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"
52 const char *file_name;
55 int n_variable_records, n_variables;
58 size_t n_var_widths, allocated_var_widths;
60 enum integer_format integer_format;
61 enum float_format float_format;
63 enum compression compression;
67 static void read_header (struct sfm_reader *);
68 static void read_variable_record (struct sfm_reader *);
69 static void read_value_label_record (struct sfm_reader *);
70 static void read_document_record (struct sfm_reader *);
71 static void read_extension_record (struct sfm_reader *);
72 static void read_machine_integer_info (struct sfm_reader *,
73 size_t size, size_t count);
74 static void read_machine_float_info (struct sfm_reader *,
75 size_t size, size_t count);
76 static void read_extra_product_info (struct sfm_reader *,
77 size_t size, size_t count);
78 static void read_mrsets (struct sfm_reader *, size_t size, size_t count);
79 static void read_display_parameters (struct sfm_reader *,
80 size_t size, size_t count);
81 static void read_long_var_name_map (struct sfm_reader *r,
82 size_t size, size_t count);
83 static void read_long_string_map (struct sfm_reader *r,
84 size_t size, size_t count);
85 static void read_datafile_attributes (struct sfm_reader *r,
86 size_t size, size_t count);
87 static void read_variable_attributes (struct sfm_reader *r,
88 size_t size, size_t count);
89 static void read_ncases64 (struct sfm_reader *, size_t size, size_t count);
90 static void read_character_encoding (struct sfm_reader *r,
91 size_t size, size_t count);
92 static void read_long_string_value_labels (struct sfm_reader *r,
93 size_t size, size_t count);
94 static void read_long_string_missing_values (struct sfm_reader *r,
95 size_t size, size_t count);
96 static void read_unknown_extension (struct sfm_reader *,
97 size_t size, size_t count);
98 static void read_simple_compressed_data (struct sfm_reader *, int max_cases);
99 static void read_zlib_compressed_data (struct sfm_reader *);
101 static struct text_record *open_text_record (
102 struct sfm_reader *, size_t size);
103 static void close_text_record (struct text_record *);
104 static bool read_variable_to_value_pair (struct text_record *,
105 char **key, char **value);
106 static char *text_tokenize (struct text_record *, int delimiter);
107 static bool text_match (struct text_record *text, int c);
108 static const char *text_parse_counted_string (struct text_record *);
109 static size_t text_pos (const struct text_record *);
110 static const char *text_get_all (const struct text_record *);
112 static void usage (void);
113 static void sys_warn (struct sfm_reader *, const char *, ...)
114 PRINTF_FORMAT (2, 3);
115 static void sys_error (struct sfm_reader *, const char *, ...)
119 static void read_bytes (struct sfm_reader *, void *, size_t);
120 static bool try_read_bytes (struct sfm_reader *, void *, size_t);
121 static int read_int (struct sfm_reader *);
122 static int64_t read_int64 (struct sfm_reader *);
123 static double read_float (struct sfm_reader *);
124 static void read_string (struct sfm_reader *, char *, size_t);
125 static void skip_bytes (struct sfm_reader *, size_t);
126 static void trim_spaces (char *);
128 static void print_string (const char *s, size_t len);
131 main (int argc, char *argv[])
137 set_program_name (argv[0]);
141 static const struct option long_options[] =
143 { "data", optional_argument, NULL, 'd' },
144 { "help", no_argument, NULL, 'h' },
145 { "version", no_argument, NULL, 'v' },
146 { NULL, 0, NULL, 0 },
151 c = getopt_long (argc, argv, "d::hv", long_options, NULL);
158 max_cases = optarg ? atoi (optarg) : INT_MAX;
162 version_etc (stdout, "pspp-dump-sav", PACKAGE_NAME, PACKAGE_VERSION,
163 "Ben Pfaff", "John Darrington", NULL_SENTINEL);
176 error (1, 0, "at least one non-option argument is required; "
177 "use --help for help");
179 for (i = optind; i < argc; i++)
183 r.file_name = argv[i];
184 r.file = fopen (r.file_name, "rb");
186 error (EXIT_FAILURE, errno, "error opening `%s'", r.file_name);
187 r.n_variable_records = 0;
190 r.allocated_var_widths = 0;
192 r.compression = COMP_NONE;
194 if (argc - optind > 1)
195 printf ("Reading \"%s\":\n", r.file_name);
198 while ((rec_type = read_int (&r)) != 999)
203 read_variable_record (&r);
207 read_value_label_record (&r);
211 sys_error (&r, "Misplaced type 4 record.");
214 read_document_record (&r);
218 read_extension_record (&r);
222 sys_error (&r, "Unrecognized record type %d.", rec_type);
225 printf ("%08llx: end-of-dictionary record "
226 "(first byte of data at %08llx)\n",
227 (long long int) ftello (r.file),
228 (long long int) ftello (r.file) + 4);
230 if (r.compression == COMP_SIMPLE)
233 read_simple_compressed_data (&r, max_cases);
235 else if (r.compression == COMP_ZLIB)
236 read_zlib_compressed_data (&r);
245 read_header (struct sfm_reader *r)
248 char eye_catcher[61];
249 uint8_t raw_layout_code[4];
252 int32_t weight_index;
255 char creation_date[10];
256 char creation_time[9];
260 read_string (r, rec_type, sizeof rec_type);
261 read_string (r, eye_catcher, sizeof eye_catcher);
263 if (!strcmp ("$FL2", rec_type))
265 else if (!strcmp ("$FL3", rec_type))
268 sys_error (r, "This is not an SPSS system file.");
270 /* Identify integer format. */
271 read_bytes (r, raw_layout_code, sizeof raw_layout_code);
272 if ((!integer_identify (2, raw_layout_code, sizeof raw_layout_code,
274 && !integer_identify (3, raw_layout_code, sizeof raw_layout_code,
276 || (r->integer_format != INTEGER_MSB_FIRST
277 && r->integer_format != INTEGER_LSB_FIRST))
278 sys_error (r, "This is not an SPSS system file.");
279 layout_code = integer_get (r->integer_format,
280 raw_layout_code, sizeof raw_layout_code);
282 read_int (r); /* Nominal case size (not actually useful). */
283 compressed = read_int (r);
284 weight_index = read_int (r);
285 ncases = read_int (r);
290 r->compression = COMP_NONE;
291 else if (compressed == 1)
292 r->compression = COMP_SIMPLE;
293 else if (compressed != 0)
294 sys_error (r, "SAV file header has invalid compression value "
295 "%"PRId32".", compressed);
300 r->compression = COMP_ZLIB;
302 sys_error (r, "ZSAV file header has invalid compression value "
303 "%"PRId32".", compressed);
306 /* Identify floating-point format and obtain compression bias. */
307 read_bytes (r, raw_bias, sizeof raw_bias);
308 if (float_identify (100.0, raw_bias, sizeof raw_bias, &r->float_format) == 0)
310 sys_warn (r, "Compression bias is not the usual value of 100, or system "
311 "file uses unrecognized floating-point format.");
312 if (r->integer_format == INTEGER_MSB_FIRST)
313 r->float_format = FLOAT_IEEE_DOUBLE_BE;
315 r->float_format = FLOAT_IEEE_DOUBLE_LE;
317 r->bias = float_get_double (r->float_format, raw_bias);
319 read_string (r, creation_date, sizeof creation_date);
320 read_string (r, creation_time, sizeof creation_time);
321 read_string (r, file_label, sizeof file_label);
322 trim_spaces (file_label);
325 printf ("File header record:\n");
326 printf ("\t%17s: %s\n", "Product name", eye_catcher);
327 printf ("\t%17s: %"PRId32"\n", "Layout code", layout_code);
328 printf ("\t%17s: %"PRId32" (%s)\n", "Compressed",
330 r->compression == COMP_NONE ? "no compression"
331 : r->compression == COMP_SIMPLE ? "simple compression"
332 : r->compression == COMP_ZLIB ? "ZLIB compression"
334 printf ("\t%17s: %"PRId32"\n", "Weight index", weight_index);
335 printf ("\t%17s: %"PRId32"\n", "Number of cases", ncases);
336 printf ("\t%17s: %.*g\n", "Compression bias", DBL_DIG + 1, r->bias);
337 printf ("\t%17s: %s\n", "Creation date", creation_date);
338 printf ("\t%17s: %s\n", "Creation time", creation_time);
339 printf ("\t%17s: \"%s\"\n", "File label", file_label);
343 format_name (int format)
345 switch ((format >> 16) & 0xff)
348 case 2: return "AHEX";
349 case 3: return "COMMA";
350 case 4: return "DOLLAR";
353 case 7: return "PIBHEX";
355 case 9: return "PIB";
356 case 10: return "PK";
357 case 11: return "RB";
358 case 12: return "RBHEX";
362 case 20: return "DATE";
363 case 21: return "TIME";
364 case 22: return "DATETIME";
365 case 23: return "ADATE";
366 case 24: return "JDATE";
367 case 25: return "DTIME";
368 case 26: return "WKDAY";
369 case 27: return "MONTH";
370 case 28: return "MOYR";
371 case 29: return "QYR";
372 case 30: return "WKYR";
373 case 31: return "PCT";
374 case 32: return "DOT";
375 case 33: return "CCA";
376 case 34: return "CCB";
377 case 35: return "CCC";
378 case 36: return "CCD";
379 case 37: return "CCE";
380 case 38: return "EDATE";
381 case 39: return "SDATE";
382 default: return "invalid";
386 /* Reads a variable (type 2) record from R and adds the
387 corresponding variable to DICT.
388 Also skips past additional variable records for long string
391 read_variable_record (struct sfm_reader *r)
394 int has_variable_label;
395 int missing_value_code;
400 printf ("%08llx: variable record #%d\n",
401 (long long int) ftello (r->file), r->n_variable_records++);
403 width = read_int (r);
404 has_variable_label = read_int (r);
405 missing_value_code = read_int (r);
406 print_format = read_int (r);
407 write_format = read_int (r);
408 read_string (r, name, sizeof name);
409 name[strcspn (name, " ")] = '\0';
414 if (r->n_var_widths >= r->allocated_var_widths)
415 r->var_widths = x2nrealloc (r->var_widths, &r->allocated_var_widths,
416 sizeof *r->var_widths);
417 r->var_widths[r->n_var_widths++] = width;
419 printf ("\tWidth: %d (%s)\n",
422 : width == 0 ? "numeric"
423 : "long string continuation record");
424 printf ("\tVariable label: %d\n", has_variable_label);
425 printf ("\tMissing values code: %d (%s)\n", missing_value_code,
426 (missing_value_code == 0 ? "no missing values"
427 : missing_value_code == 1 ? "one missing value"
428 : missing_value_code == 2 ? "two missing values"
429 : missing_value_code == 3 ? "three missing values"
430 : missing_value_code == -2 ? "one missing value range"
431 : missing_value_code == -3 ? "one missing value, one range"
433 printf ("\tPrint format: %06x (%s%d.%d)\n",
434 print_format, format_name (print_format),
435 (print_format >> 8) & 0xff, print_format & 0xff);
436 printf ("\tWrite format: %06x (%s%d.%d)\n",
437 write_format, format_name (write_format),
438 (write_format >> 8) & 0xff, write_format & 0xff);
439 printf ("\tName: %s\n", name);
441 /* Get variable label, if any. */
442 if (has_variable_label != 0 && has_variable_label != 1)
443 sys_error (r, "Variable label indicator field is not 0 or 1.");
444 if (has_variable_label == 1)
446 long long int offset = ftello (r->file);
447 size_t len, read_len;
452 /* Read up to 255 bytes of label. */
453 read_len = MIN (sizeof label - 1, len);
454 read_string (r, label, read_len + 1);
455 printf("\t%08llx Variable label: \"%s\"\n", offset, label);
457 /* Skip unread label bytes. */
458 skip_bytes (r, len - read_len);
460 /* Skip label padding up to multiple of 4 bytes. */
461 skip_bytes (r, ROUND_UP (len, 4) - len);
464 /* Set missing values. */
465 if (missing_value_code != 0)
469 printf ("\t%08llx Missing values:", (long long int) ftello (r->file));
472 if (missing_value_code < -3 || missing_value_code > 3
473 || missing_value_code == -1)
474 sys_error (r, "Numeric missing value indicator field is not "
475 "-3, -2, 0, 1, 2, or 3.");
476 if (missing_value_code < 0)
478 double low = read_float (r);
479 double high = read_float (r);
480 printf (" %.*g...%.*g", DBL_DIG + 1, low, DBL_DIG + 1, high);
481 missing_value_code = -missing_value_code - 2;
483 for (i = 0; i < missing_value_code; i++)
484 printf (" %.*g", DBL_DIG + 1, read_float (r));
488 if (missing_value_code < 1 || missing_value_code > 3)
489 sys_error (r, "String missing value indicator field is not "
491 for (i = 0; i < missing_value_code; i++)
494 read_string (r, string, sizeof string);
495 printf (" \"%s\"", string);
503 print_untyped_value (struct sfm_reader *r, char raw_value[8])
508 value = float_get_double (r->float_format, raw_value);
509 for (n_printable = 0; n_printable < 8; n_printable++)
510 if (!isprint (raw_value[n_printable]))
513 printf ("%.*g/\"%.*s\"", DBL_DIG + 1, value, n_printable, raw_value);
516 /* Reads value labels from sysfile R and inserts them into the
517 associated dictionary. */
519 read_value_label_record (struct sfm_reader *r)
521 int label_cnt, var_cnt;
524 printf ("%08llx: value labels record\n", (long long int) ftello (r->file));
526 /* Read number of labels. */
527 label_cnt = read_int (r);
528 for (i = 0; i < label_cnt; i++)
531 unsigned char label_len;
535 read_bytes (r, raw_value, sizeof raw_value);
537 /* Read label length. */
538 read_bytes (r, &label_len, sizeof label_len);
539 padded_len = ROUND_UP (label_len + 1, 8);
541 /* Read label, padding. */
542 read_bytes (r, label, padded_len - 1);
543 label[label_len] = 0;
546 print_untyped_value (r, raw_value);
547 printf (": \"%s\"\n", label);
550 /* Now, read the type 4 record that has the list of variables
551 to which the value labels are to be applied. */
553 /* Read record type of type 4 record. */
554 if (read_int (r) != 4)
555 sys_error (r, "Variable index record (type 4) does not immediately "
556 "follow value label record (type 3) as it should.");
558 /* Read number of variables associated with value label from type 4
560 printf ("\t%08llx: apply to variables", (long long int) ftello (r->file));
561 var_cnt = read_int (r);
562 for (i = 0; i < var_cnt; i++)
563 printf (" #%d", read_int (r));
568 read_document_record (struct sfm_reader *r)
573 printf ("%08llx: document record\n", (long long int) ftello (r->file));
574 n_lines = read_int (r);
575 printf ("\t%d lines of documents\n", n_lines);
577 for (i = 0; i < n_lines; i++)
580 printf ("\t%08llx: ", (long long int) ftello (r->file));
581 read_string (r, line, sizeof line);
583 printf ("line %d: \"%s\"\n", i, line);
588 read_extension_record (struct sfm_reader *r)
590 long long int offset = ftello (r->file);
591 int subtype = read_int (r);
592 size_t size = read_int (r);
593 size_t count = read_int (r);
594 size_t bytes = size * count;
596 printf ("%08llx: Record 7, subtype %d, size=%zu, count=%zu\n",
597 offset, subtype, size, count);
602 read_machine_integer_info (r, size, count);
606 read_machine_float_info (r, size, count);
610 /* Variable sets information. We don't use these yet.
611 They only apply to GUIs; see VARSETS on the APPLY
612 DICTIONARY command in SPSS documentation. */
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");
739 read_extra_product_info (struct sfm_reader *r,
740 size_t size, size_t count)
742 struct text_record *text;
745 printf ("%08llx: extra product info\n", (long long int) ftello (r->file));
746 text = open_text_record (r, size * count);
747 s = text_get_all (text);
748 print_string (s, strlen (s));
749 close_text_record (text);
752 /* Read record type 7, subtype 7. */
754 read_mrsets (struct sfm_reader *r, size_t size, size_t count)
756 struct text_record *text;
758 printf ("%08llx: multiple response sets\n",
759 (long long int) ftello (r->file));
760 text = open_text_record (r, size * count);
764 enum { MRSET_MC, MRSET_MD } type;
765 bool cat_label_from_counted_values = false;
766 bool label_from_var_label = false;
769 const char *variables;
771 while (text_match (text, '\n'))
774 name = text_tokenize (text, '=');
778 if (text_match (text, 'C'))
782 if (!text_match (text, ' '))
784 sys_warn (r, "missing space following 'C' at offset %zu "
785 "in mrsets record", text_pos (text));
789 else if (text_match (text, 'D'))
793 else if (text_match (text, 'E'))
798 cat_label_from_counted_values = true;
800 if (!text_match (text, ' '))
802 sys_warn (r, "Missing space following `%c' at offset %zu "
803 "in MRSETS record", 'E', text_pos (text));
807 number = text_tokenize (text, ' ');
808 if (!strcmp (number, "11"))
809 label_from_var_label = true;
810 else if (strcmp (number, "1"))
811 sys_warn (r, "Unexpected label source value `%s' "
812 "following `E' at offset %zu in MRSETS record",
813 number, text_pos (text));
818 sys_warn (r, "missing `C', `D', or `E' at offset %zu "
819 "in mrsets record", text_pos (text));
823 if (type == MRSET_MD)
825 counted = text_parse_counted_string (text);
830 label = text_parse_counted_string (text);
834 variables = text_tokenize (text, '\n');
835 if (variables == NULL)
837 sys_warn (r, "missing variable names following label "
838 "at offset %zu in mrsets record", text_pos (text));
842 printf ("\t\"%s\": multiple %s set",
843 name, type == MRSET_MC ? "category" : "dichotomy");
845 printf (", counted value \"%s\"", counted);
846 if (cat_label_from_counted_values)
847 printf (", category labels from counted values");
848 if (label[0] != '\0')
849 printf (", label \"%s\"", label);
850 if (label_from_var_label)
851 printf (", label from variable label");
852 printf(", variables \"%s\"\n", variables);
854 close_text_record (text);
857 /* Read record type 7, subtype 11. */
859 read_display_parameters (struct sfm_reader *r, size_t size, size_t count)
865 printf ("%08llx: variable display parameters\n",
866 (long long int) ftello (r->file));
869 sys_warn (r, "Bad size %zu on extension 11.", size);
870 skip_bytes (r, size * count);
874 n_vars = r->n_variables;
875 if (count == 3 * n_vars)
876 includes_width = true;
877 else if (count == 2 * n_vars)
878 includes_width = false;
881 sys_warn (r, "Extension 11 has bad count %zu (for %zu variables.",
883 skip_bytes (r, size * count);
887 for (i = 0; i < n_vars; ++i)
889 int measure = read_int (r);
890 int width = includes_width ? read_int (r) : 0;
891 int align = read_int (r);
893 printf ("\tVar #%zu: measure=%d (%s)", i, measure,
894 (measure == 1 ? "nominal"
895 : measure == 2 ? "ordinal"
896 : measure == 3 ? "scale"
899 printf (", width=%d", width);
900 printf (", align=%d (%s)\n", align,
902 : align == 1 ? "right"
903 : align == 2 ? "centre"
908 /* Reads record type 7, subtype 13, which gives the long name
909 that corresponds to each short name. */
911 read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count)
913 struct text_record *text;
917 printf ("%08llx: long variable names (short => long)\n",
918 (long long int) ftello (r->file));
919 text = open_text_record (r, size * count);
920 while (read_variable_to_value_pair (text, &var, &long_name))
921 printf ("\t%s => %s\n", var, long_name);
922 close_text_record (text);
925 /* Reads record type 7, subtype 14, which gives the real length
926 of each very long string. Rearranges DICT accordingly. */
928 read_long_string_map (struct sfm_reader *r, size_t size, size_t count)
930 struct text_record *text;
934 printf ("%08llx: very long strings (variable => length)\n",
935 (long long int) ftello (r->file));
936 text = open_text_record (r, size * count);
937 while (read_variable_to_value_pair (text, &var, &length_s))
938 printf ("\t%s => %d\n", var, atoi (length_s));
939 close_text_record (text);
943 read_attributes (struct sfm_reader *r, struct text_record *text,
944 const char *variable)
951 key = text_tokenize (text, '(');
955 for (index = 1; ; index++)
957 /* Parse the value. */
958 const char *value = text_tokenize (text, '\n');
961 sys_warn (r, "%s: Error parsing attribute value %s[%d]",
962 variable, key, index);
965 if (strlen (value) < 2
966 || value[0] != '\'' || value[strlen (value) - 1] != '\'')
967 sys_warn (r, "%s: Attribute value %s[%d] is not quoted: %s",
968 variable, key, index, value);
970 printf ("\t%s: %s[%d] = \"%.*s\"\n",
971 variable, key, index, (int) strlen (value) - 2, value + 1);
973 /* Was this the last value for this attribute? */
974 if (text_match (text, ')'))
978 if (text_match (text, '/'))
983 /* Read extended number of cases record. */
985 read_ncases64 (struct sfm_reader *r, size_t size, size_t count)
987 int64_t unknown, ncases64;
991 sys_warn (r, "Bad size %zu for extended number of cases.", size);
992 skip_bytes (r, size * count);
997 sys_warn (r, "Bad count %zu for extended number of cases.", size);
998 skip_bytes (r, size * count);
1001 unknown = read_int64 (r);
1002 ncases64 = read_int64 (r);
1003 printf ("%08llx: extended number of cases: "
1004 "unknown=%"PRId64", ncases64=%"PRId64"\n",
1005 (long long int) ftello (r->file), unknown, ncases64);
1009 read_datafile_attributes (struct sfm_reader *r, size_t size, size_t count)
1011 struct text_record *text;
1013 printf ("%08llx: datafile attributes\n", (long long int) ftello (r->file));
1014 text = open_text_record (r, size * count);
1015 read_attributes (r, text, "datafile");
1016 close_text_record (text);
1020 read_character_encoding (struct sfm_reader *r, size_t size, size_t count)
1022 long long int posn = ftello (r->file);
1023 char *encoding = xcalloc (size, count + 1);
1024 read_string (r, encoding, count + 1);
1026 printf ("%08llx: Character Encoding: %s\n", posn, encoding);
1030 read_long_string_value_labels (struct sfm_reader *r, size_t size, size_t count)
1032 long long int start = ftello (r->file);
1034 printf ("%08llx: long string value labels\n", start);
1035 while (ftello (r->file) - start < size * count)
1037 long long posn = ftello (r->file);
1038 char var_name[ID_MAX_LEN + 1];
1044 /* Read variable name. */
1045 var_name_len = read_int (r);
1046 if (var_name_len > ID_MAX_LEN)
1047 sys_error (r, "Variable name length in long string value label "
1048 "record (%d) exceeds %d-byte limit.",
1049 var_name_len, ID_MAX_LEN);
1050 read_string (r, var_name, var_name_len + 1);
1052 /* Read width, number of values. */
1053 width = read_int (r);
1054 n_values = read_int (r);
1056 printf ("\t%08llx: %s, width %d, %d values\n",
1057 posn, var_name, width, n_values);
1060 for (i = 0; i < n_values; i++)
1068 posn = ftello (r->file);
1071 value_length = read_int (r);
1072 value = xmalloc (value_length + 1);
1073 read_string (r, value, value_length + 1);
1076 label_length = read_int (r);
1077 label = xmalloc (label_length + 1);
1078 read_string (r, label, label_length + 1);
1080 printf ("\t\t%08llx: \"%s\" (%d bytes) => \"%s\" (%d bytes)\n",
1081 posn, value, value_length, label, label_length);
1090 read_long_string_missing_values (struct sfm_reader *r,
1091 size_t size, size_t count)
1093 long long int start = ftello (r->file);
1095 printf ("%08llx: long string missing values\n", start);
1096 while (ftello (r->file) - start < size * count)
1098 long long posn = ftello (r->file);
1099 char var_name[ID_MAX_LEN + 1];
1100 uint8_t n_missing_values;
1104 /* Read variable name. */
1105 var_name_len = read_int (r);
1106 if (var_name_len > ID_MAX_LEN)
1107 sys_error (r, "Variable name length in long string value label "
1108 "record (%d) exceeds %d-byte limit.",
1109 var_name_len, ID_MAX_LEN);
1110 read_string (r, var_name, var_name_len + 1);
1112 /* Read number of values. */
1113 read_bytes (r, &n_missing_values, 1);
1115 printf ("\t%08llx: %s, %d missing values:",
1116 posn, var_name, n_missing_values);
1119 for (i = 0; i < n_missing_values; i++)
1124 posn = ftello (r->file);
1127 value_length = read_int (r);
1128 value = xmalloc (value_length + 1);
1129 read_string (r, value, value_length + 1);
1131 printf (" \"%s\"", value);
1140 hex_dump (size_t offset, const void *buffer_, size_t buffer_size)
1142 const uint8_t *buffer = buffer_;
1144 while (buffer_size > 0)
1146 size_t n = MIN (buffer_size, 16);
1149 printf ("%04zx", offset);
1150 for (i = 0; i < 16; i++)
1153 printf ("%c%02x", i == 8 ? '-' : ' ', buffer[i]);
1159 for (i = 0; i < 16; i++)
1161 unsigned char c = i < n ? buffer[i] : ' ';
1162 putchar (isprint (c) ? c : '.');
1172 /* Reads and prints any type 7 record that we don't understand. */
1174 read_unknown_extension (struct sfm_reader *r, size_t size, size_t count)
1176 unsigned char *buffer;
1179 if (size == 0 || count > 65536 / size)
1180 skip_bytes (r, size * count);
1183 buffer = xmalloc (size);
1184 for (i = 0; i < count; i++)
1186 read_bytes (r, buffer, size);
1187 hex_dump (i * size, buffer, size);
1193 buffer = xmalloc (count);
1194 read_bytes (r, buffer, count);
1195 print_string (CHAR_CAST (char *, buffer), count);
1201 read_variable_attributes (struct sfm_reader *r, size_t size, size_t count)
1203 struct text_record *text;
1205 printf ("%08llx: variable attributes\n", (long long int) ftello (r->file));
1206 text = open_text_record (r, size * count);
1209 const char *variable = text_tokenize (text, ':');
1210 if (variable == NULL || !read_attributes (r, text, variable))
1213 close_text_record (text);
1217 read_simple_compressed_data (struct sfm_reader *r, int max_cases)
1219 enum { N_OPCODES = 8 };
1220 uint8_t opcodes[N_OPCODES];
1221 long long int opcode_ofs;
1227 printf ("\n%08llx: compressed data:\n", (long long int) ftello (r->file));
1229 opcode_idx = N_OPCODES;
1232 for (case_num = 0; case_num < max_cases; case_num++)
1234 printf ("%08llx: case %d's uncompressible data begins\n",
1235 (long long int) ftello (r->file), case_num);
1236 for (i = 0; i < r->n_var_widths; )
1238 int width = r->var_widths[i];
1242 if (opcode_idx >= N_OPCODES)
1244 opcode_ofs = ftello (r->file);
1247 if (!try_read_bytes (r, opcodes, 8))
1251 read_bytes (r, opcodes, 8);
1254 opcode = opcodes[opcode_idx];
1255 printf ("%08llx: variable %d: opcode %d: ",
1256 opcode_ofs + opcode_idx, i, opcode);
1261 printf ("%.*g", DBL_DIG + 1, opcode - r->bias);
1263 printf (", but this is a string variable (width=%d)", width);
1269 printf ("ignored padding\n");
1273 printf ("end of data\n");
1277 read_bytes (r, raw_value, 8);
1278 printf ("uncompressible data: ");
1279 print_untyped_value (r, raw_value);
1287 printf (", but this is a numeric variable");
1295 printf (", but this is a string variable (width=%d)", width);
1307 read_zlib_compressed_data (struct sfm_reader *r)
1310 long long int this_ofs, next_ofs, next_len;
1311 long long int bias, zero;
1312 long long int expected_uncmp_ofs, expected_cmp_ofs;
1313 unsigned int block_size, n_blocks;
1317 ofs = ftello (r->file);
1318 printf ("\n%08llx: ZLIB compressed data header:\n", ofs);
1320 this_ofs = read_int64 (r);
1321 next_ofs = read_int64 (r);
1322 next_len = read_int64 (r);
1324 printf ("\tzheader_ofs: 0x%llx\n", this_ofs);
1325 if (this_ofs != ofs)
1326 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1327 printf ("\tztrailer_ofs: 0x%llx\n", next_ofs);
1328 printf ("\tztrailer_len: %lld\n", next_len);
1329 if (next_len < 24 || next_len % 24)
1330 printf ("\t\t(Trailer length is not a positive multiple of 24.)\n");
1332 printf ("\n%08llx: 0x%llx bytes of ZLIB compressed data\n",
1333 ofs + 8 * 3, next_ofs - (ofs + 8 * 3));
1335 skip_bytes (r, next_ofs - (ofs + 8 * 3));
1337 printf ("\n%08llx: ZLIB trailer fixed header:\n", next_ofs);
1338 bias = read_int64 (r);
1339 zero = read_int64 (r);
1340 block_size = read_int (r);
1341 n_blocks = read_int (r);
1342 printf ("\tbias: %lld\n", bias);
1343 printf ("\tzero: 0x%llx\n", zero);
1345 printf ("\t\t(Expected 0.)\n");
1346 printf ("\tblock_size: 0x%x\n", block_size);
1347 if (block_size != 0x3ff000)
1348 printf ("\t\t(Expected 0x3ff000.)\n");
1349 printf ("\tn_blocks: %u\n", n_blocks);
1350 if (n_blocks != next_len / 24 - 1)
1351 printf ("\t\t(Expected %llu.)\n", next_len / 24 - 1);
1353 expected_uncmp_ofs = ofs;
1354 expected_cmp_ofs = ofs + 24;
1355 for (i = 0; i < n_blocks; i++)
1357 long long int blockinfo_ofs = ftello (r->file);
1358 unsigned long long int uncompressed_ofs = read_int64 (r);
1359 unsigned long long int compressed_ofs = read_int64 (r);
1360 unsigned int uncompressed_size = read_int (r);
1361 unsigned int compressed_size = read_int (r);
1363 printf ("\n%08llx: ZLIB block descriptor %d\n", blockinfo_ofs, i + 1);
1365 printf ("\tuncompressed_ofs: 0x%llx\n", uncompressed_ofs);
1366 if (uncompressed_ofs != expected_uncmp_ofs)
1367 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1369 printf ("\tcompressed_ofs: 0x%llx\n", compressed_ofs);
1370 if (compressed_ofs != expected_cmp_ofs)
1371 printf ("\t\t(Expected 0x%llx.)\n", ofs + 24);
1373 printf ("\tuncompressed_size: 0x%x\n", uncompressed_size);
1374 if (i < n_blocks - 1 && uncompressed_size != block_size)
1375 printf ("\t\t(Expected 0x%x.)\n", block_size);
1377 printf ("\tcompressed_size: 0x%x\n", compressed_size);
1378 if (i == n_blocks - 1 && compressed_ofs + compressed_size != next_ofs)
1379 printf ("\t\t(This was expected to be 0x%llx.)\n",
1380 next_ofs - compressed_size);
1382 expected_uncmp_ofs += uncompressed_size;
1383 expected_cmp_ofs += compressed_size;
1387 /* Helpers for reading records that consist of structured text
1393 struct sfm_reader *reader; /* Reader. */
1394 char *buffer; /* Record contents. */
1395 size_t size; /* Size of buffer. */
1396 size_t pos; /* Current position in buffer. */
1399 /* Reads SIZE bytes into a text record for R,
1400 and returns the new text record. */
1401 static struct text_record *
1402 open_text_record (struct sfm_reader *r, size_t size)
1404 struct text_record *text = xmalloc (sizeof *text);
1405 char *buffer = xmalloc (size + 1);
1406 read_bytes (r, buffer, size);
1407 buffer[size] = '\0';
1409 text->buffer = buffer;
1415 /* Closes TEXT and frees its storage.
1416 Not really needed, because the pool will free the text record anyway,
1417 but can be used to free it earlier. */
1419 close_text_record (struct text_record *text)
1421 free (text->buffer);
1426 text_tokenize (struct text_record *text, int delimiter)
1428 size_t start = text->pos;
1429 while (text->pos < text->size
1430 && text->buffer[text->pos] != delimiter
1431 && text->buffer[text->pos] != '\0')
1433 if (start == text->pos)
1435 text->buffer[text->pos++] = '\0';
1436 return &text->buffer[start];
1440 text_match (struct text_record *text, int c)
1442 if (text->pos < text->size && text->buffer[text->pos] == c)
1451 /* Reads a integer value expressed in decimal, then a space, then a string that
1452 consists of exactly as many bytes as specified by the integer, then a space,
1453 from TEXT. Returns the string, null-terminated, as a subset of TEXT's
1454 buffer (so the caller should not free the string). */
1456 text_parse_counted_string (struct text_record *text)
1464 while (isdigit ((unsigned char) text->buffer[text->pos]))
1465 n = (n * 10) + (text->buffer[text->pos++] - '0');
1466 if (start == text->pos)
1468 sys_error (text->reader, "expecting digit at offset %zu in record",
1473 if (!text_match (text, ' '))
1475 sys_error (text->reader, "expecting space at offset %zu in record",
1480 if (text->pos + n > text->size)
1482 sys_error (text->reader, "%zu-byte string starting at offset %zu "
1483 "exceeds record length %zu", n, text->pos, text->size);
1487 s = &text->buffer[text->pos];
1490 sys_error (text->reader, "expecting space at offset %zu following "
1491 "%zu-byte string", text->pos + n, n);
1499 /* Reads a variable=value pair from TEXT.
1500 Looks up the variable in DICT and stores it into *VAR.
1501 Stores a null-terminated value into *VALUE. */
1503 read_variable_to_value_pair (struct text_record *text,
1504 char **key, char **value)
1506 *key = text_tokenize (text, '=');
1507 *value = text_tokenize (text, '\t');
1508 if (!*key || !*value)
1511 while (text->pos < text->size
1512 && (text->buffer[text->pos] == '\t'
1513 || text->buffer[text->pos] == '\0'))
1518 /* Returns the current byte offset inside the TEXT's string. */
1520 text_pos (const struct text_record *text)
1526 text_get_all (const struct text_record *text)
1528 return text->buffer;
1535 %s, a utility for dissecting system files.\n\
1536 Usage: %s [OPTION]... SYSFILE...\n\
1537 where each SYSFILE is the name of a system file.\n\
1540 --data[=MAXCASES] print (up to MAXCASES cases of) compressed data\n\
1541 --help display this help and exit\n\
1542 --version output version information and exit\n",
1543 program_name, program_name);
1546 /* Displays a corruption message. */
1548 sys_msg (struct sfm_reader *r, const char *format, va_list args)
1550 printf ("\"%s\" near offset 0x%llx: ",
1551 r->file_name, (long long int) ftello (r->file));
1552 vprintf (format, args);
1556 /* Displays a warning for the current file position. */
1558 sys_warn (struct sfm_reader *r, const char *format, ...)
1562 va_start (args, format);
1563 sys_msg (r, format, args);
1567 /* Displays an error for the current file position,
1568 marks it as in an error state,
1569 and aborts reading it using longjmp. */
1571 sys_error (struct sfm_reader *r, const char *format, ...)
1575 va_start (args, format);
1576 sys_msg (r, format, args);
1579 exit (EXIT_FAILURE);
1582 /* Reads BYTE_CNT bytes into BUF.
1583 Returns true if exactly BYTE_CNT bytes are successfully read.
1584 Aborts if an I/O error or a partial read occurs.
1585 If EOF_IS_OK, then an immediate end-of-file causes false to be
1586 returned; otherwise, immediate end-of-file causes an abort
1589 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
1590 void *buf, size_t byte_cnt)
1592 size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
1593 if (bytes_read == byte_cnt)
1595 else if (ferror (r->file))
1596 sys_error (r, "System error: %s.", strerror (errno));
1597 else if (!eof_is_ok || bytes_read != 0)
1598 sys_error (r, "Unexpected end of file.");
1603 /* Reads BYTE_CNT into BUF.
1604 Aborts upon I/O error or if end-of-file is encountered. */
1606 read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1608 read_bytes_internal (r, false, buf, byte_cnt);
1611 /* Reads BYTE_CNT bytes into BUF.
1612 Returns true if exactly BYTE_CNT bytes are successfully read.
1613 Returns false if an immediate end-of-file is encountered.
1614 Aborts if an I/O error or a partial read occurs. */
1616 try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1618 return read_bytes_internal (r, true, buf, byte_cnt);
1621 /* Reads a 32-bit signed integer from R and returns its value in
1624 read_int (struct sfm_reader *r)
1627 read_bytes (r, integer, sizeof integer);
1628 return integer_get (r->integer_format, integer, sizeof integer);
1631 /* Reads a 64-bit signed integer from R and returns its value in
1634 read_int64 (struct sfm_reader *r)
1637 read_bytes (r, integer, sizeof integer);
1638 return integer_get (r->integer_format, integer, sizeof integer);
1641 /* Reads a 64-bit floating-point number from R and returns its
1642 value in host format. */
1644 read_float (struct sfm_reader *r)
1647 read_bytes (r, number, sizeof number);
1648 return float_get_double (r->float_format, number);
1651 /* Reads exactly SIZE - 1 bytes into BUFFER
1652 and stores a null byte into BUFFER[SIZE - 1]. */
1654 read_string (struct sfm_reader *r, char *buffer, size_t size)
1657 read_bytes (r, buffer, size - 1);
1658 buffer[size - 1] = '\0';
1661 /* Skips BYTES bytes forward in R. */
1663 skip_bytes (struct sfm_reader *r, size_t bytes)
1668 size_t chunk = MIN (sizeof buffer, bytes);
1669 read_bytes (r, buffer, chunk);
1675 trim_spaces (char *s)
1677 char *end = strchr (s, '\0');
1678 while (end > s && end[-1] == ' ')
1684 print_string (const char *s, size_t len)
1686 if (memchr (s, 0, len) == 0)
1690 for (i = 0; i < len; i++)
1692 unsigned char c = s[i];
1696 else if (c == '\n' || isprint (c))
1699 printf ("\\%02x", c);
1704 hex_dump (0, s, len);