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 enum { MAX_LABEL_LEN = 65536 };
449 size_t len = read_int (r);
450 size_t read_len = MIN (MAX_LABEL_LEN, len);
451 char *label = xmalloc (read_len + 1);
452 read_string (r, label, read_len + 1);
453 printf("\t%08llx Variable label: \"%s\"\n", offset, label);
456 /* Skip label padding up to multiple of 4 bytes. */
457 skip_bytes (r, ROUND_UP (len, 4) - len);
460 /* Set missing values. */
461 if (missing_value_code != 0)
465 printf ("\t%08llx Missing values:", (long long int) ftello (r->file));
468 if (missing_value_code < -3 || missing_value_code > 3
469 || missing_value_code == -1)
470 sys_error (r, "Numeric missing value indicator field is not "
471 "-3, -2, 0, 1, 2, or 3.");
472 if (missing_value_code < 0)
474 double low = read_float (r);
475 double high = read_float (r);
476 printf (" %.*g...%.*g", DBL_DIG + 1, low, DBL_DIG + 1, high);
477 missing_value_code = -missing_value_code - 2;
479 for (i = 0; i < missing_value_code; i++)
480 printf (" %.*g", DBL_DIG + 1, read_float (r));
484 if (missing_value_code < 1 || missing_value_code > 3)
485 sys_error (r, "String missing value indicator field is not "
487 for (i = 0; i < missing_value_code; i++)
490 read_string (r, string, sizeof string);
491 printf (" \"%s\"", string);
499 print_untyped_value (struct sfm_reader *r, char raw_value[8])
504 value = float_get_double (r->float_format, raw_value);
505 for (n_printable = 0; n_printable < 8; n_printable++)
506 if (!isprint (raw_value[n_printable]))
509 printf ("%.*g/\"%.*s\"", DBL_DIG + 1, value, n_printable, raw_value);
512 /* Reads value labels from sysfile R and inserts them into the
513 associated dictionary. */
515 read_value_label_record (struct sfm_reader *r)
517 int label_cnt, var_cnt;
520 printf ("%08llx: value labels record\n", (long long int) ftello (r->file));
522 /* Read number of labels. */
523 label_cnt = read_int (r);
524 for (i = 0; i < label_cnt; i++)
527 unsigned char label_len;
531 read_bytes (r, raw_value, sizeof raw_value);
533 /* Read label length. */
534 read_bytes (r, &label_len, sizeof label_len);
535 padded_len = ROUND_UP (label_len + 1, 8);
537 /* Read label, padding. */
538 read_bytes (r, label, padded_len - 1);
539 label[label_len] = 0;
542 print_untyped_value (r, raw_value);
543 printf (": \"%s\"\n", label);
546 /* Now, read the type 4 record that has the list of variables
547 to which the value labels are to be applied. */
549 /* Read record type of type 4 record. */
550 if (read_int (r) != 4)
551 sys_error (r, "Variable index record (type 4) does not immediately "
552 "follow value label record (type 3) as it should.");
554 /* Read number of variables associated with value label from type 4
556 printf ("\t%08llx: apply to variables", (long long int) ftello (r->file));
557 var_cnt = read_int (r);
558 for (i = 0; i < var_cnt; i++)
559 printf (" #%d", read_int (r));
564 read_document_record (struct sfm_reader *r)
569 printf ("%08llx: document record\n", (long long int) ftello (r->file));
570 n_lines = read_int (r);
571 printf ("\t%d lines of documents\n", n_lines);
573 for (i = 0; i < n_lines; i++)
576 printf ("\t%08llx: ", (long long int) ftello (r->file));
577 read_string (r, line, sizeof line);
579 printf ("line %d: \"%s\"\n", i, line);
584 read_extension_record (struct sfm_reader *r)
586 long long int offset = ftello (r->file);
587 int subtype = read_int (r);
588 size_t size = read_int (r);
589 size_t count = read_int (r);
590 size_t bytes = size * count;
592 printf ("%08llx: Record 7, subtype %d, size=%zu, count=%zu\n",
593 offset, subtype, size, count);
598 read_machine_integer_info (r, size, count);
602 read_machine_float_info (r, size, count);
606 /* DATE variable information. We don't use it yet, but we
612 read_mrsets (r, size, count);
616 read_extra_product_info (r, size, count);
620 read_display_parameters (r, size, count);
624 read_long_var_name_map (r, size, count);
628 read_long_string_map (r, size, count);
632 read_ncases64 (r, size, count);
636 read_datafile_attributes (r, size, count);
640 read_variable_attributes (r, size, count);
644 read_character_encoding (r, size, count);
648 read_long_string_value_labels (r, size, count);
652 read_long_string_missing_values (r, size, count);
656 sys_warn (r, "Unrecognized record type 7, subtype %d.", subtype);
657 read_unknown_extension (r, size, count);
661 skip_bytes (r, bytes);
665 read_machine_integer_info (struct sfm_reader *r, size_t size, size_t count)
667 long long int offset = ftello (r->file);
668 int version_major = read_int (r);
669 int version_minor = read_int (r);
670 int version_revision = read_int (r);
671 int machine_code = read_int (r);
672 int float_representation = read_int (r);
673 int compression_code = read_int (r);
674 int integer_representation = read_int (r);
675 int character_code = read_int (r);
677 printf ("%08llx: machine integer info\n", offset);
678 if (size != 4 || count != 8)
679 sys_error (r, "Bad size (%zu) or count (%zu) field on record type 7, "
680 "subtype 3.", size, count);
682 printf ("\tVersion: %d.%d.%d\n",
683 version_major, version_minor, version_revision);
684 printf ("\tMachine code: %d\n", machine_code);
685 printf ("\tFloating point representation: %d (%s)\n",
686 float_representation,
687 float_representation == 1 ? "IEEE 754"
688 : float_representation == 2 ? "IBM 370"
689 : float_representation == 3 ? "DEC VAX"
691 printf ("\tCompression code: %d\n", compression_code);
692 printf ("\tEndianness: %d (%s)\n", integer_representation,
693 integer_representation == 1 ? "big"
694 : integer_representation == 2 ? "little" : "unknown");
695 printf ("\tCharacter code: %d\n", character_code);
698 /* Read record type 7, subtype 4. */
700 read_machine_float_info (struct sfm_reader *r, size_t size, size_t count)
702 long long int offset = ftello (r->file);
703 double sysmis = read_float (r);
704 double highest = read_float (r);
705 double lowest = read_float (r);
707 printf ("%08llx: machine float info\n", offset);
708 if (size != 8 || count != 3)
709 sys_error (r, "Bad size (%zu) or count (%zu) on extension 4.",
712 printf ("\tsysmis: %.*g (%a)\n", DBL_DIG + 1, sysmis, sysmis);
713 if (sysmis != SYSMIS)
714 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
715 DBL_DIG + 1, sysmis, sysmis, "SYSMIS");
717 printf ("\thighest: %.*g (%a)\n", DBL_DIG + 1, highest, highest);
718 if (highest != HIGHEST)
719 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
720 DBL_DIG + 1, highest, highest, "HIGHEST");
722 printf ("\tlowest: %.*g (%a)\n", DBL_DIG + 1, lowest, lowest);
723 if (lowest != LOWEST && lowest != SYSMIS)
724 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
725 DBL_DIG + 1, lowest, lowest, "LOWEST");
729 read_extra_product_info (struct sfm_reader *r,
730 size_t size, size_t count)
732 struct text_record *text;
735 printf ("%08llx: extra product info\n", (long long int) ftello (r->file));
736 text = open_text_record (r, size * count);
737 s = text_get_all (text);
738 print_string (s, strlen (s));
739 close_text_record (text);
742 /* Read record type 7, subtype 7. */
744 read_mrsets (struct sfm_reader *r, size_t size, size_t count)
746 struct text_record *text;
748 printf ("%08llx: multiple response sets\n",
749 (long long int) ftello (r->file));
750 text = open_text_record (r, size * count);
754 enum { MRSET_MC, MRSET_MD } type;
755 bool cat_label_from_counted_values = false;
756 bool label_from_var_label = false;
759 const char *variables;
761 while (text_match (text, '\n'))
764 name = text_tokenize (text, '=');
768 if (text_match (text, 'C'))
772 if (!text_match (text, ' '))
774 sys_warn (r, "missing space following 'C' at offset %zu "
775 "in mrsets record", text_pos (text));
779 else if (text_match (text, 'D'))
783 else if (text_match (text, 'E'))
788 cat_label_from_counted_values = true;
790 if (!text_match (text, ' '))
792 sys_warn (r, "Missing space following `%c' at offset %zu "
793 "in MRSETS record", 'E', text_pos (text));
797 number = text_tokenize (text, ' ');
798 if (!strcmp (number, "11"))
799 label_from_var_label = true;
800 else if (strcmp (number, "1"))
801 sys_warn (r, "Unexpected label source value `%s' "
802 "following `E' at offset %zu in MRSETS record",
803 number, text_pos (text));
808 sys_warn (r, "missing `C', `D', or `E' at offset %zu "
809 "in mrsets record", text_pos (text));
813 if (type == MRSET_MD)
815 counted = text_parse_counted_string (text);
820 label = text_parse_counted_string (text);
824 variables = text_tokenize (text, '\n');
826 printf ("\t\"%s\": multiple %s set",
827 name, type == MRSET_MC ? "category" : "dichotomy");
829 printf (", counted value \"%s\"", counted);
830 if (cat_label_from_counted_values)
831 printf (", category labels from counted values");
832 if (label[0] != '\0')
833 printf (", label \"%s\"", label);
834 if (label_from_var_label)
835 printf (", label from variable label");
836 if (variables != NULL)
837 printf(", variables \"%s\"\n", variables);
839 printf(", no variables\n");
841 close_text_record (text);
844 /* Read record type 7, subtype 11. */
846 read_display_parameters (struct sfm_reader *r, size_t size, size_t count)
852 printf ("%08llx: variable display parameters\n",
853 (long long int) ftello (r->file));
856 sys_warn (r, "Bad size %zu on extension 11.", size);
857 skip_bytes (r, size * count);
861 n_vars = r->n_variables;
862 if (count == 3 * n_vars)
863 includes_width = true;
864 else if (count == 2 * n_vars)
865 includes_width = false;
868 sys_warn (r, "Extension 11 has bad count %zu (for %zu variables.",
870 skip_bytes (r, size * count);
874 for (i = 0; i < n_vars; ++i)
876 int measure = read_int (r);
877 int width = includes_width ? read_int (r) : 0;
878 int align = read_int (r);
880 printf ("\tVar #%zu: measure=%d (%s)", i, measure,
881 (measure == 1 ? "nominal"
882 : measure == 2 ? "ordinal"
883 : measure == 3 ? "scale"
886 printf (", width=%d", width);
887 printf (", align=%d (%s)\n", align,
889 : align == 1 ? "right"
890 : align == 2 ? "centre"
895 /* Reads record type 7, subtype 13, which gives the long name
896 that corresponds to each short name. */
898 read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count)
900 struct text_record *text;
904 printf ("%08llx: long variable names (short => long)\n",
905 (long long int) ftello (r->file));
906 text = open_text_record (r, size * count);
907 while (read_variable_to_value_pair (text, &var, &long_name))
908 printf ("\t%s => %s\n", var, long_name);
909 close_text_record (text);
912 /* Reads record type 7, subtype 14, which gives the real length
913 of each very long string. Rearranges DICT accordingly. */
915 read_long_string_map (struct sfm_reader *r, size_t size, size_t count)
917 struct text_record *text;
921 printf ("%08llx: very long strings (variable => length)\n",
922 (long long int) ftello (r->file));
923 text = open_text_record (r, size * count);
924 while (read_variable_to_value_pair (text, &var, &length_s))
925 printf ("\t%s => %d\n", var, atoi (length_s));
926 close_text_record (text);
930 read_attributes (struct sfm_reader *r, struct text_record *text,
931 const char *variable)
938 key = text_tokenize (text, '(');
942 for (index = 1; ; index++)
944 /* Parse the value. */
945 const char *value = text_tokenize (text, '\n');
948 sys_warn (r, "%s: Error parsing attribute value %s[%d]",
949 variable, key, index);
952 if (strlen (value) < 2
953 || value[0] != '\'' || value[strlen (value) - 1] != '\'')
954 sys_warn (r, "%s: Attribute value %s[%d] is not quoted: %s",
955 variable, key, index, value);
957 printf ("\t%s: %s[%d] = \"%.*s\"\n",
958 variable, key, index, (int) strlen (value) - 2, value + 1);
960 /* Was this the last value for this attribute? */
961 if (text_match (text, ')'))
965 if (text_match (text, '/'))
970 /* Read extended number of cases record. */
972 read_ncases64 (struct sfm_reader *r, size_t size, size_t count)
974 int64_t unknown, ncases64;
978 sys_warn (r, "Bad size %zu for extended number of cases.", size);
979 skip_bytes (r, size * count);
984 sys_warn (r, "Bad count %zu for extended number of cases.", size);
985 skip_bytes (r, size * count);
988 unknown = read_int64 (r);
989 ncases64 = read_int64 (r);
990 printf ("%08llx: extended number of cases: "
991 "unknown=%"PRId64", ncases64=%"PRId64"\n",
992 (long long int) ftello (r->file), unknown, ncases64);
996 read_datafile_attributes (struct sfm_reader *r, size_t size, size_t count)
998 struct text_record *text;
1000 printf ("%08llx: datafile attributes\n", (long long int) ftello (r->file));
1001 text = open_text_record (r, size * count);
1002 read_attributes (r, text, "datafile");
1003 close_text_record (text);
1007 read_character_encoding (struct sfm_reader *r, size_t size, size_t count)
1009 long long int posn = ftello (r->file);
1010 char *encoding = xcalloc (size, count + 1);
1011 read_string (r, encoding, count + 1);
1013 printf ("%08llx: Character Encoding: %s\n", posn, encoding);
1017 read_long_string_value_labels (struct sfm_reader *r, size_t size, size_t count)
1019 long long int start = ftello (r->file);
1021 printf ("%08llx: long string value labels\n", start);
1022 while (ftello (r->file) - start < size * count)
1024 long long posn = ftello (r->file);
1025 char var_name[ID_MAX_LEN + 1];
1031 /* Read variable name. */
1032 var_name_len = read_int (r);
1033 if (var_name_len > ID_MAX_LEN)
1034 sys_error (r, "Variable name length in long string value label "
1035 "record (%d) exceeds %d-byte limit.",
1036 var_name_len, ID_MAX_LEN);
1037 read_string (r, var_name, var_name_len + 1);
1039 /* Read width, number of values. */
1040 width = read_int (r);
1041 n_values = read_int (r);
1043 printf ("\t%08llx: %s, width %d, %d values\n",
1044 posn, var_name, width, n_values);
1047 for (i = 0; i < n_values; i++)
1055 posn = ftello (r->file);
1058 value_length = read_int (r);
1059 value = xmalloc (value_length + 1);
1060 read_string (r, value, value_length + 1);
1063 label_length = read_int (r);
1064 label = xmalloc (label_length + 1);
1065 read_string (r, label, label_length + 1);
1067 printf ("\t\t%08llx: \"%s\" (%d bytes) => \"%s\" (%d bytes)\n",
1068 posn, value, value_length, label, label_length);
1077 read_long_string_missing_values (struct sfm_reader *r,
1078 size_t size, size_t count)
1080 long long int start = ftello (r->file);
1082 printf ("%08llx: long string missing values\n", start);
1083 while (ftello (r->file) - start < size * count)
1085 long long posn = ftello (r->file);
1086 char var_name[ID_MAX_LEN + 1];
1087 uint8_t n_missing_values;
1091 /* Read variable name. */
1092 var_name_len = read_int (r);
1093 if (var_name_len > ID_MAX_LEN)
1094 sys_error (r, "Variable name length in long string value label "
1095 "record (%d) exceeds %d-byte limit.",
1096 var_name_len, ID_MAX_LEN);
1097 read_string (r, var_name, var_name_len + 1);
1099 /* Read number of values. */
1100 read_bytes (r, &n_missing_values, 1);
1102 printf ("\t%08llx: %s, %d missing values:",
1103 posn, var_name, n_missing_values);
1106 for (i = 0; i < n_missing_values; i++)
1111 posn = ftello (r->file);
1114 value_length = read_int (r);
1115 value = xmalloc (value_length + 1);
1116 read_string (r, value, value_length + 1);
1118 printf (" \"%s\"", value);
1127 hex_dump (size_t offset, const void *buffer_, size_t buffer_size)
1129 const uint8_t *buffer = buffer_;
1131 while (buffer_size > 0)
1133 size_t n = MIN (buffer_size, 16);
1136 printf ("%04zx", offset);
1137 for (i = 0; i < 16; i++)
1140 printf ("%c%02x", i == 8 ? '-' : ' ', buffer[i]);
1146 for (i = 0; i < 16; i++)
1148 unsigned char c = i < n ? buffer[i] : ' ';
1149 putchar (isprint (c) ? c : '.');
1159 /* Reads and prints any type 7 record that we don't understand. */
1161 read_unknown_extension (struct sfm_reader *r, size_t size, size_t count)
1163 unsigned char *buffer;
1166 if (size == 0 || count > 65536 / size)
1167 skip_bytes (r, size * count);
1170 buffer = xmalloc (size);
1171 for (i = 0; i < count; i++)
1173 read_bytes (r, buffer, size);
1174 hex_dump (i * size, buffer, size);
1180 buffer = xmalloc (count);
1181 read_bytes (r, buffer, count);
1182 print_string (CHAR_CAST (char *, buffer), count);
1188 read_variable_attributes (struct sfm_reader *r, size_t size, size_t count)
1190 struct text_record *text;
1192 printf ("%08llx: variable attributes\n", (long long int) ftello (r->file));
1193 text = open_text_record (r, size * count);
1196 const char *variable = text_tokenize (text, ':');
1197 if (variable == NULL || !read_attributes (r, text, variable))
1200 close_text_record (text);
1204 read_simple_compressed_data (struct sfm_reader *r, int max_cases)
1206 enum { N_OPCODES = 8 };
1207 uint8_t opcodes[N_OPCODES];
1208 long long int opcode_ofs;
1214 printf ("\n%08llx: compressed data:\n", (long long int) ftello (r->file));
1216 opcode_idx = N_OPCODES;
1219 for (case_num = 0; case_num < max_cases; case_num++)
1221 printf ("%08llx: case %d's uncompressible data begins\n",
1222 (long long int) ftello (r->file), case_num);
1223 for (i = 0; i < r->n_var_widths; )
1225 int width = r->var_widths[i];
1229 if (opcode_idx >= N_OPCODES)
1231 opcode_ofs = ftello (r->file);
1234 if (!try_read_bytes (r, opcodes, 8))
1238 read_bytes (r, opcodes, 8);
1241 opcode = opcodes[opcode_idx];
1242 printf ("%08llx: variable %d: opcode %d: ",
1243 opcode_ofs + opcode_idx, i, opcode);
1248 printf ("%.*g", DBL_DIG + 1, opcode - r->bias);
1250 printf (", but this is a string variable (width=%d)", width);
1256 printf ("ignored padding\n");
1260 printf ("end of data\n");
1264 read_bytes (r, raw_value, 8);
1265 printf ("uncompressible data: ");
1266 print_untyped_value (r, raw_value);
1274 printf (", but this is a numeric variable");
1282 printf (", but this is a string variable (width=%d)", width);
1294 read_zlib_compressed_data (struct sfm_reader *r)
1297 long long int this_ofs, next_ofs, next_len;
1298 long long int bias, zero;
1299 long long int expected_uncmp_ofs, expected_cmp_ofs;
1300 unsigned int block_size, n_blocks;
1304 ofs = ftello (r->file);
1305 printf ("\n%08llx: ZLIB compressed data header:\n", ofs);
1307 this_ofs = read_int64 (r);
1308 next_ofs = read_int64 (r);
1309 next_len = read_int64 (r);
1311 printf ("\tzheader_ofs: 0x%llx\n", this_ofs);
1312 if (this_ofs != ofs)
1313 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1314 printf ("\tztrailer_ofs: 0x%llx\n", next_ofs);
1315 printf ("\tztrailer_len: %lld\n", next_len);
1316 if (next_len < 24 || next_len % 24)
1317 printf ("\t\t(Trailer length is not a positive multiple of 24.)\n");
1319 printf ("\n%08llx: 0x%llx bytes of ZLIB compressed data\n",
1320 ofs + 8 * 3, next_ofs - (ofs + 8 * 3));
1322 skip_bytes (r, next_ofs - (ofs + 8 * 3));
1324 printf ("\n%08llx: ZLIB trailer fixed header:\n", next_ofs);
1325 bias = read_int64 (r);
1326 zero = read_int64 (r);
1327 block_size = read_int (r);
1328 n_blocks = read_int (r);
1329 printf ("\tbias: %lld\n", bias);
1330 printf ("\tzero: 0x%llx\n", zero);
1332 printf ("\t\t(Expected 0.)\n");
1333 printf ("\tblock_size: 0x%x\n", block_size);
1334 if (block_size != 0x3ff000)
1335 printf ("\t\t(Expected 0x3ff000.)\n");
1336 printf ("\tn_blocks: %u\n", n_blocks);
1337 if (n_blocks != next_len / 24 - 1)
1338 printf ("\t\t(Expected %llu.)\n", next_len / 24 - 1);
1340 expected_uncmp_ofs = ofs;
1341 expected_cmp_ofs = ofs + 24;
1342 for (i = 0; i < n_blocks; i++)
1344 long long int blockinfo_ofs = ftello (r->file);
1345 unsigned long long int uncompressed_ofs = read_int64 (r);
1346 unsigned long long int compressed_ofs = read_int64 (r);
1347 unsigned int uncompressed_size = read_int (r);
1348 unsigned int compressed_size = read_int (r);
1350 printf ("\n%08llx: ZLIB block descriptor %d\n", blockinfo_ofs, i + 1);
1352 printf ("\tuncompressed_ofs: 0x%llx\n", uncompressed_ofs);
1353 if (uncompressed_ofs != expected_uncmp_ofs)
1354 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1356 printf ("\tcompressed_ofs: 0x%llx\n", compressed_ofs);
1357 if (compressed_ofs != expected_cmp_ofs)
1358 printf ("\t\t(Expected 0x%llx.)\n", ofs + 24);
1360 printf ("\tuncompressed_size: 0x%x\n", uncompressed_size);
1361 if (i < n_blocks - 1 && uncompressed_size != block_size)
1362 printf ("\t\t(Expected 0x%x.)\n", block_size);
1364 printf ("\tcompressed_size: 0x%x\n", compressed_size);
1365 if (i == n_blocks - 1 && compressed_ofs + compressed_size != next_ofs)
1366 printf ("\t\t(This was expected to be 0x%llx.)\n",
1367 next_ofs - compressed_size);
1369 expected_uncmp_ofs += uncompressed_size;
1370 expected_cmp_ofs += compressed_size;
1374 /* Helpers for reading records that consist of structured text
1380 struct sfm_reader *reader; /* Reader. */
1381 char *buffer; /* Record contents. */
1382 size_t size; /* Size of buffer. */
1383 size_t pos; /* Current position in buffer. */
1386 /* Reads SIZE bytes into a text record for R,
1387 and returns the new text record. */
1388 static struct text_record *
1389 open_text_record (struct sfm_reader *r, size_t size)
1391 struct text_record *text = xmalloc (sizeof *text);
1392 char *buffer = xmalloc (size + 1);
1393 read_bytes (r, buffer, size);
1394 buffer[size] = '\0';
1396 text->buffer = buffer;
1402 /* Closes TEXT and frees its storage.
1403 Not really needed, because the pool will free the text record anyway,
1404 but can be used to free it earlier. */
1406 close_text_record (struct text_record *text)
1408 free (text->buffer);
1413 text_tokenize (struct text_record *text, int delimiter)
1415 size_t start = text->pos;
1416 while (text->pos < text->size
1417 && text->buffer[text->pos] != delimiter
1418 && text->buffer[text->pos] != '\0')
1420 if (start == text->pos)
1422 text->buffer[text->pos++] = '\0';
1423 return &text->buffer[start];
1427 text_match (struct text_record *text, int c)
1429 if (text->pos < text->size && text->buffer[text->pos] == c)
1438 /* Reads a integer value expressed in decimal, then a space, then a string that
1439 consists of exactly as many bytes as specified by the integer, then a space,
1440 from TEXT. Returns the string, null-terminated, as a subset of TEXT's
1441 buffer (so the caller should not free the string). */
1443 text_parse_counted_string (struct text_record *text)
1451 while (isdigit ((unsigned char) text->buffer[text->pos]))
1452 n = (n * 10) + (text->buffer[text->pos++] - '0');
1453 if (start == text->pos)
1455 sys_error (text->reader, "expecting digit at offset %zu in record",
1460 if (!text_match (text, ' '))
1462 sys_error (text->reader, "expecting space at offset %zu in record",
1467 if (text->pos + n > text->size)
1469 sys_error (text->reader, "%zu-byte string starting at offset %zu "
1470 "exceeds record length %zu", n, text->pos, text->size);
1474 s = &text->buffer[text->pos];
1477 sys_error (text->reader, "expecting space at offset %zu following "
1478 "%zu-byte string", text->pos + n, n);
1486 /* Reads a variable=value pair from TEXT.
1487 Looks up the variable in DICT and stores it into *VAR.
1488 Stores a null-terminated value into *VALUE. */
1490 read_variable_to_value_pair (struct text_record *text,
1491 char **key, char **value)
1493 *key = text_tokenize (text, '=');
1494 *value = text_tokenize (text, '\t');
1495 if (!*key || !*value)
1498 while (text->pos < text->size
1499 && (text->buffer[text->pos] == '\t'
1500 || text->buffer[text->pos] == '\0'))
1505 /* Returns the current byte offset inside the TEXT's string. */
1507 text_pos (const struct text_record *text)
1513 text_get_all (const struct text_record *text)
1515 return text->buffer;
1522 %s, a utility for dissecting system files.\n\
1523 Usage: %s [OPTION]... SYSFILE...\n\
1524 where each SYSFILE is the name of a system file.\n\
1527 --data[=MAXCASES] print (up to MAXCASES cases of) compressed data\n\
1528 --help display this help and exit\n\
1529 --version output version information and exit\n",
1530 program_name, program_name);
1533 /* Displays a corruption message. */
1535 sys_msg (struct sfm_reader *r, const char *format, va_list args)
1537 printf ("\"%s\" near offset 0x%llx: ",
1538 r->file_name, (long long int) ftello (r->file));
1539 vprintf (format, args);
1543 /* Displays a warning for the current file position. */
1545 sys_warn (struct sfm_reader *r, const char *format, ...)
1549 va_start (args, format);
1550 sys_msg (r, format, args);
1554 /* Displays an error for the current file position,
1555 marks it as in an error state,
1556 and aborts reading it using longjmp. */
1558 sys_error (struct sfm_reader *r, const char *format, ...)
1562 va_start (args, format);
1563 sys_msg (r, format, args);
1566 exit (EXIT_FAILURE);
1569 /* Reads BYTE_CNT bytes into BUF.
1570 Returns true if exactly BYTE_CNT bytes are successfully read.
1571 Aborts if an I/O error or a partial read occurs.
1572 If EOF_IS_OK, then an immediate end-of-file causes false to be
1573 returned; otherwise, immediate end-of-file causes an abort
1576 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
1577 void *buf, size_t byte_cnt)
1579 size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
1580 if (bytes_read == byte_cnt)
1582 else if (ferror (r->file))
1583 sys_error (r, "System error: %s.", strerror (errno));
1584 else if (!eof_is_ok || bytes_read != 0)
1585 sys_error (r, "Unexpected end of file.");
1590 /* Reads BYTE_CNT into BUF.
1591 Aborts upon I/O error or if end-of-file is encountered. */
1593 read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1595 read_bytes_internal (r, false, buf, byte_cnt);
1598 /* Reads BYTE_CNT bytes into BUF.
1599 Returns true if exactly BYTE_CNT bytes are successfully read.
1600 Returns false if an immediate end-of-file is encountered.
1601 Aborts if an I/O error or a partial read occurs. */
1603 try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1605 return read_bytes_internal (r, true, buf, byte_cnt);
1608 /* Reads a 32-bit signed integer from R and returns its value in
1611 read_int (struct sfm_reader *r)
1614 read_bytes (r, integer, sizeof integer);
1615 return integer_get (r->integer_format, integer, sizeof integer);
1618 /* Reads a 64-bit signed integer from R and returns its value in
1621 read_int64 (struct sfm_reader *r)
1624 read_bytes (r, integer, sizeof integer);
1625 return integer_get (r->integer_format, integer, sizeof integer);
1628 /* Reads a 64-bit floating-point number from R and returns its
1629 value in host format. */
1631 read_float (struct sfm_reader *r)
1634 read_bytes (r, number, sizeof number);
1635 return float_get_double (r->float_format, number);
1638 /* Reads exactly SIZE - 1 bytes into BUFFER
1639 and stores a null byte into BUFFER[SIZE - 1]. */
1641 read_string (struct sfm_reader *r, char *buffer, size_t size)
1644 read_bytes (r, buffer, size - 1);
1645 buffer[size - 1] = '\0';
1648 /* Skips BYTES bytes forward in R. */
1650 skip_bytes (struct sfm_reader *r, size_t bytes)
1655 size_t chunk = MIN (sizeof buffer, bytes);
1656 read_bytes (r, buffer, chunk);
1662 trim_spaces (char *s)
1664 char *end = strchr (s, '\0');
1665 while (end > s && end[-1] == ' ')
1671 print_string (const char *s, size_t len)
1673 if (memchr (s, 0, len) == 0)
1677 for (i = 0; i < len; i++)
1679 unsigned char c = s[i];
1683 else if (c == '\n' || isprint (c))
1686 printf ("\\%02x", c);
1691 hex_dump (0, s, len);