end of the variable entries */
- if (sv.rec_type != 2)
- {
- buf_unread(r, sizeof sv);
- r->value_cnt = i;
- break;
- }
+ documents = pool_nmalloc (r->pool, line_cnt + 1, DOC_LINE_LENGTH);
+ read_string (r, documents, DOC_LINE_LENGTH * line_cnt + 1);
+ if (strlen (documents) == DOC_LINE_LENGTH * line_cnt)
+ dict_set_documents (dict, documents);
+ else
+ sys_error (r, _("Document line contains null byte."));
+ pool_free (r->pool, documents);
+}
- if ( -1 == r->value_cnt )
- {
- *var_by_idx = xnrealloc (*var_by_idx, i + 1, sizeof **var_by_idx);
- r->vars = xnrealloc (r->vars, i + 1, sizeof *r->vars);
- }
+/* Read a type 7 extension record. */
+static void
+read_extension_record (struct sfm_reader *r, struct dictionary *dict,
+ struct sfm_read_info *info)
+{
+ int subtype = read_int (r);
+ size_t size = read_int (r);
+ size_t count = read_int (r);
+ size_t bytes = size * count;
+
+ /* Check that SIZE * COUNT + 1 doesn't overflow. Adding 1
+ allows an extra byte for a null terminator, used by some
+ extension processing routines. */
+ if (size != 0 && size_overflow_p (xsum (1, xtimes (count, size))))
+ sys_error (r, "Record type 7 subtype %d too large.", subtype);
+
+ switch (subtype)
+ {
+ case 3:
+ read_machine_integer_info (r, size, count, info);
+ return;
+
+ case 4:
+ read_machine_float_info (r, size, count);
+ return;
+
+ case 5:
+ /* Variable sets information. We don't use these yet.
+ They only apply to GUIs; see VARSETS on the APPLY
+ DICTIONARY command in SPSS documentation. */
+ break;
+
+ case 6:
+ /* DATE variable information. We don't use it yet, but we
+ should. */
+ break;
+
+ case 7:
+ /* Unknown purpose. */
+ break;
+
+ case 11:
+ read_display_parameters (r, size, count, dict);
+ return;
+
+ case 13:
+ read_long_var_name_map (r, size, count, dict);
+ return;
+
+ case 14:
+ read_long_string_map (r, size, count, dict);
+ return;
+
+ case 16:
+ /* New in SPSS v14? Unknown purpose. */
+ break;
+
+ case 17:
+ /* Text field that defines variable attributes. New in
+ SPSS 14. */
+ break;
+
+ default:
+ sys_warn (r, _("Unrecognized record type 7, subtype %d."), subtype);
+ break;
+ }
- /* If there was a long string previously, make sure that the
- continuations are present; otherwise make sure there aren't
- any. */
- if (long_string_count)
- {
- if (sv.type != -1)
- lose ((ME, _("%s: position %d: String variable does not have "
- "proper number of continuation records."),
- fh_get_file_name (r->fh), i));
+ skip_bytes (r, bytes);
+}
+/* Read record type 7, subtype 3. */
+static void
+read_machine_integer_info (struct sfm_reader *r, size_t size, size_t count,
+ struct sfm_read_info *info)
+{
+ int version_major = read_int (r);
+ int version_minor = read_int (r);
+ int version_revision = read_int (r);
+ int machine_code UNUSED = read_int (r);
+ int float_representation = read_int (r);
+ int compression_code UNUSED = read_int (r);
+ int integer_representation = read_int (r);
+ int character_code UNUSED = read_int (r);
+
+ int expected_float_format;
+ int expected_integer_format;
+
+ if (size != 4 || count != 8)
+ sys_error (r, _("Bad size (%zu) or count (%zu) field on record type 7, "
+ "subtype 3."),
+ size, count);
+
+ /* Save version info. */
+ info->version_major = version_major;
+ info->version_minor = version_minor;
+ info->version_revision = version_revision;
+
+ /* Check floating point format. */
+ if (r->float_format == FLOAT_IEEE_DOUBLE_BE
+ || r->float_format == FLOAT_IEEE_DOUBLE_LE)
+ expected_float_format = 1;
+ else if (r->float_format == FLOAT_Z_LONG)
+ expected_float_format = 2;
+ else if (r->float_format == FLOAT_VAX_G || r->float_format == FLOAT_VAX_D)
+ expected_float_format = 3;
+ else
+ NOT_REACHED ();
+ if (float_representation != expected_float_format)
+ sys_error (r, _("Floating-point representation indicated by "
+ "system file (%d) differs from expected (%d)."),
+ r->float_format, expected_float_format);
+
+ /* Check integer format. */
+ if (r->integer_format == INTEGER_MSB_FIRST)
+ expected_integer_format = 1;
+ else if (r->integer_format == INTEGER_LSB_FIRST)
+ expected_integer_format = 2;
+ else
+ NOT_REACHED ();
+ if (integer_representation != expected_integer_format)
+ {
+ static const char *endian[] = {N_("little-endian"), N_("big-endian")};
+ sys_warn (r, _("Integer format indicated by system file (%s) "
+ "differs from expected (%s)."),
+ gettext (endian[integer_representation == 1]),
+ gettext (endian[expected_integer_format == 1]));
+ }
+}
- r->vars[i].width = -1;
- (*var_by_idx)[i] = NULL;
- long_string_count--;
- continue;
- }
- else if (sv.type == -1)
- lose ((ME, _("%s: position %d: Superfluous long string continuation "
- "record."),
- fh_get_file_name (r->fh), i));
-
- /* Check fields for validity. */
- if (sv.type < 0 || sv.type > 255)
- lose ((ME, _("%s: position %d: Bad variable type code %d."),
- fh_get_file_name (r->fh), i, sv.type));
- if (sv.has_var_label != 0 && sv.has_var_label != 1)
- lose ((ME, _("%s: position %d: Variable label indicator field is not "
- "0 or 1."), fh_get_file_name (r->fh), i));
- if (sv.n_missing_values < -3 || sv.n_missing_values > 3
- || sv.n_missing_values == -1)
- lose ((ME, _("%s: position %d: Missing value indicator field is not "
- "-3, -2, 0, 1, 2, or 3."), fh_get_file_name (r->fh), i));
-
- /* Copy first character of variable name. */
- if (sv.name[0] == '@' || sv.name[0] == '#')
- lose ((ME, _("%s: position %d: Variable name begins with invalid "
- "character."),
- fh_get_file_name (r->fh), i));
-
- name[0] = sv.name[0];
-
- /* Copy remaining characters of variable name. */
- for (j = 1; j < SHORT_NAME_LEN; j++)
- {
- int c = (unsigned char) sv.name[j];
+/* Read record type 7, subtype 4. */
+static void
+read_machine_float_info (struct sfm_reader *r, size_t size, size_t count)
+{
+ double sysmis = read_float (r);
+ double highest = read_float (r);
+ double lowest = read_float (r);
+
+ if (size != 8 || count != 3)
+ sys_error (r, _("Bad size (%zu) or count (%zu) on extension 4."),
+ size, count);
+
+ if (sysmis != SYSMIS)
+ sys_warn (r, _("File specifies unexpected value %g as SYSMIS."), sysmis);
+ if (highest != HIGHEST)
+ sys_warn (r, _("File specifies unexpected value %g as HIGHEST."), highest);
+ if (lowest != LOWEST)
+ sys_warn (r, _("File specifies unexpected value %g as LOWEST."), lowest);
+}
- if (c == ' ')
- break;
- else
- name[j] = c;
- }
- name[j] = 0;
+/* Read record type 7, subtype 11, which specifies how variables
+ should be displayed in GUI environments. */
+static void
+read_display_parameters (struct sfm_reader *r, size_t size, size_t count,
+ struct dictionary *dict)
+{
+ const size_t n_vars = count / 3 ;
+ bool warned = false;
+ int i;
- if ( ! var_is_plausible_name(name, false) )
- lose ((ME, _("%s: Invalid variable name `%s' within system file."),
- fh_get_file_name (r->fh), name));
+ if (count % 3 || n_vars != dict_get_var_cnt (dict))
+ sys_error (r, _("Bad size (%zu) or count (%zu) on extension 11."),
+ size, count);
- /* Create variable. */
- vv = (*var_by_idx)[i] = dict_create_var (dict, name, sv.type);
- if (vv == NULL)
- lose ((ME, _("%s: Duplicate variable name `%s' within system file."),
- fh_get_file_name (r->fh), name));
+ for (i = 0; i < n_vars; ++i)
+ {
+ struct variable *v = dict_get_var (dict, i);
+ int measure = read_int (r);
+ int width = read_int (r);
+ int align = read_int (r);
+
+ /* SPSS 14 sometimes seems to set string variables' measure
+ to zero. */
+ if (0 == measure && var_is_alpha (v))
+ measure = 1;
+
+ /* Older versions (SPSS 9.0) sometimes set the display width
+ to zero. This causes confusion especially in the GUI */
+ if (0 == width)
+ width = 8;
+
+ if (measure < 1 || measure > 3 || align < 0 || align > 2)
+ {
+ if (!warned)
+ sys_warn (r, _("Invalid variable display parameters. "
+ "Default parameters substituted."));
+ warned = true;
+ continue;
+ }
- var_set_short_name (vv, vv->name);
+ var_set_measure (v, (measure == 1 ? MEASURE_NOMINAL
+ : measure == 2 ? MEASURE_ORDINAL
+ : MEASURE_SCALE));
+ var_set_display_width (v, width);
+ var_set_alignment (v, (align == 0 ? ALIGN_LEFT
+ : align == 1 ? ALIGN_RIGHT
+ : ALIGN_CENTRE));
+ }
+}
- /* Case reading data. */
- nv = sv.type == 0 ? 1 : DIV_RND_UP (sv.type, sizeof (flt64));
- long_string_count = nv - 1;
- next_value += nv;
+/* Reads record type 7, subtype 13, which gives the long name
+ that corresponds to each short name. Modifies variable names
+ in DICT accordingly. */
+static void
+read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count,
+ struct dictionary *dict)
+{
+ struct variable_to_value_map *map;
+ struct variable *var;
+ char *long_name;
+ int warning_cnt = 0;
+
+ map = open_variable_to_value_map (r, size * count);
+ while (read_variable_to_value_map (r, dict, map, &var, &long_name,
+ &warning_cnt))
+ {
+ char **short_names;
+ size_t short_name_cnt;
+ size_t i;
- /* Get variable label, if any. */
- if (sv.has_var_label == 1)
- {
- /* Disk buffer. */
- int32_t len;
-
- /* Read length of label. */
- assertive_buf_read (r, &len, sizeof len, 0);
- if (r->reverse_endian)
- bswap_int32 (&len);
-
- /* Check len. */
- if (len < 0 || len > 255)
- lose ((ME, _("%s: Variable %s indicates variable label of invalid "
- "length %d."),
- fh_get_file_name (r->fh), vv->name, len));
-
- if ( len != 0 )
- {
- /* Read label into variable structure. */
- vv->label = buf_read (r, NULL, ROUND_UP (len, sizeof (int32_t)), len + 1);
- if (vv->label == NULL)
- goto error;
- vv->label[len] = '\0';
- }
- }
+ /* Validate long name. */
+ if (!var_is_valid_name (long_name, false))
+ {
+ sys_warn (r, _("Long variable mapping from %s to invalid "
+ "variable name `%s'."),
+ var_get_name (var), long_name);
+ continue;
+ }
- /* Set missing values. */
- if (sv.n_missing_values != 0)
- {
- flt64 mv[3];
- int mv_cnt = abs (sv.n_missing_values);
-
- if (vv->width > MAX_SHORT_STRING)
- lose ((ME, _("%s: Long string variable %s may not have missing "
- "values."),
- fh_get_file_name (r->fh), vv->name));
-
- assertive_buf_read (r, mv, sizeof *mv * mv_cnt, 0);
-
- if (r->reverse_endian && vv->type == NUMERIC)
- for (j = 0; j < mv_cnt; j++)
- bswap_flt64 (&mv[j]);
-
- if (sv.n_missing_values > 0)
- {
- for (j = 0; j < sv.n_missing_values; j++)
- if (vv->type == NUMERIC)
- mv_add_num (&vv->miss, mv[j]);
- else
- mv_add_str (&vv->miss, (char *) &mv[j]);
- }
- else
- {
- if (vv->type == ALPHA)
- lose ((ME, _("%s: String variable %s may not have missing "
- "values specified as a range."),
- fh_get_file_name (r->fh), vv->name));
-
- if (mv[0] == r->lowest)
- mv_add_num_range (&vv->miss, LOWEST, mv[1]);
- else if (mv[1] == r->highest)
- mv_add_num_range (&vv->miss, mv[0], HIGHEST);
- else
- mv_add_num_range (&vv->miss, mv[0], mv[1]);
-
- if (sv.n_missing_values == -3)
- mv_add_num (&vv->miss, mv[2]);
- }
- }
+ /* Identify any duplicates. */
+ if (strcasecmp (var_get_short_name (var, 0), long_name)
+ && dict_lookup_var (dict, long_name) != NULL)
+ {
+ sys_warn (r, _("Duplicate long variable name `%s' "
+ "within system file."), long_name);
+ continue;
+ }
- if (!parse_format_spec (r, sv.print, &vv->print, vv)
- || !parse_format_spec (r, sv.write, &vv->write, vv))
- goto error;
+ /* Renaming a variable may clear its short names, but we
+ want to retain them, so we save them and re-set them
+ afterward. */
+ short_name_cnt = var_get_short_name_cnt (var);
+ short_names = xnmalloc (short_name_cnt, sizeof *short_names);
+ for (i = 0; i < short_name_cnt; i++)
+ {
+ const char *s = var_get_short_name (var, i);
+ short_names[i] = s != NULL ? xstrdup (s) : NULL;
+ }
- r->vars[i].width = vv->width;
- r->vars[i].fv = vv->fv;
+ /* Set long name. */
+ dict_rename_var (dict, var, long_name);
+ /* Restore short names. */
+ for (i = 0; i < short_name_cnt; i++)
+ {
+ var_set_short_name (var, i, short_names[i]);
+ free (short_names[i]);
+ }
+ free (short_names);
}
-
- /* Some consistency checks. */
- if (long_string_count != 0)
- lose ((ME, _("%s: Long string continuation records omitted at end of "
- "dictionary."),
- fh_get_file_name (r->fh)));
-
- if (next_value != r->value_cnt)
- corrupt_msg(MW, _("%s: System file header indicates %d variable positions but "
- "%d were read from file."),
- fh_get_file_name (r->fh), r->value_cnt, next_value);
-
-
- return 1;
-
-error:
- return 0;
+ close_variable_to_value_map (r, map);
+ r->has_long_var_names = true;
}
-/* Translates the format spec from sysfile format to internal
- format. */
-static int
-parse_format_spec (struct sfm_reader *r, int32_t s,
- struct fmt_spec *f, const struct variable *v)
+/* Reads record type 7, subtype 14, which gives the real length
+ of each very long string. Rearranges DICT accordingly. */
+static void
+read_long_string_map (struct sfm_reader *r, size_t size, size_t count,
+ struct dictionary *dict)
{
- f->type = translate_fmt ((s >> 16) & 0xff);
- if (f->type == -1)
- lose ((ME, _("%s: Bad format specifier byte (%d)."),
- fh_get_file_name (r->fh), (s >> 16) & 0xff));
- f->w = (s >> 8) & 0xff;
- f->d = s & 0xff;
-
- if ((v->type == ALPHA) ^ ((formats[f->type].cat & FCAT_STRING) != 0))
- lose ((ME, _("%s: %s variable %s has %s format specifier %s."),
- fh_get_file_name (r->fh),
- v->type == ALPHA ? _("String") : _("Numeric"),
- v->name,
- formats[f->type].cat & FCAT_STRING ? _("string") : _("numeric"),
- formats[f->type].name));
-
- if (!check_output_specifier (f, false)
- || !check_specifier_width (f, v->width, false))
+ struct variable_to_value_map *map;
+ struct variable *var;
+ char *length_s;
+ int warning_cnt = 0;
+
+ map = open_variable_to_value_map (r, size * count);
+ while (read_variable_to_value_map (r, dict, map, &var, &length_s,
+ &warning_cnt))
{
- msg (ME, _("%s variable %s has invalid format specifier %s."),
- v->type == NUMERIC ? _("Numeric") : _("String"),
- v->name, fmt_to_string (f));
- *f = v->type == NUMERIC ? f8_2 : make_output_format (FMT_A, v->width, 0);
- }
- return 1;
+ size_t idx = var_get_dict_index (var);
+ long int length;
+ int segment_cnt;
+ int i;
-error:
- return 0;
+ /* Get length. */
+ length = strtol (length_s, NULL, 10);
+ if (length < 1 || length > MAX_STRING)
+ {
+ sys_warn (r, _("%s listed as string of invalid length %s "
+ "in very length string record."),
+ var_get_name (var), length_s);
+ continue;
+ }
+
+ /* Check segments. */
+ segment_cnt = sfm_width_to_segments (length);
+ if (segment_cnt == 1)
+ {
+ sys_warn (r, _("%s listed in very long string record with width %s, "
+ "which requires only one segment."),
+ var_get_name (var), length_s);
+ continue;
+ }
+ if (idx + segment_cnt > dict_get_var_cnt (dict))
+ sys_error (r, _("Very long string %s overflows dictionary."),
+ var_get_name (var));
+
+ /* Get the short names from the segments and check their
+ lengths. */
+ for (i = 0; i < segment_cnt; i++)
+ {
+ struct variable *seg = dict_get_var (dict, idx + i);
+ int alloc_width = sfm_segment_alloc_width (length, i);
+ int width = var_get_width (seg);
+
+ if (i > 0)
+ var_set_short_name (var, i, var_get_short_name (seg, 0));
+ if (ROUND_UP (width, 8) != ROUND_UP (alloc_width, 8))
+ sys_error (r, _("Very long string with width %ld has segment %d "
+ "of width %d (expected %d)"),
+ length, i, width, alloc_width);
+ }
+ dict_delete_consecutive_vars (dict, idx + 1, segment_cnt - 1);
+ var_set_width (var, length);
+ }
+ close_variable_to_value_map (r, map);
+ dict_compact_values (dict);
}
/* Reads value labels from sysfile H and inserts them into the
associated dictionary. */
-int
+static void
read_value_labels (struct sfm_reader *r,
- struct dictionary *dict, struct variable **var_by_idx)
+ struct dictionary *dict, struct variable **var_by_value_idx)
{
- struct label
+ struct pool *subpool;
+
+ struct label
{
char raw_value[8]; /* Value as uninterpreted bytes. */
union value value; /* Value. */
};
struct label *labels = NULL;
- int32_t n_labels; /* Number of labels. */
+ int label_cnt; /* Number of labels. */
struct variable **var = NULL; /* Associated variables. */
- int32_t n_vars; /* Number of associated variables. */
+ int var_cnt; /* Number of associated variables. */
int i;
- /* First step: read the contents of the type 3 record and record its
- contents. Note that we can't do much with the data since we
- don't know yet whether it is of numeric or string type. */
+ subpool = pool_create_subpool (r->pool);
+
+ /* Read the type 3 record and record its contents. We can't do
+ much with the data yet because we don't know whether it is
+ of numeric or string type. */
/* Read number of labels. */
- assertive_buf_read (r, &n_labels, sizeof n_labels, 0);
- if (r->reverse_endian)
- bswap_int32 (&n_labels);
-
- if ( n_labels >= ((int32_t) ~0) / sizeof *labels)
- {
- corrupt_msg(MW, _("%s: Invalid number of labels: %d. Ignoring labels."),
- fh_get_file_name (r->fh), n_labels);
- n_labels = 0;
- }
+ label_cnt = read_int (r);
- /* Allocate memory. */
- labels = xcalloc (n_labels, sizeof *labels);
- for (i = 0; i < n_labels; i++)
- labels[i].label = NULL;
+ if (size_overflow_p (xtimes (label_cnt, sizeof *labels)))
+ {
+ sys_warn (r, _("Invalid number of labels: %d. Ignoring labels."),
+ label_cnt);
+ label_cnt = 0;
+ }
/* Read each value/label tuple into labels[]. */
- for (i = 0; i < n_labels; i++)
+ labels = pool_nalloc (subpool, label_cnt, sizeof *labels);
+ for (i = 0; i < label_cnt; i++)
{
struct label *label = labels + i;
unsigned char label_len;
size_t padded_len;
/* Read value. */
- assertive_buf_read (r, label->raw_value, sizeof label->raw_value, 0);
+ read_bytes (r, label->raw_value, sizeof label->raw_value);
/* Read label length. */
- assertive_buf_read (r, &label_len, sizeof label_len, 0);
- padded_len = ROUND_UP (label_len + 1, sizeof (flt64));
+ read_bytes (r, &label_len, sizeof label_len);
+ padded_len = ROUND_UP (label_len + 1, 8);
/* Read label, padding. */
- label->label = xmalloc (padded_len + 1);
- assertive_buf_read (r, label->label, padded_len - 1, 0);
+ label->label = pool_alloc (subpool, padded_len + 1);
+ read_bytes (r, label->label, padded_len - 1);
label->label[label_len] = 0;
}
- /* Second step: Read the type 4 record that has the list of
- variables to which the value labels are to be applied. */
+ /* Now, read the type 4 record that has the list of variables
+ to which the value labels are to be applied. */
/* Read record type of type 4 record. */
- {
- int32_t rec_type;
-
- assertive_buf_read (r, &rec_type, sizeof rec_type, 0);
- if (r->reverse_endian)
- bswap_int32 (&rec_type);
-
- if (rec_type != 4)
- lose ((ME, _("%s: Variable index record (type 4) does not immediately "
- "follow value label record (type 3) as it should."),
- fh_get_file_name (r->fh)));
- }
+ if (read_int (r) != 4)
+ sys_error (r, _("Variable index record (type 4) does not immediately "
+ "follow value label record (type 3) as it should."));
/* Read number of variables associated with value label from type 4
record. */
- assertive_buf_read (r, &n_vars, sizeof n_vars, 0);
- if (r->reverse_endian)
- bswap_int32 (&n_vars);
- if (n_vars < 1 || n_vars > dict_get_var_cnt (dict))
- lose ((ME, _("%s: Number of variables associated with a value label (%d) "
- "is not between 1 and the number of variables (%d)."),
- fh_get_file_name (r->fh), n_vars, dict_get_var_cnt (dict)));
+ var_cnt = read_int (r);
+ if (var_cnt < 1 || var_cnt > dict_get_var_cnt (dict))
+ sys_error (r, _("Number of variables associated with a value label (%d) "
+ "is not between 1 and the number of variables (%zu)."),
+ var_cnt, dict_get_var_cnt (dict));
/* Read the list of variables. */
- var = xnmalloc (n_vars, sizeof *var);
- for (i = 0; i < n_vars; i++)
+ var = pool_nalloc (subpool, var_cnt, sizeof *var);
+ for (i = 0; i < var_cnt; i++)
{
- int32_t var_idx;
- struct variable *v;
-
- /* Read variable index, check range. */
- assertive_buf_read (r, &var_idx, sizeof var_idx, 0);
- if (r->reverse_endian)
- bswap_int32 (&var_idx);
- if (var_idx < 1 || var_idx > r->value_cnt)
- lose ((ME, _("%s: Variable index associated with value label (%d) is "
- "not between 1 and the number of values (%d)."),
- fh_get_file_name (r->fh), var_idx, r->value_cnt));
-
- /* Make sure it's a real variable. */
- v = var_by_idx[var_idx - 1];
- if (v == NULL)
- lose ((ME, _("%s: Variable index associated with value label (%d) "
- "refers to a continuation of a string variable, not to "
- "an actual variable."),
- fh_get_file_name (r->fh), var_idx));
- if (v->type == ALPHA && v->width > MAX_SHORT_STRING)
- lose ((ME, _("%s: Value labels are not allowed on long string "
- "variables (%s)."),
- fh_get_file_name (r->fh), v->name));
-
- /* Add it to the list of variables. */
- var[i] = v;
+ var[i] = lookup_var_by_value_idx (r, var_by_value_idx, read_int (r));
+ if (var_is_long_string (var[i]))
+ sys_error (r, _("Value labels are not allowed on long string "
+ "variables (%s)."), var_get_name (var[i]));
}
/* Type check the variables. */
- for (i = 1; i < n_vars; i++)
- if (var[i]->type != var[0]->type)
- lose ((ME, _("%s: Variables associated with value label are not all of "
- "identical type. Variable %s has %s type, but variable "
- "%s has %s type."),
- fh_get_file_name (r->fh),
- var[0]->name, var[0]->type == ALPHA ? _("string") : _("numeric"),
- var[i]->name, var[i]->type == ALPHA ? _("string") : _("numeric")));
+ for (i = 1; i < var_cnt; i++)
+ if (var_get_type (var[i]) != var_get_type (var[0]))
+ sys_error (r, _("Variables associated with value label are not all of "
+ "identical type. Variable %s is %s, but variable "
+ "%s is %s."),
+ var_get_name (var[0]),
+ var_is_numeric (var[0]) ? _("numeric") : _("string"),
+ var_get_name (var[i]),
+ var_is_numeric (var[i]) ? _("numeric") : _("string"));
/* Fill in labels[].value, now that we know the desired type. */
- for (i = 0; i < n_labels; i++)
+ for (i = 0; i < label_cnt; i++)
{
struct label *label = labels + i;
-
- if (var[0]->type == ALPHA)
- {
- const int copy_len = min (sizeof label->raw_value,
- sizeof label->label);
- memcpy (label->value.s, label->raw_value, copy_len);
- } else {
- flt64 f;
- assert (sizeof f == sizeof label->raw_value);
- memcpy (&f, label->raw_value, sizeof f);
- if (r->reverse_endian)
- bswap_flt64 (&f);
- label->value.f = f;
- }
+
+ if (var_is_alpha (var[0]))
+ buf_copy_rpad (label->value.s, sizeof label->value.s,
+ label->raw_value, sizeof label->raw_value);
+ else
+ label->value.f = float_get_double (r->float_format, label->raw_value);
}
-
- /* Assign the value_label's to each variable. */
- for (i = 0; i < n_vars; i++)
+
+ /* Assign the `value_label's to each variable. */
+ for (i = 0; i < var_cnt; i++)
{
struct variable *v = var[i];
int j;
/* Add each label to the variable. */
- for (j = 0; j < n_labels; j++)
+ for (j = 0; j < label_cnt; j++)
{
- struct label *label = labels + j;
- if (!val_labs_replace (v->val_labs, label->value, label->label))
- continue;
-
- if (var[0]->type == NUMERIC)
- msg (MW, _("%s: File contains duplicate label for value %g for "
- "variable %s."),
- fh_get_file_name (r->fh), label->value.f, v->name);
- else
- msg (MW, _("%s: File contains duplicate label for value `%.*s' "
- "for variable %s."),
- fh_get_file_name (r->fh), v->width, label->value.s, v->name);
+ struct label *label = &labels[j];
+ if (!var_add_value_label (v, &label->value, label->label))
+ {
+ if (var_is_numeric (var[0]))
+ sys_warn (r, _("Duplicate value label for %g on %s."),
+ label->value.f, var_get_name (v));
+ else
+ sys_warn (r, _("Duplicate value label for \"%.*s\" on %s."),
+ var_get_width (v), label->value.s,
+ var_get_name (v));
+ }
}
}
- for (i = 0; i < n_labels; i++)
- free (labels[i].label);
- free (labels);
- free (var);
- return 1;
-
-error:
- if (labels)
- {
- for (i = 0; i < n_labels; i++)
- free (labels[i].label);
- free (labels);
- }
- free (var);
- return 0;
+ pool_destroy (subpool);
}
-
-/* Reads BYTE_CNT bytes from the file represented by H. If BUF is
- non-NULL, uses that as the buffer; otherwise allocates at least
- MIN_ALLOC bytes. Returns a pointer to the buffer on success, NULL
- on failure. */
-static void *
-buf_read (struct sfm_reader *r, void *buf, size_t byte_cnt, size_t min_alloc)
+\f
+/* Case reader. */
+
+static void partial_record (struct sfm_reader *r)
+ NO_RETURN;
+
+static void read_error (struct casereader *, const struct sfm_reader *);
+
+static bool read_case_number (struct sfm_reader *, double *);
+static bool read_case_string (struct sfm_reader *, char *, size_t);
+static int read_opcode (struct sfm_reader *);
+static bool read_compressed_number (struct sfm_reader *, double *);
+static bool read_compressed_string (struct sfm_reader *, char *);
+static bool read_whole_strings (struct sfm_reader *, char *, size_t);
+static bool skip_whole_strings (struct sfm_reader *, size_t);
+
+/* Reads one case from READER's file into C. Returns true only
+ if successful. */
+static bool
+sys_file_casereader_read (struct casereader *reader, void *r_,
+ struct ccase *c)
{
- assert (r);
+ struct sfm_reader *r = r_;
+ int i;
- if (buf == NULL && byte_cnt > 0 )
- buf = xmalloc (max (byte_cnt, min_alloc));
+ if (r->error)
+ return false;
- if ( byte_cnt == 0 )
- return buf;
+ case_create (c, r->value_cnt);
+ if (setjmp (r->bail_out))
+ {
+ casereader_force_error (reader);
+ case_destroy (c);
+ return false;
+ }
-
- if (1 != fread (buf, byte_cnt, 1, r->file))
+ for (i = 0; i < r->sfm_var_cnt; i++)
{
- if (ferror (r->file))
- msg (ME, _("%s: Reading system file: %s."),
- fh_get_file_name (r->fh), strerror (errno));
+ struct sfm_var *sv = &r->sfm_vars[i];
+ union value *v = case_data_rw_idx (c, sv->case_index);
+
+ if (sv->width == 0)
+ {
+ if (!read_case_number (r, &v->f))
+ goto eof;
+ }
else
- corrupt_msg (ME, _("%s: Unexpected end of file."),
- fh_get_file_name (r->fh));
- r->ok = false;
- return NULL;
+ {
+ if (!read_case_string (r, v->s + sv->offset, sv->width))
+ goto eof;
+ if (!skip_whole_strings (r, ROUND_DOWN (sv->padding, 8)))
+ partial_record (r);
+ }
}
- return buf;
+ return true;
+
+eof:
+ case_destroy (c);
+ if (i != 0)
+ partial_record (r);
+ if (r->case_cnt != -1)
+ read_error (reader, r);
+ return false;
}
-/* Winds the reader BYTE_CNT bytes back in the reader stream. */
-void
-buf_unread(struct sfm_reader *r, size_t byte_cnt)
+/* Issues an error that R ends in a partial record. */
+static void
+partial_record (struct sfm_reader *r)
{
- assert(byte_cnt > 0);
-
- if ( 0 != fseek(r->file, -byte_cnt, SEEK_CUR))
- {
- msg (ME, _("%s: Seeking system file: %s."),
- fh_get_file_name (r->fh), strerror (errno));
- }
+ sys_error (r, _("File ends in partial case."));
}
-/* Reads a document record, type 6, from system file R, and sets up
- the documents and n_documents fields in the associated
- dictionary. */
-static int
-read_documents (struct sfm_reader *r, struct dictionary *dict)
+/* Issues an error that an unspecified error occurred SFM, and
+ marks R tainted. */
+static void
+read_error (struct casereader *r, const struct sfm_reader *sfm)
{
- int32_t line_cnt;
- char *documents;
-
- if (dict_get_documents (dict) != NULL)
- lose ((ME, _("%s: System file contains multiple "
- "type 6 (document) records."),
- fh_get_file_name (r->fh)));
+ msg (ME, _("Error reading case from file %s."), fh_get_name (sfm->fh));
+ casereader_force_error (r);
+}
- assertive_buf_read (r, &line_cnt, sizeof line_cnt, 0);
- if (line_cnt <= 0)
- lose ((ME, _("%s: Number of document lines (%ld) "
- "must be greater than 0."),
- fh_get_file_name (r->fh), (long) line_cnt));
-
- documents = buf_read (r, NULL, 80 * line_cnt, line_cnt * 80 + 1);
- /* FIXME? Run through asciify. */
- if (documents == NULL)
- return 0;
- documents[80 * line_cnt] = '\0';
- dict_set_documents (dict, documents);
- free (documents);
- return 1;
-
-error:
- return 0;
+/* Reads a number from R and stores its value in *D.
+ If R is compressed, reads a compressed number;
+ otherwise, reads a number in the regular way.
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+read_case_number (struct sfm_reader *r, double *d)
+{
+ if (!r->compressed)
+ {
+ uint8_t number[8];
+ if (!try_read_bytes (r, number, sizeof number))
+ return false;
+ float_convert (r->float_format, number, FLOAT_NATIVE_DOUBLE, d);
+ return true;
+ }
+ else
+ return read_compressed_number (r, d);
}
-\f
-/* Data reader. */
-/* Reads compressed data into H->BUF and sets other pointers
- appropriately. Returns nonzero only if both no errors occur and
- data was read. */
-static int
-buffer_input (struct sfm_reader *r)
+/* Reads LENGTH string bytes from R into S.
+ Always reads a multiple of 8 bytes; if LENGTH is not a
+ multiple of 8, then extra bytes are read and discarded without
+ being written to S.
+ Reads compressed strings if S is compressed.
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+read_case_string (struct sfm_reader *r, char *s, size_t length)
{
- size_t amt;
+ size_t whole = ROUND_DOWN (length, 8);
+ size_t partial = length % 8;
- if (!r->ok)
- return false;
- if (r->buf == NULL)
- r->buf = xnmalloc (128, sizeof *r->buf);
- amt = fread (r->buf, sizeof *r->buf, 128, r->file);
- if (ferror (r->file))
+ if (whole)
{
- msg (ME, _("%s: Error reading file: %s."),
- fh_get_file_name (r->fh), strerror (errno));
- r->ok = false;
- return 0;
+ if (!read_whole_strings (r, s, whole))
+ return false;
}
- r->ptr = r->buf;
- r->end = &r->buf[amt];
- return amt;
+
+ if (partial)
+ {
+ char bounce[8];
+ if (!read_whole_strings (r, bounce, sizeof bounce))
+ {
+ if (whole)
+ partial_record (r);
+ return false;
+ }
+ memcpy (s + whole, bounce, partial);
+ }
+
+ return true;
}
-/* Reads a single case consisting of compressed data from system
- file H into the array BUF[] according to reader R, and
- returns nonzero only if successful. */
-/* Data in system files is compressed in this manner. Data
- values are grouped into sets of eight ("octets"). Each value
- in an octet has one instruction byte that are output together.
- Each instruction byte gives a value for that byte or indicates
- that the value can be found following the instructions. */
+/* Reads and returns the next compression opcode from R. */
static int
-read_compressed_data (struct sfm_reader *r, flt64 *buf)
+read_opcode (struct sfm_reader *r)
{
- const unsigned char *p_end = r->x + sizeof (flt64);
- unsigned char *p = r->y;
-
- const flt64 *buf_beg = buf;
- const flt64 *buf_end = &buf[r->value_cnt];
-
+ assert (r->compressed);
for (;;)
{
- for (; p < p_end; p++){
- switch (*p)
- {
- case 0:
- /* Code 0 is ignored. */
- continue;
- case 252:
- /* Code 252 is end of file. */
- if (buf_beg == buf)
- return 0;
- lose ((ME, _("%s: Compressed data is corrupted. Data ends "
- "in partial case."),
- fh_get_file_name (r->fh)));
- case 253:
- /* Code 253 indicates that the value is stored explicitly
- following the instruction bytes. */
- if (r->ptr == NULL || r->ptr >= r->end)
- if (!buffer_input (r))
- lose ((ME, _("%s: Unexpected end of file."),
- fh_get_file_name (r->fh)));
- memcpy (buf++, r->ptr++, sizeof *buf);
- if (buf >= buf_end)
- goto success;
- break;
- case 254:
- /* Code 254 indicates a string that is all blanks. */
- memset (buf++, ' ', sizeof *buf);
- if (buf >= buf_end)
- goto success;
- break;
- case 255:
- /* Code 255 indicates the system-missing value. */
- *buf = r->sysmis;
- if (r->reverse_endian)
- bswap_flt64 (buf);
- buf++;
- if (buf >= buf_end)
- goto success;
- break;
- default:
- /* Codes 1 through 251 inclusive are taken to indicate a
- value of (BYTE - BIAS), where BYTE is the byte's value
- and BIAS is the compression bias (generally 100.0). */
- *buf = *p - r->bias;
- if (r->reverse_endian)
- bswap_flt64 (buf);
- buf++;
- if (buf >= buf_end)
- goto success;
- break;
- }
- }
- /* We have reached the end of this instruction octet. Read
- another. */
- if (r->ptr == NULL || r->ptr >= r->end)
+ int opcode;
+ if (r->opcode_idx >= sizeof r->opcodes)
{
- if (!buffer_input (r))
- {
- if (buf_beg != buf)
- lose ((ME, _("%s: Unexpected end of file."),
- fh_get_file_name (r->fh)));
- else
- return 0;
- }
+ if (!try_read_bytes (r, r->opcodes, sizeof r->opcodes))
+ return -1;
+ r->opcode_idx = 0;
}
- memcpy (r->x, r->ptr++, sizeof *buf);
- p = r->x;
+ opcode = r->opcodes[r->opcode_idx++];
+
+ if (opcode != 0)
+ return opcode;
}
+}
+
+/* Reads a compressed number from R and stores its value in D.
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+read_compressed_number (struct sfm_reader *r, double *d)
+{
+ int opcode = read_opcode (r);
+ switch (opcode)
+ {
+ case -1:
+ case 252:
+ return false;
- abort ();
+ case 253:
+ *d = read_float (r);
+ break;
-success:
- /* We have filled up an entire record. Update state and return
- successfully. */
- r->y = ++p;
- return 1;
+ case 254:
+ sys_error (r, _("Compressed data is corrupt."));
-error:
- /* I/O error. */
- r->ok = false;
- return 0;
+ case 255:
+ *d = SYSMIS;
+ break;
+
+ default:
+ *d = opcode - r->bias;
+ break;
+ }
+
+ return true;
}
-/* Reads one case from READER's file into C. Returns nonzero
- only if successful. */
-int
-sfm_read_case (struct sfm_reader *r, struct ccase *c)
+/* Reads a compressed 8-byte string segment from R and stores it
+ in DST.
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+read_compressed_string (struct sfm_reader *r, char *dst)
{
- if (!r->ok)
- return 0;
-
- if (!r->compressed && sizeof (flt64) == sizeof (double))
+ switch (read_opcode (r))
{
- /* Fast path: external and internal representations are the
- same, except possibly for endianness or SYSMIS. Read
- directly into the case's buffer, then fix up any minor
- details as needed. */
- if (!fread_ok (r, case_data_all_rw (c),
- sizeof (union value) * r->value_cnt))
- return 0;
-
- /* Fix up endianness if needed. */
- if (r->reverse_endian)
- {
- int i;
-
- for (i = 0; i < r->value_cnt; i++)
- if (r->vars[i].width == 0)
- bswap_flt64 (&case_data_rw (c, r->vars[i].fv)->f);
- }
+ case -1:
+ case 252:
+ return false;
- /* Fix up SYSMIS values if needed.
- I don't think this will ever actually kick in, but it
- can't hurt. */
- if (r->sysmis != SYSMIS)
- {
- int i;
-
- for (i = 0; i < r->value_cnt; i++)
- if (r->vars[i].width == 0 && case_num (c, i) == r->sysmis)
- case_data_rw (c, r->vars[i].fv)->f = SYSMIS;
- }
+ case 253:
+ read_bytes (r, dst, 8);
+ break;
+
+ case 254:
+ memset (dst, ' ', 8);
+ break;
+
+ default:
+ sys_error (r, _("Compressed data is corrupt."));
+ }
+
+ return true;
+}
+
+/* Reads LENGTH string bytes from R into S.
+ LENGTH must be a multiple of 8.
+ Reads compressed strings if S is compressed.
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+read_whole_strings (struct sfm_reader *r, char *s, size_t length)
+{
+ assert (length % 8 == 0);
+ if (!r->compressed)
+ return try_read_bytes (r, s, length);
+ else
+ {
+ size_t ofs;
+ for (ofs = 0; ofs < length; ofs += 8)
+ if (!read_compressed_string (r, s + ofs))
+ {
+ if (ofs != 0)
+ partial_record (r);
+ return false;
+ }
+ return true;
}
- else
+}
+
+/* Skips LENGTH string bytes from R.
+ LENGTH must be a multiple of 8.
+ (LENGTH is also limited to 1024, but that's only because the
+ current caller never needs more than that many bytes.)
+ Returns true if successful, false if end of file is
+ reached immediately. */
+static bool
+skip_whole_strings (struct sfm_reader *r, size_t length)
+{
+ char buffer[1024];
+ assert (length < sizeof buffer);
+ return read_whole_strings (r, buffer, length);
+}
+\f
+/* Creates and returns a table that can be used for translating a value
+ index into a case to a "struct variable *" for DICT. Multiple
+ system file fields reference variables this way.
+
+ This table must be created before processing the very long
+ string extension record, because that record causes some
+ values to be deleted from the case and the dictionary to be
+ compacted. */
+static struct variable **
+make_var_by_value_idx (struct sfm_reader *r, struct dictionary *dict)
+{
+ struct variable **var_by_value_idx;
+ int value_idx = 0;
+ int i;
+
+ var_by_value_idx = pool_nmalloc (r->pool,
+ r->oct_cnt, sizeof *var_by_value_idx);
+ for (i = 0; i < dict_get_var_cnt (dict); i++)
{
- /* Slow path: internal and external representations differ.
- Read into a bounce buffer, then copy to C. */
- flt64 *bounce;
- flt64 *bounce_cur;
- size_t bounce_size;
- int read_ok;
- int i;
+ struct variable *v = dict_get_var (dict, i);
+ int nv = var_is_numeric (v) ? 1 : DIV_RND_UP (var_get_width (v), 8);
+ int j;
- bounce_size = sizeof *bounce * r->value_cnt;
- bounce = bounce_cur = local_alloc (bounce_size);
+ var_by_value_idx[value_idx++] = v;
+ for (j = 1; j < nv; j++)
+ var_by_value_idx[value_idx++] = NULL;
+ }
+ assert (value_idx == r->oct_cnt);
- if (!r->compressed)
- read_ok = fread_ok (r, bounce, bounce_size);
- else
- read_ok = read_compressed_data (r, bounce);
- if (!read_ok)
- {
- local_free (bounce);
- return 0;
- }
+ return var_by_value_idx;
+}
- for (i = 0; i < r->value_cnt; i++)
- {
- struct sfm_var *v = &r->vars[i];
+/* Returns the "struct variable" corresponding to the given
+ 1-basd VALUE_IDX in VAR_BY_VALUE_IDX. Verifies that the index
+ is valid. */
+static struct variable *
+lookup_var_by_value_idx (struct sfm_reader *r,
+ struct variable **var_by_value_idx, int value_idx)
+{
+ struct variable *var;
- if (v->width == 0)
- {
- flt64 f = *bounce_cur++;
- if (r->reverse_endian)
- bswap_flt64 (&f);
- case_data_rw (c, v->fv)->f = f == r->sysmis ? SYSMIS : f;
- }
- else if (v->width != -1)
- {
- memcpy (case_data_rw (c, v->fv)->s, bounce_cur, v->width);
- bounce_cur += DIV_RND_UP (v->width, sizeof (flt64));
- }
- }
+ if (value_idx < 1 || value_idx > r->oct_cnt)
+ sys_error (r, _("Variable index %d not in valid range 1...%d."),
+ value_idx, r->oct_cnt);
+
+ var = var_by_value_idx[value_idx - 1];
+ if (var == NULL)
+ sys_error (r, _("Variable index %d refers to long string "
+ "continuation."),
+ value_idx);
+
+ return var;
+}
- local_free (bounce);
+/* Returns the variable in D with the given SHORT_NAME,
+ or a null pointer if there is none. */
+static struct variable *
+lookup_var_by_short_name (struct dictionary *d, const char *short_name)
+{
+ struct variable *var;
+ size_t var_cnt;
+ size_t i;
+
+ /* First try looking up by full name. This often succeeds. */
+ var = dict_lookup_var (d, short_name);
+ if (var != NULL && !strcasecmp (var_get_short_name (var, 0), short_name))
+ return var;
+
+ /* Iterate through the whole dictionary as a fallback. */
+ var_cnt = dict_get_var_cnt (d);
+ for (i = 0; i < var_cnt; i++)
+ {
+ var = dict_get_var (d, i);
+ if (!strcasecmp (var_get_short_name (var, 0), short_name))
+ return var;
}
- return 1;
+
+ return NULL;
}
+\f
+/* Helpers for reading records that contain "variable=value"
+ pairs. */
-static int
-fread_ok (struct sfm_reader *r, void *buffer, size_t byte_cnt)
+/* State. */
+struct variable_to_value_map
+ {
+ struct substring buffer; /* Record contents. */
+ size_t pos; /* Current position in buffer. */
+ };
+
+/* Reads SIZE bytes into a "variable=value" map for R,
+ and returns the map. */
+static struct variable_to_value_map *
+open_variable_to_value_map (struct sfm_reader *r, size_t size)
{
- size_t read_bytes = fread (buffer, 1, byte_cnt, r->file);
+ struct variable_to_value_map *map = pool_alloc (r->pool, sizeof *map);
+ char *buffer = pool_malloc (r->pool, size + 1);
+ read_bytes (r, buffer, size);
+ map->buffer = ss_buffer (buffer, size);
+ map->pos = 0;
+ return map;
+}
- if (read_bytes == byte_cnt)
- return 1;
- else
+/* Closes MAP and frees its storage.
+ Not really needed, because the pool will free the map anyway,
+ but can be used to free it earlier. */
+static void
+close_variable_to_value_map (struct sfm_reader *r,
+ struct variable_to_value_map *map)
+{
+ pool_free (r->pool, ss_data (map->buffer));
+}
+
+/* Reads the next variable=value pair from MAP.
+ Looks up the variable in DICT and stores it into *VAR.
+ Stores a null-terminated value into *VALUE. */
+static bool
+read_variable_to_value_map (struct sfm_reader *r, struct dictionary *dict,
+ struct variable_to_value_map *map,
+ struct variable **var, char **value,
+ int *warning_cnt)
+{
+ int max_warnings = 5;
+
+ for (;;)
{
- if (ferror (r->file))
+ struct substring short_name_ss, value_ss;
+
+ if (!ss_tokenize (map->buffer, ss_cstr ("="), &map->pos, &short_name_ss)
+ || !ss_tokenize (map->buffer, ss_buffer ("\t\0", 2), &map->pos,
+ &value_ss))
{
- msg (ME, _("%s: Reading system file: %s."),
- fh_get_file_name (r->fh), strerror (errno));
- r->ok = false;
+ if (*warning_cnt > max_warnings)
+ sys_warn (r, _("Suppressed %d additional variable map warnings."),
+ *warning_cnt - max_warnings);
+ return false;
}
- else if (read_bytes != 0)
+
+ map->pos += ss_span (ss_substr (map->buffer, map->pos, SIZE_MAX),
+ ss_buffer ("\t\0", 2));
+
+ ss_data (short_name_ss)[ss_length (short_name_ss)] = '\0';
+ *var = lookup_var_by_short_name (dict, ss_data (short_name_ss));
+ if (*var == NULL)
{
- msg (ME, _("%s: Partial record at end of system file."),
- fh_get_file_name (r->fh));
- r->ok = false;
+ if (++*warning_cnt <= max_warnings)
+ sys_warn (r, _("Variable map refers to unknown variable %s."),
+ ss_data (short_name_ss));
+ continue;
}
- return 0;
+
+ ss_data (value_ss)[ss_length (value_ss)] = '\0';
+ *value = ss_data (value_ss);
+
+ return true;
}
}
\f
-/* Returns true if an I/O error has occurred on READER, false
- otherwise. */
-bool
-sfm_read_error (const struct sfm_reader *reader)
+/* Messages. */
+
+/* Displays a corruption message. */
+static void
+sys_msg (struct sfm_reader *r, int class, const char *format, va_list args)
{
- return !reader->ok;
+ struct msg m;
+ struct string text;
+
+ ds_init_empty (&text);
+ ds_put_format (&text, "\"%s\" near offset 0x%lx: ",
+ fh_get_file_name (r->fh), (unsigned long) ftell (r->file));
+ ds_put_vformat (&text, format, args);
+
+ m.category = msg_class_to_category (class);
+ m.severity = msg_class_to_severity (class);
+ m.where.file_name = NULL;
+ m.where.line_number = 0;
+ m.text = ds_cstr (&text);
+
+ msg_emit (&m);
}
-/* Returns true if FILE is an SPSS system file,
- false otherwise. */
-bool
-sfm_detect (FILE *file)
+/* Displays a warning for the current file position. */
+static void
+sys_warn (struct sfm_reader *r, const char *format, ...)
{
- struct sysfile_header hdr;
+ va_list args;
- if (fread (&hdr, sizeof hdr, 1, file) != 1)
- return false;
- if (strncmp ("$FL2", hdr.rec_type, 4))
+ va_start (args, format);
+ sys_msg (r, MW, format, args);
+ va_end (args);
+}
+
+/* Displays an error for the current file position,
+ marks it as in an error state,
+ and aborts reading it using longjmp. */
+static void
+sys_error (struct sfm_reader *r, const char *format, ...)
+{
+ va_list args;
+
+ va_start (args, format);
+ sys_msg (r, ME, format, args);
+ va_end (args);
+
+ r->error = true;
+ longjmp (r->bail_out, 1);
+}
+\f
+/* Reads BYTE_CNT bytes into BUF.
+ Returns true if exactly BYTE_CNT bytes are successfully read.
+ Aborts if an I/O error or a partial read occurs.
+ If EOF_IS_OK, then an immediate end-of-file causes false to be
+ returned; otherwise, immediate end-of-file causes an abort
+ too. */
+static inline bool
+read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
+ void *buf, size_t byte_cnt)
+{
+ size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
+ if (bytes_read == byte_cnt)
+ return true;
+ else if (ferror (r->file))
+ sys_error (r, _("System error: %s."), strerror (errno));
+ else if (!eof_is_ok || bytes_read != 0)
+ sys_error (r, _("Unexpected end of file."));
+ else
return false;
- return true;
}
+/* Reads BYTE_CNT into BUF.
+ Aborts upon I/O error or if end-of-file is encountered. */
+static void
+read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
+{
+ read_bytes_internal (r, false, buf, byte_cnt);
+}
+
+/* Reads BYTE_CNT bytes into BUF.
+ Returns true if exactly BYTE_CNT bytes are successfully read.
+ Returns false if an immediate end-of-file is encountered.
+ Aborts if an I/O error or a partial read occurs. */
+static bool
+try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
+{
+ return read_bytes_internal (r, true, buf, byte_cnt);
+}
+
+/* Reads a 32-bit signed integer from R and returns its value in
+ host format. */
+static int
+read_int (struct sfm_reader *r)
+{
+ uint8_t integer[4];
+ read_bytes (r, integer, sizeof integer);
+ return integer_get (r->integer_format, integer, sizeof integer);
+}
+
+/* Reads a 64-bit floating-point number from R and returns its
+ value in host format. */
+static double
+read_float (struct sfm_reader *r)
+{
+ uint8_t number[8];
+ read_bytes (r, number, sizeof number);
+ return float_get_double (r->float_format, number);
+}
+
+/* Reads exactly SIZE - 1 bytes into BUFFER
+ and stores a null byte into BUFFER[SIZE - 1]. */
+static void
+read_string (struct sfm_reader *r, char *buffer, size_t size)
+{
+ assert (size > 0);
+ read_bytes (r, buffer, size - 1);
+ buffer[size - 1] = '\0';
+}
+
+/* Skips BYTES bytes forward in R. */
+static void
+skip_bytes (struct sfm_reader *r, size_t bytes)
+{
+ while (bytes > 0)
+ {
+ char buffer[1024];
+ size_t chunk = MIN (sizeof buffer, bytes);
+ read_bytes (r, buffer, chunk);
+ bytes -= chunk;
+ }
+}
+\f
+static struct casereader_class sys_file_casereader_class =
+ {
+ sys_file_casereader_read,
+ sys_file_casereader_destroy,
+ NULL,
+ NULL,
+ };