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 case 40: return "MTIME";
383 case 41: return "YMDHMS";
384 default: return "invalid";
388 /* Reads a variable (type 2) record from R and adds the
389 corresponding variable to DICT.
390 Also skips past additional variable records for long string
393 read_variable_record (struct sfm_reader *r)
396 int has_variable_label;
397 int missing_value_code;
402 printf ("%08llx: variable record #%d\n",
403 (long long int) ftello (r->file), r->n_variable_records++);
405 width = read_int (r);
406 has_variable_label = read_int (r);
407 missing_value_code = read_int (r);
408 print_format = read_int (r);
409 write_format = read_int (r);
410 read_string (r, name, sizeof name);
411 name[strcspn (name, " ")] = '\0';
416 if (r->n_var_widths >= r->allocated_var_widths)
417 r->var_widths = x2nrealloc (r->var_widths, &r->allocated_var_widths,
418 sizeof *r->var_widths);
419 r->var_widths[r->n_var_widths++] = width;
421 printf ("\tWidth: %d (%s)\n",
424 : width == 0 ? "numeric"
425 : "long string continuation record");
426 printf ("\tVariable label: %d\n", has_variable_label);
427 printf ("\tMissing values code: %d (%s)\n", missing_value_code,
428 (missing_value_code == 0 ? "no missing values"
429 : missing_value_code == 1 ? "one missing value"
430 : missing_value_code == 2 ? "two missing values"
431 : missing_value_code == 3 ? "three missing values"
432 : missing_value_code == -2 ? "one missing value range"
433 : missing_value_code == -3 ? "one missing value, one range"
435 printf ("\tPrint format: %06x (%s%d.%d)\n",
436 print_format, format_name (print_format),
437 (print_format >> 8) & 0xff, print_format & 0xff);
438 printf ("\tWrite format: %06x (%s%d.%d)\n",
439 write_format, format_name (write_format),
440 (write_format >> 8) & 0xff, write_format & 0xff);
441 printf ("\tName: %s\n", name);
443 /* Get variable label, if any. */
444 if (has_variable_label != 0 && has_variable_label != 1)
445 sys_error (r, "Variable label indicator field is not 0 or 1.");
446 if (has_variable_label == 1)
448 long long int offset = ftello (r->file);
449 enum { MAX_LABEL_LEN = 65536 };
451 size_t len = read_int (r);
452 size_t read_len = MIN (MAX_LABEL_LEN, len);
453 char *label = xmalloc (read_len + 1);
454 read_string (r, label, read_len + 1);
455 printf("\t%08llx Variable label: \"%s\"\n", offset, label);
458 /* Skip label padding up to multiple of 4 bytes. */
459 skip_bytes (r, ROUND_UP (len, 4) - len);
462 /* Set missing values. */
463 if (missing_value_code != 0)
467 printf ("\t%08llx Missing values:", (long long int) ftello (r->file));
470 if (missing_value_code < -3 || missing_value_code > 3
471 || missing_value_code == -1)
472 sys_error (r, "Numeric missing value indicator field is not "
473 "-3, -2, 0, 1, 2, or 3.");
474 if (missing_value_code < 0)
476 double low = read_float (r);
477 double high = read_float (r);
478 printf (" %.*g...%.*g", DBL_DIG + 1, low, DBL_DIG + 1, high);
479 missing_value_code = -missing_value_code - 2;
481 for (i = 0; i < missing_value_code; i++)
482 printf (" %.*g", DBL_DIG + 1, read_float (r));
486 if (missing_value_code < 1 || missing_value_code > 3)
487 sys_error (r, "String missing value indicator field is not "
489 for (i = 0; i < missing_value_code; i++)
492 read_string (r, string, sizeof string);
493 printf (" \"%s\"", string);
501 print_untyped_value (struct sfm_reader *r, char raw_value[8])
506 value = float_get_double (r->float_format, raw_value);
507 for (n_printable = 0; n_printable < 8; n_printable++)
508 if (!isprint (raw_value[n_printable]))
511 printf ("%.*g/\"%.*s\"", DBL_DIG + 1, value, n_printable, raw_value);
514 /* Reads value labels from sysfile R and inserts them into the
515 associated dictionary. */
517 read_value_label_record (struct sfm_reader *r)
519 int label_cnt, var_cnt;
522 printf ("%08llx: value labels record\n", (long long int) ftello (r->file));
524 /* Read number of labels. */
525 label_cnt = read_int (r);
526 for (i = 0; i < label_cnt; i++)
529 unsigned char label_len;
533 read_bytes (r, raw_value, sizeof raw_value);
535 /* Read label length. */
536 read_bytes (r, &label_len, sizeof label_len);
537 padded_len = ROUND_UP (label_len + 1, 8);
539 /* Read label, padding. */
540 read_bytes (r, label, padded_len - 1);
541 label[label_len] = 0;
544 print_untyped_value (r, raw_value);
545 printf (": \"%s\"\n", label);
548 /* Now, read the type 4 record that has the list of variables
549 to which the value labels are to be applied. */
551 /* Read record type of type 4 record. */
552 if (read_int (r) != 4)
553 sys_error (r, "Variable index record (type 4) does not immediately "
554 "follow value label record (type 3) as it should.");
556 /* Read number of variables associated with value label from type 4
558 printf ("\t%08llx: apply to variables", (long long int) ftello (r->file));
559 var_cnt = read_int (r);
560 for (i = 0; i < var_cnt; i++)
561 printf (" #%d", read_int (r));
566 read_document_record (struct sfm_reader *r)
571 printf ("%08llx: document record\n", (long long int) ftello (r->file));
572 n_lines = read_int (r);
573 printf ("\t%d lines of documents\n", n_lines);
575 for (i = 0; i < n_lines; i++)
578 printf ("\t%08llx: ", (long long int) ftello (r->file));
579 read_string (r, line, sizeof line);
581 printf ("line %d: \"%s\"\n", i, line);
586 read_extension_record (struct sfm_reader *r)
588 long long int offset = ftello (r->file);
589 int subtype = read_int (r);
590 size_t size = read_int (r);
591 size_t count = read_int (r);
592 size_t bytes = size * count;
594 printf ("%08llx: Record 7, subtype %d, size=%zu, count=%zu\n",
595 offset, subtype, size, count);
600 read_machine_integer_info (r, size, count);
604 read_machine_float_info (r, size, count);
608 /* DATE variable information. We don't use it yet, but we
614 read_mrsets (r, size, count);
618 read_extra_product_info (r, size, count);
622 read_display_parameters (r, size, count);
626 read_long_var_name_map (r, size, count);
630 read_long_string_map (r, size, count);
634 read_ncases64 (r, size, count);
638 read_datafile_attributes (r, size, count);
642 read_variable_attributes (r, size, count);
646 read_character_encoding (r, size, count);
650 read_long_string_value_labels (r, size, count);
654 read_long_string_missing_values (r, size, count);
658 sys_warn (r, "Unrecognized record type 7, subtype %d.", subtype);
659 read_unknown_extension (r, size, count);
663 skip_bytes (r, bytes);
667 read_machine_integer_info (struct sfm_reader *r, size_t size, size_t count)
669 long long int offset = ftello (r->file);
670 int version_major = read_int (r);
671 int version_minor = read_int (r);
672 int version_revision = read_int (r);
673 int machine_code = read_int (r);
674 int float_representation = read_int (r);
675 int compression_code = read_int (r);
676 int integer_representation = read_int (r);
677 int character_code = read_int (r);
679 printf ("%08llx: machine integer info\n", offset);
680 if (size != 4 || count != 8)
681 sys_error (r, "Bad size (%zu) or count (%zu) field on record type 7, "
682 "subtype 3.", size, count);
684 printf ("\tVersion: %d.%d.%d\n",
685 version_major, version_minor, version_revision);
686 printf ("\tMachine code: %d\n", machine_code);
687 printf ("\tFloating point representation: %d (%s)\n",
688 float_representation,
689 float_representation == 1 ? "IEEE 754"
690 : float_representation == 2 ? "IBM 370"
691 : float_representation == 3 ? "DEC VAX"
693 printf ("\tCompression code: %d\n", compression_code);
694 printf ("\tEndianness: %d (%s)\n", integer_representation,
695 integer_representation == 1 ? "big"
696 : integer_representation == 2 ? "little" : "unknown");
697 printf ("\tCharacter code: %d\n", character_code);
700 /* Read record type 7, subtype 4. */
702 read_machine_float_info (struct sfm_reader *r, size_t size, size_t count)
704 long long int offset = ftello (r->file);
705 double sysmis = read_float (r);
706 double highest = read_float (r);
707 double lowest = read_float (r);
709 printf ("%08llx: machine float info\n", offset);
710 if (size != 8 || count != 3)
711 sys_error (r, "Bad size (%zu) or count (%zu) on extension 4.",
714 printf ("\tsysmis: %.*g (%a)\n", DBL_DIG + 1, sysmis, sysmis);
715 if (sysmis != SYSMIS)
716 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
717 DBL_DIG + 1, sysmis, sysmis, "SYSMIS");
719 printf ("\thighest: %.*g (%a)\n", DBL_DIG + 1, highest, highest);
720 if (highest != HIGHEST)
721 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
722 DBL_DIG + 1, highest, highest, "HIGHEST");
724 printf ("\tlowest: %.*g (%a)\n", DBL_DIG + 1, lowest, lowest);
725 if (lowest != LOWEST && lowest != SYSMIS)
726 sys_warn (r, "File specifies unexpected value %.*g (%a) as %s.",
727 DBL_DIG + 1, lowest, lowest, "LOWEST");
731 read_extra_product_info (struct sfm_reader *r,
732 size_t size, size_t count)
734 struct text_record *text;
737 printf ("%08llx: extra product info\n", (long long int) ftello (r->file));
738 text = open_text_record (r, size * count);
739 s = text_get_all (text);
740 print_string (s, strlen (s));
741 close_text_record (text);
744 /* Read record type 7, subtype 7. */
746 read_mrsets (struct sfm_reader *r, size_t size, size_t count)
748 struct text_record *text;
750 printf ("%08llx: multiple response sets\n",
751 (long long int) ftello (r->file));
752 text = open_text_record (r, size * count);
756 enum { MRSET_MC, MRSET_MD } type;
757 bool cat_label_from_counted_values = false;
758 bool label_from_var_label = false;
761 const char *variables;
763 while (text_match (text, '\n'))
766 name = text_tokenize (text, '=');
770 if (text_match (text, 'C'))
774 if (!text_match (text, ' '))
776 sys_warn (r, "missing space following 'C' at offset %zu "
777 "in mrsets record", text_pos (text));
781 else if (text_match (text, 'D'))
785 else if (text_match (text, 'E'))
790 cat_label_from_counted_values = true;
792 if (!text_match (text, ' '))
794 sys_warn (r, "Missing space following `%c' at offset %zu "
795 "in MRSETS record", 'E', text_pos (text));
799 number = text_tokenize (text, ' ');
801 sys_warn (r, "Missing label source value "
802 "following `E' at offset %zu in MRSETS record",
804 else if (!strcmp (number, "11"))
805 label_from_var_label = true;
806 else if (strcmp (number, "1"))
807 sys_warn (r, "Unexpected label source value `%s' "
808 "following `E' at offset %zu in MRSETS record",
809 number, text_pos (text));
814 sys_warn (r, "missing `C', `D', or `E' at offset %zu "
815 "in mrsets record", text_pos (text));
819 if (type == MRSET_MD)
821 counted = text_parse_counted_string (text);
826 label = text_parse_counted_string (text);
830 variables = text_tokenize (text, '\n');
832 printf ("\t\"%s\": multiple %s set",
833 name, type == MRSET_MC ? "category" : "dichotomy");
835 printf (", counted value \"%s\"", counted);
836 if (cat_label_from_counted_values)
837 printf (", category labels from counted values");
838 if (label[0] != '\0')
839 printf (", label \"%s\"", label);
840 if (label_from_var_label)
841 printf (", label from variable label");
842 if (variables != NULL)
843 printf(", variables \"%s\"\n", variables);
845 printf(", no variables\n");
847 close_text_record (text);
850 /* Read record type 7, subtype 11. */
852 read_display_parameters (struct sfm_reader *r, size_t size, size_t count)
858 printf ("%08llx: variable display parameters\n",
859 (long long int) ftello (r->file));
862 sys_warn (r, "Bad size %zu on extension 11.", size);
863 skip_bytes (r, size * count);
867 n_vars = r->n_variables;
868 if (count == 3 * n_vars)
869 includes_width = true;
870 else if (count == 2 * n_vars)
871 includes_width = false;
874 sys_warn (r, "Extension 11 has bad count %zu (for %zu variables.",
876 skip_bytes (r, size * count);
880 for (i = 0; i < n_vars; ++i)
882 int measure = read_int (r);
883 int width = includes_width ? read_int (r) : 0;
884 int align = read_int (r);
886 printf ("\tVar #%zu: measure=%d (%s)", i, measure,
887 (measure == 1 ? "nominal"
888 : measure == 2 ? "ordinal"
889 : measure == 3 ? "scale"
892 printf (", width=%d", width);
893 printf (", align=%d (%s)\n", align,
895 : align == 1 ? "right"
896 : align == 2 ? "centre"
901 /* Reads record type 7, subtype 13, which gives the long name
902 that corresponds to each short name. */
904 read_long_var_name_map (struct sfm_reader *r, size_t size, size_t count)
906 struct text_record *text;
910 printf ("%08llx: long variable names (short => long)\n",
911 (long long int) ftello (r->file));
912 text = open_text_record (r, size * count);
913 while (read_variable_to_value_pair (text, &var, &long_name))
914 printf ("\t%s => %s\n", var, long_name);
915 close_text_record (text);
918 /* Reads record type 7, subtype 14, which gives the real length
919 of each very long string. Rearranges DICT accordingly. */
921 read_long_string_map (struct sfm_reader *r, size_t size, size_t count)
923 struct text_record *text;
927 printf ("%08llx: very long strings (variable => length)\n",
928 (long long int) ftello (r->file));
929 text = open_text_record (r, size * count);
930 while (read_variable_to_value_pair (text, &var, &length_s))
931 printf ("\t%s => %d\n", var, atoi (length_s));
932 close_text_record (text);
936 read_attributes (struct sfm_reader *r, struct text_record *text,
937 const char *variable)
944 key = text_tokenize (text, '(');
948 for (index = 1; ; index++)
950 /* Parse the value. */
951 const char *value = text_tokenize (text, '\n');
954 sys_warn (r, "%s: Error parsing attribute value %s[%d]",
955 variable, key, index);
958 if (strlen (value) < 2
959 || value[0] != '\'' || value[strlen (value) - 1] != '\'')
960 sys_warn (r, "%s: Attribute value %s[%d] is not quoted: %s",
961 variable, key, index, value);
963 printf ("\t%s: %s[%d] = \"%.*s\"\n",
964 variable, key, index, (int) strlen (value) - 2, value + 1);
966 /* Was this the last value for this attribute? */
967 if (text_match (text, ')'))
971 if (text_match (text, '/'))
976 /* Read extended number of cases record. */
978 read_ncases64 (struct sfm_reader *r, size_t size, size_t count)
980 int64_t unknown, ncases64;
984 sys_warn (r, "Bad size %zu for extended number of cases.", size);
985 skip_bytes (r, size * count);
990 sys_warn (r, "Bad count %zu for extended number of cases.", size);
991 skip_bytes (r, size * count);
994 unknown = read_int64 (r);
995 ncases64 = read_int64 (r);
996 printf ("%08llx: extended number of cases: "
997 "unknown=%"PRId64", ncases64=%"PRId64"\n",
998 (long long int) ftello (r->file), unknown, ncases64);
1002 read_datafile_attributes (struct sfm_reader *r, size_t size, size_t count)
1004 struct text_record *text;
1006 printf ("%08llx: datafile attributes\n", (long long int) ftello (r->file));
1007 text = open_text_record (r, size * count);
1008 read_attributes (r, text, "datafile");
1009 close_text_record (text);
1013 read_character_encoding (struct sfm_reader *r, size_t size, size_t count)
1015 long long int posn = ftello (r->file);
1016 char *encoding = xcalloc (size, count + 1);
1017 read_string (r, encoding, count + 1);
1019 printf ("%08llx: Character Encoding: %s\n", posn, encoding);
1023 read_long_string_value_labels (struct sfm_reader *r, size_t size, size_t count)
1025 long long int start = ftello (r->file);
1027 printf ("%08llx: long string value labels\n", start);
1028 while (ftello (r->file) - start < size * count)
1030 long long posn = ftello (r->file);
1031 char var_name[ID_MAX_LEN + 1];
1037 /* Read variable name. */
1038 var_name_len = read_int (r);
1039 if (var_name_len > ID_MAX_LEN)
1040 sys_error (r, "Variable name length in long string value label "
1041 "record (%d) exceeds %d-byte limit.",
1042 var_name_len, ID_MAX_LEN);
1043 read_string (r, var_name, var_name_len + 1);
1045 /* Read width, number of values. */
1046 width = read_int (r);
1047 n_values = read_int (r);
1049 printf ("\t%08llx: %s, width %d, %d values\n",
1050 posn, var_name, width, n_values);
1053 for (i = 0; i < n_values; i++)
1061 posn = ftello (r->file);
1064 value_length = read_int (r);
1065 value = xmalloc (value_length + 1);
1066 read_string (r, value, value_length + 1);
1069 label_length = read_int (r);
1070 label = xmalloc (label_length + 1);
1071 read_string (r, label, label_length + 1);
1073 printf ("\t\t%08llx: \"%s\" (%d bytes) => \"%s\" (%d bytes)\n",
1074 posn, value, value_length, label, label_length);
1083 read_long_string_missing_values (struct sfm_reader *r,
1084 size_t size, size_t count)
1086 long long int start = ftello (r->file);
1088 printf ("%08llx: long string missing values\n", start);
1089 while (ftello (r->file) - start < size * count)
1091 long long posn = ftello (r->file);
1092 char var_name[ID_MAX_LEN + 1];
1093 uint8_t n_missing_values;
1097 /* Read variable name. */
1098 var_name_len = read_int (r);
1099 if (var_name_len > ID_MAX_LEN)
1100 sys_error (r, "Variable name length in long string value label "
1101 "record (%d) exceeds %d-byte limit.",
1102 var_name_len, ID_MAX_LEN);
1103 read_string (r, var_name, var_name_len + 1);
1105 /* Read number of values. */
1106 read_bytes (r, &n_missing_values, 1);
1108 printf ("\t%08llx: %s, %d missing values:",
1109 posn, var_name, n_missing_values);
1112 for (i = 0; i < n_missing_values; i++)
1117 posn = ftello (r->file);
1120 value_length = read_int (r);
1121 value = xmalloc (value_length + 1);
1122 read_string (r, value, value_length + 1);
1124 printf (" \"%s\"", value);
1133 hex_dump (size_t offset, const void *buffer_, size_t buffer_size)
1135 const uint8_t *buffer = buffer_;
1137 while (buffer_size > 0)
1139 size_t n = MIN (buffer_size, 16);
1142 printf ("%04zx", offset);
1143 for (i = 0; i < 16; i++)
1146 printf ("%c%02x", i == 8 ? '-' : ' ', buffer[i]);
1152 for (i = 0; i < 16; i++)
1154 unsigned char c = i < n ? buffer[i] : ' ';
1155 putchar (isprint (c) ? c : '.');
1165 /* Reads and prints any type 7 record that we don't understand. */
1167 read_unknown_extension (struct sfm_reader *r, size_t size, size_t count)
1169 unsigned char *buffer;
1172 if (size == 0 || count > 65536 / size)
1173 skip_bytes (r, size * count);
1176 buffer = xmalloc (size);
1177 for (i = 0; i < count; i++)
1179 read_bytes (r, buffer, size);
1180 hex_dump (i * size, buffer, size);
1186 buffer = xmalloc (count);
1187 read_bytes (r, buffer, count);
1188 print_string (CHAR_CAST (char *, buffer), count);
1194 read_variable_attributes (struct sfm_reader *r, size_t size, size_t count)
1196 struct text_record *text;
1198 printf ("%08llx: variable attributes\n", (long long int) ftello (r->file));
1199 text = open_text_record (r, size * count);
1202 const char *variable = text_tokenize (text, ':');
1203 if (variable == NULL || !read_attributes (r, text, variable))
1206 close_text_record (text);
1210 read_simple_compressed_data (struct sfm_reader *r, int max_cases)
1212 enum { N_OPCODES = 8 };
1213 uint8_t opcodes[N_OPCODES];
1214 long long int opcode_ofs;
1220 printf ("\n%08llx: compressed data:\n", (long long int) ftello (r->file));
1222 opcode_idx = N_OPCODES;
1225 for (case_num = 0; case_num < max_cases; case_num++)
1227 printf ("%08llx: case %d's uncompressible data begins\n",
1228 (long long int) ftello (r->file), case_num);
1229 for (i = 0; i < r->n_var_widths; )
1231 int width = r->var_widths[i];
1235 if (opcode_idx >= N_OPCODES)
1237 opcode_ofs = ftello (r->file);
1240 if (!try_read_bytes (r, opcodes, 8))
1244 read_bytes (r, opcodes, 8);
1247 opcode = opcodes[opcode_idx];
1248 printf ("%08llx: variable %d: opcode %d: ",
1249 opcode_ofs + opcode_idx, i, opcode);
1254 printf ("%.*g", DBL_DIG + 1, opcode - r->bias);
1256 printf (", but this is a string variable (width=%d)", width);
1262 printf ("ignored padding\n");
1266 printf ("end of data\n");
1270 read_bytes (r, raw_value, 8);
1271 printf ("uncompressible data: ");
1272 print_untyped_value (r, raw_value);
1280 printf (", but this is a numeric variable");
1288 printf (", but this is a string variable (width=%d)", width);
1300 read_zlib_compressed_data (struct sfm_reader *r)
1303 long long int this_ofs, next_ofs, next_len;
1304 long long int bias, zero;
1305 long long int expected_uncmp_ofs, expected_cmp_ofs;
1306 unsigned int block_size, n_blocks;
1310 ofs = ftello (r->file);
1311 printf ("\n%08llx: ZLIB compressed data header:\n", ofs);
1313 this_ofs = read_int64 (r);
1314 next_ofs = read_int64 (r);
1315 next_len = read_int64 (r);
1317 printf ("\tzheader_ofs: 0x%llx\n", this_ofs);
1318 if (this_ofs != ofs)
1319 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1320 printf ("\tztrailer_ofs: 0x%llx\n", next_ofs);
1321 printf ("\tztrailer_len: %lld\n", next_len);
1322 if (next_len < 24 || next_len % 24)
1323 printf ("\t\t(Trailer length is not a positive multiple of 24.)\n");
1325 printf ("\n%08llx: 0x%llx bytes of ZLIB compressed data\n",
1326 ofs + 8 * 3, next_ofs - (ofs + 8 * 3));
1328 skip_bytes (r, next_ofs - (ofs + 8 * 3));
1330 printf ("\n%08llx: ZLIB trailer fixed header:\n", next_ofs);
1331 bias = read_int64 (r);
1332 zero = read_int64 (r);
1333 block_size = read_int (r);
1334 n_blocks = read_int (r);
1335 printf ("\tbias: %lld\n", bias);
1336 printf ("\tzero: 0x%llx\n", zero);
1338 printf ("\t\t(Expected 0.)\n");
1339 printf ("\tblock_size: 0x%x\n", block_size);
1340 if (block_size != 0x3ff000)
1341 printf ("\t\t(Expected 0x3ff000.)\n");
1342 printf ("\tn_blocks: %u\n", n_blocks);
1343 if (n_blocks != next_len / 24 - 1)
1344 printf ("\t\t(Expected %llu.)\n", next_len / 24 - 1);
1346 expected_uncmp_ofs = ofs;
1347 expected_cmp_ofs = ofs + 24;
1348 for (i = 0; i < n_blocks; i++)
1350 long long int blockinfo_ofs = ftello (r->file);
1351 unsigned long long int uncompressed_ofs = read_int64 (r);
1352 unsigned long long int compressed_ofs = read_int64 (r);
1353 unsigned int uncompressed_size = read_int (r);
1354 unsigned int compressed_size = read_int (r);
1356 printf ("\n%08llx: ZLIB block descriptor %d\n", blockinfo_ofs, i + 1);
1358 printf ("\tuncompressed_ofs: 0x%llx\n", uncompressed_ofs);
1359 if (uncompressed_ofs != expected_uncmp_ofs)
1360 printf ("\t\t(Expected 0x%llx.)\n", ofs);
1362 printf ("\tcompressed_ofs: 0x%llx\n", compressed_ofs);
1363 if (compressed_ofs != expected_cmp_ofs)
1364 printf ("\t\t(Expected 0x%llx.)\n", ofs + 24);
1366 printf ("\tuncompressed_size: 0x%x\n", uncompressed_size);
1367 if (i < n_blocks - 1 && uncompressed_size != block_size)
1368 printf ("\t\t(Expected 0x%x.)\n", block_size);
1370 printf ("\tcompressed_size: 0x%x\n", compressed_size);
1371 if (i == n_blocks - 1 && compressed_ofs + compressed_size != next_ofs)
1372 printf ("\t\t(This was expected to be 0x%llx.)\n",
1373 next_ofs - compressed_size);
1375 expected_uncmp_ofs += uncompressed_size;
1376 expected_cmp_ofs += compressed_size;
1380 /* Helpers for reading records that consist of structured text
1386 struct sfm_reader *reader; /* Reader. */
1387 char *buffer; /* Record contents. */
1388 size_t size; /* Size of buffer. */
1389 size_t pos; /* Current position in buffer. */
1392 /* Reads SIZE bytes into a text record for R,
1393 and returns the new text record. */
1394 static struct text_record *
1395 open_text_record (struct sfm_reader *r, size_t size)
1397 struct text_record *text = xmalloc (sizeof *text);
1398 char *buffer = xmalloc (size + 1);
1399 read_bytes (r, buffer, size);
1400 buffer[size] = '\0';
1402 text->buffer = buffer;
1408 /* Closes TEXT and frees its storage.
1409 Not really needed, because the pool will free the text record anyway,
1410 but can be used to free it earlier. */
1412 close_text_record (struct text_record *text)
1414 free (text->buffer);
1419 text_tokenize (struct text_record *text, int delimiter)
1421 size_t start = text->pos;
1422 while (text->pos < text->size
1423 && text->buffer[text->pos] != delimiter
1424 && text->buffer[text->pos] != '\0')
1426 if (start == text->pos)
1428 text->buffer[text->pos++] = '\0';
1429 return &text->buffer[start];
1433 text_match (struct text_record *text, int c)
1435 if (text->pos < text->size && text->buffer[text->pos] == c)
1444 /* Reads a integer value expressed in decimal, then a space, then a string that
1445 consists of exactly as many bytes as specified by the integer, then a space,
1446 from TEXT. Returns the string, null-terminated, as a subset of TEXT's
1447 buffer (so the caller should not free the string). */
1449 text_parse_counted_string (struct text_record *text)
1457 while (isdigit ((unsigned char) text->buffer[text->pos]))
1458 n = (n * 10) + (text->buffer[text->pos++] - '0');
1459 if (start == text->pos)
1461 sys_error (text->reader, "expecting digit at offset %zu in record",
1466 if (!text_match (text, ' '))
1468 sys_error (text->reader, "expecting space at offset %zu in record",
1473 if (text->pos + n > text->size)
1475 sys_error (text->reader, "%zu-byte string starting at offset %zu "
1476 "exceeds record length %zu", n, text->pos, text->size);
1480 s = &text->buffer[text->pos];
1483 sys_error (text->reader, "expecting space at offset %zu following "
1484 "%zu-byte string", text->pos + n, n);
1492 /* Reads a variable=value pair from TEXT.
1493 Looks up the variable in DICT and stores it into *VAR.
1494 Stores a null-terminated value into *VALUE. */
1496 read_variable_to_value_pair (struct text_record *text,
1497 char **key, char **value)
1499 *key = text_tokenize (text, '=');
1500 *value = text_tokenize (text, '\t');
1501 if (!*key || !*value)
1504 while (text->pos < text->size
1505 && (text->buffer[text->pos] == '\t'
1506 || text->buffer[text->pos] == '\0'))
1511 /* Returns the current byte offset inside the TEXT's string. */
1513 text_pos (const struct text_record *text)
1519 text_get_all (const struct text_record *text)
1521 return text->buffer;
1528 %s, a utility for dissecting system files.\n\
1529 Usage: %s [OPTION]... SYSFILE...\n\
1530 where each SYSFILE is the name of a system file.\n\
1533 --data[=MAXCASES] print (up to MAXCASES cases of) compressed data\n\
1534 --help display this help and exit\n\
1535 --version output version information and exit\n",
1536 program_name, program_name);
1539 /* Displays a corruption message. */
1541 sys_msg (struct sfm_reader *r, const char *format, va_list args)
1543 printf ("\"%s\" near offset 0x%llx: ",
1544 r->file_name, (long long int) ftello (r->file));
1545 vprintf (format, args);
1549 /* Displays a warning for the current file position. */
1551 sys_warn (struct sfm_reader *r, const char *format, ...)
1555 va_start (args, format);
1556 sys_msg (r, format, args);
1560 /* Displays an error for the current file position,
1561 marks it as in an error state,
1562 and aborts reading it using longjmp. */
1564 sys_error (struct sfm_reader *r, const char *format, ...)
1568 va_start (args, format);
1569 sys_msg (r, format, args);
1572 exit (EXIT_FAILURE);
1575 /* Reads BYTE_CNT bytes into BUF.
1576 Returns true if exactly BYTE_CNT bytes are successfully read.
1577 Aborts if an I/O error or a partial read occurs.
1578 If EOF_IS_OK, then an immediate end-of-file causes false to be
1579 returned; otherwise, immediate end-of-file causes an abort
1582 read_bytes_internal (struct sfm_reader *r, bool eof_is_ok,
1583 void *buf, size_t byte_cnt)
1585 size_t bytes_read = fread (buf, 1, byte_cnt, r->file);
1586 if (bytes_read == byte_cnt)
1588 else if (ferror (r->file))
1589 sys_error (r, "System error: %s.", strerror (errno));
1590 else if (!eof_is_ok || bytes_read != 0)
1591 sys_error (r, "Unexpected end of file.");
1596 /* Reads BYTE_CNT into BUF.
1597 Aborts upon I/O error or if end-of-file is encountered. */
1599 read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1601 read_bytes_internal (r, false, buf, byte_cnt);
1604 /* Reads BYTE_CNT bytes into BUF.
1605 Returns true if exactly BYTE_CNT bytes are successfully read.
1606 Returns false if an immediate end-of-file is encountered.
1607 Aborts if an I/O error or a partial read occurs. */
1609 try_read_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
1611 return read_bytes_internal (r, true, buf, byte_cnt);
1614 /* Reads a 32-bit signed integer from R and returns its value in
1617 read_int (struct sfm_reader *r)
1620 read_bytes (r, integer, sizeof integer);
1621 return integer_get (r->integer_format, integer, sizeof integer);
1624 /* Reads a 64-bit signed integer from R and returns its value in
1627 read_int64 (struct sfm_reader *r)
1630 read_bytes (r, integer, sizeof integer);
1631 return integer_get (r->integer_format, integer, sizeof integer);
1634 /* Reads a 64-bit floating-point number from R and returns its
1635 value in host format. */
1637 read_float (struct sfm_reader *r)
1640 read_bytes (r, number, sizeof number);
1641 return float_get_double (r->float_format, number);
1644 /* Reads exactly SIZE - 1 bytes into BUFFER
1645 and stores a null byte into BUFFER[SIZE - 1]. */
1647 read_string (struct sfm_reader *r, char *buffer, size_t size)
1650 read_bytes (r, buffer, size - 1);
1651 buffer[size - 1] = '\0';
1654 /* Skips BYTES bytes forward in R. */
1656 skip_bytes (struct sfm_reader *r, size_t bytes)
1661 size_t chunk = MIN (sizeof buffer, bytes);
1662 read_bytes (r, buffer, chunk);
1668 trim_spaces (char *s)
1670 char *end = strchr (s, '\0');
1671 while (end > s && end[-1] == ' ')
1677 print_string (const char *s, size_t len)
1679 if (memchr (s, 0, len) == 0)
1683 for (i = 0; i < len; i++)
1685 unsigned char c = s[i];
1689 else if (c == '\n' || isprint (c))
1692 printf ("\\%02x", c);
1697 hex_dump (0, s, len);