+#include <assert.h>
#include <float.h>
#include <stdbool.h>
#include <stdint.h>
#include <string.h>
#include <sys/stat.h>
#include <unistd.h>
+#include "u8-mbtouc.h"
static uint8_t *data;
static size_t n;
}
+static bool __attribute__((unused))
+all_utf8(const char *p_)
+{
+ const uint8_t *p = (const uint8_t *) p_;
+ size_t len = strlen ((char *) p);
+ for (size_t ofs = 0, mblen; ofs < len; ofs += mblen)
+ {
+ ucs4_t uc;
+
+ mblen = u8_mbtouc (&uc, p + ofs, len - ofs);
+ if ((uc < 32 && uc != '\n') || uc == 127 || uc == 0xfffd)
+ return false;
+ }
+ return true;
+}
+
static char *
get_string(const char *where)
{
}
static void
-dump_value_31(FILE *stream)
+dump_optional_value(FILE *stream)
{
if (match_byte (0x31))
{
if (match_byte (3))
{
char *text = get_string();
- dump_value_31(stream);
+ dump_optional_value(stream);
char *identifier = get_string();
char *text_eng = get_string();
fprintf (stream, "<string c=\"%s\"", text_eng);
fprintf (stream, "/>\n");
if (!match_byte (0))
match_byte_assert(1);
-
}
else if (match_byte (5))
{
- dump_value_31(stream);
+ dump_optional_value(stream);
char *name = get_string ();
char *label = get_string ();
fprintf (stream, "<variable name=\"%s\"", name);
char *var, *vallab;
double value;
- dump_value_31 (stream);
+ dump_optional_value (stream);
format = get_u32 ();
value = get_double ();
var = get_string ();
unsigned int format;
double value;
- dump_value_31(stream);
+ dump_optional_value(stream);
format = get_u32 ();
value = get_double ();
fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
}
else
{
- dump_value_31(stream);
+ dump_optional_value(stream);
char *base = get_string();
int x = get_u32();
fprintf (stdout, " ");
printf ("<category>\n");
dump_value (stdout, level + 1);
- match_u32(1);
+ match_byte(1);
+ match_byte(0);
+ match_byte(0);
+ match_byte(0);
if (match_u32 (1))
match_byte (0);
static void
dump_title(void)
{
- pos = 0x27;
printf ("<title-local>\n");
dump_value(stdout, 0);
match_byte(1);
else
match_byte_assert(0x58);
-
int n_footnotes = get_u32();
for (int i = 0; i < n_footnotes; i++)
{
printf ("<footnote index=\"%d\">\n", i);
dump_value(stdout, 0);
+ /* Custom footnote marker string. */
if (match_byte (0x31))
- {
- /* Custom footnote marker string. */
- match_byte_assert(3);
- get_string();
- match_byte_assert(0x58);
- match_u32_assert(0);
- get_string();
- }
+ dump_value(stdout, 0);
else
match_byte_assert (0x58);
- printf("(%d)\n", get_u32());
+ get_u32 ();
printf ("</footnote>\n");
}
}
match_byte_assert(0);
if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10) && !match_byte(0x70))
match_byte_assert(0x50);
- if (!match_byte(0x41))
- match_byte_assert(0x51);
+ 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))
- match_u32_assert(0xfaad);
+ {
+ if (i == 7)
+ match_u32_assert(0xfaad);
+ else
+ match_u32_assert(0);
+ }
if (!match_u32(0) && !match_u32(1) && !match_u32(2))
match_u32_assert(3);
if (version > 1)
{
- /* These seem unlikely to be correct too. */
if (i != 3)
{
if (!match_u32(8))
int count = get_u32();
pos += 4 * count;
- printf ("<encoding>%s</encoding>\n", get_string ());
+ const char *encoding = get_string();
+ printf ("<encoding>%s</encoding>\n", encoding);
if (!match_u32(0))
match_u32_assert(UINT32_MAX);
}
else
match_u32_assert(UINT32_MAX);
+
+ int decimal = data[pos];
+ int grouping = data[pos + 1];
if (match_byte('.'))
{
- if (!match_byte(','))
+ if (!match_byte(',') && !match_byte('\''))
match_byte_assert(' ');
}
else
if (!match_byte('.') && !match_byte(' '))
match_byte_assert(0);
}
+ printf("<format decimal=\"%c\" grouping=\"", decimal);
+ if (grouping)
+ putchar(grouping);
+ printf("\"/>\n");
if (match_u32(5))
{
for (int i = 0; i < 5; i++)
- get_string();
+ printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
}
else
match_u32_assert(0);
exit(1);
}
- if (argc > 1)
+ if (argc != 2)
{
- if (!strcmp(argv[1], "title0"))
- {
- pos = 0x27;
- if (match_byte (0x03)
- || (match_byte (0x05) && match_byte (0x58)))
- printf ("%s\n", get_string());
- else
- printf ("<unknown>\n");
- return 0;
- }
- else if (!strcmp(argv[1], "title"))
- {
- dump_title();
- exit(0);
- }
- else if (!strcmp(argv[1], "titleraw"))
- {
- const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
- start = 0x27;
- n = find(fonts, sizeof fonts - 1);
- }
- else if (!strcmp(argv[1], "fonts"))
- {
- const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
- const char styles[] = "\xf0\0\0\0";
- start = find(fonts, sizeof fonts - 1);
- n = find(styles, sizeof styles - 1);
- }
- else if (!strcmp(argv[1], "styles"))
- {
- const char styles[] = "\xf0\0\0\0";
- const char dimensions[] = "-,,,.\0";
- start = find(styles, sizeof styles - 1);
- n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
- }
- else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
- {
- pos = 0;
- match_byte_assert(1);
- match_byte_assert(0);
-
- /* 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_value_31(). */
- version = get_u32();
- pos -= 4;
- if (!match_u32(1))
- match_u32_assert(3);
+ fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
+ exit (1);
+ }
- match_byte_assert(1);
- if (!match_byte(0))
- match_byte_assert(1);
- match_byte_assert(0);
- match_byte_assert(0);
- if (!match_byte(0))
- match_byte_assert(1);
- pos++;
- match_byte_assert(0);
- match_byte_assert(0);
- match_byte_assert(0);
- dump_title ();
- dump_fonts();
- dump_dims ();
- dump_data ();
- match_byte (1);
- if (pos != n)
- {
- fprintf (stderr, "%x / %x\n", pos, n);
- exit(1);
- }
- exit(0);
- }
+ if (!strcmp(argv[1], "title0"))
+ {
+ pos = 0x27;
+ if (match_byte (0x03)
+ || (match_byte (0x05) && match_byte (0x58)))
+ printf ("%s\n", get_string());
else
+ printf ("<unknown>\n");
+ return 0;
+ }
+ else if (!strcmp(argv[1], "title"))
+ {
+ pos = 0x27;
+ dump_title();
+ exit(0);
+ }
+ else if (!strcmp(argv[1], "titleraw"))
+ {
+ const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
+ start = 0x27;
+ n = find(fonts, sizeof fonts - 1);
+ }
+ else if (!strcmp(argv[1], "fonts"))
+ {
+ const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
+ const char styles[] = "\xf0\0\0\0";
+ start = find(fonts, sizeof fonts - 1);
+ n = find(styles, sizeof styles - 1);
+ }
+ else if (!strcmp(argv[1], "styles"))
+ {
+ const char styles[] = "\xf0\0\0\0";
+ const char dimensions[] = "-,,,.\0";
+ start = find(styles, sizeof styles - 1);
+ n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
+ }
+ else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
+ {
+ pos = 0;
+ match_byte_assert(1);
+ match_byte_assert(0);
+
+ /* 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(). */
+ version = get_u32();
+ pos -= 4;
+ if (!match_u32(1))
+ match_u32_assert(3);
+
+ match_byte_assert(1);
+ if (!match_byte(0))
+ match_byte_assert(1);
+
+ /* Offset 8. */
+ match_byte_assert(0);
+ match_byte_assert(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
+
+ /* Offset 11. */
+ pos++;
+ match_byte_assert(0);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ /* Offset 15. */
+ pos++;
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ /* Offset 19. */
+ pos++;
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ /* Offset 23. */
+ pos++;
+ if (!match_byte(0))
+ match_byte_assert(1);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ /* Offset 27. */
+ 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;
+
+ /* Offset 35. */
+ pos += 4;
+
+ dump_title ();
+ dump_fonts();
+ dump_dims ();
+ dump_data ();
+ match_byte (1);
+ if (pos != n)
{
- fprintf (stderr, "unknown section %s\n", argv[1]);
+ fprintf (stderr, "%x / %x\n", pos, n);
exit(1);
}
+ exit(0);
}
- else
- start = 0x27;
+ else if (!strcmp(argv[1], "raw"))
+ {
+ start = 0x27;
- dump_raw(stdout, start, n);
+ dump_raw(stdout, start, n);
+ }
+ else
+ {
+ fprintf (stderr, "unknown section %s\n", argv[1]);
+ exit(1);
+ }
return 0;
}
projects / pspp / blobdiff