}
#define get_string() get_string(WHERE)
+static int
+get_end(void)
+{
+ int len = get_u32();
+ return pos + len;
+}
+
static char *
dump_counted_string(void)
{
char *s = NULL;
- int inner_end = pos + get_u32();
+ int inner_end = get_end();
if (pos != inner_end)
{
match_u32_assert(0);
return s;
}
+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_style(void)
+{
+ 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? */
+}
+
static char *
dump_nested_string(void)
{
match_byte_assert (0);
match_byte_assert (0);
- int outer_end = pos + get_u32();
+ int outer_end = get_end();
s = dump_counted_string();
- match_byte_assert(0x58);
+ if (match_byte(0x31))
+ dump_style();
+ else
+ match_byte_assert(0x58);
match_byte_assert(0x58);
if (pos != outer_end)
{
return;
}
- int outer_end = pos + get_u32();
-
+ int outer_end = get_end();
+
/* This counted-string appears to be a template string,
e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
dump_counted_string();
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();
else
match_byte_assert(0x58);
if (match_byte(0x31))
else if (match_u32 (2))
{
fprintf(stream, "(special 2)");
- if (!match_byte(0))
+ if (!match_byte(0) && !match_byte(1))
match_byte_assert(2);
match_byte_assert(0);
if (!match_u32 (2) && !match_u32(1))
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);
+ if (!match_u32(0) && !match_u32(1))
+ match_u32_assert(2);
}
else
{
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 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 *encoding = get_string();
- printf ("<encoding>%s</encoding>\n", encoding);
+ const char *locale = get_string();
+ printf ("<locale>%s</locale>\n", locale);
if (!match_u32(0))
match_u32_assert(UINT32_MAX);
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_assert(0x9b);
match_byte_assert(7);
match_byte_assert(0);
match_byte_assert(0);
}
else
match_u32_assert(0);
- int skip = get_u32();
- pos += skip;
+
+ /* The last chunk is an outer envelope that contains two inner envelopes.
+ The second inner envelope has some interesting data like the encoding and
+ the locale. */
+ if (version == 3)
+ {
+ int outer_end = get_end();
+
+ /* First inner envelope: byte*33 int[n] int*[n]. */
+ pos = get_end();
+
+ /* Second inner envelope. */
+ assert(get_end() == outer_end);
+
+ match_byte_assert(1);
+ match_byte_assert(0);
+ if (!match_byte(3) && !match_byte(4))
+ match_byte_assert(5);
+ match_byte_assert(0);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ printf("<command>%s</command>\n", get_string());
+ printf("<subcommand>%s</subcommand>\n", get_string());
+ printf("<language>%s</language>\n", get_string());
+ printf("<charset>%s</charset>\n", get_string());
+ printf("<locale>%s</locale>\n", get_string());
+
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
+ if (!match_byte(0))
+ match_byte_assert(1);
+
+ if (!match_byte(0x97) && !match_byte(0x98)
+ && !match_byte(0x99) && !match_byte(0x9a))
+ match_byte_assert(0x9b);
+ match_byte_assert(7);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ if (match_byte('.'))
+ {
+ if (!match_byte(',') && !match_byte('\''))
+ match_byte_assert(' ');
+ }
+ else
+ {
+ match_byte_assert(',');
+ if (!match_byte('.') && !match_byte(' '))
+ match_byte_assert(0);
+ }
+
+ pos += 8;
+ match_byte_assert(1);
+
+ if (outer_end - pos > 6)
+ {
+ /* There might be a pair of strings representing a dataset and
+ datafile name, or there might be a set of custom currency strings.
+ The custom currency strings start with a pair of integers, so we
+ can distinguish these from a string by checking for a null byte; a
+ small 32-bit integer will always contain a null and a text string
+ never will. */
+ int save_pos = pos;
+ int len = get_u32();
+ bool has_dataset = !memchr(&data[pos], '\0', len);
+ pos = save_pos;
+
+ if (has_dataset)
+ {
+ printf("<dataset>%s</dataset>\n", get_string());
+ printf("<datafile>%s</datafile>\n", get_string());
+
+ match_u32_assert(0);
+ get_u32();
+ match_u32_assert(0);
+ }
+ }
+
+ if (match_u32(5))
+ {
+ for (int i = 0; i < 5; i++)
+ printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
+ }
+ else
+ match_u32_assert(0);
+
+ match_byte_assert(0x2e);
+ if (!match_byte(0))
+ match_byte_assert(1);
+
+ if (pos < outer_end)
+ {
+ match_u32_assert(2000000);
+ match_u32_assert(0);
+ }
+ assert(pos == outer_end);
+
+ pos = outer_end;
+ }
+ else
+ match_u32_assert(0);
}
int
projects / pspp / blobdiff