}
#define match_byte_assert(b) match_byte_assert(b, WHERE)
+static void
+dump_raw(FILE *stream, int start, int end, const char *separator)
+{
+ for (size_t i = start; i < end; )
+ {
+ if (i + 5 <= n
+ && data[i]
+ //&& !data[i + 1]
+ && !data[i + 2]
+ && !data[i + 3]
+ && i + 4 + data[i] + data[i + 1] * 256 <= end
+ && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
+ {
+ fprintf(stream, "%s\"", separator);
+ fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
+ fputs("\" ", stream);
+
+ i += 4 + data[i] + data[i + 1] * 256;
+ }
+ else if (i + 12 <= end
+ && data[i + 1] == 40
+ && data[i + 2] == 5
+ && data[i + 3] == 0)
+ {
+ double d;
+
+ memcpy (&d, &data[i + 4], 8);
+ fprintf (stream, "F40.%d(%.*f)%s", data[i], data[i], d, separator);
+ i += 12;
+ }
+ else if (i + 12 <= end
+ && data[i + 1] == 40
+ && data[i + 2] == 31
+ && data[i + 3] == 0)
+ {
+ double d;
+
+ memcpy (&d, &data[i + 4], 8);
+ fprintf (stream, "PCT40.%d(%.*f)%s", data[i], data[i], d, separator);
+ i += 12;
+ }
+ else if (i + 4 <= end
+ && (data[i] && data[i] != 88 && data[i] != 0x41)
+ && !data[i + 1]
+ && !data[i + 2]
+ && !data[i + 3])
+ {
+ fprintf (stream, "i%d ", data[i]);
+ i += 4;
+ }
+ else
+ {
+ fprintf(stream, "%02x ", data[i]);
+ i++;
+ }
+ }
+
+
+}
+
static char *
get_string(const char *where)
{
}
#define get_string() get_string(WHERE)
+static void
+dump_nested(void)
+{
+ int subn = get_u32 ();
+#if 0
+ fprintf (stderr, "nested %d bytes: ", subn);
+ dump_raw(stderr, pos, pos + subn, "");
+ putc('\n', stderr);
+#endif
+ pos += subn;
+}
+
static void
dump_value_31(void)
{
get_string();
else
match_u32_assert (0);
- int subn = get_u32 ();
- printf ("nested %d bytes", subn);
- pos += subn;
+ dump_nested();
}
else if (match_u32 (1))
{
printf("(footnote %d) ", get_u32());
match_byte_assert (0);
match_byte_assert (0);
- int subn = get_u32 ();
- printf ("nested %d bytes", subn);
- pos += subn;
+ dump_nested();
}
else if (match_u32 (2))
{
match_u32_assert(1);
match_byte_assert(0);
match_byte_assert(0);
- int subn = get_u32 ();
- printf ("nested %d bytes", subn);
- pos += subn;
+ dump_nested();
}
else
{
match_byte_assert(0);
match_byte_assert(1);
match_byte_assert(0);
- int subn = get_u32 ();
- printf ("nested %d bytes, ", subn);
- pos += subn;
- subn = get_u32 ();
- printf ("nested %d bytes, ", subn);
- pos += subn;
+ dump_nested();
+ dump_nested();
}
}
else
match_byte_assert (0x58);
}
-static void dump_value(int level, bool match1);
-static void dump_substs(void (*dump)(int level), int level);
-static void dump_value__(int level, bool match1);
-
static void
-dump_value__with_preskip(int level)
+dump_value__(int level, bool match1)
{
+ for (int i = 0; i <= level; i++)
+ printf (" ");
+
match_byte(0);
match_byte(0);
match_byte(0);
match_byte(0);
- dump_value__ (level, false);
- putchar('\n');
-}
-static void
-dump_value__(int level, bool match1)
-{
if (match_byte (3))
{
char *s1 = get_string();
match_byte (1);
}
else
- dump_substs(dump_value__with_preskip, level + 1);
+ {
+ dump_value_31();
+
+ char *base = get_string();
+ int x = get_u32();
+ printf ("\"%s\" with %d variables:\n", base, x);
+ for (int i = 0; i < x; i++)
+ {
+ int y = get_u32();
+ if (!y)
+ y = 1;
+ else
+ match_u32_assert(0);
+ for (int j = 0; j <= level; j++)
+ printf (" ");
+ printf("variable %d has %d values:\n", i, y);
+ for (int j = 0; j < y; j++)
+ {
+ dump_value__ (level + 1, false);
+ putchar('\n');
+ }
+ }
+ }
}
-static void
-dump_value(int level, bool match1)
+static int
+compare_int(const void *a_, const void *b_)
{
- for (int i = 0; i <= level; i++)
- printf (" ");
-
- match_byte (0);
- match_byte (0);
- match_byte (0);
- match_byte (0);
- dump_value__(level, match1);
- match_byte(0);
- match_byte(0);
- match_byte(0);
+ const int *a = a_;
+ const int *b = b_;
+ return *a < *b ? -1 : *a > *b;
}
static void
-dump_dim_value(int level)
+check_permutation(int *a, int n, const char *name)
{
- for (int i = 0; i <= level; i++)
- printf (" ");
-
- match_byte(0);
- match_byte(0);
- match_byte(0);
- match_byte(0);
- dump_value__(level, true);
+ int b[n];
+ memcpy(b, a, n * sizeof *a);
+ qsort(b, n, sizeof *b, compare_int);
+ for (int i = 0; i < n; i++)
+ if (b[i] != i)
+ {
+ fprintf(stderr, "bad %s permutation:", name);
+ for (int i = 0; i < n; i++)
+ fprintf(stderr, " %d", a[i]);
+ putc('\n', stderr);
+ exit(1);
+ }
}
static void
-dump_category(int level)
+dump_category(int level, int *indexes, int *n_indexes)
{
- dump_value (level, true);
+ dump_value__ (level, true);
+ match_byte(0);
+ match_byte(0);
+ match_byte(0);
if (match_u32 (1))
match_byte (0);
else if (!match_u32(2))
match_u32_assert (0);
- get_u32 ();
-
+ int indx = get_u32();
int n_categories = get_u32();
+ if (indx != -1)
+ {
+ if (n_categories != 0)
+ {
+ fprintf(stderr, "index not -1 but subcategories\n");
+ exit(1);
+ }
+ indexes[(*n_indexes)++] = indx;
+ }
if (n_categories > 0)
printf (", %d subcategories:", n_categories);
+ else
+ printf (", index %d", indx);
printf("\n");
for (int i = 0; i < n_categories; i++)
- dump_category (level + 1);
+ dump_category (level + 1, indexes, n_indexes);
}
static void
{
int n_categories;
printf("next dim\n");
- dump_dim_value(0);
+ dump_value__ (0, false);
/* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
pos++;
match_u32_assert(2);
if (!match_byte(0))
match_byte_assert(1);
- get_u32();
- match_byte(0);
- match_byte(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(1);
+ static int dim_indx = 0;
+ match_u32_assert(dim_indx++);
n_categories = get_u32();
printf("%d nested categories\n", n_categories);
+
+ int indexes[1024];
+ int n_indexes = 0;
for (int i = 0; i < n_categories; i++)
- dump_category (0);
+ dump_category (0, indexes, &n_indexes);
+ check_permutation(indexes, n_indexes, "categories");
}
int n_dims;
}
}
-static void
-dump_substs(void (*dump)(int level), int level)
-{
- dump_value_31();
-
- char *base = get_string();
- int x = get_u32();
- printf ("\"%s\" with %d variables:\n", base, x);
- for (int i = 0; i < x; i++)
- {
- int y = get_u32();
- if (!y)
- y = 1;
- else
- match_u32_assert(0);
- for (int j = 0; j <= level; j++)
- printf (" ");
- printf("variable %d has %d values:\n", i, y);
- for (int j = 0; j < y; j++)
- {
- dump (level+1);
- putchar('\n');
- }
- }
-}
-
-static void
-dump_data_value(int level)
-{
- for (int i = 0; i <= level; i++)
- printf (" ");
-
- match_byte(0);
- match_byte(0);
- match_byte(0);
- match_byte(0);
- if (data[pos] <= 5)
- dump_value__(0, false);
- else
- dump_substs (dump_data_value, level + 1);
-}
-
static void
dump_data(void)
{
-#if 1
- int a[16];
- for (int i = 0; i < 3 + n_dims; i++)
+ /* The first three numbers add to the number of dimensions. */
+ int t = get_u32();
+ t += get_u32();
+ match_u32_assert(n_dims - t);
+
+ /* The next n_dims numbers are a permutation of the dimension numbers. */
+ int a[n_dims];
+ for (int i = 0; i < n_dims; i++)
a[i] = get_u32();
- printf ("data intro:");
- for (int i = 0; i < 3 + n_dims; i++)
- printf(" %d", a[i]);
- printf("\n");
-#else
- fprintf (stderr,"data intro (%d dims):", n_dims);
- for (int i = 0; i < 3+n_dims; i++)
- fprintf (stderr," %d", get_u32());
- fprintf(stderr,"\n");
-#endif
+ check_permutation(a, n_dims, "dimensions");
+
int x = get_u32();
printf ("%d data values, starting at %08x\n", x, pos);
for (int i = 0; i < x; i++)
{
printf("%08x, index %d:\n", pos, get_u32());
match_u32_assert(0);
- dump_data_value(0);
+ dump_value__(0, false);
putchar('\n');
}
}
-static void
-dump_title_value(int level)
-{
- for (int i = 0; i <= level; i++)
- printf (" ");
-
- match_byte (0);
- match_byte (0);
- match_byte (0);
- match_byte (0);
- if (data[pos] <= 5)
- dump_value__(level, true);
- else
- dump_substs(dump_title_value, level + 1);
-}
-
-static void
-dump_footnote_value(int level)
-{
- for (int i = 0; i <= level; i++)
- printf (" ");
-
- match_byte (0);
- match_byte (0);
- match_byte (0);
- match_byte (0);
- if (data[pos] <= 5)
- dump_value__(level, false);
- else
- dump_substs(dump_footnote_value, level + 1);
-}
-
static void
dump_title(void)
{
pos = 0x27;
- dump_title_value(0); putchar('\n');
- dump_title_value(0); putchar('\n');
+ dump_value__(0, true); putchar('\n');
+ dump_value__(0, true); putchar('\n');
match_byte_assert(0x31);
- dump_title_value(0); putchar('\n');
+ dump_value__(0, true); putchar('\n');
match_byte(0);
match_byte_assert(0x58);
if (match_byte(0x31))
{
- dump_footnote_value(0); putchar('\n');
+ dump_value__(0, false); putchar('\n');
}
else
match_byte_assert(0x58);
for (int i = 0; i < n_footnotes; i++)
{
printf("footnote %d:\n", i);
- dump_footnote_value(0);
+ dump_value__(0, false);
if (match_byte (0x31))
{
/* Custom footnote marker string. */
match_byte_assert(0x50);
if (!match_byte(0x41))
match_byte_assert(0x51);
- pos += 13;
+ if (!match_u32(0))
+ match_u32_assert(1);
+ 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))
+ match_u32_assert(0xfaad);
+
+ if (!match_u32(0) && !match_u32(1) && !match_u32(2))
+ match_u32_assert(3);
printf ("%s, ", get_string());
printf ("%s, ", get_string());
match_u32_assert(0);
match_u32_assert(0);
- pos++;
+ match_byte_assert(0);
get_u32();
get_u32();
get_u32();
else
start = 0x27;
- for (size_t i = start; i < n; )
- {
- if (i + 5 <= n
- && data[i]
- //&& !data[i + 1]
- && !data[i + 2]
- && !data[i + 3]
- && i + 4 + data[i] + data[i + 1] * 256 <= n
- && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
- {
- fputs("\n\"", stdout);
- fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stdout);
- fputs("\" ", stdout);
-
- i += 4 + data[i] + data[i + 1] * 256;
- }
- else if (i + 12 <= n
- && data[i + 1] == 40
- && data[i + 2] == 5
- && data[i + 3] == 0)
- {
- double d;
-
- memcpy (&d, &data[i + 4], 8);
- printf ("F40.%d(%.*f)\n", data[i], data[i], d);
- i += 12;
- }
- else if (i + 12 <= n
- && data[i + 1] == 40
- && data[i + 2] == 31
- && data[i + 3] == 0)
- {
- double d;
-
- memcpy (&d, &data[i + 4], 8);
- printf ("PCT40.%d(%.*f)\n", data[i], data[i], d);
- i += 12;
- }
- else if (i + 4 <= n
- && (data[i] && data[i] != 88 && data[i] != 0x41)
- && !data[i + 1]
- && !data[i + 2]
- && !data[i + 3])
- {
- printf ("i%d ", data[i]);
- i += 4;
- }
- else
- {
- printf("%02x ", data[i]);
- i++;
- }
- }
+ dump_raw(stdout, start, n, "\n");
return 0;
}
projects / pspp / blobdiff