return x;
}
+static unsigned int
+get_u16(void)
+{
+ uint16_t x;
+ memcpy(&x, &data[pos], 2);
+ pos += 2;
+ return x;
+}
+
static double
get_double(void)
{
return pos + len;
}
+static void __attribute__((unused))
+hex_dump(int ofs, int n)
+{
+ for (int i = 0; i < n; i++)
+ {
+ int c = data[ofs + i];
+#if 1
+ if (i && !(i % 16))
+ fprintf(stderr, "-");
+ else
+ fprintf(stderr, " ");
+#endif
+ fprintf(stderr, "%02x", c);
+ //fprintf(stderr, "%c", c >= 32 && c < 127 ? c : '.');
+ }
+ fprintf(stderr, "\n");
+}
+
static char *
dump_counted_string(void)
{
int inner_end = get_end();
if (pos != inner_end)
{
- match_u32_assert(0);
+ if (match_u32(5))
+ {
+ match_u32_assert(0);
+ match_byte_assert(0x58);
+ }
+ else
+ match_u32_assert(0);
if (match_byte(0x31))
s = get_string();
else
return s;
}
+static void
+dump_style(FILE *stream)
+{
+ match_byte(1);
+ match_byte(0);
+ match_byte(0);
+ match_byte(0);
+ match_byte_assert(1);
+ char *fg = get_string(); /* foreground */
+ char *bg = get_string(); /* background */
+ char *font = get_string(); /* font */
+ int size = data[pos];
+ if (!match_byte(14))
+ match_byte_assert(12); /* size? */
+ fprintf(stream, " fgcolor=\"%s\" bgcolor=\"%s\" font=\"%s\" size=\"%d\"",
+ fg, bg, font, size);
+}
+
static char *
-dump_nested_string(void)
+dump_nested_string(FILE *stream)
{
char *s = NULL;
match_byte_assert (0);
int outer_end = get_end();
s = dump_counted_string();
- match_byte_assert(0x58);
+ if (s)
+ fprintf(stream, " \"%s\"", s);
+ if (match_byte(0x31))
+ dump_style(stream);
+ else
+ match_byte_assert(0x58);
match_byte_assert(0x58);
if (pos != outer_end)
{
}
static void
-dump_optional_value(FILE *stream)
+dump_value_modifier(FILE *stream)
{
if (match_byte (0x31))
{
if (match_u32 (0))
{
+ fprintf(stream, "<special0");
if (match_u32 (1))
{
/* Corpus frequencies:
124 "a"
12 "b"
8 "a, b"
+
+ The given text is appended to the cell in a subscript font.
*/
- get_string();
+ fprintf(stream, " subscript=\"%s\"", get_string());
}
else
match_u32_assert (0);
match_u32_assert(10);
match_byte(0);
match_byte(0);
+ fprintf(stream, "/>\n");
return;
}
/* This counted-string appears to be a template string,
e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
- dump_counted_string();
+ char *template = dump_counted_string();
+ if (template)
+ fprintf(stream, " template=\"%s\"", template);
if (match_byte(0x31))
- {
- /* Only one example in the corpus. */
- match_byte(1);
- match_byte(0);
- match_byte(0);
- match_byte(0);
- match_byte_assert(1);
- get_string(); /* foreground */
- get_string(); /* background */
- get_string(); /* font */
- if (!match_byte(14))
- match_byte_assert(12); /* size? */
- }
+ dump_style(stream);
else
match_byte_assert(0x58);
if (match_byte(0x31))
fprintf(stderr, "outer end discrepancy\n");
exit(1);
}
- }
- else if (match_u32 (1))
- {
- fprintf(stream, "(footnote %d) ", get_u32());
- dump_nested_string();
- }
- else if (match_u32 (2))
- {
- fprintf(stream, "(special 2)");
- if (!match_byte(0))
- match_byte_assert(2);
- match_byte_assert(0);
- if (!match_u32 (2) && !match_u32(1))
- match_u32_assert(3);
- dump_nested_string();
+ fprintf(stream, "/>\n");
}
else
{
- match_u32_assert(3);
- fprintf(stream, "(special 3)");
- match_byte_assert(0);
+ int count = get_u32();
+ fprintf(stream, "<footnote-ref indexes=\"");
+ for (int i = 0; i < count; i++)
+ {
+ if (i)
+ putc(' ', stream);
+ fprintf(stream, "%d", get_u16());
+ }
+ putc('"', stream);
match_byte_assert(0);
- match_byte_assert(1);
match_byte_assert(0);
- match_u32_assert(2);
- dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+ dump_nested_string(stream);
+ fprintf(stream, "/>\n");
}
}
else
unsigned int format;
double value;
- dump_optional_value(stream);
+ dump_value_modifier(stream);
format = get_u32 ();
value = get_double ();
fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
char *var, *vallab;
double value;
- dump_optional_value (stream);
+ dump_value_modifier (stream);
format = get_u32 ();
value = get_double ();
var = get_string ();
if (var[0])
fprintf (stream, " variable=\"%s\"", var);
if (vallab[0])
- fprintf (stream, " label=\"%s\"/>\n", vallab);
+ fprintf (stream, " label=\"%s\"", vallab);
fprintf (stream, "/>\n");
if (!match_byte (1) && !match_byte(2))
match_byte_assert (3);
else if (match_byte (3))
{
char *text = get_string();
- dump_optional_value(stream);
+ dump_value_modifier(stream);
char *identifier = get_string();
char *text_eng = get_string();
fprintf (stream, "<string c=\"%s\"", text_eng);
unsigned int format;
char *var, *vallab, *value;
- dump_optional_value(stream);
+ dump_value_modifier(stream);
format = get_u32 ();
vallab = get_string ();
var = get_string ();
}
else if (match_byte (5))
{
- dump_optional_value(stream);
+ dump_value_modifier(stream);
char *name = get_string ();
char *label = get_string ();
fprintf (stream, "<variable name=\"%s\"", name);
}
else
{
- dump_optional_value(stream);
+ dump_value_modifier(stream);
char *base = get_string();
int x = get_u32();
}
static void
-dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_indexes)
+dump_category(FILE *stream, int level, int **indexes, int *allocated_indexes,
+ int *n_indexes)
{
for (int i = 0; i <= level; i++)
fprintf (stream, " ");
fprintf(stderr, "index not -1 but subcategories\n");
exit(1);
}
- if (*n_indexes >= max_indexes)
+ if (*n_indexes >= *allocated_indexes)
{
- fprintf(stderr, "too many categories (increase max_indexes)\n");
- exit(1);
+ *allocated_indexes = *allocated_indexes ? 2 * *allocated_indexes : 16;
+ *indexes = realloc(*indexes, *allocated_indexes * sizeof **indexes);
}
- indexes[(*n_indexes)++] = indx;
+ (*indexes)[(*n_indexes)++] = indx;
}
int expected_unindexed = indx == -1;
fprintf (stream, "<category-index>%d</category-index>\n", indx);
}
for (int i = 0; i < n_categories; i++)
- dump_category (stream, level + 1, indexes, n_indexes, max_indexes);
+ dump_category (stream, level + 1, indexes, allocated_indexes, n_indexes);
for (int i = 0; i <= level; i++)
fprintf (stream, " ");
printf ("</category>\n");
match_u32_assert(indx);
n_categories = get_u32();
- int indexes[2048];
+ int *indexes = NULL;
int n_indexes = 0;
+ int allocated_indexes = 0;
for (int i = 0; i < n_categories; i++)
- dump_category (stdout, 0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
+ dump_category (stdout, 0, &indexes, &allocated_indexes, &n_indexes);
check_permutation(indexes, n_indexes, "categories");
fprintf (stdout, "</dimension>\n");
match_byte(1);
printf ("</title-c>\n");
- match_byte(0);
- match_byte_assert(0x58);
+ if (match_byte(0x31))
+ {
+ printf ("<user-caption>\n");
+ dump_value(stdout, 0);
+ printf ("</user-caption>\n");
+ }
+ else
+ match_byte_assert(0x58);
if (match_byte(0x31))
{
printf ("<caption>\n");
}
}
-static void __attribute__((unused))
-hex_dump(int ofs, int n)
-{
- for (int i = 0; i < n; i++)
- {
- int c = data[ofs + i];
-#if 1
- if (i && !(i % 16))
- fprintf(stderr, "-");
- else
- fprintf(stderr, " ");
-#endif
- fprintf(stderr, "%02x", c);
- //fprintf(stderr, "%c", c >= 32 && c < 127 ? c : '.');
- }
- fprintf(stderr, "\n");
-}
-
static void
dump_fonts(void)
{
match_byte_assert(0);
/* OK, this seems really unlikely to be totally correct, but it matches my corpus... */
- if (!match_u32(0) && !match_u32(2))
+ if (!match_u32(0) && !match_u32(2) && !match_u32(4))
{
if (i == 7)
match_u32_assert(0xfaad);
match_u32_assert(3);
printf (" fgcolor=\"%s\"", get_string());
printf (" bgcolor=\"%s\"", get_string());
+ if (!match_byte(0))
+ match_byte_assert(1);
match_u32_assert(0);
- match_u32_assert(0);
- match_byte_assert(0);
+ char *othercolor = get_string();
+ if (othercolor[0])
+ printf(" othercolor=\"%s\"", othercolor);
if (version > 1)
{
if (i != 3)
{
- if (!match_u32(8))
- match_u32_assert(5);
- if (!match_u32(10) && !match_u32(11) && !match_u32(5))
- match_u32_assert(9);
- if (!match_u32(0))
- match_u32_assert(1);
+ pos += 12;
}
else
{
match_u32_assert(240);
pos += 240;
- match_u32_assert(18);
- pos += 18;
-
- if (match_u32(117))
- pos += 117;
- else if (match_u32(142))
- pos += 142;
- else if (match_u32(143))
- pos += 143;
- else if (match_u32(150))
- pos += 150;
- else
+ int skip = get_u32();
+ assert(skip == 18 || skip == 25);
+ pos += skip;
+
+ int x3 = get_u32();
+ if (version == 3)
{
- match_u32_assert(16);
- pos += 16;
+ assert(x3 >= 117);
+ int len = data[pos + 0x34];
+ if (len)
+ printf("<tablelook>%.*s</tablelook>\n", len, &data[pos + 0x35]);
}
+ pos += x3;
int count = get_u32();
pos += 4 * count;
const char *locale = get_string();
printf ("<locale>%s</locale>\n", locale);
- if (!match_u32(0))
- match_u32_assert(UINT32_MAX);
+ get_u32(); /* Seen: 0, UINT32_MAX, 2, 3, 4, 5, 6, 8, 9, 21, 24. */
if (!match_byte(0))
match_byte_assert(1);
match_byte_assert(0);
match_byte_assert(1);
if (version > 1)
{
- if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
- match_byte_assert(0x9a);
+ if (!match_byte(0x97) && !match_byte(0x98)
+ && !match_byte(0x99) && !match_byte(0x9a) && !match_byte(0x9b))
+ match_byte_assert(0x9c);
match_byte_assert(7);
match_byte_assert(0);
match_byte_assert(0);
}
else
- match_u32_assert(UINT32_MAX);
+ {
+ printf("%x\n", get_u32());
+ }
int decimal = data[pos];
int grouping = data[pos + 1];
else
{
match_byte_assert(',');
- if (!match_byte('.') && !match_byte(' '))
+ if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
match_byte_assert(0);
}
printf("<format decimal=\"%c\" grouping=\"", decimal);
match_byte_assert(1);
match_byte_assert(0);
- if (!match_byte(3))
- match_byte_assert(4);
+ if (!match_byte(3) && !match_byte(4))
+ match_byte_assert(5);
match_byte_assert(0);
match_byte_assert(0);
match_byte_assert(0);
if (!match_byte(0))
match_byte_assert(1);
- if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
- match_byte_assert(0x9a);
+ if (!match_byte(0x97) && !match_byte(0x98)
+ && !match_byte(0x99) && !match_byte(0x9a) && !match_byte(0x9b))
+ match_byte_assert(0x9c);
match_byte_assert(7);
match_byte_assert(0);
match_byte_assert(0);
else
{
match_byte_assert(',');
- if (!match_byte('.') && !match_byte(' '))
+ if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
match_byte_assert(0);
}
if (!match_byte(0))
match_byte_assert(1);
+ if (pos < outer_end)
+ {
+ printf("<seed>%d</seed>\n", get_u32());
+ match_u32_assert(0);
+ }
assert(pos == outer_end);
pos = outer_end;
}
else
- match_u32_assert(0);
+ {
+ pos = get_end();
+ }
}
int
/* This might be a version number of some kind, because value 1 seems
to only appear in an SPV file that also required its own weird
- special cases in dump_optional_value(). */
+ special cases in dump_value_modifier(). */
version = get_u32();
pos -= 4;
if (!match_u32(1))
/* Offset 8. */
match_byte_assert(0);
- match_byte_assert(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
if (!match_byte(0))
match_byte_assert(1);
/* Offset 27. */
pos++;
pos++;
- match_byte_assert(0);
- match_byte_assert(0);
+ pos++;
+ pos++;
/* Offset 31.