14 all_ascii(const uint8_t *p, size_t n)
16 for (size_t i = 0; i < n; i++)
17 if (p[i] < 32 || p[i] > 126)
23 try_find(const char *target, size_t target_len)
25 const uint8_t *pos = (const uint8_t *) memmem (data, n, target, target_len);
26 return pos ? pos - data : 0;
30 find(const char *target, size_t target_len)
32 size_t pos = try_find(target, target_len);
35 fprintf (stderr, "not found\n");
44 #define STR(x) XSTR(x)
45 #define WHERE __FILE__":" STR(__LINE__)
51 memcpy(&x, &data[pos], 4);
60 memcpy(&x, &data[pos], 8);
75 match_u32_assert(uint32_t x, const char *where)
77 unsigned int y = get_u32();
80 fprintf(stderr, "%s: 0x%x: expected i%u, got i%u\n", where, pos - 4, x, y);
84 #define match_u32_assert(x) match_u32_assert(x, WHERE)
89 if (pos < n && data[pos] == b)
99 match_byte_assert(uint8_t b, const char *where)
103 fprintf(stderr, "%s: 0x%x: expected %02x, got %02x\n", where, pos, b, data[pos]);
107 #define match_byte_assert(b) match_byte_assert(b, WHERE)
110 newline(FILE *stream, int pos)
112 fprintf(stream, "\n%08x: ", pos);
116 dump_raw(FILE *stream, int start, int end)
118 for (size_t i = start; i < end; )
125 && i + 4 + data[i] + data[i + 1] * 256 <= end
126 && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
129 fprintf(stream, "\"");
130 fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
131 fputs("\" ", stream);
133 i += 4 + data[i] + data[i + 1] * 256;
135 else if (i + 12 <= end
142 memcpy (&d, &data[i + 4], 8);
143 fprintf (stream, "F40.%d(%.*f)", data[i], data[i], d);
147 else if (i + 12 <= end
154 memcpy (&d, &data[i + 4], 8);
155 fprintf (stream, "PCT40.%d(%.*f)", data[i], data[i], d);
159 else if (i + 4 <= end
160 && (data[i] && data[i] != 88 && data[i] != 0x41)
165 fprintf (stream, "i%d ", data[i]);
170 fprintf(stream, "%02x ", data[i]);
179 get_string(const char *where)
182 /*data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0*/
183 /*&& all_ascii(&data[pos + 4], data[pos])*/)
185 int len = data[pos] + data[pos + 1] * 256;
186 char *s = malloc(len + 1);
188 memcpy(s, &data[pos + 4], len);
195 fprintf(stderr, "%s: 0x%x: expected string\n", where, pos);
199 #define get_string() get_string(WHERE)
202 dump_nested_string(void)
206 match_byte_assert (0);
207 match_byte_assert (0);
208 int outer_end = pos + get_u32();
209 int inner_end = pos + get_u32();
210 if (pos != inner_end)
213 if (match_byte(0x31))
216 match_byte_assert(0x58);
217 if (pos != inner_end)
219 fprintf(stderr, "inner end discrepancy\n");
223 match_byte_assert(0x58);
224 match_byte_assert(0x58);
225 if (pos != outer_end)
227 fprintf(stderr, "outer end discrepancy\n");
235 dump_value_31(FILE *stream)
237 if (match_byte (0x31))
243 /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
247 match_u32_assert (0);
249 int outer_end = pos + get_u32();
250 int inner_end = pos + get_u32();
252 if (match_byte(0x31))
254 /* Appears to be a template string, e.g. '^1 cells (^2) expf < 5. Min exp = ^3...'.
255 Probably doesn't actually appear in output because many examples look unpolished,
256 e.g. 'partial list cases value ^1 shown upper...' */
260 match_byte_assert(0x58);
261 if (pos != inner_end)
263 fprintf(stderr, "inner end discrepancy\n");
267 if (match_byte(0x31))
269 /* Only one example in the corpus. */
274 match_byte_assert(1);
275 get_string(); /* foreground */
276 get_string(); /* background */
277 get_string(); /* font */
279 match_byte_assert(12); /* size? */
282 match_byte_assert(0x58);
283 match_byte_assert(0x58);
284 if (pos != outer_end)
286 fprintf(stderr, "outer end discrepancy\n");
290 else if (match_u32 (1))
292 fprintf(stream, "(footnote %d) ", get_u32());
293 dump_nested_string();
295 else if (match_u32 (2))
297 fprintf(stream, "(special 2)");
298 match_byte_assert(0);
299 match_byte_assert(0);
302 dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
307 fprintf(stream, "(special 3)");
308 match_byte_assert(0);
309 match_byte_assert(0);
310 match_byte_assert(1);
311 match_byte_assert(0);
313 dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
317 match_byte_assert (0x58);
321 format_to_string (int type)
327 case 2: return "AHEX";
328 case 3: return "COMMA";
329 case 4: return "DOLLAR";
330 case 5: case 40: return "F";
332 case 7: return "PIBHEX";
334 case 9: return "PIB";
335 case 10: return "PK";
336 case 11: return "RB";
337 case 12: return "RBHEX";
341 case 20: return "DATE";
342 case 21: return "TIME";
343 case 22: return "DATETIME";
344 case 23: return "ADATE";
345 case 24: return "JDATE";
346 case 25: return "DTIME";
347 case 26: return "WKDAY";
348 case 27: return "MONTH";
349 case 28: return "MOYR";
350 case 29: return "QYR";
351 case 30: return "WKYR";
352 case 31: return "PCT";
353 case 32: return "DOT";
354 case 33: return "CCA";
355 case 34: return "CCB";
356 case 35: return "CCC";
357 case 36: return "CCD";
358 case 37: return "CCE";
359 case 38: return "EDATE";
360 case 39: return "SDATE";
363 sprintf(tmp, "<%d>", type);
369 dump_value(FILE *stream, int level, bool match1)
376 for (int i = 0; i <= level; i++)
377 fprintf (stream, " ");
381 char *text = get_string();
382 dump_value_31(stream);
383 char *identifier = get_string();
384 char *text_eng = get_string();
385 fprintf (stream, "<string c=\"%s\"", text_eng);
387 fprintf (stream, " identifier=\"%s\"", identifier);
388 if (strcmp(text_eng, text))
389 fprintf (stream, " local=\"%s\"", text);
390 fprintf (stream, "/>\n");
392 match_byte_assert(1);
396 else if (match_byte (5))
398 dump_value_31(stream);
399 char *name = get_string ();
400 char *label = get_string ();
401 fprintf (stream, "<variable name=\"%s\"", name);
403 fprintf (stream, " label=\"%s\"", label);
404 fprintf (stream, "/>\n");
405 if (!match_byte(1) && !match_byte(2))
406 match_byte_assert(3);
408 else if (match_byte (2))
414 match_byte_assert (0x58);
416 value = get_double ();
418 vallab = get_string ();
419 fprintf (stream, "<numeric-datum value=\"%.*g\" format=\"%s%d.%d\"",
420 DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
422 fprintf (stream, " variable=\"%s\"", var);
424 fprintf (stream, " label=\"%s\"/>\n", vallab);
425 fprintf (stream, "/>\n");
426 if (!match_byte (1) && !match_byte(2))
427 match_byte_assert (3);
429 else if (match_byte (4))
432 char *var, *vallab, *value;
434 match_byte_assert (0x58);
436 vallab = get_string ();
438 if (!match_byte(1) && !match_byte(2))
439 match_byte_assert (3);
440 value = get_string ();
441 fprintf (stream, "<string-datum value=\"%s\" format=\"%s%d.%d\"",
442 value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
444 fprintf (stream, " variable=\"%s\"", var);
446 fprintf (stream, " label=\"%s\"/>\n", vallab);
447 fprintf (stream, "/>\n");
449 else if (match_byte (1))
454 dump_value_31(stream);
456 value = get_double ();
457 fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
458 DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
464 dump_value_31(stream);
466 char *base = get_string();
468 fprintf (stream, "<template format=\"%s\">\n", base);
469 for (int i = 0; i < x; i++)
476 for (int j = 0; j <= level + 1; j++)
477 fprintf (stream, " ");
478 fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
479 for (int j = 0; j < y; j++)
480 dump_value (stream, level + 2, false);
481 for (int j = 0; j <= level + 1; j++)
482 fprintf (stream, " ");
483 fprintf (stream, "</substitution>\n");
485 for (int j = 0; j <= level; j++)
486 fprintf (stream, " ");
487 fprintf (stream, "</template>\n");
492 compare_int(const void *a_, const void *b_)
496 return *a < *b ? -1 : *a > *b;
500 check_permutation(int *a, int n, const char *name)
503 memcpy(b, a, n * sizeof *a);
504 qsort(b, n, sizeof *b, compare_int);
505 for (int i = 0; i < n; i++)
508 fprintf(stderr, "bad %s permutation:", name);
509 for (int i = 0; i < n; i++)
510 fprintf(stderr, " %d", a[i]);
517 dump_category(int level, int *indexes, int *n_indexes)
519 for (int i = 0; i <= level; i++)
520 fprintf (stdout, " ");
521 printf ("<category>\n");
522 dump_value (stdout, level + 1, true);
529 else if (match_byte (1))
533 match_u32_assert (1);
536 else if (!match_u32(2))
537 match_u32_assert (0);
539 int indx = get_u32();
540 int n_categories = get_u32();
543 if (n_categories != 0)
545 fprintf(stderr, "index not -1 but subcategories\n");
548 indexes[(*n_indexes)++] = indx;
550 if (n_categories == 0)
552 for (int i = 0; i <= level + 1; i++)
553 fprintf (stdout, " ");
554 fprintf (stdout, "<category-index>%d</category-index>\n", indx);
556 for (int i = 0; i < n_categories; i++)
557 dump_category (level + 1, indexes, n_indexes);
558 for (int i = 0; i <= level; i++)
559 fprintf (stdout, " ");
560 printf ("</category>\n");
568 printf ("<dimension index=\"%d\">\n", indx);
569 dump_value (stdout, 0, false);
571 /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
574 if (!match_byte(0) && !match_byte(1))
575 match_byte_assert(2);
579 match_byte_assert(1);
581 match_byte_assert(1);
582 match_byte_assert(1);
583 match_u32_assert(indx);
584 n_categories = get_u32();
588 for (int i = 0; i < n_categories; i++)
589 dump_category (0, indexes, &n_indexes);
590 check_permutation(indexes, n_indexes, "categories");
592 fprintf (stdout, "</dimension>\n");
600 for (int i = 0; i < n_dims; i++)
607 /* The first three numbers add to the number of dimensions. */
610 match_u32_assert(n_dims - t);
612 /* The next n_dims numbers are a permutation of the dimension numbers. */
614 for (int i = 0; i < n_dims; i++)
616 check_permutation(a, n_dims, "dimensions");
620 for (int i = 0; i < x; i++)
622 printf (" <datum index=\"%d\">\n", get_u32());
624 dump_value(stdout, 1, false);
625 fprintf (stdout, " </datum>\n");
627 printf ("</data>\n");
634 printf ("<title-local>\n");
635 dump_value(stdout, 0, true);
636 printf ("</title-local>\n");
638 printf ("<subtype>\n");
639 dump_value(stdout, 0, true);
640 printf ("</subtype>\n");
642 match_byte_assert(0x31);
644 printf ("<title-c>\n");
645 dump_value(stdout, 0, true);
646 printf ("</title-c>\n");
649 match_byte_assert(0x58);
650 if (match_byte(0x31))
652 printf ("<caption>\n");
653 dump_value(stdout, 0, false);
654 printf ("</caption>\n");
657 match_byte_assert(0x58);
660 int n_footnotes = get_u32();
661 for (int i = 0; i < n_footnotes; i++)
663 printf ("<footnote index=\"%d\">\n", i);
664 dump_value(stdout, 0, false);
665 if (match_byte (0x31))
667 /* Custom footnote marker string. */
668 match_byte_assert(3);
670 match_byte_assert(0x58);
675 match_byte_assert (0x58);
676 printf("(%d)\n", get_u32());
677 printf ("</footnote>\n");
685 for (int i = 1; i <= 8; i++)
687 printf ("<style index=\"%d\"", i);
688 match_byte_assert(i);
689 match_byte_assert(0x31);
690 printf(" font=\"%s\"", get_string());
691 match_byte_assert(0);
692 match_byte_assert(0);
693 if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10))
694 match_byte_assert(0x50);
695 if (!match_byte(0x41))
696 match_byte_assert(0x51);
699 match_byte_assert(0);
701 /* OK, this seems really unlikely to be totally correct, but it matches my corpus... */
702 if (!match_u32(0) && !match_u32(2))
703 match_u32_assert(0xfaad);
705 if (!match_u32(0) && !match_u32(1) && !match_u32(2))
707 printf (" fgcolor=\"%s\"", get_string());
708 printf (" bgcolor=\"%s\"", get_string());
711 match_byte_assert(0);
713 /* These seem unlikely to be correct too. */
718 match_u32_assert(11);
724 if (!match_u32(-1) && !match_u32(8))
725 match_u32_assert(24);
726 if (!match_u32(-1) && !match_u32(2))
730 /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
736 match_u32_assert(240);
739 match_u32_assert(18);
746 match_u32_assert(142);
750 int count = get_u32();
753 printf ("<encoding>%s</encoding>\n", get_string ());
756 match_u32_assert(UINT32_MAX);
758 match_byte_assert(1);
759 match_byte_assert(0);
761 match_byte_assert(1);
762 if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
763 match_byte_assert(0x9a);
764 match_byte_assert(7);
765 match_byte_assert(0);
766 match_byte_assert(0);
768 match_byte_assert(',');
771 match_byte_assert(',');
772 if (!match_byte('.'))
773 match_byte_assert(' ');
776 for (int i = 0; i < 5; i++)
782 main(int argc, char *argv[])
787 if (isatty(STDIN_FILENO))
789 fprintf(stderr, "redirect stdin from a .bin file\n");
792 if (fstat(STDIN_FILENO, &s))
804 if (read(STDIN_FILENO, data, n) != n)
812 if (!strcmp(argv[1], "title0"))
815 if (match_byte (0x03)
816 || (match_byte (0x05) && match_byte (0x58)))
817 printf ("%s\n", get_string());
819 printf ("<unknown>\n");
822 else if (!strcmp(argv[1], "title"))
827 else if (!strcmp(argv[1], "titleraw"))
829 const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
831 n = find(fonts, sizeof fonts - 1);
833 else if (!strcmp(argv[1], "fonts"))
835 const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
836 const char styles[] = "\xf0\0\0\0";
837 start = find(fonts, sizeof fonts - 1);
838 n = find(styles, sizeof styles - 1);
840 else if (!strcmp(argv[1], "styles"))
842 const char styles[] = "\xf0\0\0\0";
843 const char dimensions[] = "-,,,.\0";
844 start = find(styles, sizeof styles - 1);
845 n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
847 else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
850 match_byte_assert(1);
851 match_byte_assert(0);
853 match_byte_assert(1);
855 match_byte_assert(1);
856 match_byte_assert(0);
857 match_byte_assert(0);
859 match_byte_assert(1);
861 match_byte_assert(0);
862 match_byte_assert(0);
863 match_byte_assert(0);
871 fprintf (stderr, "%x / %x\n", pos, n);
878 fprintf (stderr, "unknown section %s\n", argv[1]);
885 dump_raw(stdout, start, n);