1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc.
3 Written by Ben Pfaff <blp@gnu.org>.
5 This program is free software; you can redistribute it and/or
6 modify it under the terms of the GNU General Public License as
7 published by the Free Software Foundation; either version 2 of the
8 License, or (at your option) any later version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
28 #include <libpspp/alloc.h>
29 #include <libpspp/assertion.h>
30 #include <libpspp/message.h>
31 #include <libpspp/compiler.h>
32 #include <libpspp/magic.h>
33 #include <libpspp/misc.h>
34 #include <libpspp/str.h>
35 #include <libpspp/hash.h>
36 #include <libpspp/array.h>
38 #include "sys-file-reader.h"
39 #include "sfm-private.h"
41 #include "dictionary.h"
42 #include "file-handle-def.h"
43 #include "file-name.h"
45 #include "value-labels.h"
50 #define _(msgid) gettext (msgid)
52 /* System file reader. */
55 struct file_handle *fh; /* File handle. */
56 FILE *file; /* File stream. */
58 int reverse_endian; /* 1=file has endianness opposite us. */
59 int value_cnt; /* Number of `union values's per case. */
60 long case_cnt; /* Number of cases, -1 if unknown. */
61 int compressed; /* 1=compressed, 0=not compressed. */
62 double bias; /* Compression bias, usually 100.0. */
63 int weight_idx; /* 0-based index of weighting variable, or -1. */
64 bool ok; /* False after an I/O error or corrupt data. */
65 bool has_vls; /* True if the file has one or more Very Long Strings*/
71 /* File's special constants. */
76 /* Decompression buffer. */
77 flt64 *buf; /* Buffer data. */
78 flt64 *ptr; /* Current location in buffer. */
79 flt64 *end; /* End of buffer data. */
81 /* Compression instruction octet. */
82 unsigned char x[8]; /* Current instruction octet. */
83 unsigned char *y; /* Location in current instruction octet. */
86 /* A variable in a system file. */
89 int width; /* 0=numeric, otherwise string width. */
90 int fv; /* Index into case. */
95 /* Swap bytes *A and *B. */
97 bswap (char *a, char *b)
104 /* Reverse the byte order of 32-bit integer *X. */
106 bswap_int32 (int32_t *x_)
108 char *x = (char *) x_;
109 bswap (x + 0, x + 3);
110 bswap (x + 1, x + 2);
113 /* Reverse the byte order of 64-bit floating point *X. */
115 bswap_flt64 (flt64 *x_)
117 char *x = (char *) x_;
118 bswap (x + 0, x + 7);
119 bswap (x + 1, x + 6);
120 bswap (x + 2, x + 5);
121 bswap (x + 3, x + 4);
125 corrupt_msg (int class, const char *format,...)
126 PRINTF_FORMAT (2, 3);
128 /* Displays a corrupt sysfile error. */
130 corrupt_msg (int class, const char *format,...)
136 ds_init_cstr (&text, _("corrupt system file: "));
137 va_start (args, format);
138 ds_put_vformat (&text, format, args);
141 m.category = msg_class_to_category (class);
142 m.severity = msg_class_to_severity (class);
143 m.where.file_name = NULL;
144 m.where.line_number = 0;
145 m.text = ds_cstr (&text);
150 /* Closes a system file after we're done with it. */
152 sfm_close_reader (struct sfm_reader *r)
159 if (fn_close (fh_get_file_name (r->fh), r->file) == EOF)
160 msg (ME, _("%s: Closing system file: %s."),
161 fh_get_file_name (r->fh), strerror (errno));
166 fh_close (r->fh, "system file", "rs");
173 /* Dictionary reader. */
175 static void buf_unread(struct sfm_reader *r, size_t byte_cnt);
177 static void *buf_read (struct sfm_reader *, void *buf, size_t byte_cnt,
180 static int read_header (struct sfm_reader *,
181 struct dictionary *, struct sfm_read_info *);
182 static int parse_format_spec (struct sfm_reader *, int32_t,
183 struct fmt_spec *, const struct variable *);
184 static int read_value_labels (struct sfm_reader *, struct dictionary *,
185 struct variable **var_by_idx);
186 static int read_variables (struct sfm_reader *,
187 struct dictionary *, struct variable ***var_by_idx);
188 static int read_machine_int32_info (struct sfm_reader *, int size, int count);
189 static int read_machine_flt64_info (struct sfm_reader *, int size, int count);
190 static int read_documents (struct sfm_reader *, struct dictionary *);
192 static int fread_ok (struct sfm_reader *, void *, size_t);
194 /* Displays the message X with corrupt_msg, then jumps to the error
202 /* Calls buf_read with the specified arguments, and jumps to
203 error if the read fails. */
204 #define assertive_buf_read(a,b,c,d) \
206 if (!buf_read (a,b,c,d)) \
218 pair_sn_compare(const void *_p1, const void *_p2, const void *aux UNUSED)
222 const struct name_pair *p1 = _p1;
223 const struct name_pair *p2 = _p2;
225 char buf1[SHORT_NAME_LEN + 1];
226 char buf2[SHORT_NAME_LEN + 1];
228 memset(buf1, 0, SHORT_NAME_LEN + 1);
229 memset(buf2, 0, SHORT_NAME_LEN + 1);
231 for (i = 0 ; i <= SHORT_NAME_LEN ; ++i )
233 buf1[i] = p1->shortname[i];
234 if ( '\0' == buf1[i])
238 for (i = 0 ; i <= SHORT_NAME_LEN ; ++i )
240 buf2[i] = p2->shortname[i];
241 if ( '\0' == buf2[i])
245 return strncmp(buf1, buf2, SHORT_NAME_LEN);
249 pair_sn_hash(const void *_p, const void *aux UNUSED)
252 const struct name_pair *p = _p;
253 char buf[SHORT_NAME_LEN + 1];
255 memset(buf, 0, SHORT_NAME_LEN + 1);
256 for (i = 0 ; i <= SHORT_NAME_LEN ; ++i )
258 buf[i] = p->shortname[i];
263 return hsh_hash_bytes(buf, strlen(buf));
267 pair_sn_free(void *p, const void *aux UNUSED)
274 /* Opens the system file designated by file handle FH for
275 reading. Reads the system file's dictionary into *DICT.
276 If INFO is non-null, then it receives additional info about the
279 sfm_open_reader (struct file_handle *fh, struct dictionary **dict,
280 struct sfm_read_info *info)
282 struct sfm_reader *r = NULL;
283 struct variable **var_by_idx = NULL;
285 /* The data in record 7(14) */
286 char *subrec14data = 0;
288 /* A hash table of long variable names indexed by short name */
289 struct hsh_table *short_to_long = NULL;
291 *dict = dict_create ();
292 if (!fh_open (fh, FH_REF_FILE, "system file", "rs"))
295 /* Create and initialize reader. */
296 r = xmalloc (sizeof *r);
298 r->file = fn_open (fh_get_file_name (fh), "rb");
300 r->reverse_endian = 0;
311 r->sysmis = -FLT64_MAX;
312 r->highest = FLT64_MAX;
313 r->lowest = second_lowest_flt64;
315 r->buf = r->ptr = r->end = NULL;
316 r->y = r->x + sizeof r->x;
318 /* Check that file open succeeded. */
321 msg (ME, _("An error occurred while opening \"%s\" for reading "
322 "as a system file: %s."),
323 fh_get_file_name (r->fh), strerror (errno));
327 /* Read header and variables. */
328 if (!read_header (r, *dict, info) || !read_variables (r, *dict, &var_by_idx))
332 /* Handle weighting. */
333 if (r->weight_idx != -1)
335 struct variable *weight_var;
337 if (r->weight_idx < 0 || r->weight_idx >= r->value_cnt)
338 lose ((ME, _("%s: Index of weighting variable (%d) is not between 0 "
339 "and number of elements per case (%d)."),
340 fh_get_file_name (r->fh), r->weight_idx, r->value_cnt));
343 weight_var = var_by_idx[r->weight_idx];
345 if (weight_var == NULL)
347 _("%s: Weighting variable may not be a continuation of "
348 "a long string variable."), fh_get_file_name (fh)));
349 else if (weight_var->type == ALPHA)
350 lose ((ME, _("%s: Weighting variable may not be a string variable."),
351 fh_get_file_name (fh)));
353 dict_set_weight (*dict, weight_var);
356 dict_set_weight (*dict, NULL);
358 /* Read records of types 3, 4, 6, and 7. */
363 assertive_buf_read (r, &rec_type, sizeof rec_type, 0);
364 if (r->reverse_endian)
365 bswap_int32 (&rec_type);
371 if (!read_value_labels (r, *dict, var_by_idx))
376 lose ((ME, _("%s: Orphaned variable index record (type 4). Type 4 "
377 "records must always immediately follow type 3 "
379 fh_get_file_name (r->fh)));
382 if (!read_documents (r, *dict))
393 } ATTRIBUTE((packed))
399 assertive_buf_read (r, &data, sizeof data, 0);
400 if (r->reverse_endian)
402 bswap_int32 (&data.subtype);
403 bswap_int32 (&data.size);
404 bswap_int32 (&data.count);
406 bytes = data.size * data.count;
408 if (bytes < data.size || bytes < data.count)
409 lose ((ME, "%s: Record type %d subtype %d too large.",
410 fh_get_file_name (r->fh), rec_type, data.subtype));
412 switch (data.subtype)
415 if (!read_machine_int32_info (r, data.size, data.count))
420 if (!read_machine_flt64_info (r, data.size, data.count))
425 case 6: /* ?? Used by SPSS 8.0. */
429 case 11: /* Variable display parameters */
431 const int n_vars = data.count / 3 ;
433 if ( data.count % 3 || n_vars != dict_get_var_cnt(*dict) )
435 msg (MW, _("%s: Invalid subrecord length. "
436 "Record: 7; Subrecord: 11"),
437 fh_get_file_name (r->fh));
442 for ( i = 0 ; i < min(n_vars, dict_get_var_cnt(*dict)) ; ++i )
449 } ATTRIBUTE((packed))
454 assertive_buf_read (r, ¶ms, sizeof(params), 0);
456 if ( ! measure_is_valid(params.measure)
458 ! alignment_is_valid(params.align))
461 _("%s: Invalid variable display parameters. Default parameters substituted."),
462 fh_get_file_name(r->fh));
466 v = dict_get_var(*dict, i);
468 v->measure = params.measure;
469 v->display_width = params.width;
470 v->alignment = params.align;
475 case 13: /* SPSS 12.0 Long variable name map */
478 char *save_ptr = NULL;
482 subrec14data = xmalloc (bytes + 1);
483 if (!buf_read (r, subrec14data, bytes, 0))
487 subrec14data[bytes] = '\0';
489 short_to_long = hsh_create(4,
496 for (short_name = strtok_r (subrec14data, "=", &save_ptr), idx = 0;
498 short_name = strtok_r (NULL, "=", &save_ptr), idx++)
500 struct name_pair *pair ;
501 char *long_name = strtok_r (NULL, "\t", &save_ptr);
504 /* Validate long name. */
505 if (long_name == NULL)
507 msg (MW, _("%s: Trailing garbage in long variable "
509 fh_get_file_name (r->fh));
512 if (!var_is_valid_name (long_name, false))
514 msg (MW, _("%s: Long variable mapping to invalid "
515 "variable name `%s'."),
516 fh_get_file_name (r->fh), long_name);
520 /* Find variable using short name. */
521 v = dict_lookup_var (*dict, short_name);
524 msg (MW, _("%s: Long variable mapping for "
525 "nonexistent variable %s."),
526 fh_get_file_name (r->fh), short_name);
530 /* Identify any duplicates. */
531 if ( compare_var_names(short_name, long_name, 0) &&
532 NULL != dict_lookup_var (*dict, long_name))
533 lose ((ME, _("%s: Duplicate long variable name `%s' "
534 "within system file."),
535 fh_get_file_name (r->fh), long_name));
539 Renaming a variable may clear the short
540 name, but we want to retain it, so
541 re-set it explicitly. */
542 dict_rename_var (*dict, v, long_name);
543 var_set_short_name (v, short_name);
545 pair = xmalloc(sizeof *pair);
546 pair->shortname = short_name;
547 pair->longname = long_name;
548 hsh_insert(short_to_long, pair);
550 /* This messes up the processing of subtype 14 (below).
551 I'm not sure if it is needed anyway, so I'm removing it for
552 now. If it's needed, then it will need to be done after all the
553 records have been processed. --- JMD 27 April 2006
556 /* For compatibility, make sure dictionary
557 is in long variable name map order. In
558 the common case, this has no effect,
559 because the dictionary and the long
560 variable name map are already in the
562 dict_reorder_var (*dict, v, idx);
572 bool eq_seen = false;
576 char *buffer = xmalloc (bytes + 1);
577 if (!buf_read (r, buffer, bytes, 0))
582 buffer[bytes] = '\0';
586 /* Note: SPSS v13 terminates this record with 00,
587 whereas SPSS v14 terminates it with 00 09. We must
589 for(i = 0; i < bytes ; ++i)
592 static char name[SHORT_NAME_LEN + 1] = {0};
593 static char len_str[6] ={0};
602 length = strtol(len_str, 0, 10);
603 if ( length != LONG_MAX && length != LONG_MIN)
605 char *lookup_name = name;
612 struct name_pair pair;
615 pair.shortname = name;
616 p = hsh_find(short_to_long, &pair);
618 lookup_name = p->longname;
621 v = dict_lookup_var(*dict, lookup_name);
625 _("%s: No variable called %s but it is listed in length table."),
626 fh_get_file_name (r->fh), lookup_name);
633 if ( v->width > EFFECTIVE_LONG_STRING_LENGTH )
634 l -= EFFECTIVE_LONG_STRING_LENGTH;
641 struct variable *v_next;
642 v_next = dict_get_var(*dict, idx + 1);
644 if ( v_next->width > EFFECTIVE_LONG_STRING_LENGTH )
645 l -= EFFECTIVE_LONG_STRING_LENGTH;
649 dict_delete_var(*dict, v_next);
652 assert ( length > MAX_LONG_STRING );
655 v->print.w = v->width;
656 v->write.w = v->width;
657 v->nv = DIV_RND_UP (length, MAX_SHORT_STRING);
660 memset(name, 0, SHORT_NAME_LEN+1);
661 memset(len_str, 0, 6);
668 len_str[j] = buffer[i];
676 dict_compact_values(*dict);
681 msg (MW, _("%s: Unrecognized record type 7, subtype %d "
682 "encountered in system file."),
683 fh_get_file_name (r->fh), data.subtype);
689 void *x = buf_read (r, NULL, data.size * data.count, 0);
701 assertive_buf_read (r, &filler, sizeof filler, 0);
707 corrupt_msg(MW, _("%s: Unrecognized record type %d."),
708 fh_get_file_name (r->fh), rec_type);
713 /* Come here on successful completion. */
715 /* Create an index of dictionary variable widths for
716 sfm_read_case to use. We cannot use the `struct variables'
717 from the dictionary we created, because the caller owns the
718 dictionary and may destroy or modify its variables. */
722 r->var_cnt = dict_get_var_cnt (*dict);
723 r->vars = xnmalloc (r->var_cnt, sizeof *r->vars);
724 for (i = 0; i < r->var_cnt; i++)
726 struct variable *v = dict_get_var (*dict, i);
727 struct sfm_var *sv = &r->vars[i];
728 sv->width = v->width;
734 hsh_destroy(short_to_long);
739 /* Come here on unsuccessful completion. */
740 sfm_close_reader (r);
742 hsh_destroy(short_to_long);
746 dict_destroy (*dict);
752 /* Read record type 7, subtype 3. */
754 read_machine_int32_info (struct sfm_reader *r, int size, int count)
761 if (size != sizeof (int32_t) || count != 8)
762 lose ((ME, _("%s: Bad size (%d) or count (%d) field on record type 7, "
763 "subtype 3. Expected size %d, count 8."),
764 fh_get_file_name (r->fh), size, count, sizeof (int32_t)));
766 assertive_buf_read (r, data, sizeof data, 0);
767 if (r->reverse_endian)
768 for (i = 0; i < 8; i++)
769 bswap_int32 (&data[i]);
773 lose ((ME, _("%s: Floating-point representation in system file is not "
774 "IEEE-754. PSPP cannot convert between floating-point "
776 fh_get_file_name (r->fh)));
778 #error Add support for your floating-point format.
781 #ifdef WORDS_BIGENDIAN
786 if (r->reverse_endian)
788 if (file_bigendian ^ (data[6] == 1))
789 lose ((ME, _("%s: File-indicated endianness (%s) does not match "
790 "endianness intuited from file header (%s)."),
791 fh_get_file_name (r->fh),
792 file_bigendian ? _("big-endian") : _("little-endian"),
793 data[6] == 1 ? _("big-endian") : (data[6] == 2 ? _("little-endian")
796 /* PORTME: Character representation code. */
797 if (data[7] != 2 && data[7] != 3)
798 lose ((ME, _("%s: File-indicated character representation code (%s) is "
800 fh_get_file_name (r->fh),
801 (data[7] == 1 ? "EBCDIC"
802 : (data[7] == 4 ? _("DEC Kanji") : _("Unknown")))));
810 /* Read record type 7, subtype 4. */
812 read_machine_flt64_info (struct sfm_reader *r, int size, int count)
817 if (size != sizeof (flt64) || count != 3)
818 lose ((ME, _("%s: Bad size (%d) or count (%d) field on record type 7, "
819 "subtype 4. Expected size %d, count 8."),
820 fh_get_file_name (r->fh), size, count, sizeof (flt64)));
822 assertive_buf_read (r, data, sizeof data, 0);
823 if (r->reverse_endian)
824 for (i = 0; i < 3; i++)
825 bswap_flt64 (&data[i]);
827 if (data[0] != SYSMIS || data[1] != FLT64_MAX
828 || data[2] != second_lowest_flt64)
831 r->highest = data[1];
833 msg (MW, _("%s: File-indicated value is different from internal value "
834 "for at least one of the three system values. SYSMIS: "
835 "indicated %g, expected %g; HIGHEST: %g, %g; LOWEST: "
837 fh_get_file_name (r->fh), (double) data[0], (double) SYSMIS,
838 (double) data[1], (double) FLT64_MAX,
839 (double) data[2], (double) second_lowest_flt64);
849 read_header (struct sfm_reader *r,
850 struct dictionary *dict, struct sfm_read_info *info)
852 struct sysfile_header hdr; /* Disk buffer. */
853 char prod_name[sizeof hdr.prod_name + 1]; /* Buffer for product name. */
854 int skip_amt = 0; /* Amount of product name to omit. */
857 /* Read header, check magic. */
858 assertive_buf_read (r, &hdr, sizeof hdr, 0);
859 if (strncmp ("$FL2", hdr.rec_type, 4) != 0)
860 lose ((ME, _("%s: Bad magic. Proper system files begin with "
861 "the four characters `$FL2'. This file will not be read."),
862 fh_get_file_name (r->fh)));
864 /* Check eye-category.her string. */
865 memcpy (prod_name, hdr.prod_name, sizeof hdr.prod_name);
866 for (i = 0; i < 60; i++)
867 if (!c_isprint ((unsigned char) prod_name[i]))
869 for (i = 59; i >= 0; i--)
870 if (!c_isgraph ((unsigned char) prod_name[i]))
875 prod_name[60] = '\0';
879 static const char *prefix[N_PREFIXES] =
881 "@(#) SPSS DATA FILE",
887 for (i = 0; i < N_PREFIXES; i++)
888 if (!strncmp (prefix[i], hdr.prod_name, strlen (prefix[i])))
890 skip_amt = strlen (prefix[i]);
895 /* Check endianness. */
896 if (hdr.layout_code == 2)
897 r->reverse_endian = 0;
900 bswap_int32 (&hdr.layout_code);
901 if (hdr.layout_code != 2)
902 lose ((ME, _("%s: File layout code has unexpected value %d. Value "
903 "should be 2, in big-endian or little-endian format."),
904 fh_get_file_name (r->fh), hdr.layout_code));
906 r->reverse_endian = 1;
907 bswap_int32 (&hdr.nominal_case_size);
908 bswap_int32 (&hdr.compress);
909 bswap_int32 (&hdr.weight_idx);
910 bswap_int32 (&hdr.case_cnt);
911 bswap_flt64 (&hdr.bias);
915 /* Copy basic info and verify correctness. */
916 r->value_cnt = hdr.nominal_case_size;
918 /* If value count is ridiculous, then force it to -1 (a
920 if ( r->value_cnt < 0 ||
921 r->value_cnt > (INT_MAX / (int) sizeof (union value) / 2))
924 r->compressed = hdr.compress;
926 r->weight_idx = hdr.weight_idx - 1;
928 r->case_cnt = hdr.case_cnt;
929 if (r->case_cnt < -1 || r->case_cnt > INT_MAX / 2)
931 _("%s: Number of cases in file (%ld) is not between -1 and %d."),
932 fh_get_file_name (r->fh), (long) r->case_cnt, INT_MAX / 2));
935 if (r->bias != 100.0)
936 corrupt_msg (MW, _("%s: Compression bias (%g) is not the usual "
938 fh_get_file_name (r->fh), r->bias);
940 /* Make a file label only on the condition that the given label is
941 not all spaces or nulls. */
945 for (i = sizeof hdr.file_label - 1; i >= 0; i--)
947 if (!c_isspace ((unsigned char) hdr.file_label[i])
948 && hdr.file_label[i] != 0)
950 char *label = xmalloc (i + 2);
951 memcpy (label, hdr.file_label, i + 1);
953 dict_set_label (dict, label);
964 memcpy (info->creation_date, hdr.creation_date, 9);
965 info->creation_date[9] = 0;
967 memcpy (info->creation_time, hdr.creation_time, 8);
968 info->creation_time[8] = 0;
970 #ifdef WORDS_BIGENDIAN
971 info->big_endian = !r->reverse_endian;
973 info->big_endian = r->reverse_endian;
976 info->compressed = hdr.compress;
978 info->case_cnt = hdr.case_cnt;
980 for (cp = &prod_name[skip_amt]; cp < &prod_name[60]; cp++)
981 if (c_isgraph ((unsigned char) *cp))
983 strcpy (info->product, cp);
992 /* Reads most of the dictionary from file H; also fills in the
993 associated VAR_BY_IDX array. */
995 read_variables (struct sfm_reader *r,
996 struct dictionary *dict, struct variable ***var_by_idx)
1000 struct sysfile_variable sv; /* Disk buffer. */
1001 int long_string_count = 0; /* # of long string continuation
1002 records still expected. */
1003 int next_value = 0; /* Index to next `value' structure. */
1010 /* Read in the entry for each variable and use the info to
1011 initialize the dictionary. */
1014 struct variable *vv;
1015 char name[SHORT_NAME_LEN + 1];
1019 assertive_buf_read (r, &sv, sizeof sv, 0);
1021 if (r->reverse_endian)
1023 bswap_int32 (&sv.rec_type);
1024 bswap_int32 (&sv.type);
1025 bswap_int32 (&sv.has_var_label);
1026 bswap_int32 (&sv.n_missing_values);
1027 bswap_int32 (&sv.print);
1028 bswap_int32 (&sv.write);
1031 /* We've come to the end of the variable entries */
1032 if (sv.rec_type != 2)
1034 buf_unread(r, sizeof sv);
1039 *var_by_idx = xnrealloc (*var_by_idx, i + 1, sizeof **var_by_idx);
1041 /* If there was a long string previously, make sure that the
1042 continuations are present; otherwise make sure there aren't
1044 if (long_string_count)
1047 lose ((ME, _("%s: position %d: String variable does not have "
1048 "proper number of continuation records."),
1049 fh_get_file_name (r->fh), i));
1052 (*var_by_idx)[i] = NULL;
1053 long_string_count--;
1056 else if (sv.type == -1)
1057 lose ((ME, _("%s: position %d: Superfluous long string continuation "
1059 fh_get_file_name (r->fh), i));
1061 /* Check fields for validity. */
1062 if (sv.type < 0 || sv.type > 255)
1063 lose ((ME, _("%s: position %d: Bad variable type code %d."),
1064 fh_get_file_name (r->fh), i, sv.type));
1065 if (sv.has_var_label != 0 && sv.has_var_label != 1)
1066 lose ((ME, _("%s: position %d: Variable label indicator field is not "
1067 "0 or 1."), fh_get_file_name (r->fh), i));
1068 if (sv.n_missing_values < -3 || sv.n_missing_values > 3
1069 || sv.n_missing_values == -1)
1070 lose ((ME, _("%s: position %d: Missing value indicator field is not "
1071 "-3, -2, 0, 1, 2, or 3."), fh_get_file_name (r->fh), i));
1073 /* Copy first character of variable name. */
1074 if (sv.name[0] == '@' || sv.name[0] == '#')
1075 lose ((ME, _("%s: position %d: Variable name begins with invalid "
1077 fh_get_file_name (r->fh), i));
1079 name[0] = sv.name[0];
1081 /* Copy remaining characters of variable name. */
1082 for (j = 1; j < SHORT_NAME_LEN; j++)
1084 int c = (unsigned char) sv.name[j];
1093 if ( ! var_is_plausible_name(name, false) )
1094 lose ((ME, _("%s: Invalid variable name `%s' within system file."),
1095 fh_get_file_name (r->fh), name));
1097 /* Create variable. */
1098 vv = (*var_by_idx)[i] = dict_create_var (dict, name, sv.type);
1100 lose ((ME, _("%s: Duplicate variable name `%s' within system file."),
1101 fh_get_file_name (r->fh), name));
1103 /* Set the short name the same as the long name */
1104 var_set_short_name (vv, vv->name);
1106 /* Case reading data. */
1107 nv = sv.type == 0 ? 1 : DIV_RND_UP (sv.type, sizeof (flt64));
1108 long_string_count = nv - 1;
1111 /* Get variable label, if any. */
1112 if (sv.has_var_label == 1)
1117 /* Read length of label. */
1118 assertive_buf_read (r, &len, sizeof len, 0);
1119 if (r->reverse_endian)
1123 if (len < 0 || len > 255)
1124 lose ((ME, _("%s: Variable %s indicates variable label of invalid "
1126 fh_get_file_name (r->fh), vv->name, len));
1130 /* Read label into variable structure. */
1131 vv->label = buf_read (r, NULL, ROUND_UP (len, sizeof (int32_t)), len + 1);
1132 if (vv->label == NULL)
1134 vv->label[len] = '\0';
1138 /* Set missing values. */
1139 if (sv.n_missing_values != 0)
1142 int mv_cnt = abs (sv.n_missing_values);
1144 if (vv->width > MAX_SHORT_STRING)
1145 lose ((ME, _("%s: Long string variable %s may not have missing "
1147 fh_get_file_name (r->fh), vv->name));
1149 assertive_buf_read (r, mv, sizeof *mv * mv_cnt, 0);
1151 if (r->reverse_endian && vv->type == NUMERIC)
1152 for (j = 0; j < mv_cnt; j++)
1153 bswap_flt64 (&mv[j]);
1155 if (sv.n_missing_values > 0)
1157 for (j = 0; j < sv.n_missing_values; j++)
1158 if (vv->type == NUMERIC)
1159 mv_add_num (&vv->miss, mv[j]);
1161 mv_add_str (&vv->miss, (char *) &mv[j]);
1165 if (vv->type == ALPHA)
1166 lose ((ME, _("%s: String variable %s may not have missing "
1167 "values specified as a range."),
1168 fh_get_file_name (r->fh), vv->name));
1170 if (mv[0] == r->lowest)
1171 mv_add_num_range (&vv->miss, LOWEST, mv[1]);
1172 else if (mv[1] == r->highest)
1173 mv_add_num_range (&vv->miss, mv[0], HIGHEST);
1175 mv_add_num_range (&vv->miss, mv[0], mv[1]);
1177 if (sv.n_missing_values == -3)
1178 mv_add_num (&vv->miss, mv[2]);
1182 if (!parse_format_spec (r, sv.print, &vv->print, vv)
1183 || !parse_format_spec (r, sv.write, &vv->write, vv))
1187 /* Some consistency checks. */
1188 if (long_string_count != 0)
1189 lose ((ME, _("%s: Long string continuation records omitted at end of "
1191 fh_get_file_name (r->fh)));
1193 if (next_value != r->value_cnt)
1194 corrupt_msg(MW, _("%s: System file header indicates %d variable positions but "
1195 "%d were read from file."),
1196 fh_get_file_name (r->fh), r->value_cnt, next_value);
1205 /* Translates the format spec from sysfile format to internal
1208 parse_format_spec (struct sfm_reader *r, int32_t s,
1209 struct fmt_spec *f, const struct variable *v)
1211 f->type = translate_fmt ((s >> 16) & 0xff);
1213 lose ((ME, _("%s: Bad format specifier byte (%d)."),
1214 fh_get_file_name (r->fh), (s >> 16) & 0xff));
1215 f->w = (s >> 8) & 0xff;
1218 if ((v->type == ALPHA) ^ ((formats[f->type].cat & FCAT_STRING) != 0))
1219 lose ((ME, _("%s: %s variable %s has %s format specifier %s."),
1220 fh_get_file_name (r->fh),
1221 v->type == ALPHA ? _("String") : _("Numeric"),
1223 formats[f->type].cat & FCAT_STRING ? _("string") : _("numeric"),
1224 formats[f->type].name));
1226 if (!check_output_specifier (f, false)
1227 || !check_specifier_width (f, v->width, false))
1229 msg (ME, _("%s variable %s has invalid format specifier %s."),
1230 v->type == NUMERIC ? _("Numeric") : _("String"),
1231 v->name, fmt_to_string (f));
1232 *f = v->type == NUMERIC ? f8_2 : make_output_format (FMT_A, v->width, 0);
1240 /* Reads value labels from sysfile H and inserts them into the
1241 associated dictionary. */
1243 read_value_labels (struct sfm_reader *r,
1244 struct dictionary *dict, struct variable **var_by_idx)
1248 char raw_value[8]; /* Value as uninterpreted bytes. */
1249 union value value; /* Value. */
1250 char *label; /* Null-terminated label string. */
1253 struct label *labels = NULL;
1254 int32_t n_labels; /* Number of labels. */
1256 struct variable **var = NULL; /* Associated variables. */
1257 int32_t n_vars; /* Number of associated variables. */
1261 /* First step: read the contents of the type 3 record and record its
1262 contents. Note that we can't do much with the data since we
1263 don't know yet whether it is of numeric or string type. */
1265 /* Read number of labels. */
1266 assertive_buf_read (r, &n_labels, sizeof n_labels, 0);
1267 if (r->reverse_endian)
1268 bswap_int32 (&n_labels);
1270 if ( n_labels >= ((int32_t) ~0) / sizeof *labels)
1272 corrupt_msg(MW, _("%s: Invalid number of labels: %d. Ignoring labels."),
1273 fh_get_file_name (r->fh), n_labels);
1277 /* Allocate memory. */
1278 labels = xcalloc (n_labels, sizeof *labels);
1279 for (i = 0; i < n_labels; i++)
1280 labels[i].label = NULL;
1282 /* Read each value/label tuple into labels[]. */
1283 for (i = 0; i < n_labels; i++)
1285 struct label *label = labels + i;
1286 unsigned char label_len;
1290 assertive_buf_read (r, label->raw_value, sizeof label->raw_value, 0);
1292 /* Read label length. */
1293 assertive_buf_read (r, &label_len, sizeof label_len, 0);
1294 padded_len = ROUND_UP (label_len + 1, sizeof (flt64));
1296 /* Read label, padding. */
1297 label->label = xmalloc (padded_len + 1);
1298 assertive_buf_read (r, label->label, padded_len - 1, 0);
1299 label->label[label_len] = 0;
1302 /* Second step: Read the type 4 record that has the list of
1303 variables to which the value labels are to be applied. */
1305 /* Read record type of type 4 record. */
1309 assertive_buf_read (r, &rec_type, sizeof rec_type, 0);
1310 if (r->reverse_endian)
1311 bswap_int32 (&rec_type);
1314 lose ((ME, _("%s: Variable index record (type 4) does not immediately "
1315 "follow value label record (type 3) as it should."),
1316 fh_get_file_name (r->fh)));
1319 /* Read number of variables associated with value label from type 4
1321 assertive_buf_read (r, &n_vars, sizeof n_vars, 0);
1322 if (r->reverse_endian)
1323 bswap_int32 (&n_vars);
1324 if (n_vars < 1 || n_vars > dict_get_var_cnt (dict))
1325 lose ((ME, _("%s: Number of variables associated with a value label (%d) "
1326 "is not between 1 and the number of variables (%d)."),
1327 fh_get_file_name (r->fh), n_vars, dict_get_var_cnt (dict)));
1329 /* Read the list of variables. */
1330 var = xnmalloc (n_vars, sizeof *var);
1331 for (i = 0; i < n_vars; i++)
1336 /* Read variable index, check range. */
1337 assertive_buf_read (r, &var_idx, sizeof var_idx, 0);
1338 if (r->reverse_endian)
1339 bswap_int32 (&var_idx);
1340 if (var_idx < 1 || var_idx > r->value_cnt)
1341 lose ((ME, _("%s: Variable index associated with value label (%d) is "
1342 "not between 1 and the number of values (%d)."),
1343 fh_get_file_name (r->fh), var_idx, r->value_cnt));
1345 /* Make sure it's a real variable. */
1346 v = var_by_idx[var_idx - 1];
1348 lose ((ME, _("%s: Variable index associated with value label (%d) "
1349 "refers to a continuation of a string variable, not to "
1350 "an actual variable."),
1351 fh_get_file_name (r->fh), var_idx));
1352 if (v->type == ALPHA && v->width > MAX_SHORT_STRING)
1353 lose ((ME, _("%s: Value labels are not allowed on long string "
1355 fh_get_file_name (r->fh), v->name));
1357 /* Add it to the list of variables. */
1361 /* Type check the variables. */
1362 for (i = 1; i < n_vars; i++)
1363 if (var[i]->type != var[0]->type)
1364 lose ((ME, _("%s: Variables associated with value label are not all of "
1365 "identical type. Variable %s has %s type, but variable "
1367 fh_get_file_name (r->fh),
1368 var[0]->name, var[0]->type == ALPHA ? _("string") : _("numeric"),
1369 var[i]->name, var[i]->type == ALPHA ? _("string") : _("numeric")));
1371 /* Fill in labels[].value, now that we know the desired type. */
1372 for (i = 0; i < n_labels; i++)
1374 struct label *label = labels + i;
1376 if (var[0]->type == ALPHA)
1378 const int copy_len = min (sizeof label->raw_value,
1379 sizeof label->label);
1380 memcpy (label->value.s, label->raw_value, copy_len);
1383 assert (sizeof f == sizeof label->raw_value);
1384 memcpy (&f, label->raw_value, sizeof f);
1385 if (r->reverse_endian)
1391 /* Assign the value_label's to each variable. */
1392 for (i = 0; i < n_vars; i++)
1394 struct variable *v = var[i];
1397 /* Add each label to the variable. */
1398 for (j = 0; j < n_labels; j++)
1400 struct label *label = labels + j;
1401 if (!val_labs_replace (v->val_labs, label->value, label->label))
1404 if (var[0]->type == NUMERIC)
1405 msg (MW, _("%s: File contains duplicate label for value %g for "
1407 fh_get_file_name (r->fh), label->value.f, v->name);
1409 msg (MW, _("%s: File contains duplicate label for value `%.*s' "
1410 "for variable %s."),
1411 fh_get_file_name (r->fh), v->width, label->value.s, v->name);
1415 for (i = 0; i < n_labels; i++)
1416 free (labels[i].label);
1424 for (i = 0; i < n_labels; i++)
1425 free (labels[i].label);
1432 /* Reads BYTE_CNT bytes from the file represented by H. If BUF is
1433 non-NULL, uses that as the buffer; otherwise allocates at least
1434 MIN_ALLOC bytes. Returns a pointer to the buffer on success, NULL
1437 buf_read (struct sfm_reader *r, void *buf, size_t byte_cnt, size_t min_alloc)
1441 if (buf == NULL && byte_cnt > 0 )
1442 buf = xmalloc (max (byte_cnt, min_alloc));
1444 if ( byte_cnt == 0 )
1448 if (1 != fread (buf, byte_cnt, 1, r->file))
1450 if (ferror (r->file))
1451 msg (ME, _("%s: Reading system file: %s."),
1452 fh_get_file_name (r->fh), strerror (errno));
1454 corrupt_msg (ME, _("%s: Unexpected end of file."),
1455 fh_get_file_name (r->fh));
1463 /* Winds the reader BYTE_CNT bytes back in the reader stream. */
1465 buf_unread(struct sfm_reader *r, size_t byte_cnt)
1467 assert(byte_cnt > 0);
1469 if ( 0 != fseek(r->file, -byte_cnt, SEEK_CUR))
1471 msg (ME, _("%s: Seeking system file: %s."),
1472 fh_get_file_name (r->fh), strerror (errno));
1476 /* Reads a document record, type 6, from system file R, and sets up
1477 the documents and n_documents fields in the associated
1480 read_documents (struct sfm_reader *r, struct dictionary *dict)
1485 if (dict_get_documents (dict) != NULL)
1486 lose ((ME, _("%s: System file contains multiple "
1487 "type 6 (document) records."),
1488 fh_get_file_name (r->fh)));
1490 assertive_buf_read (r, &line_cnt, sizeof line_cnt, 0);
1492 lose ((ME, _("%s: Number of document lines (%ld) "
1493 "must be greater than 0."),
1494 fh_get_file_name (r->fh), (long) line_cnt));
1496 documents = buf_read (r, NULL, 80 * line_cnt, line_cnt * 80 + 1);
1497 /* FIXME? Run through asciify. */
1498 if (documents == NULL)
1500 documents[80 * line_cnt] = '\0';
1501 dict_set_documents (dict, documents);
1511 /* Reads compressed data into H->BUF and sets other pointers
1512 appropriately. Returns nonzero only if both no errors occur and
1515 buffer_input (struct sfm_reader *r)
1522 r->buf = xnmalloc (128, sizeof *r->buf);
1523 amt = fread (r->buf, sizeof *r->buf, 128, r->file);
1524 if (ferror (r->file))
1526 msg (ME, _("%s: Error reading file: %s."),
1527 fh_get_file_name (r->fh), strerror (errno));
1532 r->end = &r->buf[amt];
1536 /* Reads a single case consisting of compressed data from system
1537 file H into the array BUF[] according to reader R, and
1538 returns nonzero only if successful. */
1539 /* Data in system files is compressed in this manner. Data
1540 values are grouped into sets of eight ("octets"). Each value
1541 in an octet has one instruction byte that are output together.
1542 Each instruction byte gives a value for that byte or indicates
1543 that the value can be found following the instructions. */
1545 read_compressed_data (struct sfm_reader *r, flt64 *buf)
1547 const unsigned char *p_end = r->x + sizeof (flt64);
1548 unsigned char *p = r->y;
1550 const flt64 *buf_beg = buf;
1551 const flt64 *buf_end = &buf[r->value_cnt];
1555 for (; p < p_end; p++){
1559 /* Code 0 is ignored. */
1562 /* Code 252 is end of file. */
1565 lose ((ME, _("%s: Compressed data is corrupted. Data ends "
1566 "in partial case."),
1567 fh_get_file_name (r->fh)));
1569 /* Code 253 indicates that the value is stored explicitly
1570 following the instruction bytes. */
1571 if (r->ptr == NULL || r->ptr >= r->end)
1572 if (!buffer_input (r))
1573 lose ((ME, _("%s: Unexpected end of file."),
1574 fh_get_file_name (r->fh)));
1575 memcpy (buf++, r->ptr++, sizeof *buf);
1580 /* Code 254 indicates a string that is all blanks. */
1581 memset (buf++, ' ', sizeof *buf);
1586 /* Code 255 indicates the system-missing value. */
1588 if (r->reverse_endian)
1595 /* Codes 1 through 251 inclusive are taken to indicate a
1596 value of (BYTE - BIAS), where BYTE is the byte's value
1597 and BIAS is the compression bias (generally 100.0). */
1598 *buf = *p - r->bias;
1599 if (r->reverse_endian)
1607 /* We have reached the end of this instruction octet. Read
1609 if (r->ptr == NULL || r->ptr >= r->end)
1611 if (!buffer_input (r))
1614 lose ((ME, _("%s: Unexpected end of file."),
1615 fh_get_file_name (r->fh)));
1620 memcpy (r->x, r->ptr++, sizeof *buf);
1627 /* We have filled up an entire record. Update state and return
1638 /* Reads one case from READER's file into C. Returns nonzero
1639 only if successful. */
1641 sfm_read_case (struct sfm_reader *r, struct ccase *c)
1646 if (!r->compressed && sizeof (flt64) == sizeof (double) && ! r->has_vls)
1648 /* Fast path: external and internal representations are the
1649 same, except possibly for endianness or SYSMIS. Read
1650 directly into the case's buffer, then fix up any minor
1651 details as needed. */
1652 if (!fread_ok (r, case_data_all_rw (c),
1653 sizeof (union value) * r->value_cnt))
1656 /* Fix up endianness if needed. */
1657 if (r->reverse_endian)
1661 for (i = 0; i < r->var_cnt; i++)
1662 if (r->vars[i].width == 0)
1663 bswap_flt64 (&case_data_rw (c, r->vars[i].fv)->f);
1666 /* Fix up SYSMIS values if needed.
1667 I don't think this will ever actually kick in, but it
1669 if (r->sysmis != SYSMIS)
1673 for (i = 0; i < r->var_cnt; i++)
1674 if (r->vars[i].width == 0 && case_num (c, i) == r->sysmis)
1675 case_data_rw (c, r->vars[i].fv)->f = SYSMIS;
1680 /* Slow path: internal and external representations differ.
1681 Read into a bounce buffer, then copy to C. */
1688 bounce_size = sizeof *bounce * r->value_cnt;
1689 bounce = bounce_cur = local_alloc (bounce_size);
1691 memset(bounce, 0, bounce_size);
1694 read_ok = fread_ok (r, bounce, bounce_size);
1696 read_ok = read_compressed_data (r, bounce);
1699 local_free (bounce);
1703 for (i = 0; i < r->var_cnt; i++)
1705 struct sfm_var *sv = &r->vars[i];
1709 flt64 f = *bounce_cur++;
1710 if (r->reverse_endian)
1712 case_data_rw (c, sv->fv)->f = f == r->sysmis ? SYSMIS : f;
1716 flt64 *bc_start = bounce_cur;
1718 while (ofs < sv->width )
1720 const int chunk = MIN (MAX_LONG_STRING, sv->width - ofs);
1721 memcpy (case_data_rw (c, sv->fv)->s + ofs, bounce_cur, chunk);
1723 bounce_cur += DIV_RND_UP (chunk, sizeof (flt64));
1727 bounce_cur = bc_start + width_to_bytes(sv->width) / sizeof(flt64);
1731 local_free (bounce);
1737 fread_ok (struct sfm_reader *r, void *buffer, size_t byte_cnt)
1739 size_t read_bytes = fread (buffer, 1, byte_cnt, r->file);
1741 if (read_bytes == byte_cnt)
1745 if (ferror (r->file))
1747 msg (ME, _("%s: Reading system file: %s."),
1748 fh_get_file_name (r->fh), strerror (errno));
1751 else if (read_bytes != 0)
1753 msg (ME, _("%s: Partial record at end of system file."),
1754 fh_get_file_name (r->fh));
1761 /* Returns true if an I/O error has occurred on READER, false
1764 sfm_read_error (const struct sfm_reader *reader)
1769 /* Returns true if FILE is an SPSS system file,
1772 sfm_detect (FILE *file)
1774 struct sysfile_header hdr;
1776 if (fread (&hdr, sizeof hdr, 1, file) != 1)
1778 if (strncmp ("$FL2", hdr.rec_type, 4))