#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>
+#include <time.h>
#include <unistd.h>
#include "u8-mbtouc.h"
#define STR(x) XSTR(x)
#define WHERE __FILE__":" STR(__LINE__)
+static uint8_t
+get_byte(void)
+{
+ return data[pos++];
+}
+
static unsigned int
get_u32(void)
{
return x;
}
+static unsigned long long int
+get_u64(void)
+{
+ uint64_t x;
+ memcpy(&x, &data[pos], 8);
+ pos += 8;
+ return x;
+}
+
+static unsigned int
+get_be32(void)
+{
+ uint32_t x;
+ x = (data[pos] << 24) | (data[pos + 1] << 16) | (data[pos + 2] << 8) | data[pos + 3];
+ pos += 4;
+ 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 x;
}
+static double __attribute__((unused))
+get_float(void)
+{
+ float x;
+ memcpy(&x, &data[pos], 4);
+ pos += 4;
+ return x;
+}
+
static bool
match_u32(uint32_t x)
{
}
#define match_u32_assert(x) match_u32_assert(x, WHERE)
+static bool __attribute__((unused))
+match_u64(uint64_t x)
+{
+ if (get_u64() == x)
+ return true;
+ pos -= 8;
+ return false;
+}
+
+static void __attribute__((unused))
+match_u64_assert(uint64_t x, const char *where)
+{
+ unsigned long long int y = get_u64();
+ if (x != y)
+ {
+ fprintf(stderr, "%s: 0x%x: expected u64:%llu, got u64:%llu\n", where, pos - 8, x, y);
+ exit(1);
+ }
+}
+#define match_u64_assert(x) match_u64_assert(x, WHERE)
+
+static bool __attribute__((unused))
+match_be32(uint32_t x)
+{
+ if (get_be32() == x)
+ return true;
+ pos -= 4;
+ return false;
+}
+
+static void
+match_be32_assert(uint32_t x, const char *where)
+{
+ unsigned int y = get_be32();
+ if (x != y)
+ {
+ fprintf(stderr, "%s: 0x%x: expected be%u, got be%u\n", where, pos - 4, x, y);
+ exit(1);
+ }
+}
+#define match_be32_assert(x) match_be32_assert(x, WHERE)
+
static bool
match_byte(uint8_t b)
{
}
#define get_string() get_string(WHERE)
+static char *
+get_string_be(const char *where)
+{
+ if (1
+ /*data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0*/
+ /*&& all_ascii(&data[pos + 4], data[pos])*/)
+ {
+ int len = data[pos + 2] * 256 + data[pos + 3];
+ char *s = malloc(len + 1);
+
+ memcpy(s, &data[pos + 4], len);
+ s[len] = 0;
+ pos += 4 + len;
+ return s;
+ }
+ else
+ {
+ fprintf(stderr, "%s: 0x%x: expected string\n", where, pos);
+ exit(1);
+ }
+}
+#define get_string_be() get_string_be(WHERE)
+
static int
get_end(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))
+ printf("-");
+ else
+ printf(" ");
+#endif
+ printf("%02x", c);
+ }
+ for (int i = 0; i < n; i++)
+ {
+ int c = data[ofs + i];
+ printf("%c", c >= 32 && c < 127 ? c : '.');
+ }
+}
+
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(i);
match_byte_assert(0x31);
printf(" font=\"%s\"", get_string());
- match_byte_assert(0);
- match_byte_assert(0);
- if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10) && !match_byte(0x70))
- match_byte_assert(0x50);
- match_byte_assert(0x41);
- if (!match_u32(0) && !match_u32(1))
- match_u32_assert(2);
- 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 (i == 7)
- match_u32_assert(0xfaad);
- else
- match_u32_assert(0);
- }
+ printf(" size=\"%gpt\"", get_float());
+
+ int style = get_u32();
+ if (style & 1)
+ printf(" bold=\"true\"");
+ if (style & 2)
+ printf(" italic=\"true\"");
+
+ bool underline = data[pos++];
+ if (underline)
+ printf(" underline=\"true\"");
+
+ int halign = get_u32();
+ printf("\nhalign=%d\n", halign);
+
+ int valign = get_u32();
+ printf("\nvalign=%d\n", valign);
- if (!match_u32(0) && !match_u32(1) && !match_u32(2))
- 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);
- }
- else
- {
- get_u32();
- if (!match_u32(-1) && !match_u32(8))
- match_u32_assert(24);
- if (!match_u32(-1) && !match_u32(2))
- match_u32_assert(3);
- }
-
- /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
- get_u32();
+ printf("\nfonts:");
+ for (int i = 0; i < 4; i++)
+ printf(" %2d", get_u32());
+ printf("\n");
}
printf ("/>\n");
}
- 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 x1 = get_u32();
+ int x1_end = pos + x1;
+ printf("<borders>\n");
+ match_be32_assert(1);
+ int n_borders = get_be32();
+ for (int i = 0; i < n_borders; i++)
{
- match_u32_assert(16);
- pos += 16;
+ int type = get_be32();
+ int stroke = get_be32();
+ int color = get_be32();
+ printf(" <border type=\"%d\" stroke=\"%s\" color=\"#%06x\"/>\n",
+ type,
+ (stroke == 0 ? "none"
+ : stroke == 1 ? "solid"
+ : stroke == 2 ? "dashed"
+ : stroke == 3 ? "thick"
+ : stroke == 4 ? "thin"
+ : stroke == 5 ? "double"
+ : "<error>"),
+ color);
}
+ bool grid = get_byte();
+ pos += 3;
+ printf(" <grid show=\"%s\"/>\n", grid ? "yes" : "no");
+ printf("</borders>\n");
+ assert(pos == x1_end);
+
+ int skip = get_u32();
+ assert(skip == 18 || skip == 25);
+ pos += skip;
+
+ int x3 = get_u32();
+ int x3_end = pos + x3;
+ if (version == 3)
+ {
+ match_be32_assert(1);
+ get_be32();
+ printf("<settings layer=\"%d\"", get_be32());
+ if (!get_byte())
+ printf(" skipempty=\"false\"");
+ if (!get_byte())
+ printf(" showdimensionincorner=\"false\"");
+ if (!get_byte())
+ printf(" markers=\"numeric\"");
+ if (!get_byte())
+ printf(" footnoteposition=\"subscript\"");
+ get_byte();
+ int nbytes = get_be32();
+ printf("\n");
+ hex_dump(pos, nbytes);
+ printf("\n");
+ pos += nbytes;
+ get_string_be();
+ char *look = get_string_be();
+ if (look[0])
+ printf(" look=\"%s\"", look);
+ printf(">\n");
+ }
+ pos = x3_end;
+ /* Manual column widths, if present. */
int count = get_u32();
- pos += 4 * count;
+ printf("<columnwidths>");
+ for (int i = 0; i < count; i++)
+ {
+ if (i)
+ putchar(' ');
+ printf("%d", get_u32());
+ }
+ printf("</columnwidths>\n");
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);
if (!match_byte(0))
match_byte_assert(1);
- if (version > 1)
- {
- if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
- match_byte_assert(0x9a);
- match_byte_assert(7);
- match_byte_assert(0);
- match_byte_assert(0);
- }
- else
- match_u32_assert(UINT32_MAX);
+ printf("<epoch>%d</epoch>\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);
int outer_end = get_end();
/* First inner envelope: byte*33 int[n] int*[n]. */
- pos = get_end();
+ int inner_len = get_u32();
+ int inner_end = pos + inner_len;
+ int array_start = pos + 33;
+ match_byte_assert(0);
+ pos++; /* 0, 1, 10 seen. */
+ match_byte_assert(0);
+ pos++; /* 0...11 seen. */
+ if (!match_byte(0) && !match_byte(1) && !match_byte(2))
+ match_byte_assert(3);
+ if (!match_byte(0) && !match_byte(2))
+ match_byte_assert(3);
+ if (!match_u64(0))
+ match_u64_assert(UINT64_MAX);
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_byte_assert(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(1);
+ pos = array_start;
+#if 1
+ printf("widths:");
+ while (pos < inner_end)
+ printf(" %d", get_u32());
+ printf("\n");
+#endif
+ pos = inner_end;;
/* Second inner envelope. */
assert(get_end() == outer_end);
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);
- match_byte_assert(7);
- match_byte_assert(0);
- match_byte_assert(0);
+ printf("<epoch2>%d</epoch2>\n", get_u32());
if (match_byte('.'))
{
else
{
match_byte_assert(',');
- if (!match_byte('.') && !match_byte(' '))
+ if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
match_byte_assert(0);
}
- pos += 8;
- match_byte_assert(1);
+ printf ("small: %g\n", get_double());
+ match_byte_assert(1);
if (outer_end - pos > 6)
{
/* There might be a pair of strings representing a dataset and
printf("<datafile>%s</datafile>\n", get_string());
match_u32_assert(0);
- get_u32();
+
+ time_t date = get_u32();
+ struct tm tm = *localtime(&date);
+ char s[128];
+ strftime(s, sizeof s, "%a, %d %b %Y %H:%M:%S %z", &tm);
+ printf("<date>%s</date>\n", s);
+
match_u32_assert(0);
}
}
else
match_u32_assert(0);
- match_byte_assert(0x2e);
+ match_byte_assert('.');
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);
match_byte_assert(0);
/* Offset 27. */
+#if 1
pos++;
pos++;
- match_byte_assert(0);
- match_byte_assert(0);
-
- /* Offset 31.
-
- This is the tableId, e.g. -4154297861994971133 would be 0xdca00003.
- We don't have enough context to validate it. */
- pos += 4;
+ pos++;
+ pos++;
+#else
+ printf("\nheader:");
+ for (int i = 0; i < 4; i++)
+ printf(" %02x", data[pos++]);
+ printf("\n");
+#endif
- /* Offset 35. */
- pos += 4;
+ /* Offset 31. */
+ printf("<tableid>%lld</tableid>", get_u64());
dump_title ();
dump_fonts();
projects / pspp / blobdiff