1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 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/>. */
28 #include "data/val-type.h"
29 #include "libpspp/cast.h"
30 #include "libpspp/compiler.h"
31 #include "libpspp/float-format.h"
32 #include "libpspp/integer-format.h"
33 #include "libpspp/misc.h"
36 #include "gl/minmax.h"
37 #include "gl/progname.h"
38 #include "gl/version-etc.h"
39 #include "gl/xalloc.h"
53 const char *file_name;
56 int n_variable_records, n_variables;
59 size_t n_var_widths, allocated_var_widths;
61 enum integer_format integer_format;
62 enum float_format float_format;
64 enum compression compression;
68 static void read_header (struct sfm_reader *);
69 static void read_variable_record (struct sfm_reader *);
70 static void read_value_label_record (struct sfm_reader *);
71 static void read_document_record (struct sfm_reader *);
72 static void read_extension_record (struct sfm_reader *);
73 static void read_machine_integer_info (struct sfm_reader *,
74 size_t size, size_t count);
75 static void read_machine_float_info (struct sfm_reader *,
76 size_t size, size_t count);
77 static void read_extra_product_info (struct sfm_reader *,
78 size_t size, size_t count);
79 static void read_variable_sets (struct sfm_reader *, size_t size, size_t count);
80 static void read_mrsets (struct sfm_reader *, size_t size, size_t count);
81 static void read_display_parameters (struct sfm_reader *,
82 size_t size, size_t count);
83 static void read_long_var_name_map (struct sfm_reader *r,
84 size_t size, size_t count);
85 static void read_long_string_map (struct sfm_reader *r,
86 size_t size, size_t count);
87 static void read_datafile_attributes (struct sfm_reader *r,
88 size_t size, size_t count);
89 static void read_variable_attributes (struct sfm_reader *r,
90 size_t size, size_t count);
91 static void read_ncases64 (struct sfm_reader *, size_t size, size_t count);
92 static void read_character_encoding (struct sfm_reader *r,
93 size_t size, size_t count);
94 static void read_long_string_value_labels (struct sfm_reader *r,
95 size_t size, size_t count);
96 static void read_long_string_missing_values (struct sfm_reader *r,
97 size_t size, size_t count);
98 static void read_unknown_extension (struct sfm_reader *,
99 size_t size, size_t count);
100 static void read_simple_compressed_data (struct sfm_reader *, int max_cases);
101 static void read_zlib_compressed_data (struct sfm_reader *);
103 static struct text_record *open_text_record (
104 struct sfm_reader *, size_t size, size_t count);
105 static void close_text_record (struct text_record *);
106 static bool read_variable_to_value_pair (struct text_record *,
107 char **key, char **value);
108 static char *text_tokenize (struct text_record *, int delimiter);
109 static bool text_match (struct text_record *text, int c);
110 static const char *text_parse_counted_string (struct text_record *);
111 static size_t text_pos (const struct text_record *);
112 static const char *text_get_all (const struct text_record *);
114 static void usage (void);
115 static void sys_warn (struct sfm_reader *, const char *, ...)
116 PRINTF_FORMAT (2, 3);
117 static void sys_error (struct sfm_reader *, const char *, ...)
121 static void read_bytes (struct sfm_reader *, void *, size_t);
122 static bool try_read_bytes (struct sfm_reader *, void *, size_t);
123 static int read_int (struct sfm_reader *);
124 static int64_t read_int64 (struct sfm_reader *);
125 static double read_float (struct sfm_reader *);
126 static void read_string (struct sfm_reader *, char *, size_t);
127 static void skip_bytes (struct sfm_reader *, size_t);
128 static void trim_spaces (char *);
130 static void print_string (const char *s, size_t len);
133 main (int argc, char *argv[])
139 set_program_name (argv[0]);
143 static const struct option long_options[] =
145 { "data", optional_argument, NULL, 'd' },
146 { "help", no_argument, NULL, 'h' },
147 { "version", no_argument, NULL, 'v' },
148 { NULL, 0, NULL, 0 },
153 c = getopt_long (argc, argv, "d::hv", long_options, NULL);
160 max_cases = optarg ? atoi (optarg) : INT_MAX;
164 version_etc (stdout, "pspp-dump-sav", PACKAGE_NAME, PACKAGE_VERSION,
165 "Ben Pfaff", "John Darrington", NULL_SENTINEL);
178 error (1, 0, "at least one non-option argument is required; "
179 "use --help for help");
181 for (i = optind; i < argc; i++)
185 r.file_name = argv[i];
186 r.file = fopen (r.file_name, "rb");
188 error (EXIT_FAILURE, errno, "error opening `%s'", r.file_name);
189 r.n_variable_records = 0;
192 r.allocated_var_widths = 0;
194 r.compression = COMP_NONE;
196 if (argc - optind > 1)
197 printf ("Reading \"%s\":\n", r.file_name);
200 while ((rec_type = read_int (&r)) != 999)
205 read_variable_record (&r);
209 read_value_label_record (&r);
213 sys_error (&r, "Misplaced type 4 record.");
216 read_document_record (&r);
220 read_extension_record (&r);
224 sys_error (&r, "Unrecognized record type %d.", rec_type);
227 printf ("%08llx: end-of-dictionary record "
228 "(first byte of data at %08llx)\n",
229 (long long int) ftello (r.file),
230 (long long int) ftello (r.file) + 4);
232 if (r.compression == COMP_SIMPLE)
235 read_simple_compressed_data (&r, max_cases);
237 else if (r.compression == COMP_ZLIB)
238 read_zlib_compressed_data (&r);
249 read_header (struct sfm_reader *r)
252 char eye_catcher[61];
253 uint8_t raw_layout_code[4];
256 int32_t weight_index;
259 char creation_date[10];
260 char creation_time[9];
264 read_string (r, rec_type, sizeof rec_type);
265 read_string (r, eye_catcher, sizeof eye_catcher);
267 if (!strcmp ("$FL2", rec_type))
269 else if (!strcmp ("$FL3", rec_type))
272 sys_error (r, "This is not an SPSS system file.");
274 /* Identify integer format. */
275 read_bytes (r, raw_layout_code, sizeof raw_layout_code);
276 if ((!integer_identify (2, raw_layout_code, sizeof raw_layout_code,
278 && !integer_identify (3, raw_layout_code, sizeof raw_layout_code,
280 || (r->integer_format != INTEGER_MSB_FIRST
281 && r->integer_format != INTEGER_LSB_FIRST))
282 sys_error (r, "This is not an SPSS system file.");
283 layout_code = integer_get (r->integer_format,
284 raw_layout_code, sizeof raw_layout_code);
286 read_int (r); /* Nominal case size (not actually useful). */
287 compressed = read_int (r);
288 weight_index = read_int (r);
289 ncases = read_int (r);
294 r->compression = COMP_NONE;
295 else if (compressed == 1)
296 r->compression = COMP_SIMPLE;
297 else if (compressed != 0)
298 sys_error (r, "SAV file header has invalid compression value "
299 "%"PRId32".", compressed);
304 r->compression = COMP_ZLIB;
306 sys_error (r, "ZSAV file header has invalid compression value "
307 "%"PRId32".", compressed);
310 /* Identify floating-point format and obtain compression bias. */
311 read_bytes (r, raw_bias, sizeof raw_bias);
312 if (float_identify (100.0, raw_bias, sizeof raw_bias, &r->float_format) == 0)
314 sys_warn (r, "Compression bias is not the usual value of 100, or system "
315 "file uses unrecognized floating-point format.");
316 if (r->integer_format == INTEGER_MSB_FIRST)
317 r->float_format = FLOAT_IEEE_DOUBLE_BE;
319 r->float_format = FLOAT_IEEE_DOUBLE_LE;
321 r->bias = float_get_double (r->float_format, raw_bias);
323 read_string (r, creation_date, sizeof creation_date);
324 read_string (r, creation_time, sizeof creation_time);
325 read_string (r, file_label, sizeof file_label);
326 trim_spaces (file_label);
329 printf ("File header record:\n");
330 printf ("\t%17s: %s\n", "Product name", eye_catcher);
331 printf ("\t%17s: %"PRId32"\n", "Layout code", layout_code);
332 printf ("\t%17s: %"PRId32" (%s)\n", "Compressed",
334 r->compression == COMP_NONE ? "no compression"
335 : r->compression == COMP_SIMPLE ? "simple compression"
336 : r->compression == COMP_ZLIB ? "ZLIB compression"
338 printf ("\t%17s: %"PRId32"\n", "Weight index", weight_index);
339 printf ("\t%17s: %"PRId32"\n", "Number of cases", ncases);
340 printf ("\t%17s: %.*g\n", "Compression bias", DBL_DIG + 1, r->bias);
341 printf ("\t%17s: %s\n", "Creation date", creation_date);
342 printf ("\t%17s: %s\n", "Creation time", creation_time);
343 printf ("\t%17s: \"%s\"\n", "File label", file_label);
347 format_name (int format)
349 switch ((format >> 16) & 0xff)
352 case 2: return "AHEX";
353 case 3: return "COMMA";
354 case 4: return "DOLLAR";
357 case 7: return "PIBHEX";
359 case 9: return "PIB";
360 case 10: return "PK";
361 case 11: return "RB";
362 case 12: return "RBHEX";
366 case 20: return "DATE";
367 case 21: return "TIME";
368 case 22: return "DATETIME";
369 case 23: return "ADATE";
370 case 24: return "JDATE";
371 case 25: return "DTIME";
372 case 26: return "WKDAY";
373 case 27: return "MONTH";
374 case 28: return "MOYR";
375 case 29: return "QYR";
376 case 30: return "WKYR";
377 case 31: return "PCT";
378 case 32: return "DOT";
379 case 33: return "CCA";
380 case 34: return "CCB";
381 case 35: return "CCC";
382 case 36: return "CCD";
383 case 37: return "CCE";
384 case 38: return "EDATE";
385 case 39: return "SDATE";
386 case 40: return "MTIME";
387 case 41: return "YMDHMS";
388 default: return "invalid";
392 /* Reads a variable (type 2) record from R and adds the
393 corresponding variable to DICT.
394 Also skips past additional variable records for long string
397 read_variable_record (struct sfm_reader *r)
400 int has_variable_label;
401 int missing_value_code;
406 printf ("%08llx: variable record #%d\n",
407 (long long int) ftello (r->file), ++r->n_variable_records);
409 width = read_int (r);
410 has_variable_label = read_int (r);
411 missing_value_code = read_int (r);
412 print_format = read_int (r);
413 write_format = read_int (r);
414 read_string (r, name, sizeof name);
415 name[strcspn (name, " ")] = '\0';
420 if (r->n_var_widths >= r->allocated_var_widths)
421 r->var_widths = x2nrealloc (r->var_widths, &r->allocated_var_widths,
422 sizeof *r->var_widths);
423 r->var_widths[r->n_var_widths++] = width;
425 printf ("\tWidth: %d (%s)\n",
428 : width == 0 ? "numeric"
429 : "long string continuation record");
430 printf ("\tVariable label: %d\n", has_variable_label);
431 printf ("\tMissing values code: %d (%s)\n", missing_value_code,
432 (missing_value_code == 0 ? "no missing values"
433 : missing_value_code == 1 ? "one missing value"
434 : missing_value_code == 2 ? "two missing values"
435 : missing_value_code == 3 ? "three missing values"
436 : missing_value_code == -2 ? "one missing value range"
437 : missing_value_code == -3 ? "one missing value, one range"
439 printf ("\tPrint format: %06x (%s%d.%d)\n",
440 print_format, format_name (print_format),
441 (print_format >> 8) & 0xff, print_format & 0xff);
442 printf ("\tWrite format: %06x (%s%d.%d)\n",
443 write_format, format_name (write_format),
444 (write_format >> 8) & 0xff, write_format & 0xff);
445 printf ("\tName: %s\n", name);
447 /* Get variable label, if any. */
448 if (has_variable_label != 0 && has_variable_label != 1)
449 sys_error (r, "Variable label indicator field is not 0 or 1.");
450 if (has_variable_label == 1)
452 long long int offset = ftello (r->file);
453 enum { MAX_LABEL_LEN = 65536 };
455 size_t len = read_int (r);
456 size_t read_len = MIN (MAX_LABEL_LEN, len);
457 char *label = xmalloc (read_len + 1);
458 read_string (r, label, read_len + 1);
459 printf("\t%08llx Variable label: \"%s\"\n", offset, label);
462 /* Skip label padding up to multiple of 4 bytes. */
463 skip_bytes (r, ROUND_UP (len, 4) - len);
466 /* Set missing values. */
467 if (missing_value_code != 0)
471 printf ("\t%08llx Missing values:", (long long int) ftello (r->file));
474 if (missing_value_code < -3 || missing_value_code > 3
475 || missing_value_code == -1)
476 sys_error (r, "Numeric missing value indicator field is not "
477 "-3, -2, 0, 1, 2, or 3.");
478 if (missing_value_code < 0)
480 double low = read_float (r);
481 double high = read_float (r);
482 printf (" %.*g...%.*g", DBL_DIG + 1, low, DBL_DIG + 1, high);
483 missing_value_code = -missing_value_code - 2;
485 for (i = 0; i < missing_value_code; i++)
486 printf (" %.*g", DBL_DIG + 1, read_float (r));
490 if (missing_value_code < 1 || missing_value_code > 3)
491 sys_error (r, "String missing value indicator field is not "
493 for (i = 0; i < missing_value_code; i++)
496 read_string (r, string, sizeof string);
497 printf (" \"%s\"", string);
505 print_untyped_value (struct sfm_reader *r, char raw_value[8])
510 value = float_get_double (r->float_format, raw_value);
511 for (n_printable = 0; n_printable < 8; n_printable++)
512 if (!isprint (raw_value[n_printable]))
515 printf ("%.*g/\"%.*s\"", DBL_DIG + 1, value, n_printable, raw_value);
518 /* Reads value labels from sysfile R and inserts them into the
519 associated dictionary. */
521 read_value_label_record (struct sfm_reader *r)
523 int n_labels, n_vars;
526 printf ("%08llx: value labels record\n", (long long int) ftello (r->file));
528 /* Read number of labels. */
529 n_labels = read_int (r);
530 for (i = 0; i < n_labels; i++)
533 unsigned char label_len;
537 read_bytes (r, raw_value, sizeof raw_value);
539 /* Read label length. */
540 read_bytes (r, &label_len, sizeof label_len);
541 padded_len = ROUND_UP (label_len + 1, 8);
543 /* Read label, padding. */
544 read_bytes (r, label, padded_len - 1);
545 label[label_len] = 0;
548 print_untyped_value (r, raw_value);
549 printf (": \"%s\"\n", label);
552 /* Now, read the type 4 record that has the list of variables
553 to which the value labels are to be applied. */
555 /* Read record type of type 4 record. */
556 if (read_int (r) != 4)
557 sys_error (r, "Variable index record (type 4) does not immediately "
558 "follow value label record (type 3) as it should.");
560 /* Read number of variables associated with value label from type 4
562 printf ("\t%08llx: apply to variables", (long long int) ftello (r->file));
563 n_vars = read_int (r);
564 for (i = 0; i < n_vars; i++)
565 printf (" #%d", read_int (r));
570 read_document_record (struct sfm_reader *r)
575 printf ("%08llx: document record\n", (long long int) ftello (r->file));
576 n_lines = read_int (r);
577 printf ("\t%d lines of documents\n", n_lines);
579 for (i = 0; i < n_lines; i++)
582 printf ("\t%08llx: ", (long long int) ftello (r->file));
583 read_string (r, line, sizeof line);
585 printf ("line %d: \"%s\"\n", i, line);
590 read_extension_record (struct sfm_reader *r)
592 long long int offset = ftello (r->file);
593 int subtype = read_int (r);
594 size_t size = read_int (r);
595 size_t count = read_int (r);
596 size_t bytes = size * count;
598 printf ("%08llx: Record 7, subtype %d, size=%zu, count=%zu\n",
599 offset, subtype, size, count);
604 read_machine_integer_info (r, size, count);
608 read_machine_float_info (r, size, count);
612 read_variable_sets (r, size, count);
616 /* DATE variable information. We don't use it yet, but we
622 read_mrsets (r, size, count);
626 read_extra_product_info (r, size, count);
630 read_display_parameters (r, size, count);
634 read_long_var_name_map (r, size, count);
638 read_long_string_map (r, size, count);
642 read_ncases64 (r, size, count);
646 read_datafile_attributes (r, size, count);
650 read_variable_attributes (r, size, count);
654 read_character_encoding (r, size, count);
658 read_long_string_value_labels (r, size, count);
662 read_long_string_missing_values (r, size, count);
666 sys_warn (r, "Unrecognized record type 7, subtype %d.", subtype);
667 read_unknown_extension (r, size, count);
671 skip_bytes (r, bytes);
675 read_machine_integer_info (struct sfm_reader *r, size_t size, size_t count)
677 long long int offset = ftello (r->file);
678 int version_major = read_int (r);
679 int version_minor = read_int (r);
680 int version_revision = read_int (r);
681 int machine_code = read_int (r);
682 int float_representation = read_int (r);
683 int compression_code = read_int (r);
684 int integer_representation = read_int (r);
685 int character_code = read_int (r);
687 printf ("%08llx: machine integer info\n", offset);
688 if (size != 4 || count != 8)
689 sys_error (r, "Bad size (%zu) or count (%zu) field on record type 7, "
690 "subtype 3.", size, count);
692 printf ("\tVersion: %d.%d.%d\n",
693 version_major, version_minor, version_revision);
694 printf ("\tMachine code: %d\n", machine_code);
695 printf ("\tFloating point representation: %d (%s)\n",
696 float_representation,
697 float_representation == 1 ? "IEEE 754"
698 : float_representation == 2 ? "IBM 370"
699 : float_representation == 3 ? "DEC VAX"
701 printf ("\tCompression code: %d\n", compression_code);
702 printf ("\tEndianness: %d (%s)\n", integer_representation,
703 integer_representation == 1 ? "big"
704 : integer_representation == 2 ? "little" : "unknown");
705 printf ("\tCharacter code: %d\n", character_code);
708 /* Read record type 7, subtype 4. */
710 read_machine_float_info (struct sfm_reader *r, size_t size, size_t count)
712 long long int offset = ftello (r->file);
713 double sysmis = read_float (r);
714 double highest = read_float (r);
715 double lowest = read_float (r);
717 printf ("%08llx: machine float info\n", offset);
718 if (size != 8 || count != 3)
719 sys_error (r, "Bad size (%zu) or count (%zu) on extension 4.",
722 printf ("\tsysmis: %.*g (%a)\n", DBL_DIG + 1, sysmis, sysmis);
723 if (sysmis != SYSMIS)
724 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
725 DBL_DIG + 1, sysmis, sysmis, "SYSMIS");
727 printf ("\thighest: %.*g (%a)\n", DBL_DIG + 1, highest, highest);
728 if (highest != HIGHEST)
729 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
730 DBL_DIG + 1, highest, highest, "HIGHEST");
732 printf ("\tlowest: %.*g (%a)\n", DBL_DIG + 1, lowest, lowest);
733 if (lowest != LOWEST && lowest != SYSMIS)
734 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
735 DBL_DIG + 1, lowest, lowest, "LOWEST");
738 /* Read record type 7, subtype 5. */
740 read_variable_sets (struct sfm_reader *r, size_t size, size_t count)
742 printf ("%08llx: variable sets\n", (long long int) ftello (r->file));
743 struct text_record *text = open_text_record (r, size, count);
746 while (text_match (text, '\n'))
749 const char *set = text_tokenize (text, '=');
753 /* Always present even for an empty set. */
754 text_match (text, ' ');
756 char *variables = text_tokenize (text, '\n');
758 printf ("\tset \"%s\" is empty\n", set);
761 size_t length = strlen (variables);
762 if (variables[length - 1] == '\r')
763 variables[length - 1] = '\0';
764 printf ("\tset \"%s\" contains \"%s\"\n", set, variables);
767 close_text_record (text);
771 read_extra_product_info (struct sfm_reader *r,
772 size_t size, size_t count)
774 struct text_record *text;
777 printf ("%08llx: extra product info\n", (long long int) ftello (r->file));
778 text = open_text_record (r, size, count);
779 s = text_get_all (text);
780 print_string (s, strlen (s));
781 close_text_record (text);
784 /* Read record type 7, subtype 7. */
786 read_mrsets (struct sfm_reader *r, size_t size, size_t count)
788 struct text_record *text;
790 printf ("%08llx: multiple response sets\n",
791 (long long int) ftello (r->file));
792 text = open_text_record (r, size, count);
796 enum { MRSET_MC, MRSET_MD } type;
797 bool cat_label_from_counted_values = false;
798 bool label_from_var_label = false;
801 const char *variables;
803 while (text_match (text, '\n'))
806 name = text_tokenize (text, '=');
810 if (text_match (text, 'C'))
814 if (!text_match (text, ' '))
816 sys_warn (r, "missing space following 'C' at offset %zu "
817 "in mrsets record", text_pos (text));
821 else if (text_match (text, 'D'))
825 else if (text_match (text, 'E'))
830 cat_label_from_counted_values = true;
832 if (!text_match (text, ' '))
834 sys_warn (r, "Missing space following `%c' at offset %zu "
835 "in MRSETS record", 'E', text_pos (text));
839 number = text_tokenize (text, ' ');
841 sys_warn (r, "Missing label source value "
842 "following `E' at offset %zu in MRSETS record",
844 else if (!strcmp (number, "11"))
845 label_from_var_label = true;
846 else if (strcmp (number, "1"))
847 sys_warn (r, "Unexpected label source value `%s' "
848 "following `E' at offset %zu in MRSETS record",
849 number, text_pos (text));
854 sys_warn (r, "missing `C', `D', or `E' at offset %zu "
855 "in mrsets record", text_pos (text));
859 if (type == MRSET_MD)
861 counted = text_parse_counted_string (text);
866 label = text_parse_counted_string (text);
870 variables = text_tokenize (text, '\n');
872 printf ("\t\"%s\": multiple %s set",
873 name, type == MRSET_MC ? "category" : "dichotomy");
875 printf (", counted value \"%s\"", counted);
876 if (cat_label_from_counted_values)
877 printf (", category labels from counted values");
878 if (label[0] != '\0')
879 printf (", label \"%s\"", label);
880 if (label_from_var_label)
881 printf (", label from variable label");
882 if (variables != NULL)
883 printf(", variables \"%s\"\n", variables);
885 printf(", no variables\n");
887 close_text_record (text);
890 /* Read record type 7, subtype 11. */
892 read_display_parameters (struct sfm_reader *r, size_t size, size_t count)
898 printf ("%08llx: variable display parameters\n",
899 (long long int) ftello (r->file));
902 sys_warn (r, "Bad size %zu on extension 11.", size);
903 skip_bytes (r, size * count);
907 n_vars = r->n_variables;
908 if (count == 3 * n_vars)
909 includes_width = true;
910 else if (count == 2 * n_vars)
911 includes_width = false;
914 sys_warn (r, "Extension 11 has bad count %zu (for %zu variables.",
916 skip_bytes (r, size * count);
920 for (i = 0; i < n_vars; ++i)
922 int measure = read_int (r);
923 int width = includes_width ? read_int (r) : 0;
924 int align = read_int (r);
926 printf ("\tVar #%zu: measure=%d (%s)", i, measure,
927 (measure == 1 ? "nominal"
928 : measure == 2 ? "ordinal"
929 : measure == 3 ? "scale"
932 printf (", width=%d", width);
933 printf (", align=%d (%s)\n", align,
935 : align == 1 ? "right"
936 : align == 2 ? "centre"
941 /* Reads record type 7, subtype 13, which gives the long name
942 that corresponds to each short name. */
944 read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count)
946 struct text_record *text;
950 printf ("%08llx: long variable names (short => long)\n",
951 (long long int) ftello (r->file));
952 text = open_text_record (r, size, count);
953 while (read_variable_to_value_pair (text, &var, &long_name))
954 printf ("\t%s => %s\n", var, long_name);
955 close_text_record (text);
958 /* Reads record type 7, subtype 14, which gives the real length
959 of each very long string. Rearranges DICT accordingly. */
961 read_long_string_map (struct sfm_reader *r, size_t size, size_t count)
963 struct text_record *text;
967 printf ("%08llx: very long strings (variable => length)\n",
968 (long long int) ftello (r->file));
969 text = open_text_record (r, size, count);
970 while (read_variable_to_value_pair (text, &var, &length_s))
971 printf ("\t%s => %d\n", var, atoi (length_s));
972 close_text_record (text);
976 read_attributes (struct sfm_reader *r, struct text_record *text,
977 const char *variable)
984 key = text_tokenize (text, '(');
988 for (index = 1; ; index++)
990 /* Parse the value. */
991 const char *value = text_tokenize (text, '\n');
994 sys_warn (r, "%s: Error parsing attribute value %s[%d]",
995 variable, key, index);
998 if (strlen (value) < 2
999 || value[0] != '\'' || value[strlen (value) - 1] != '\'')
1000 sys_warn (r, "%s: Attribute value %s[%d] is not quoted: %s",
1001 variable, key, index, value);
1003 printf ("\t%s: %s[%d] = \"%.*s\"\n",
1004 variable, key, index, (int) strlen (value) - 2, value + 1);
1006 /* Was this the last value for this attribute? */
1007 if (text_match (text, ')'))
1011 if (text_match (text, '/'))
1016 /* Read extended number of cases record. */
1018 read_ncases64 (struct sfm_reader *r, size_t size, size_t count)
1020 int64_t unknown, ncases64;
1024 sys_warn (r, "Bad size %zu for extended number of cases.", size);
1025 skip_bytes (r, size * count);
1030 sys_warn (r, "Bad count %zu for extended number of cases.", size);
1031 skip_bytes (r, size * count);
1034 unknown = read_int64 (r);
1035 ncases64 = read_int64 (r);
1036 printf ("%08llx: extended number of cases: "
1037 "unknown=%"PRId64", ncases64=%"PRId64"\n",
1038 (long long int) ftello (r->file), unknown, ncases64);
1042 read_datafile_attributes (struct sfm_reader *r, size_t size, size_t count)
1044 struct text_record *text;
1046 printf ("%08llx: datafile attributes\n", (long long int) ftello (r->file));
1047 text = open_text_record (r, size, count);
1048 read_attributes (r, text, "datafile");
1049 close_text_record (text);
1053 read_character_encoding (struct sfm_reader *r, size_t size, size_t count)
1055 long long int posn = ftello (r->file);
1056 char *encoding = xcalloc (size, count + 1);
1057 read_string (r, encoding, count + 1);
1059 printf ("%08llx: Character Encoding: %s\n", posn, encoding);
1065 read_long_string_value_labels (struct sfm_reader *r, size_t size, size_t count)
1067 long long int start = ftello (r->file);
1069 printf ("%08llx: long string value labels\n", start);
1070 while (ftello (r->file) - start < size * count)
1072 long long posn = ftello (r->file);
1073 char var_name[ID_MAX_LEN + 1];
1079 /* Read variable name. */
1080 var_name_len = read_int (r);
1081 if (var_name_len > ID_MAX_LEN)
1082 sys_error (r, "Variable name length in long string value label "
1083 "record (%d) exceeds %d-byte limit.",
1084 var_name_len, ID_MAX_LEN);
1085 read_string (r, var_name, var_name_len + 1);
1087 /* Read width, number of values. */
1088 width = read_int (r);
1089 n_values = read_int (r);
1091 printf ("\t%08llx: %s, width %d, %d values\n",
1092 posn, var_name, width, n_values);
1095 for (i = 0; i < n_values; i++)
1103 posn = ftello (r->file);
1106 value_length = read_int (r);
1107 value = xmalloc (value_length + 1);
1108 read_string (r, value, value_length + 1);
1111 label_length = read_int (r);
1112 label = xmalloc (label_length + 1);
1113 read_string (r, label, label_length + 1);
1115 printf ("\t\t%08llx: \"%s\" (%d bytes) => \"%s\" (%d bytes)\n",
1116 posn, value, value_length, label, label_length);
1125 read_long_string_missing_values (struct sfm_reader *r,
1126 size_t size, size_t count)
1128 long long int start = ftello (r->file);
1130 printf ("%08llx: long string missing values\n", start);
1131 while (ftello (r->file) - start < size * count)
1133 long long posn = ftello (r->file);
1135 /* Read variable name. */
1136 int var_name_len = read_int (r);
1137 if (var_name_len > ID_MAX_LEN)
1138 sys_error (r, "Variable name length in long string value label "
1139 "record (%d) exceeds %d-byte limit.",
1140 var_name_len, ID_MAX_LEN);
1141 char var_name[ID_MAX_LEN + 1];
1142 read_string (r, var_name, var_name_len + 1);
1144 /* Read number of values. */
1145 uint8_t n_missing_values;
1146 read_bytes (r, &n_missing_values, 1);
1148 /* Read value length. */
1149 int value_length = read_int (r);
1151 printf ("\t%08llx: %s, %d missing values, each %d bytes:",
1152 posn, var_name, n_missing_values, value_length);
1155 for (int i = 0; i < n_missing_values; i++)
1157 posn = ftello (r->file);
1160 char *value = xmalloc (value_length + 1);
1161 read_string (r, value, value_length + 1);
1163 printf (" \"%s\"", value);
1172 hex_dump (size_t offset, const void *buffer_, size_t buffer_size)
1174 const uint8_t *buffer = buffer_;
1176 while (buffer_size > 0)
1178 size_t n = MIN (buffer_size, 16);
1181 printf ("%04zx", offset);
1182 for (i = 0; i < 16; i++)
1185 printf ("%c%02x", i == 8 ? '-' : ' ', buffer[i]);
1191 for (i = 0; i < 16; i++)
1193 unsigned char c = i < n ? buffer[i] : ' ';
1194 putchar (isprint (c) ? c : '.');
1204 /* Reads and prints any type 7 record that we don't understand. */
1206 read_unknown_extension (struct sfm_reader *r, size_t size, size_t count)
1208 unsigned char *buffer;
1211 if (size == 0 || count > 65536 / size)
1212 skip_bytes (r, size * count);
1215 buffer = xmalloc (size);
1216 for (i = 0; i < count; i++)
1218 read_bytes (r, buffer, size);
1219 hex_dump (i * size, buffer, size);
1225 buffer = xmalloc (count);
1226 read_bytes (r, buffer, count);
1227 print_string (CHAR_CAST (char *, buffer), count);
1233 read_variable_attributes (struct sfm_reader *r, size_t size, size_t count)
1235 struct text_record *text;
1237 printf ("%08llx: variable attributes\n", (long long int) ftello (r->file));
1238 text = open_text_record (r, size, count);
1241 const char *variable = text_tokenize (text, ':');
1242 if (variable == NULL || !read_attributes (r, text, variable))
1245 close_text_record (text);
1249 read_simple_compressed_data (struct sfm_reader *r, int max_cases)
1251 enum { N_OPCODES = 8 };
1252 uint8_t opcodes[N_OPCODES];
1253 long long int opcode_ofs;
1259 printf ("\n%08llx: compressed data:\n", (long long int) ftello (r->file));
1261 opcode_idx = N_OPCODES;
1264 for (case_num = 0; case_num < max_cases; case_num++)
1266 printf ("%08llx: case %d's uncompressible data begins\n",
1267 (long long int) ftello (r->file), case_num);
1268 for (i = 0; i < r->n_var_widths;)
1270 int width = r->var_widths[i];
1274 if (opcode_idx >= N_OPCODES)
1276 opcode_ofs = ftello (r->file);
1279 if (!try_read_bytes (r, opcodes, 8))
1283 read_bytes (r, opcodes, 8);
1286 opcode = opcodes[opcode_idx];
1287 printf ("%08llx: variable %d: opcode %d: ",
1288 opcode_ofs + opcode_idx, i, opcode);
1293 printf ("%.*g", DBL_DIG + 1, opcode - r->bias);
1295 printf (", but this is a string variable (width=%d)", width);
1301 printf ("ignored padding\n");
1305 printf ("end of data\n");
1309 read_bytes (r, raw_value, 8);
1310 printf ("uncompressible data: ");
1311 print_untyped_value (r, raw_value);
1319 printf (", but this is a numeric variable");
1327 printf (", but this is a string variable (width=%d)", width);
1339 read_zlib_compressed_data (struct sfm_reader *r)
1342 long long int this_ofs, next_ofs, next_len;
1343 long long int bias, zero;
1344 long long int expected_uncmp_ofs, expected_cmp_ofs;
1345 unsigned int block_size, n_blocks;
1349 ofs = ftello (r->file);
1350 printf ("\n%08llx: ZLIB compressed data header:\n", ofs);
1352 this_ofs = read_int64 (r);
1353 next_ofs = read_int64 (r);
1354 next_len = read_int64 (r);
1356 printf ("\tzheader_ofs: 0x%llx\n", this_ofs);
1357 if (this_ofs != ofs)
1358 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1359 printf ("\tztrailer_ofs: 0x%llx\n", next_ofs);
1360 printf ("\tztrailer_len: %lld\n", next_len);
1361 if (next_len < 24 || next_len % 24)
1362 printf ("\t\t(Trailer length is not a positive multiple of 24.)\n");
1364 printf ("\n%08llx: 0x%llx bytes of ZLIB compressed data\n",
1365 ofs + 8 * 3, next_ofs - (ofs + 8 * 3));
1367 skip_bytes (r, next_ofs - (ofs + 8 * 3));
1369 printf ("\n%08llx: ZLIB trailer fixed header:\n", next_ofs);
1370 bias = read_int64 (r);
1371 zero = read_int64 (r);
1372 block_size = read_int (r);
1373 n_blocks = read_int (r);
1374 printf ("\tbias: %lld\n", bias);
1375 printf ("\tzero: 0x%llx\n", zero);
1377 printf ("\t\t(Expected 0.)\n");
1378 printf ("\tblock_size: 0x%x\n", block_size);
1379 if (block_size != 0x3ff000)
1380 printf ("\t\t(Expected 0x3ff000.)\n");
1381 printf ("\tn_blocks: %u\n", n_blocks);
1382 if (n_blocks != next_len / 24 - 1)
1383 printf ("\t\t(Expected %llu.)\n", next_len / 24 - 1);
1385 expected_uncmp_ofs = ofs;
1386 expected_cmp_ofs = ofs + 24;
1387 for (i = 0; i < n_blocks; i++)
1389 long long int blockinfo_ofs = ftello (r->file);
1390 unsigned long long int uncompressed_ofs = read_int64 (r);
1391 unsigned long long int compressed_ofs = read_int64 (r);
1392 unsigned int uncompressed_size = read_int (r);
1393 unsigned int compressed_size = read_int (r);
1395 printf ("\n%08llx: ZLIB block descriptor %d\n", blockinfo_ofs, i + 1);
1397 printf ("\tuncompressed_ofs: 0x%llx\n", uncompressed_ofs);
1398 if (uncompressed_ofs != expected_uncmp_ofs)
1399 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1401 printf ("\tcompressed_ofs: 0x%llx\n", compressed_ofs);
1402 if (compressed_ofs != expected_cmp_ofs)
1403 printf ("\t\t(Expected 0x%llx.)\n", ofs + 24);
1405 printf ("\tuncompressed_size: 0x%x\n", uncompressed_size);
1406 if (i < n_blocks - 1 && uncompressed_size != block_size)
1407 printf ("\t\t(Expected 0x%x.)\n", block_size);
1409 printf ("\tcompressed_size: 0x%x\n", compressed_size);
1410 if (i == n_blocks - 1 && compressed_ofs + compressed_size != next_ofs)
1411 printf ("\t\t(This was expected to be 0x%llx.)\n",
1412 next_ofs - compressed_size);
1414 expected_uncmp_ofs += uncompressed_size;
1415 expected_cmp_ofs += compressed_size;
1419 /* Helpers for reading records that consist of structured text
1425 struct sfm_reader *reader; /* Reader. */
1426 char *buffer; /* Record contents. */
1427 size_t size; /* Size of buffer. */
1428 size_t pos; /* Current position in buffer. */
1431 /* Reads SIZE * COUNT bytes into a text record for R,
1432 and returns the new text record. */
1433 static struct text_record *
1434 open_text_record (struct sfm_reader *r, size_t size, size_t count)
1436 struct text_record *text = xmalloc (sizeof *text);
1438 if (size_overflow_p (xsum (1, xtimes (size, count))))
1439 sys_error (r, "Extension record too large.");
1441 size_t n_bytes = size * count;
1442 char *buffer = xmalloc (n_bytes + 1);
1443 read_bytes (r, buffer, n_bytes);
1444 buffer[n_bytes] = '\0';
1446 text->buffer = buffer;
1447 text->size = n_bytes;
1452 /* Closes TEXT and frees its storage.
1453 Not really needed, because the pool will free the text record anyway,
1454 but can be used to free it earlier. */
1456 close_text_record (struct text_record *text)
1458 free (text->buffer);
1463 text_tokenize (struct text_record *text, int delimiter)
1465 size_t start = text->pos;
1466 while (text->pos < text->size
1467 && text->buffer[text->pos] != delimiter
1468 && text->buffer[text->pos] != '\0')
1470 if (start == text->pos)
1472 text->buffer[text->pos++] = '\0';
1473 return &text->buffer[start];
1477 text_match (struct text_record *text, int c)
1479 if (text->pos < text->size && text->buffer[text->pos] == c)
1488 /* Reads a integer value expressed in decimal, then a space, then a string that
1489 consists of exactly as many bytes as specified by the integer, then a space,
1490 from TEXT. Returns the string, null-terminated, as a subset of TEXT's
1491 buffer (so the caller should not free the string). */
1493 text_parse_counted_string (struct text_record *text)
1501 while (isdigit ((unsigned char) text->buffer[text->pos]))
1502 n = (n * 10) + (text->buffer[text->pos++] - '0');
1503 if (start == text->pos)
1505 sys_error (text->reader, "expecting digit at offset %zu in record",
1510 if (!text_match (text, ' '))
1512 sys_error (text->reader, "expecting space at offset %zu in record",
1517 if (text->pos + n > text->size)
1519 sys_error (text->reader, "%zu-byte string starting at offset %zu "
1520 "exceeds record length %zu", n, text->pos, text->size);
1524 s = &text->buffer[text->pos];
1527 sys_error (text->reader, "expecting space at offset %zu following "
1528 "%zu-byte string", text->pos + n, n);
1536 /* Reads a variable=value pair from TEXT.
1537 Looks up the variable in DICT and stores it into *VAR.
1538 Stores a null-terminated value into *VALUE. */
1540 read_variable_to_value_pair (struct text_record *text,
1541 char **key, char **value)
1543 *key = text_tokenize (text, '=');
1544 *value = text_tokenize (text, '\t');
1545 if (!*key || !*value)
1548 while (text->pos < text->size
1549 && (text->buffer[text->pos] == '\t'
1550 || text->buffer[text->pos] == '\0'))
1555 /* Returns the current byte offset inside the TEXT's string. */
1557 text_pos (const struct text_record *text)
1563 text_get_all (const struct text_record *text)
1565 return text->buffer;
1572 %s, a utility for dissecting system files.\n\
1573 Usage: %s [OPTION]... SYSFILE...\n\
1574 where each SYSFILE is the name of a system file.\n\
1577 --data[=MAXCASES] print (up to MAXCASES cases of) compressed data\n\
1578 --help display this help and exit\n\
1579 --version output version information and exit\n",
1580 program_name, program_name);
1583 /* Displays a corruption message. */
1585 sys_msg (struct sfm_reader *r, const char *format, va_list args)
1587 printf ("\"%s\" near offset 0x%llx: ",
1588 r->file_name, (long long int) ftello (r->file));
1589 vprintf (format, args);
1593 /* Displays a warning for the current file position. */
1595 sys_warn (struct sfm_reader *r, const char *format, ...)
1599 va_start (args, format);
1600 sys_msg (r, format, args);
1604 /* Displays an error for the current file position,
1605 marks it as in an error state,
1606 and aborts reading it using longjmp. */
1608 sys_error (struct sfm_reader *r, const char *format, ...)
1612 va_start (args, format);
1613 sys_msg (r, format, args);
1616 exit (EXIT_FAILURE);
1619 /* Reads BYTE_CNT bytes into BUF.
1620 Returns true if exactly BYTE_CNT bytes are successfully read.
1621 Aborts if an I/O error or a partial read occurs.
1622 If EOF_IS_OK, then an immediate end-of-file causes false to be
1623 returned; otherwise, immediate end-of-file causes an abort
1626 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
1627 void *buf, size_t n_bytes)
1629 size_t bytes_read = fread (buf, 1, n_bytes, r->file);
1630 if (bytes_read == n_bytes)
1632 else if (ferror (r->file))
1633 sys_error (r, "System error: %s.", strerror (errno));
1634 else if (!eof_is_ok || bytes_read != 0)
1635 sys_error (r, "Unexpected end of file.");
1640 /* Reads BYTE_CNT into BUF.
1641 Aborts upon I/O error or if end-of-file is encountered. */
1643 read_bytes (struct sfm_reader *r, void *buf, size_t n_bytes)
1645 read_bytes_internal (r, false, buf, n_bytes);
1648 /* Reads BYTE_CNT bytes into BUF.
1649 Returns true if exactly BYTE_CNT bytes are successfully read.
1650 Returns false if an immediate end-of-file is encountered.
1651 Aborts if an I/O error or a partial read occurs. */
1653 try_read_bytes (struct sfm_reader *r, void *buf, size_t n_bytes)
1655 return read_bytes_internal (r, true, buf, n_bytes);
1658 /* Reads a 32-bit signed integer from R and returns its value in
1661 read_int (struct sfm_reader *r)
1664 read_bytes (r, integer, sizeof integer);
1665 return integer_get (r->integer_format, integer, sizeof integer);
1668 /* Reads a 64-bit signed integer from R and returns its value in
1671 read_int64 (struct sfm_reader *r)
1674 read_bytes (r, integer, sizeof integer);
1675 return integer_get (r->integer_format, integer, sizeof integer);
1678 /* Reads a 64-bit floating-point number from R and returns its
1679 value in host format. */
1681 read_float (struct sfm_reader *r)
1684 read_bytes (r, number, sizeof number);
1685 return float_get_double (r->float_format, number);
1688 /* Reads exactly SIZE - 1 bytes into BUFFER
1689 and stores a null byte into BUFFER[SIZE - 1]. */
1691 read_string (struct sfm_reader *r, char *buffer, size_t size)
1694 read_bytes (r, buffer, size - 1);
1695 buffer[size - 1] = '\0';
1698 /* Skips BYTES bytes forward in R. */
1700 skip_bytes (struct sfm_reader *r, size_t bytes)
1705 size_t chunk = MIN (sizeof buffer, bytes);
1706 read_bytes (r, buffer, chunk);
1712 trim_spaces (char *s)
1714 char *end = strchr (s, '\0');
1715 while (end > s && end[-1] == ' ')
1721 print_string (const char *s, size_t len)
1723 if (memchr (s, 0, len) == 0)
1727 for (i = 0; i < len; i++)
1729 unsigned char c = s[i];
1733 else if (c == '\n' || isprint (c))
1736 printf ("\\%02x", c);
1741 hex_dump (0, s, len);