1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2006, 2007 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include <data/sys-file-reader.h>
20 #include <data/sys-file-private.h>
28 #include <libpspp/assertion.h>
29 #include <libpspp/message.h>
30 #include <libpspp/compiler.h>
31 #include <libpspp/misc.h>
32 #include <libpspp/pool.h>
33 #include <libpspp/str.h>
34 #include <libpspp/hash.h>
35 #include <libpspp/array.h>
37 #include <data/case.h>
38 #include <data/casereader-provider.h>
39 #include <data/casereader.h>
40 #include <data/dictionary.h>
41 #include <data/file-handle-def.h>
42 #include <data/file-name.h>
43 #include <data/format.h>
44 #include <data/missing-values.h>
45 #include <data/short-names.h>
46 #include <data/value-labels.h>
47 #include <data/variable.h>
48 #include <data/value.h>
53 #include "unlocked-io.h"
58 #define _(msgid) gettext (msgid)
59 #define N_(msgid) (msgid)
61 /* System file reader. */
64 /* Resource tracking. */
65 struct pool *pool; /* All system file state. */
66 jmp_buf bail_out; /* longjmp() target for error handling. */
69 struct file_handle *fh; /* File handle. */
70 struct fh_lock *lock; /* Mutual exclusion for file handle. */
71 FILE *file; /* File stream. */
72 bool error; /* I/O or corruption error? */
73 size_t value_cnt; /* Number of "union value"s in struct case. */
76 enum integer_format integer_format; /* On-disk integer format. */
77 enum float_format float_format; /* On-disk floating point format. */
78 int oct_cnt; /* Number of 8-byte units per case. */
79 struct sfm_var *sfm_vars; /* Variables. */
80 size_t sfm_var_cnt; /* Number of variables. */
81 casenumber case_cnt; /* Number of cases */
82 bool has_long_var_names; /* File has a long variable name map */
85 bool compressed; /* File is compressed? */
86 double bias; /* Compression bias, usually 100.0. */
87 uint8_t opcodes[8]; /* Current block of opcodes. */
88 size_t opcode_idx; /* Next opcode to interpret, 8 if none left. */
91 static struct casereader_class sys_file_casereader_class;
93 static bool close_reader (struct sfm_reader *);
95 static struct variable **make_var_by_value_idx (struct sfm_reader *,
97 static struct variable *lookup_var_by_value_idx (struct sfm_reader *,
101 static void sys_warn (struct sfm_reader *, const char *, ...)
102 PRINTF_FORMAT (2, 3);
104 static void sys_error (struct sfm_reader *, const char *, ...)
108 static void read_bytes (struct sfm_reader *, void *, size_t);
109 static bool try_read_bytes (struct sfm_reader *, void *, size_t);
110 static int read_int (struct sfm_reader *);
111 static double read_float (struct sfm_reader *);
112 static void read_string (struct sfm_reader *, char *, size_t);
113 static void skip_bytes (struct sfm_reader *, size_t);
115 static struct variable_to_value_map *open_variable_to_value_map (
116 struct sfm_reader *, size_t size);
117 static void close_variable_to_value_map (struct sfm_reader *r,
118 struct variable_to_value_map *);
119 static bool read_variable_to_value_map (struct sfm_reader *,
121 struct variable_to_value_map *,
122 struct variable **var, char **value,
125 static bool close_reader (struct sfm_reader *r);
127 /* Dictionary reader. */
135 static void read_header (struct sfm_reader *, struct dictionary *,
136 int *weight_idx, int *claimed_oct_cnt,
137 struct sfm_read_info *);
138 static void read_variable_record (struct sfm_reader *, struct dictionary *,
139 int *format_warning_cnt);
140 static void parse_format_spec (struct sfm_reader *, unsigned int,
141 enum which_format, struct variable *,
142 int *format_warning_cnt);
143 static void setup_weight (struct sfm_reader *, int weight_idx,
144 struct variable **var_by_value_idx,
145 struct dictionary *);
146 static void read_documents (struct sfm_reader *, struct dictionary *);
147 static void read_value_labels (struct sfm_reader *, struct dictionary *,
148 struct variable **var_by_value_idx);
150 static void read_extension_record (struct sfm_reader *, struct dictionary *,
151 struct sfm_read_info *);
152 static void read_machine_integer_info (struct sfm_reader *,
153 size_t size, size_t count,
154 struct sfm_read_info *);
155 static void read_machine_float_info (struct sfm_reader *,
156 size_t size, size_t count);
157 static void read_display_parameters (struct sfm_reader *,
158 size_t size, size_t count,
159 struct dictionary *);
160 static void read_long_var_name_map (struct sfm_reader *,
161 size_t size, size_t count,
162 struct dictionary *);
163 static void read_long_string_map (struct sfm_reader *,
164 size_t size, size_t count,
165 struct dictionary *);
168 /* Opens the system file designated by file handle FH for
169 reading. Reads the system file's dictionary into *DICT.
170 If INFO is non-null, then it receives additional info about the
173 sfm_open_reader (struct file_handle *fh, struct dictionary **dict,
174 struct sfm_read_info *volatile info)
176 struct sfm_reader *volatile r = NULL;
177 struct variable **var_by_value_idx;
178 struct sfm_read_info local_info;
179 int format_warning_cnt = 0;
184 *dict = dict_create ();
186 /* Create and initialize reader. */
187 r = pool_create_container (struct sfm_reader, pool);
193 r->has_long_var_names = false;
194 r->opcode_idx = sizeof r->opcodes;
196 /* TRANSLATORS: this fragment will be interpolated into
197 messages in fh_lock() that identify types of files. */
198 r->lock = fh_lock (fh, FH_REF_FILE, N_("system file"), FH_ACC_READ, false);
202 r->file = fn_open (fh_get_file_name (fh), "rb");
205 msg (ME, _("Error opening \"%s\" for reading as a system file: %s."),
206 fh_get_file_name (r->fh), strerror (errno));
210 /* Initialize info. */
213 memset (info, 0, sizeof *info);
215 if (setjmp (r->bail_out))
220 read_header (r, *dict, &weight_idx, &claimed_oct_cnt, info);
222 /* Read all the variable definition records. */
223 rec_type = read_int (r);
224 while (rec_type == 2)
226 read_variable_record (r, *dict, &format_warning_cnt);
227 rec_type = read_int (r);
230 /* Figure out the case format. */
231 var_by_value_idx = make_var_by_value_idx (r, *dict);
232 setup_weight (r, weight_idx, var_by_value_idx, *dict);
234 /* Read all the rest of the dictionary records. */
235 while (rec_type != 999)
240 read_value_labels (r, *dict, var_by_value_idx);
244 sys_error (r, _("Misplaced type 4 record."));
247 read_documents (r, *dict);
251 read_extension_record (r, *dict, info);
255 sys_error (r, _("Unrecognized record type %d."), rec_type);
257 rec_type = read_int (r);
261 if ( ! r->has_long_var_names )
264 for (i = 0; i < dict_get_var_cnt (*dict); i++)
266 struct variable *var = dict_get_var (*dict, i);
267 char short_name[SHORT_NAME_LEN + 1];
268 char long_name[SHORT_NAME_LEN + 1];
270 strcpy (short_name, var_get_name (var));
272 strcpy (long_name, short_name);
273 str_lowercase (long_name);
275 /* Set long name. Renaming a variable may clear the short
276 name, but we want to retain it, so re-set it
278 dict_rename_var (*dict, var, long_name);
279 var_set_short_name (var, 0, short_name);
282 r->has_long_var_names = true;
285 /* Read record 999 data, which is just filler. */
288 /* Warn if the actual amount of data per case differs from the
289 amount that the header claims. SPSS version 13 gets this
290 wrong when very long strings are involved, so don't warn in
292 if (claimed_oct_cnt != -1 && claimed_oct_cnt != r->oct_cnt
293 && info->version_major != 13)
294 sys_warn (r, _("File header claims %d variable positions but "
295 "%d were read from file."),
296 claimed_oct_cnt, r->oct_cnt);
298 /* Create an index of dictionary variable widths for
299 sfm_read_case to use. We cannot use the `struct variable's
300 from the dictionary we created, because the caller owns the
301 dictionary and may destroy or modify its variables. */
302 sfm_dictionary_to_sfm_vars (*dict, &r->sfm_vars, &r->sfm_var_cnt);
303 pool_register (r->pool, free, r->sfm_vars);
305 pool_free (r->pool, var_by_value_idx);
306 r->value_cnt = dict_get_next_value_idx (*dict);
307 return casereader_create_sequential
309 r->case_cnt == -1 ? CASENUMBER_MAX: r->case_cnt,
310 &sys_file_casereader_class, r);
314 dict_destroy (*dict);
319 /* Closes a system file after we're done with it.
320 Returns true if an I/O error has occurred on READER, false
323 close_reader (struct sfm_reader *r)
332 if (fn_close (fh_get_file_name (r->fh), r->file) == EOF)
334 msg (ME, _("Error closing system file \"%s\": %s."),
335 fh_get_file_name (r->fh), strerror (errno));
345 pool_destroy (r->pool);
350 /* Destroys READER. */
352 sys_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
354 struct sfm_reader *r = r_;
358 /* Returns true if FILE is an SPSS system file,
361 sfm_detect (FILE *file)
365 if (fread (rec_type, 4, 1, file) != 1)
369 return !strcmp ("$FL2", rec_type);
372 /* Reads the global header of the system file.
373 Sets DICT's file label to the system file's label.
374 Sets *WEIGHT_IDX to 0 if the system file is unweighted,
375 or to the value index of the weight variable otherwise.
376 Sets *CLAIMED_OCT_CNT to the number of "octs" (8-byte units)
377 per case that the file claims to have (although it is not
379 Initializes INFO with header information. */
381 read_header (struct sfm_reader *r, struct dictionary *dict,
382 int *weight_idx, int *claimed_oct_cnt,
383 struct sfm_read_info *info)
386 char eye_catcher[61];
387 uint8_t raw_layout_code[4];
389 char creation_date[10];
390 char creation_time[9];
392 struct substring file_label_ss;
393 struct substring product;
395 read_string (r, rec_type, sizeof rec_type);
396 read_string (r, eye_catcher, sizeof eye_catcher);
398 if (strcmp ("$FL2", rec_type) != 0)
399 sys_error (r, _("This is not an SPSS system file."));
401 /* Identify integer format. */
402 read_bytes (r, raw_layout_code, sizeof raw_layout_code);
403 if ((!integer_identify (2, raw_layout_code, sizeof raw_layout_code,
405 && !integer_identify (3, raw_layout_code, sizeof raw_layout_code,
407 || (r->integer_format != INTEGER_MSB_FIRST
408 && r->integer_format != INTEGER_LSB_FIRST))
409 sys_error (r, _("This is not an SPSS system file."));
411 *claimed_oct_cnt = read_int (r);
412 if (*claimed_oct_cnt < 0 || *claimed_oct_cnt > INT_MAX / 16)
413 *claimed_oct_cnt = -1;
415 r->compressed = read_int (r) != 0;
417 *weight_idx = read_int (r);
419 r->case_cnt = read_int (r);
420 if ( r->case_cnt > INT_MAX / 2)
424 /* Identify floating-point format and obtain compression bias. */
425 read_bytes (r, raw_bias, sizeof raw_bias);
426 if (float_identify (100.0, raw_bias, sizeof raw_bias, &r->float_format) == 0)
428 sys_warn (r, _("Compression bias (%g) is not the usual "
429 "value of 100, or system file uses unrecognized "
430 "floating-point format."),
432 if (r->integer_format == INTEGER_MSB_FIRST)
433 r->float_format = FLOAT_IEEE_DOUBLE_BE;
435 r->float_format = FLOAT_IEEE_DOUBLE_LE;
437 float_convert (r->float_format, raw_bias, FLOAT_NATIVE_DOUBLE, &r->bias);
439 read_string (r, creation_date, sizeof creation_date);
440 read_string (r, creation_time, sizeof creation_time);
441 read_string (r, file_label, sizeof file_label);
444 file_label_ss = ss_cstr (file_label);
445 ss_trim (&file_label_ss, ss_cstr (" "));
446 if (!ss_is_empty (file_label_ss))
448 ss_data (file_label_ss)[ss_length (file_label_ss)] = '\0';
449 dict_set_label (dict, ss_data (file_label_ss));
452 strcpy (info->creation_date, creation_date);
453 strcpy (info->creation_time, creation_time);
454 info->integer_format = r->integer_format;
455 info->float_format = r->float_format;
456 info->compressed = r->compressed;
457 info->case_cnt = r->case_cnt;
459 product = ss_cstr (eye_catcher);
460 ss_match_string (&product, ss_cstr ("@(#) SPSS DATA FILE"));
461 ss_trim (&product, ss_cstr (" "));
462 str_copy_buf_trunc (info->product, sizeof info->product,
463 ss_data (product), ss_length (product));
466 /* Reads a variable (type 2) record from R and adds the
467 corresponding variable to DICT.
468 Also skips past additional variable records for long string
471 read_variable_record (struct sfm_reader *r, struct dictionary *dict,
472 int *format_warning_cnt)
475 int has_variable_label;
476 int missing_value_code;
481 struct variable *var;
484 width = read_int (r);
485 has_variable_label = read_int (r);
486 missing_value_code = read_int (r);
487 print_format = read_int (r);
488 write_format = read_int (r);
489 read_string (r, name, sizeof name);
490 name[strcspn (name, " ")] = '\0';
492 /* Check variable name. */
493 if (name[0] == '$' || name[0] == '#')
494 sys_error (r, "Variable name begins with invalid character `%c'.",
496 if (!var_is_plausible_name (name, false))
497 sys_error (r, _("Invalid variable name `%s'."), name);
499 /* Create variable. */
500 if (width < 0 || width > 255)
501 sys_error (r, _("Bad variable width %d."), width);
502 var = dict_create_var (dict, name, width);
505 _("Duplicate variable name `%s' within system file."),
508 /* Set the short name the same as the long name. */
509 var_set_short_name (var, 0, var_get_name (var));
511 /* Get variable label, if any. */
512 if (has_variable_label != 0 && has_variable_label != 1)
513 sys_error (r, _("Variable label indicator field is not 0 or 1."));
514 if (has_variable_label == 1)
520 if (len >= sizeof label)
521 sys_error (r, _("Variable %s has label of invalid length %zu."),
523 read_string (r, label, len + 1);
524 var_set_label (var, label);
526 skip_bytes (r, ROUND_UP (len, 4) - len);
529 /* Set missing values. */
530 if (missing_value_code != 0)
532 struct missing_values mv;
535 mv_init (&mv, var_get_width (var));
536 if (var_is_numeric (var))
538 if (missing_value_code < -3 || missing_value_code > 3
539 || missing_value_code == -1)
540 sys_error (r, _("Numeric missing value indicator field is not "
541 "-3, -2, 0, 1, 2, or 3."));
542 if (missing_value_code < 0)
544 double low = read_float (r);
545 double high = read_float (r);
546 mv_add_range (&mv, low, high);
547 missing_value_code = -missing_value_code - 2;
549 for (i = 0; i < missing_value_code; i++)
550 mv_add_num (&mv, read_float (r));
552 else if (var_get_width (var) <= MAX_SHORT_STRING)
554 if (missing_value_code < 1 || missing_value_code > 3)
555 sys_error (r, _("String missing value indicator field is not "
557 for (i = 0; i < missing_value_code; i++)
560 read_string (r, string, sizeof string);
561 mv_add_str (&mv, string);
565 sys_error (r, _("Long string variable %s may not have missing "
567 var_set_missing_values (var, &mv);
571 parse_format_spec (r, print_format, PRINT_FORMAT, var, format_warning_cnt);
572 parse_format_spec (r, write_format, WRITE_FORMAT, var, format_warning_cnt);
574 /* Account for values.
575 Skip long string continuation records, if any. */
576 nv = width == 0 ? 1 : DIV_RND_UP (width, 8);
582 for (i = 1; i < nv; i++)
584 /* Check for record type 2 and width -1. */
585 if (read_int (r) != 2 || read_int (r) != -1)
586 sys_error (r, _("Missing string continuation record."));
588 /* Skip and ignore remaining continuation data. */
589 has_variable_label = read_int (r);
590 missing_value_code = read_int (r);
591 print_format = read_int (r);
592 write_format = read_int (r);
593 read_string (r, name, sizeof name);
595 /* Variable label fields on continuation records have
596 been spotted in system files created by "SPSS Power
597 Macintosh Release 6.1". */
598 if (has_variable_label)
599 skip_bytes (r, ROUND_UP (read_int (r), 4));
604 /* Translates the format spec from sysfile format to internal
607 parse_format_spec (struct sfm_reader *r, unsigned int s,
608 enum which_format which, struct variable *v,
609 int *format_warning_cnt)
611 const int max_format_warnings = 8;
613 uint8_t raw_type = s >> 16;
619 if (!fmt_from_io (raw_type, &f.type))
620 sys_error (r, _("Unknown variable format %"PRIu8"."), raw_type);
625 ok = fmt_check_output (&f) && fmt_check_width_compat (&f, var_get_width (v));
630 if (which == PRINT_FORMAT)
631 var_set_print_format (v, &f);
633 var_set_write_format (v, &f);
635 else if (*++format_warning_cnt <= max_format_warnings)
637 char fmt_string[FMT_STRING_LEN_MAX + 1];
638 sys_warn (r, _("%s variable %s has invalid %s format %s."),
639 var_is_numeric (v) ? _("Numeric") : _("String"),
641 which == PRINT_FORMAT ? _("print") : _("write"),
642 fmt_to_string (&f, fmt_string));
644 if (*format_warning_cnt == max_format_warnings)
645 sys_warn (r, _("Suppressing further invalid format warnings."));
649 /* Sets the weighting variable in DICT to the variable
650 corresponding to the given 1-based VALUE_IDX, if VALUE_IDX is
653 setup_weight (struct sfm_reader *r, int weight_idx,
654 struct variable **var_by_value_idx, struct dictionary *dict)
658 struct variable *weight_var
659 = lookup_var_by_value_idx (r, var_by_value_idx, weight_idx);
660 if (var_is_numeric (weight_var))
661 dict_set_weight (dict, weight_var);
663 sys_error (r, _("Weighting variable must be numeric."));
667 /* Reads a document record, type 6, from system file R, and sets up
668 the documents and n_documents fields in the associated
671 read_documents (struct sfm_reader *r, struct dictionary *dict)
676 if (dict_get_documents (dict) != NULL)
677 sys_error (r, _("Multiple type 6 (document) records."));
679 line_cnt = read_int (r);
681 sys_error (r, _("Number of document lines (%d) "
682 "must be greater than 0."), line_cnt);
684 documents = pool_nmalloc (r->pool, line_cnt + 1, DOC_LINE_LENGTH);
685 read_string (r, documents, DOC_LINE_LENGTH * line_cnt + 1);
686 if (strlen (documents) == DOC_LINE_LENGTH * line_cnt)
687 dict_set_documents (dict, documents);
689 sys_error (r, _("Document line contains null byte."));
690 pool_free (r->pool, documents);
693 /* Read a type 7 extension record. */
695 read_extension_record (struct sfm_reader *r, struct dictionary *dict,
696 struct sfm_read_info *info)
698 int subtype = read_int (r);
699 size_t size = read_int (r);
700 size_t count = read_int (r);
701 size_t bytes = size * count;
703 /* Check that SIZE * COUNT + 1 doesn't overflow. Adding 1
704 allows an extra byte for a null terminator, used by some
705 extension processing routines. */
706 if (size != 0 && size_overflow_p (xsum (1, xtimes (count, size))))
707 sys_error (r, "Record type 7 subtype %d too large.", subtype);
712 read_machine_integer_info (r, size, count, info);
716 read_machine_float_info (r, size, count);
720 /* Variable sets information. We don't use these yet.
721 They only apply to GUIs; see VARSETS on the APPLY
722 DICTIONARY command in SPSS documentation. */
726 /* DATE variable information. We don't use it yet, but we
731 /* Unknown purpose. */
735 read_display_parameters (r, size, count, dict);
739 read_long_var_name_map (r, size, count, dict);
743 read_long_string_map (r, size, count, dict);
747 /* New in SPSS v14? Unknown purpose. */
751 /* Text field that defines variable attributes. New in
756 sys_warn (r, _("Unrecognized record type 7, subtype %d."), subtype);
760 skip_bytes (r, bytes);
763 /* Read record type 7, subtype 3. */
765 read_machine_integer_info (struct sfm_reader *r, size_t size, size_t count,
766 struct sfm_read_info *info)
768 int version_major = read_int (r);
769 int version_minor = read_int (r);
770 int version_revision = read_int (r);
771 int machine_code UNUSED = read_int (r);
772 int float_representation = read_int (r);
773 int compression_code UNUSED = read_int (r);
774 int integer_representation = read_int (r);
775 int character_code UNUSED = read_int (r);
777 int expected_float_format;
778 int expected_integer_format;
780 if (size != 4 || count != 8)
781 sys_error (r, _("Bad size (%zu) or count (%zu) field on record type 7, "
785 /* Save version info. */
786 info->version_major = version_major;
787 info->version_minor = version_minor;
788 info->version_revision = version_revision;
790 /* Check floating point format. */
791 if (r->float_format == FLOAT_IEEE_DOUBLE_BE
792 || r->float_format == FLOAT_IEEE_DOUBLE_LE)
793 expected_float_format = 1;
794 else if (r->float_format == FLOAT_Z_LONG)
795 expected_float_format = 2;
796 else if (r->float_format == FLOAT_VAX_G || r->float_format == FLOAT_VAX_D)
797 expected_float_format = 3;
800 if (float_representation != expected_float_format)
801 sys_error (r, _("Floating-point representation indicated by "
802 "system file (%d) differs from expected (%d)."),
803 r->float_format, expected_float_format);
805 /* Check integer format. */
806 if (r->integer_format == INTEGER_MSB_FIRST)
807 expected_integer_format = 1;
808 else if (r->integer_format == INTEGER_LSB_FIRST)
809 expected_integer_format = 2;
812 if (integer_representation != expected_integer_format)
814 static const char *endian[] = {N_("little-endian"), N_("big-endian")};
815 sys_warn (r, _("Integer format indicated by system file (%s) "
816 "differs from expected (%s)."),
817 gettext (endian[integer_representation == 1]),
818 gettext (endian[expected_integer_format == 1]));
822 /* Read record type 7, subtype 4. */
824 read_machine_float_info (struct sfm_reader *r, size_t size, size_t count)
826 double sysmis = read_float (r);
827 double highest = read_float (r);
828 double lowest = read_float (r);
830 if (size != 8 || count != 3)
831 sys_error (r, _("Bad size (%zu) or count (%zu) on extension 4."),
834 if (sysmis != SYSMIS)
835 sys_warn (r, _("File specifies unexpected value %g as SYSMIS."), sysmis);
836 if (highest != HIGHEST)
837 sys_warn (r, _("File specifies unexpected value %g as HIGHEST."), highest);
838 if (lowest != LOWEST)
839 sys_warn (r, _("File specifies unexpected value %g as LOWEST."), lowest);
842 /* Read record type 7, subtype 11, which specifies how variables
843 should be displayed in GUI environments. */
845 read_display_parameters (struct sfm_reader *r, size_t size, size_t count,
846 struct dictionary *dict)
855 sys_warn (r, _("Bad size %zu on extension 11."), size);
856 skip_bytes (r, size * count);
860 n_vars = dict_get_var_cnt (dict);
861 if (count == 3 * n_vars)
862 includes_width = true;
863 else if (count == 2 * n_vars)
864 includes_width = false;
867 sys_warn (r, _("Extension 11 has bad count %zu (for %zu variables)."),
869 skip_bytes (r, size * count);
873 for (i = 0; i < n_vars; ++i)
875 struct variable *v = dict_get_var (dict, i);
876 int measure = read_int (r);
877 int width = includes_width ? read_int (r) : 0;
878 int align = read_int (r);
880 /* SPSS 14 sometimes seems to set string variables' measure
882 if (0 == measure && var_is_alpha (v))
885 if (measure < 1 || measure > 3 || align < 0 || align > 2)
888 sys_warn (r, _("Invalid variable display parameters "
889 "for variable %zu (%s). "
890 "Default parameters substituted."),
891 i, var_get_name (v));
896 var_set_measure (v, (measure == 1 ? MEASURE_NOMINAL
897 : measure == 2 ? MEASURE_ORDINAL
899 var_set_alignment (v, (align == 0 ? ALIGN_LEFT
900 : align == 1 ? ALIGN_RIGHT
903 /* Older versions (SPSS 9.0) sometimes set the display
904 width to zero. This causes confusion in the GUI, so
905 only set the width if it is nonzero. */
907 var_set_display_width (v, width);
911 /* Reads record type 7, subtype 13, which gives the long name
912 that corresponds to each short name. Modifies variable names
913 in DICT accordingly. */
915 read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count,
916 struct dictionary *dict)
918 struct variable_to_value_map *map;
919 struct variable *var;
923 map = open_variable_to_value_map (r, size * count);
924 while (read_variable_to_value_map (r, dict, map, &var, &long_name,
928 size_t short_name_cnt;
931 /* Validate long name. */
932 if (!var_is_valid_name (long_name, false))
934 sys_warn (r, _("Long variable mapping from %s to invalid "
935 "variable name `%s'."),
936 var_get_name (var), long_name);
940 /* Identify any duplicates. */
941 if (strcasecmp (var_get_short_name (var, 0), long_name)
942 && dict_lookup_var (dict, long_name) != NULL)
944 sys_warn (r, _("Duplicate long variable name `%s' "
945 "within system file."), long_name);
949 /* Renaming a variable may clear its short names, but we
950 want to retain them, so we save them and re-set them
952 short_name_cnt = var_get_short_name_cnt (var);
953 short_names = xnmalloc (short_name_cnt, sizeof *short_names);
954 for (i = 0; i < short_name_cnt; i++)
956 const char *s = var_get_short_name (var, i);
957 short_names[i] = s != NULL ? xstrdup (s) : NULL;
961 dict_rename_var (dict, var, long_name);
963 /* Restore short names. */
964 for (i = 0; i < short_name_cnt; i++)
966 var_set_short_name (var, i, short_names[i]);
967 free (short_names[i]);
971 close_variable_to_value_map (r, map);
972 r->has_long_var_names = true;
975 /* Reads record type 7, subtype 14, which gives the real length
976 of each very long string. Rearranges DICT accordingly. */
978 read_long_string_map (struct sfm_reader *r, size_t size, size_t count,
979 struct dictionary *dict)
981 struct variable_to_value_map *map;
982 struct variable *var;
986 map = open_variable_to_value_map (r, size * count);
987 while (read_variable_to_value_map (r, dict, map, &var, &length_s,
990 size_t idx = var_get_dict_index (var);
996 length = strtol (length_s, NULL, 10);
997 if (length < 1 || length > MAX_STRING)
999 sys_warn (r, _("%s listed as string of invalid length %s "
1000 "in very length string record."),
1001 var_get_name (var), length_s);
1005 /* Check segments. */
1006 segment_cnt = sfm_width_to_segments (length);
1007 if (segment_cnt == 1)
1009 sys_warn (r, _("%s listed in very long string record with width %s, "
1010 "which requires only one segment."),
1011 var_get_name (var), length_s);
1014 if (idx + segment_cnt > dict_get_var_cnt (dict))
1015 sys_error (r, _("Very long string %s overflows dictionary."),
1016 var_get_name (var));
1018 /* Get the short names from the segments and check their
1020 for (i = 0; i < segment_cnt; i++)
1022 struct variable *seg = dict_get_var (dict, idx + i);
1023 int alloc_width = sfm_segment_alloc_width (length, i);
1024 int width = var_get_width (seg);
1027 var_set_short_name (var, i, var_get_short_name (seg, 0));
1028 if (ROUND_UP (width, 8) != ROUND_UP (alloc_width, 8))
1029 sys_error (r, _("Very long string with width %ld has segment %d "
1030 "of width %d (expected %d)"),
1031 length, i, width, alloc_width);
1033 dict_delete_consecutive_vars (dict, idx + 1, segment_cnt - 1);
1034 var_set_width (var, length);
1036 close_variable_to_value_map (r, map);
1037 dict_compact_values (dict);
1040 /* Reads value labels from sysfile H and inserts them into the
1041 associated dictionary. */
1043 read_value_labels (struct sfm_reader *r,
1044 struct dictionary *dict, struct variable **var_by_value_idx)
1046 struct pool *subpool;
1050 char raw_value[8]; /* Value as uninterpreted bytes. */
1051 union value value; /* Value. */
1052 char *label; /* Null-terminated label string. */
1055 struct label *labels = NULL;
1056 int label_cnt; /* Number of labels. */
1058 struct variable **var = NULL; /* Associated variables. */
1059 int var_cnt; /* Number of associated variables. */
1063 subpool = pool_create_subpool (r->pool);
1065 /* Read the type 3 record and record its contents. We can't do
1066 much with the data yet because we don't know whether it is
1067 of numeric or string type. */
1069 /* Read number of labels. */
1070 label_cnt = read_int (r);
1072 if (size_overflow_p (xtimes (label_cnt, sizeof *labels)))
1074 sys_warn (r, _("Invalid number of labels: %d. Ignoring labels."),
1079 /* Read each value/label tuple into labels[]. */
1080 labels = pool_nalloc (subpool, label_cnt, sizeof *labels);
1081 for (i = 0; i < label_cnt; i++)
1083 struct label *label = labels + i;
1084 unsigned char label_len;
1088 read_bytes (r, label->raw_value, sizeof label->raw_value);
1090 /* Read label length. */
1091 read_bytes (r, &label_len, sizeof label_len);
1092 padded_len = ROUND_UP (label_len + 1, 8);
1094 /* Read label, padding. */
1095 label->label = pool_alloc (subpool, padded_len + 1);
1096 read_bytes (r, label->label, padded_len - 1);
1097 label->label[label_len] = 0;
1100 /* Now, read the type 4 record that has the list of variables
1101 to which the value labels are to be applied. */
1103 /* Read record type of type 4 record. */
1104 if (read_int (r) != 4)
1105 sys_error (r, _("Variable index record (type 4) does not immediately "
1106 "follow value label record (type 3) as it should."));
1108 /* Read number of variables associated with value label from type 4
1110 var_cnt = read_int (r);
1111 if (var_cnt < 1 || var_cnt > dict_get_var_cnt (dict))
1112 sys_error (r, _("Number of variables associated with a value label (%d) "
1113 "is not between 1 and the number of variables (%zu)."),
1114 var_cnt, dict_get_var_cnt (dict));
1116 /* Read the list of variables. */
1117 var = pool_nalloc (subpool, var_cnt, sizeof *var);
1118 for (i = 0; i < var_cnt; i++)
1120 var[i] = lookup_var_by_value_idx (r, var_by_value_idx, read_int (r));
1121 if (var_is_long_string (var[i]))
1122 sys_error (r, _("Value labels are not allowed on long string "
1123 "variables (%s)."), var_get_name (var[i]));
1126 /* Type check the variables. */
1127 for (i = 1; i < var_cnt; i++)
1128 if (var_get_type (var[i]) != var_get_type (var[0]))
1129 sys_error (r, _("Variables associated with value label are not all of "
1130 "identical type. Variable %s is %s, but variable "
1132 var_get_name (var[0]),
1133 var_is_numeric (var[0]) ? _("numeric") : _("string"),
1134 var_get_name (var[i]),
1135 var_is_numeric (var[i]) ? _("numeric") : _("string"));
1137 /* Fill in labels[].value, now that we know the desired type. */
1138 for (i = 0; i < label_cnt; i++)
1140 struct label *label = labels + i;
1142 if (var_is_alpha (var[0]))
1143 buf_copy_rpad (label->value.s, sizeof label->value.s,
1144 label->raw_value, sizeof label->raw_value);
1146 label->value.f = float_get_double (r->float_format, label->raw_value);
1149 /* Assign the `value_label's to each variable. */
1150 for (i = 0; i < var_cnt; i++)
1152 struct variable *v = var[i];
1155 /* Add each label to the variable. */
1156 for (j = 0; j < label_cnt; j++)
1158 struct label *label = &labels[j];
1159 if (!var_add_value_label (v, &label->value, label->label))
1161 if (var_is_numeric (var[0]))
1162 sys_warn (r, _("Duplicate value label for %g on %s."),
1163 label->value.f, var_get_name (v));
1165 sys_warn (r, _("Duplicate value label for \"%.*s\" on %s."),
1166 var_get_width (v), label->value.s,
1172 pool_destroy (subpool);
1177 static void partial_record (struct sfm_reader *r)
1180 static void read_error (struct casereader *, const struct sfm_reader *);
1182 static bool read_case_number (struct sfm_reader *, double *);
1183 static bool read_case_string (struct sfm_reader *, char *, size_t);
1184 static int read_opcode (struct sfm_reader *);
1185 static bool read_compressed_number (struct sfm_reader *, double *);
1186 static bool read_compressed_string (struct sfm_reader *, char *);
1187 static bool read_whole_strings (struct sfm_reader *, char *, size_t);
1188 static bool skip_whole_strings (struct sfm_reader *, size_t);
1190 /* Reads one case from READER's file into C. Returns true only
1193 sys_file_casereader_read (struct casereader *reader, void *r_,
1196 struct sfm_reader *r = r_;
1202 case_create (c, r->value_cnt);
1203 if (setjmp (r->bail_out))
1205 casereader_force_error (reader);
1210 for (i = 0; i < r->sfm_var_cnt; i++)
1212 struct sfm_var *sv = &r->sfm_vars[i];
1213 union value *v = case_data_rw_idx (c, sv->case_index);
1217 if (!read_case_number (r, &v->f))
1222 if (!read_case_string (r, v->s + sv->offset, sv->width))
1224 if (!skip_whole_strings (r, ROUND_DOWN (sv->padding, 8)))
1234 if (r->case_cnt != -1)
1235 read_error (reader, r);
1239 /* Issues an error that R ends in a partial record. */
1241 partial_record (struct sfm_reader *r)
1243 sys_error (r, _("File ends in partial case."));
1246 /* Issues an error that an unspecified error occurred SFM, and
1249 read_error (struct casereader *r, const struct sfm_reader *sfm)
1251 msg (ME, _("Error reading case from file %s."), fh_get_name (sfm->fh));
1252 casereader_force_error (r);
1255 /* Reads a number from R and stores its value in *D.
1256 If R is compressed, reads a compressed number;
1257 otherwise, reads a number in the regular way.
1258 Returns true if successful, false if end of file is
1259 reached immediately. */
1261 read_case_number (struct sfm_reader *r, double *d)
1266 if (!try_read_bytes (r, number, sizeof number))
1268 float_convert (r->float_format, number, FLOAT_NATIVE_DOUBLE, d);
1272 return read_compressed_number (r, d);
1275 /* Reads LENGTH string bytes from R into S.
1276 Always reads a multiple of 8 bytes; if LENGTH is not a
1277 multiple of 8, then extra bytes are read and discarded without
1279 Reads compressed strings if S is compressed.
1280 Returns true if successful, false if end of file is
1281 reached immediately. */
1283 read_case_string (struct sfm_reader *r, char *s, size_t length)
1285 size_t whole = ROUND_DOWN (length, 8);
1286 size_t partial = length % 8;
1290 if (!read_whole_strings (r, s, whole))
1297 if (!read_whole_strings (r, bounce, sizeof bounce))
1303 memcpy (s + whole, bounce, partial);
1309 /* Reads and returns the next compression opcode from R. */
1311 read_opcode (struct sfm_reader *r)
1313 assert (r->compressed);
1317 if (r->opcode_idx >= sizeof r->opcodes)
1319 if (!try_read_bytes (r, r->opcodes, sizeof r->opcodes))
1323 opcode = r->opcodes[r->opcode_idx++];
1330 /* Reads a compressed number from R and stores its value in D.
1331 Returns true if successful, false if end of file is
1332 reached immediately. */
1334 read_compressed_number (struct sfm_reader *r, double *d)
1336 int opcode = read_opcode (r);
1344 *d = read_float (r);
1348 sys_error (r, _("Compressed data is corrupt."));
1355 *d = opcode - r->bias;
1362 /* Reads a compressed 8-byte string segment from R and stores it
1364 Returns true if successful, false if end of file is
1365 reached immediately. */
1367 read_compressed_string (struct sfm_reader *r, char *dst)
1369 switch (read_opcode (r))
1376 read_bytes (r, dst, 8);
1380 memset (dst, ' ', 8);
1384 sys_error (r, _("Compressed data is corrupt."));
1390 /* Reads LENGTH string bytes from R into S.
1391 LENGTH must be a multiple of 8.
1392 Reads compressed strings if S is compressed.
1393 Returns true if successful, false if end of file is
1394 reached immediately. */
1396 read_whole_strings (struct sfm_reader *r, char *s, size_t length)
1398 assert (length % 8 == 0);
1400 return try_read_bytes (r, s, length);
1404 for (ofs = 0; ofs < length; ofs += 8)
1405 if (!read_compressed_string (r, s + ofs))
1415 /* Skips LENGTH string bytes from R.
1416 LENGTH must be a multiple of 8.
1417 (LENGTH is also limited to 1024, but that's only because the
1418 current caller never needs more than that many bytes.)
1419 Returns true if successful, false if end of file is
1420 reached immediately. */
1422 skip_whole_strings (struct sfm_reader *r, size_t length)
1425 assert (length < sizeof buffer);
1426 return read_whole_strings (r, buffer, length);
1429 /* Creates and returns a table that can be used for translating a value
1430 index into a case to a "struct variable *" for DICT. Multiple
1431 system file fields reference variables this way.
1433 This table must be created before processing the very long
1434 string extension record, because that record causes some
1435 values to be deleted from the case and the dictionary to be
1437 static struct variable **
1438 make_var_by_value_idx (struct sfm_reader *r, struct dictionary *dict)
1440 struct variable **var_by_value_idx;
1444 var_by_value_idx = pool_nmalloc (r->pool,
1445 r->oct_cnt, sizeof *var_by_value_idx);
1446 for (i = 0; i < dict_get_var_cnt (dict); i++)
1448 struct variable *v = dict_get_var (dict, i);
1449 int nv = var_is_numeric (v) ? 1 : DIV_RND_UP (var_get_width (v), 8);
1452 var_by_value_idx[value_idx++] = v;
1453 for (j = 1; j < nv; j++)
1454 var_by_value_idx[value_idx++] = NULL;
1456 assert (value_idx == r->oct_cnt);
1458 return var_by_value_idx;
1461 /* Returns the "struct variable" corresponding to the given
1462 1-basd VALUE_IDX in VAR_BY_VALUE_IDX. Verifies that the index
1464 static struct variable *
1465 lookup_var_by_value_idx (struct sfm_reader *r,
1466 struct variable **var_by_value_idx, int value_idx)
1468 struct variable *var;
1470 if (value_idx < 1 || value_idx > r->oct_cnt)
1471 sys_error (r, _("Variable index %d not in valid range 1...%d."),
1472 value_idx, r->oct_cnt);
1474 var = var_by_value_idx[value_idx - 1];
1476 sys_error (r, _("Variable index %d refers to long string "
1483 /* Returns the variable in D with the given SHORT_NAME,
1484 or a null pointer if there is none. */
1485 static struct variable *
1486 lookup_var_by_short_name (struct dictionary *d, const char *short_name)
1488 struct variable *var;
1492 /* First try looking up by full name. This often succeeds. */
1493 var = dict_lookup_var (d, short_name);
1494 if (var != NULL && !strcasecmp (var_get_short_name (var, 0), short_name))
1497 /* Iterate through the whole dictionary as a fallback. */
1498 var_cnt = dict_get_var_cnt (d);
1499 for (i = 0; i < var_cnt; i++)
1501 var = dict_get_var (d, i);
1502 if (!strcasecmp (var_get_short_name (var, 0), short_name))
1509 /* Helpers for reading records that contain "variable=value"
1513 struct variable_to_value_map
1515 struct substring buffer; /* Record contents. */
1516 size_t pos; /* Current position in buffer. */
1519 /* Reads SIZE bytes into a "variable=value" map for R,
1520 and returns the map. */
1521 static struct variable_to_value_map *
1522 open_variable_to_value_map (struct sfm_reader *r, size_t size)
1524 struct variable_to_value_map *map = pool_alloc (r->pool, sizeof *map);
1525 char *buffer = pool_malloc (r->pool, size + 1);
1526 read_bytes (r, buffer, size);
1527 map->buffer = ss_buffer (buffer, size);
1532 /* Closes MAP and frees its storage.
1533 Not really needed, because the pool will free the map anyway,
1534 but can be used to free it earlier. */
1536 close_variable_to_value_map (struct sfm_reader *r,
1537 struct variable_to_value_map *map)
1539 pool_free (r->pool, ss_data (map->buffer));
1542 /* Reads the next variable=value pair from MAP.
1543 Looks up the variable in DICT and stores it into *VAR.
1544 Stores a null-terminated value into *VALUE. */
1546 read_variable_to_value_map (struct sfm_reader *r, struct dictionary *dict,
1547 struct variable_to_value_map *map,
1548 struct variable **var, char **value,
1551 int max_warnings = 5;
1555 struct substring short_name_ss, value_ss;
1557 if (!ss_tokenize (map->buffer, ss_cstr ("="), &map->pos, &short_name_ss)
1558 || !ss_tokenize (map->buffer, ss_buffer ("\t\0", 2), &map->pos,
1561 if (*warning_cnt > max_warnings)
1562 sys_warn (r, _("Suppressed %d additional variable map warnings."),
1563 *warning_cnt - max_warnings);
1567 map->pos += ss_span (ss_substr (map->buffer, map->pos, SIZE_MAX),
1568 ss_buffer ("\t\0", 2));
1570 ss_data (short_name_ss)[ss_length (short_name_ss)] = '\0';
1571 *var = lookup_var_by_short_name (dict, ss_data (short_name_ss));
1574 if (++*warning_cnt <= max_warnings)
1575 sys_warn (r, _("Variable map refers to unknown variable %s."),
1576 ss_data (short_name_ss));
1580 ss_data (value_ss)[ss_length (value_ss)] = '\0';
1581 *value = ss_data (value_ss);
1589 /* Displays a corruption message. */
1591 sys_msg (struct sfm_reader *r, int class, const char *format, va_list args)
1596 ds_init_empty (&text);
1597 ds_put_format (&text, "\"%s\" near offset 0x%lx: ",
1598 fh_get_file_name (r->fh), (unsigned long) ftell (r->file));
1599 ds_put_vformat (&text, format, args);
1601 m.category = msg_class_to_category (class);
1602 m.severity = msg_class_to_severity (class);
1603 m.where.file_name = NULL;
1604 m.where.line_number = 0;
1605 m.text = ds_cstr (&text);
1610 /* Displays a warning for the current file position. */
1612 sys_warn (struct sfm_reader *r, const char *format, ...)
1616 va_start (args, format);
1617 sys_msg (r, MW, format, args);
1621 /* Displays an error for the current file position,
1622 marks it as in an error state,
1623 and aborts reading it using longjmp. */
1625 sys_error (struct sfm_reader *r, const char *format, ...)
1629 va_start (args, format);
1630 sys_msg (r, ME, format, args);
1634 longjmp (r->bail_out, 1);
1637 /* Reads BYTE_CNT bytes into BUF.
1638 Returns true if exactly BYTE_CNT bytes are successfully read.
1639 Aborts if an I/O error or a partial read occurs.
1640 If EOF_IS_OK, then an immediate end-of-file causes false to be
1641 returned; otherwise, immediate end-of-file causes an abort
1644 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
1645 void *buf, size_t byte_cnt)
1647 size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
1648 if (bytes_read == byte_cnt)
1650 else if (ferror (r->file))
1651 sys_error (r, _("System error: %s."), strerror (errno));
1652 else if (!eof_is_ok || bytes_read != 0)
1653 sys_error (r, _("Unexpected end of file."));
1658 /* Reads BYTE_CNT into BUF.
1659 Aborts upon I/O error or if end-of-file is encountered. */
1661 read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1663 read_bytes_internal (r, false, buf, byte_cnt);
1666 /* Reads BYTE_CNT bytes into BUF.
1667 Returns true if exactly BYTE_CNT bytes are successfully read.
1668 Returns false if an immediate end-of-file is encountered.
1669 Aborts if an I/O error or a partial read occurs. */
1671 try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1673 return read_bytes_internal (r, true, buf, byte_cnt);
1676 /* Reads a 32-bit signed integer from R and returns its value in
1679 read_int (struct sfm_reader *r)
1682 read_bytes (r, integer, sizeof integer);
1683 return integer_get (r->integer_format, integer, sizeof integer);
1686 /* Reads a 64-bit floating-point number from R and returns its
1687 value in host format. */
1689 read_float (struct sfm_reader *r)
1692 read_bytes (r, number, sizeof number);
1693 return float_get_double (r->float_format, number);
1696 /* Reads exactly SIZE - 1 bytes into BUFFER
1697 and stores a null byte into BUFFER[SIZE - 1]. */
1699 read_string (struct sfm_reader *r, char *buffer, size_t size)
1702 read_bytes (r, buffer, size - 1);
1703 buffer[size - 1] = '\0';
1706 /* Skips BYTES bytes forward in R. */
1708 skip_bytes (struct sfm_reader *r, size_t bytes)
1713 size_t chunk = MIN (sizeof buffer, bytes);
1714 read_bytes (r, buffer, chunk);
1719 static struct casereader_class sys_file_casereader_class =
1721 sys_file_casereader_read,
1722 sys_file_casereader_destroy,