10 #include "u8-mbtouc.h"
18 all_ascii(const uint8_t *p, size_t n)
20 for (size_t i = 0; i < n; i++)
21 if (p[i] < 32 || p[i] > 126)
27 try_find(const char *target, size_t target_len)
29 const uint8_t *pos = (const uint8_t *) memmem (data, n, target, target_len);
30 return pos ? pos - data : 0;
34 find(const char *target, size_t target_len)
36 size_t pos = try_find(target, target_len);
39 fprintf (stderr, "not found\n");
48 #define STR(x) XSTR(x)
49 #define WHERE __FILE__":" STR(__LINE__)
55 memcpy(&x, &data[pos], 4);
64 memcpy(&x, &data[pos], 8);
79 match_u32_assert(uint32_t x, const char *where)
81 unsigned int y = get_u32();
84 fprintf(stderr, "%s: 0x%x: expected i%u, got i%u\n", where, pos - 4, x, y);
88 #define match_u32_assert(x) match_u32_assert(x, WHERE)
93 if (pos < n && data[pos] == b)
103 match_byte_assert(uint8_t b, const char *where)
107 fprintf(stderr, "%s: 0x%x: expected %02x, got %02x\n", where, pos, b, data[pos]);
111 #define match_byte_assert(b) match_byte_assert(b, WHERE)
114 newline(FILE *stream, int pos)
116 fprintf(stream, "\n%08x: ", pos);
120 dump_raw(FILE *stream, int start, int end)
122 for (size_t i = start; i < end; )
129 && i + 4 + data[i] + data[i + 1] * 256 <= end
130 && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
133 fprintf(stream, "\"");
134 fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
135 fputs("\" ", stream);
137 i += 4 + data[i] + data[i + 1] * 256;
139 else if (i + 12 <= end
146 memcpy (&d, &data[i + 4], 8);
147 fprintf (stream, "F40.%d(%.*f)", data[i], data[i], d);
151 else if (i + 12 <= end
158 memcpy (&d, &data[i + 4], 8);
159 fprintf (stream, "PCT40.%d(%.*f)", data[i], data[i], d);
163 else if (i + 4 <= end
164 && (data[i] && data[i] != 88 && data[i] != 0x41)
169 fprintf (stream, "i%d ", data[i]);
174 fprintf(stream, "%02x ", data[i]);
182 static bool __attribute__((unused))
183 all_utf8(const char *p_)
185 const uint8_t *p = (const uint8_t *) p_;
186 size_t len = strlen ((char *) p);
187 for (size_t ofs = 0, mblen; ofs < len; ofs += mblen)
191 mblen = u8_mbtouc (&uc, p + ofs, len - ofs);
192 if ((uc < 32 && uc != '\n') || uc == 127 || uc == 0xfffd)
199 get_string(const char *where)
202 /*data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0*/
203 /*&& all_ascii(&data[pos + 4], data[pos])*/)
205 int len = data[pos] + data[pos + 1] * 256;
206 char *s = malloc(len + 1);
208 memcpy(s, &data[pos + 4], len);
215 fprintf(stderr, "%s: 0x%x: expected string\n", where, pos);
219 #define get_string() get_string(WHERE)
222 dump_nested_string(void)
226 match_byte_assert (0);
227 match_byte_assert (0);
228 int outer_end = pos + get_u32();
229 int inner_end = pos + get_u32();
230 if (pos != inner_end)
233 if (match_byte(0x31))
236 match_byte_assert(0x58);
237 if (pos != inner_end)
239 fprintf(stderr, "inner end discrepancy\n");
243 match_byte_assert(0x58);
244 match_byte_assert(0x58);
245 if (pos != outer_end)
247 fprintf(stderr, "outer end discrepancy\n");
255 dump_optional_value(FILE *stream)
257 if (match_byte (0x31))
263 /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
267 match_u32_assert (0);
271 /* We only have one SPV file for this version (with many
278 if (!match_u32(0) && !match_u32(1) && !match_u32(2) && !match_u32(3) && !match_u32(4) && !match_u32(5) && !match_u32(6) && !match_u32(7) && !match_u32(8) && !match_u32(9))
279 match_u32_assert(10);
285 int outer_end = pos + get_u32();
286 int inner_end = pos + get_u32();
287 if (pos != inner_end)
290 if (match_byte(0x31))
292 /* Appears to be a template string, e.g. '^1 cells (^2) expf < 5. Min exp = ^3...'.
293 Probably doesn't actually appear in output because many examples look unpolished,
294 e.g. 'partial list cases value ^1 shown upper...' */
298 match_byte_assert(0x58);
299 if (pos != inner_end)
301 fprintf(stderr, "inner end discrepancy\n");
306 if (match_byte(0x31))
308 /* Only one example in the corpus. */
313 match_byte_assert(1);
314 get_string(); /* foreground */
315 get_string(); /* background */
316 get_string(); /* font */
318 match_byte_assert(12); /* size? */
321 match_byte_assert(0x58);
322 if (match_byte(0x31))
324 /* Only two SPV files have anything like this, so it's hard to
330 match_byte_assert(1);
331 match_byte_assert(0);
332 if (!match_byte(8) && !match_byte(1))
333 match_byte_assert(2);
334 match_byte_assert(0);
335 match_byte_assert(8);
336 match_byte_assert(0);
337 match_byte_assert(10);
338 match_byte_assert(0);
341 match_byte_assert(0x58);
342 if (pos != outer_end)
344 fprintf(stderr, "outer end discrepancy\n");
348 else if (match_u32 (1))
350 fprintf(stream, "(footnote %d) ", get_u32());
351 dump_nested_string();
353 else if (match_u32 (2))
355 fprintf(stream, "(special 2)");
357 match_byte_assert(2);
358 match_byte_assert(0);
359 if (!match_u32 (2) && !match_u32(1))
361 dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
366 fprintf(stream, "(special 3)");
367 match_byte_assert(0);
368 match_byte_assert(0);
369 match_byte_assert(1);
370 match_byte_assert(0);
372 dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
376 match_byte_assert (0x58);
380 format_to_string (int type)
386 case 2: return "AHEX";
387 case 3: return "COMMA";
388 case 4: return "DOLLAR";
389 case 5: case 40: return "F";
391 case 7: return "PIBHEX";
393 case 9: return "PIB";
394 case 10: return "PK";
395 case 11: return "RB";
396 case 12: return "RBHEX";
400 case 20: return "DATE";
401 case 21: return "TIME";
402 case 22: return "DATETIME";
403 case 23: return "ADATE";
404 case 24: return "JDATE";
405 case 25: return "DTIME";
406 case 26: return "WKDAY";
407 case 27: return "MONTH";
408 case 28: return "MOYR";
409 case 29: return "QYR";
410 case 30: return "WKYR";
411 case 31: return "PCT";
412 case 32: return "DOT";
413 case 33: return "CCA";
414 case 34: return "CCB";
415 case 35: return "CCC";
416 case 36: return "CCD";
417 case 37: return "CCE";
418 case 38: return "EDATE";
419 case 39: return "SDATE";
422 sprintf(tmp, "<%d>", type);
428 dump_value(FILE *stream, int level)
435 for (int i = 0; i <= level; i++)
436 fprintf (stream, " ");
440 char *text = get_string();
441 dump_optional_value(stream);
442 char *identifier = get_string();
443 char *text_eng = get_string();
444 fprintf (stream, "<string c=\"%s\"", text_eng);
446 fprintf (stream, " identifier=\"%s\"", identifier);
447 if (strcmp(text_eng, text))
448 fprintf (stream, " local=\"%s\"", text);
449 fprintf (stream, "/>\n");
451 match_byte_assert(1);
453 else if (match_byte (5))
455 dump_optional_value(stream);
456 char *name = get_string ();
457 char *label = get_string ();
458 fprintf (stream, "<variable name=\"%s\"", name);
460 fprintf (stream, " label=\"%s\"", label);
461 fprintf (stream, "/>\n");
462 if (!match_byte(1) && !match_byte(2))
463 match_byte_assert(3);
465 else if (match_byte (2))
471 dump_optional_value (stream);
473 value = get_double ();
475 vallab = get_string ();
476 fprintf (stream, "<numeric-datum value=\"%.*g\" format=\"%s%d.%d\"",
477 DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
479 fprintf (stream, " variable=\"%s\"", var);
481 fprintf (stream, " label=\"%s\"/>\n", vallab);
482 fprintf (stream, "/>\n");
483 if (!match_byte (1) && !match_byte(2))
484 match_byte_assert (3);
486 else if (match_byte (4))
489 char *var, *vallab, *value;
491 match_byte_assert (0x58);
493 vallab = get_string ();
495 if (!match_byte(1) && !match_byte(2))
496 match_byte_assert (3);
497 value = get_string ();
498 fprintf (stream, "<string-datum value=\"%s\" format=\"%s%d.%d\"",
499 value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
501 fprintf (stream, " variable=\"%s\"", var);
503 fprintf (stream, " label=\"%s\"/>\n", vallab);
504 fprintf (stream, "/>\n");
506 else if (match_byte (1))
511 dump_optional_value(stream);
513 value = get_double ();
514 fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
515 DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
519 dump_optional_value(stream);
521 char *base = get_string();
523 fprintf (stream, "<template format=\"%s\">\n", base);
524 for (int i = 0; i < x; i++)
531 for (int j = 0; j <= level + 1; j++)
532 fprintf (stream, " ");
533 fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
534 for (int j = 0; j < y; j++)
535 dump_value (stream, level + 2);
536 for (int j = 0; j <= level + 1; j++)
537 fprintf (stream, " ");
538 fprintf (stream, "</substitution>\n");
540 for (int j = 0; j <= level; j++)
541 fprintf (stream, " ");
542 fprintf (stream, "</template>\n");
547 compare_int(const void *a_, const void *b_)
551 return *a < *b ? -1 : *a > *b;
555 check_permutation(int *a, int n, const char *name)
558 memcpy(b, a, n * sizeof *a);
559 qsort(b, n, sizeof *b, compare_int);
560 for (int i = 0; i < n; i++)
563 fprintf(stderr, "bad %s permutation:", name);
564 for (int i = 0; i < n; i++)
565 fprintf(stderr, " %d", a[i]);
572 dump_category(int level, int *indexes, int *n_indexes, int max_indexes)
574 for (int i = 0; i <= level; i++)
575 fprintf (stdout, " ");
576 printf ("<category>\n");
577 dump_value (stdout, level + 1);
592 if (!match_u32 (0) && !match_u32 (1))
593 match_u32_assert (2);
596 match_u32_assert (2);
597 int len = pos - start;
599 int indx = get_u32();
600 int n_categories = get_u32();
603 fprintf (stderr, "category %d %d %d %d %d:", d0, c, d1, d2, len);
604 if (n_categories != 0)
606 fprintf(stderr, "index not -1 but subcategories\n");
609 if (*n_indexes >= max_indexes)
611 fprintf(stderr, "too many categories (increase max_indexes)\n");
614 indexes[(*n_indexes)++] = indx;
617 fprintf (stderr, "group %d %d %d %d %d:", d0, c, d1, d2, len);
618 for (int i = 0; i < len; i++)
619 fprintf (stderr, " %02x", data[start + i]);
622 if (n_categories == 0)
624 for (int i = 0; i <= level + 1; i++)
625 fprintf (stdout, " ");
626 fprintf (stdout, "<category-index>%d</category-index>\n", indx);
628 for (int i = 0; i < n_categories; i++)
629 dump_category (level + 1, indexes, n_indexes, max_indexes);
630 for (int i = 0; i <= level; i++)
631 fprintf (stdout, " ");
632 printf ("</category>\n");
640 printf ("<dimension index=\"%d\">\n", indx);
641 dump_value (stdout, 0);
643 /* This byte is usually 0 but many other values have been spotted. */
646 if (!match_byte(0) && !match_byte(1))
647 match_byte_assert(2);
651 match_byte_assert(1);
653 match_byte_assert(1);
654 match_byte_assert(1);
655 if (!match_u32(UINT32_MAX))
656 match_u32_assert(indx);
657 n_categories = get_u32();
661 for (int i = 0; i < n_categories; i++)
662 dump_category (0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
663 check_permutation(indexes, n_indexes, "categories");
665 fprintf (stdout, "</dimension>\n");
673 for (int i = 0; i < n_dims; i++)
680 /* The first three numbers add to the number of dimensions. */
683 match_u32_assert(n_dims - t);
685 /* The next n_dims numbers are a permutation of the dimension numbers. */
687 for (int i = 0; i < n_dims; i++)
689 check_permutation(a, n_dims, "dimensions");
693 for (int i = 0; i < x; i++)
695 printf (" <datum index=\"%d\">\n", get_u32());
699 dump_value(stdout, 1);
700 fprintf (stdout, " </datum>\n");
702 printf ("</data>\n");
708 printf ("<title-local>\n");
709 dump_value(stdout, 0);
711 printf ("</title-local>\n");
713 printf ("<subtype>\n");
714 dump_value(stdout, 0);
716 printf ("</subtype>\n");
718 match_byte_assert(0x31);
720 printf ("<title-c>\n");
721 dump_value(stdout, 0);
723 printf ("</title-c>\n");
726 match_byte_assert(0x58);
727 if (match_byte(0x31))
729 printf ("<caption>\n");
730 dump_value(stdout, 0);
731 printf ("</caption>\n");
734 match_byte_assert(0x58);
736 int n_footnotes = get_u32();
737 for (int i = 0; i < n_footnotes; i++)
739 printf ("<footnote index=\"%d\">\n", i);
740 dump_value(stdout, 0);
741 /* Custom footnote marker string. */
742 if (match_byte (0x31))
743 dump_value(stdout, 0);
745 match_byte_assert (0x58);
747 printf ("</footnote>\n");
755 for (int i = 1; i <= 8; i++)
757 printf ("<style index=\"%d\"", i);
758 match_byte_assert(i);
759 match_byte_assert(0x31);
760 printf(" font=\"%s\"", get_string());
761 match_byte_assert(0);
762 match_byte_assert(0);
763 if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10) && !match_byte(0x70))
764 match_byte_assert(0x50);
765 match_byte_assert(0x41);
766 if (!match_u32(0) && !match_u32(1))
768 match_byte_assert(0);
770 /* OK, this seems really unlikely to be totally correct, but it matches my corpus... */
771 if (!match_u32(0) && !match_u32(2))
774 match_u32_assert(0xfaad);
779 if (!match_u32(0) && !match_u32(1) && !match_u32(2))
781 printf (" fgcolor=\"%s\"", get_string());
782 printf (" bgcolor=\"%s\"", get_string());
785 match_byte_assert(0);
793 if (!match_u32(10) && !match_u32(11) && !match_u32(5))
801 if (!match_u32(-1) && !match_u32(8))
802 match_u32_assert(24);
803 if (!match_u32(-1) && !match_u32(2))
807 /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
814 match_u32_assert(240);
817 match_u32_assert(18);
822 else if (match_u32(142))
824 else if (match_u32(143))
826 else if (match_u32(150))
830 match_u32_assert(16);
834 int count = get_u32();
837 const char *encoding = get_string();
838 printf ("<encoding>%s</encoding>\n", encoding);
841 match_u32_assert(UINT32_MAX);
843 match_byte_assert(1);
844 match_byte_assert(0);
846 match_byte_assert(1);
849 if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
850 match_byte_assert(0x9a);
851 match_byte_assert(7);
852 match_byte_assert(0);
853 match_byte_assert(0);
856 match_u32_assert(UINT32_MAX);
858 int decimal = data[pos];
859 int grouping = data[pos + 1];
862 if (!match_byte(',') && !match_byte('\''))
863 match_byte_assert(' ');
867 match_byte_assert(',');
868 if (!match_byte('.') && !match_byte(' '))
869 match_byte_assert(0);
871 printf("<format decimal=\"%c\" grouping=\"", decimal);
877 for (int i = 0; i < 5; i++)
878 printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
882 int skip = get_u32();
887 main(int argc, char *argv[])
892 if (isatty(STDIN_FILENO))
894 fprintf(stderr, "redirect stdin from a .bin file\n");
897 if (fstat(STDIN_FILENO, &s))
909 if (read(STDIN_FILENO, data, n) != n)
917 fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
921 if (!strcmp(argv[1], "title0"))
924 if (match_byte (0x03)
925 || (match_byte (0x05) && match_byte (0x58)))
926 printf ("%s\n", get_string());
928 printf ("<unknown>\n");
931 else if (!strcmp(argv[1], "title"))
937 else if (!strcmp(argv[1], "titleraw"))
939 const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
941 n = find(fonts, sizeof fonts - 1);
943 else if (!strcmp(argv[1], "fonts"))
945 const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
946 const char styles[] = "\xf0\0\0\0";
947 start = find(fonts, sizeof fonts - 1);
948 n = find(styles, sizeof styles - 1);
950 else if (!strcmp(argv[1], "styles"))
952 const char styles[] = "\xf0\0\0\0";
953 const char dimensions[] = "-,,,.\0";
954 start = find(styles, sizeof styles - 1);
955 n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
957 else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
960 match_byte_assert(1);
961 match_byte_assert(0);
963 /* This might be a version number of some kind, because value 1 seems
964 to only appear in an SPV file that also required its own weird
965 special cases in dump_optional_value(). */
971 match_byte_assert(1);
973 match_byte_assert(1);
976 match_byte_assert(0);
977 match_byte_assert(0);
979 match_byte_assert(1);
983 match_byte_assert(0);
984 match_byte_assert(0);
985 match_byte_assert(0);
990 match_byte_assert(1);
991 match_byte_assert(0);
992 match_byte_assert(0);
997 match_byte_assert(1);
998 match_byte_assert(0);
999 match_byte_assert(0);
1004 match_byte_assert(1);
1005 match_byte_assert(0);
1006 match_byte_assert(0);
1011 match_byte_assert(0);
1012 match_byte_assert(0);
1016 This is the tableId, e.g. -4154297861994971133 would be 0xdca00003.
1017 We don't have enough context to validate it. */
1030 fprintf (stderr, "%x / %x\n", pos, n);
1035 else if (!strcmp(argv[1], "raw"))
1039 dump_raw(stdout, start, n);
1043 fprintf (stderr, "unknown section %s\n", argv[1]);