+#include <assert.h>
+#include <float.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <sys/stat.h>
#include <unistd.h>
+#include "u8-mbtouc.h"
static uint8_t *data;
static size_t n;
+int version;
+
static bool
all_ascii(const uint8_t *p, size_t n)
{
return pos ? pos - data : 0;
}
-static size_t
-try_find_tail(const char *target, size_t target_len)
-{
- size_t pos = try_find(target, target_len);
- return pos ? pos + target_len : 0;
-}
-
static size_t
find(const char *target, size_t target_len)
{
return pos;
}
-static size_t
-find_tail(const char *target, size_t target_len)
-{
- size_t pos = try_find_tail(target, target_len);
- if (!pos)
- {
- fprintf (stderr, "not found\n");
- exit(1);
- }
- return pos;
-}
-
size_t pos;
#define XSTR(x) #x
#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 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_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 match_byte_assert(b) match_byte_assert(b, WHERE)
+static void
+newline(FILE *stream, int pos)
+{
+ fprintf(stream, "\n%08x: ", pos);
+}
+
+static void
+dump_raw(FILE *stream, int start, int end)
+{
+ 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))
+ {
+ newline(stream, i);
+ fprintf(stream, "\"");
+ 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)", data[i], data[i], d);
+ i += 12;
+ newline (stream, i);
+ }
+ 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)", data[i], data[i], d);
+ i += 12;
+ newline(stream, i);
+ }
+ 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 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)
{
}
#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)
+{
+ int len = get_u32();
+ 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);
+ }
+ for (int i = 0; i < n; i++)
+ {
+ int c = data[ofs + i];
+ fprintf(stderr, "%c", c >= 32 && c < 127 ? c : '.');
+ }
+ fprintf(stderr, "\n");
+}
+
+static char *
+dump_counted_string(void)
+{
+ char *s = NULL;
+ int inner_end = get_end();
+ if (pos != inner_end)
+ {
+ 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
+ match_byte_assert(0x58);
+ if (pos != inner_end)
+ {
+ fprintf(stderr, "inner end discrepancy\n");
+ exit(1);
+ }
+ }
+ 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(FILE *stream)
+{
+ char *s = NULL;
+
+ match_byte_assert (0);
+ match_byte_assert (0);
+ int outer_end = get_end();
+ s = dump_counted_string();
+ 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)
+ {
+ fprintf(stderr, "outer end discrepancy\n");
+ exit(1);
+ }
+
+ return s;
+}
+
static void
-dump_value_31(void)
+dump_value_modifier(FILE *stream)
{
if (match_byte (0x31))
{
if (match_u32 (0))
{
+ fprintf(stream, "<special0");
if (match_u32 (1))
- get_string();
+ {
+ /* Corpus frequencies:
+ 124 "a"
+ 12 "b"
+ 8 "a, b"
+
+ The given text is appended to the cell in a subscript font.
+ */
+ fprintf(stream, " subscript=\"%s\"", get_string());
+ }
else
match_u32_assert (0);
- int subn = get_u32 ();
- printf ("nested %d bytes", subn);
- pos += subn;
- }
- 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;
- }
- else if (match_u32 (2))
- {
- printf("(special 2)");
- match_byte_assert(0);
- match_byte_assert(0);
- 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;
+
+ if (version == 1)
+ {
+ /* We only have one SPV file for this version (with many
+ tables). */
+ match_byte(0);
+ if (!match_u32(1))
+ match_u32_assert(2);
+ match_byte(0);
+ match_byte(0);
+ if (!match_u32(0) && !match_u32(1) && !match_u32(2) && !match_u32(3) && !match_u32(4) && !match_u32(5) && !match_u32(6) && !match_u32(7) && !match_u32(8) && !match_u32(9))
+ match_u32_assert(10);
+ match_byte(0);
+ match_byte(0);
+ fprintf(stream, "/>\n");
+ return;
+ }
+
+ int outer_end = get_end();
+
+ /* This counted-string appears to be a template string,
+ e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
+ char *template = dump_counted_string();
+ if (template)
+ fprintf(stream, " template=\"%s\"", template);
+
+ if (match_byte(0x31))
+ dump_style(stream);
+ else
+ match_byte_assert(0x58);
+ if (match_byte(0x31))
+ {
+ /* Only two SPV files have anything like this, so it's hard to
+ generalize. */
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_u32_assert(0);
+ match_byte_assert(1);
+ match_byte_assert(0);
+ if (!match_byte(8) && !match_byte(1))
+ match_byte_assert(2);
+ match_byte_assert(0);
+ match_byte_assert(8);
+ match_byte_assert(0);
+ match_byte_assert(10);
+ match_byte_assert(0);
+ }
+ else
+ match_byte_assert(0x58);
+ if (pos != outer_end)
+ {
+ fprintf(stderr, "outer end discrepancy\n");
+ exit(1);
+ }
+ fprintf(stream, "/>\n");
}
else
{
- match_u32_assert(3);
- printf("(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);
- int subn = get_u32 ();
- printf ("nested %d bytes, ", subn);
- pos += subn;
- subn = get_u32 ();
- printf ("nested %d bytes, ", subn);
- pos += subn;
+ dump_nested_string(stream);
+ fprintf(stream, "/>\n");
}
}
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 const char *
+format_to_string (int type)
+{
+ static char tmp[16];
+ switch (type)
+ {
+ case 1: return "A";
+ case 2: return "AHEX";
+ case 3: return "COMMA";
+ case 4: return "DOLLAR";
+ case 5: case 40: return "F";
+ case 6: return "IB";
+ case 7: return "PIBHEX";
+ case 8: return "P";
+ case 9: return "PIB";
+ case 10: return "PK";
+ case 11: return "RB";
+ case 12: return "RBHEX";
+ case 15: return "Z";
+ case 16: return "N";
+ case 17: return "E";
+ case 20: return "DATE";
+ case 21: return "TIME";
+ case 22: return "DATETIME";
+ case 23: return "ADATE";
+ case 24: return "JDATE";
+ case 25: return "DTIME";
+ case 26: return "WKDAY";
+ case 27: return "MONTH";
+ case 28: return "MOYR";
+ case 29: return "QYR";
+ case 30: return "WKYR";
+ case 31: return "PCT";
+ case 32: return "DOT";
+ case 33: return "CCA";
+ case 34: return "CCB";
+ case 35: return "CCC";
+ case 36: return "CCD";
+ case 37: return "CCE";
+ case 38: return "EDATE";
+ case 39: return "SDATE";
+ default:
+ abort();
+ sprintf(tmp, "<%d>", type);
+ return tmp;
+ }
+}
static void
-dump_value__with_preskip(int level)
+dump_value(FILE *stream, int level)
{
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();
- dump_value_31();
- char *s2 = get_string();
- char *s3 = get_string();
- if (strcmp(s1, s3))
- printf("strings \"%s\", \"%s\" and \"%s\"", s1, s2, s3);
- else
- printf("string \"%s\" and \"%s\"", s1, s2);
- if (!match_byte (0))
- match_byte_assert(1);
- if (match1)
- match_byte (1);
- }
- else if (match_byte (5))
+ for (int i = 0; i <= level; i++)
+ fprintf (stream, " ");
+
+ if (match_byte (1))
{
- dump_value_31();
- printf ("variable \"%s\"", get_string());
- get_string();
- if (!match_byte(1) && !match_byte(2))
- match_byte_assert(3);
+ unsigned int format;
+ double value;
+
+ dump_value_modifier(stream);
+ format = get_u32 ();
+ value = get_double ();
+ fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
+ DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
}
else if (match_byte (2))
{
char *var, *vallab;
double value;
- match_byte_assert (0x58);
+ dump_value_modifier (stream);
format = get_u32 ();
value = get_double ();
var = get_string ();
vallab = get_string ();
- printf ("value %g format %d(%d.%d) var \"%s\" vallab \"%s\"",
- value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
+ fprintf (stream, "<numeric-datum value=\"%.*g\" format=\"%s%d.%d\"",
+ DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
+ if (var[0])
+ fprintf (stream, " variable=\"%s\"", var);
+ if (vallab[0])
+ 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_value_modifier(stream);
+ char *identifier = get_string();
+ char *text_eng = get_string();
+ fprintf (stream, "<string c=\"%s\"", text_eng);
+ if (identifier[0])
+ fprintf (stream, " identifier=\"%s\"", identifier);
+ if (strcmp(text_eng, text))
+ fprintf (stream, " local=\"%s\"", text);
+ fprintf (stream, "/>\n");
+ if (!match_byte (0))
+ match_byte_assert(1);
+ }
else if (match_byte (4))
{
unsigned int format;
char *var, *vallab, *value;
- match_byte_assert (0x58);
+ dump_value_modifier(stream);
format = get_u32 ();
vallab = get_string ();
var = get_string ();
if (!match_byte(1) && !match_byte(2))
match_byte_assert (3);
value = get_string ();
- printf ("value \"%s\" format %d(%d.%d) var \"%s\" vallab \"%s\"",
- value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
+ fprintf (stream, "<string-datum value=\"%s\" format=\"%s%d.%d\"",
+ value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
+ if (var[0])
+ fprintf (stream, " variable=\"%s\"", var);
+ if (vallab[0])
+ fprintf (stream, " label=\"%s\"/>\n", vallab);
+ fprintf (stream, "/>\n");
}
- else if (match_byte (1))
+ else if (match_byte (5))
{
- unsigned int format;
- double value;
-
- dump_value_31();
- format = get_u32 ();
- value = get_double ();
- printf ("value %g format %d(%d.%d)", value, format >> 16, (format >> 8) & 0xff, format & 0xff);
- if (match1)
- match_byte (1);
+ dump_value_modifier(stream);
+ char *name = get_string ();
+ char *label = get_string ();
+ fprintf (stream, "<variable name=\"%s\"", name);
+ if (label[0])
+ fprintf (stream, " label=\"%s\"", label);
+ fprintf (stream, "/>\n");
+ if (!match_byte(1) && !match_byte(2))
+ match_byte_assert(3);
}
else
- dump_substs(dump_value__with_preskip, level + 1);
+ {
+ dump_value_modifier(stream);
+
+ char *base = get_string();
+ int x = get_u32();
+ fprintf (stream, "<template format=\"%s\">\n", base);
+ 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 + 1; j++)
+ fprintf (stream, " ");
+ fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
+ for (int j = 0; j < y; j++)
+ dump_value (stream, level + 2);
+ for (int j = 0; j <= level + 1; j++)
+ fprintf (stream, " ");
+ fprintf (stream, "</substitution>\n");
+ }
+ for (int j = 0; j <= level; j++)
+ fprintf (stream, " ");
+ fprintf (stream, "</template>\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 (" ");
+ const int *a = a_;
+ const int *b = b_;
+ return *a < *b ? -1 : *a > *b;
+}
- 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);
+static void
+check_permutation(int *a, int n, const char *name)
+{
+ 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_dim_value(int level)
+dump_category(FILE *stream, int level, int **indexes, int *allocated_indexes,
+ int *n_indexes)
{
for (int i = 0; i <= level; i++)
- printf (" ");
+ fprintf (stream, " ");
+ printf ("<category>\n");
+ dump_value (stream, level + 1);
- match_byte(0);
- match_byte(0);
- match_byte(0);
- match_byte(0);
- dump_value__(level, true);
-}
+ int merge = data[pos];
+ if (!match_byte(0))
+ match_byte_assert (1);
-static void
-dump_category(int level)
-{
- dump_value (level, true);
+ match_byte_assert (0);
- if (match_u32 (1))
- match_byte (0);
- else if (match_byte (1))
+ int unindexed = data[pos];
+ if (!match_byte(0))
+ match_byte_assert (1);
+
+ int x = get_u32 ();
+ pos -= 4;
+ if (!match_u32 (0))
+ match_u32_assert (2);
+
+ int indx = get_u32();
+ int n_categories = get_u32();
+ if (indx == -1)
{
- match_byte (0);
- if (!match_u32 (2))
- match_u32_assert (1);
- match_byte (0);
+ if (merge)
+ {
+ for (int i = 0; i <= level + 1; i++)
+ fprintf (stream, " ");
+ fprintf (stream, "<merge/>\n");
+ }
+ }
+ else
+ {
+ if (merge)
+ {
+ fprintf(stderr, "index not -1 but merged\n");
+ exit(1);
+ }
+ if (x != 2)
+ {
+ fprintf(stderr, "index not -1 but x != 2\n");
+ exit(1);
+ }
+ if (n_categories != 0)
+ {
+ fprintf(stderr, "index not -1 but subcategories\n");
+ exit(1);
+ }
+ if (*n_indexes >= *allocated_indexes)
+ {
+ *allocated_indexes = *allocated_indexes ? 2 * *allocated_indexes : 16;
+ *indexes = realloc(*indexes, *allocated_indexes * sizeof **indexes);
+ }
+ (*indexes)[(*n_indexes)++] = indx;
}
- else if (!match_u32(2))
- match_u32_assert (0);
- get_u32 ();
+ int expected_unindexed = indx == -1;
+ if (unindexed != expected_unindexed)
+ {
+ fprintf(stderr, "unindexed (%d) mismatch with indx (%d)\n",
+ unindexed, indx);
+ exit(1);
+ }
- int n_categories = get_u32();
- if (n_categories > 0)
- printf (", %d subcategories:", n_categories);
- printf("\n");
+ if (n_categories == 0)
+ {
+ for (int i = 0; i <= level + 1; i++)
+ fprintf (stream, " ");
+ fprintf (stream, "<category-index>%d</category-index>\n", indx);
+ }
for (int i = 0; i < n_categories; i++)
- dump_category (level + 1);
+ dump_category (stream, level + 1, indexes, allocated_indexes, n_indexes);
+ for (int i = 0; i <= level; i++)
+ fprintf (stream, " ");
+ printf ("</category>\n");
}
-static void
-dump_dim(void)
+static int
+dump_dim(int indx)
{
int n_categories;
- printf("next dim\n");
- dump_dim_value(0);
- /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
+ printf ("<dimension index=\"%d\">\n", indx);
+ dump_value (stdout, 0);
+
+ /* This byte is usually 0 but many other values have been spotted. */
pos++;
if (!match_byte(0) && !match_byte(1))
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);
+ if (!match_u32(UINT32_MAX))
+ match_u32_assert(indx);
n_categories = get_u32();
- printf("%d nested categories\n", n_categories);
+
+ int *indexes = NULL;
+ int n_indexes = 0;
+ int allocated_indexes = 0;
for (int i = 0; i < n_categories; i++)
- dump_category (0);
+ dump_category (stdout, 0, &indexes, &allocated_indexes, &n_indexes);
+ check_permutation(indexes, n_indexes, "categories");
+
+ fprintf (stdout, "</dimension>\n");
+ return n_indexes;
}
int n_dims;
+static int dim_n_cats[64];
+#define MAX_DIMS (sizeof dim_n_cats / sizeof *dim_n_cats)
+
static void
dump_dims(void)
{
n_dims = get_u32();
- printf ("%u dimensions\n", n_dims);
+ assert(n_dims < MAX_DIMS);
for (int i = 0; i < n_dims; i++)
- {
- printf("\n");
- dump_dim ();
- }
+ dim_n_cats[i] = dump_dim (i);
}
static void
-dump_substs(void (*dump)(int level), int level)
+dump_data(void)
{
- dump_value_31();
+ /* The first three numbers add to the number of dimensions. */
+ int l = get_u32();
+ int r = get_u32();
+ int c = n_dims - l - r;
+ match_u32_assert(c);
+
+ /* The next n_dims numbers are a permutation of the dimension numbers. */
+ int a[n_dims];
+ for (int i = 0; i < n_dims; i++)
+ {
+ int dim = get_u32();
+ a[i] = dim;
+
+ const char *name = i < l ? "layer" : i < l + r ? "row" : "column";
+ printf ("<%s dimension=\"%d\"/>\n", name, dim);
+ }
+ check_permutation(a, n_dims, "dimensions");
- char *base = get_string();
int x = get_u32();
- printf ("\"%s\" with %d variables:\n", base, x);
+ printf ("<data>\n");
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++)
+ unsigned int indx = get_u32();
+ printf (" <datum index=\"%d\" coords=", indx);
+
+ int coords[MAX_DIMS];
+ for (int i = n_dims; i-- > 0; )
{
- dump (level+1);
- putchar('\n');
+ coords[i] = indx % dim_n_cats[i];
+ indx /= dim_n_cats[i];
}
- }
-}
-
-static void
-dump_data_value(int level)
-{
- for (int i = 0; i <= level; i++)
- printf (" ");
+ for (int i = 0; i < n_dims; i++)
+ printf("%c%d", i ? ',' : '"', coords[i]);
- 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++)
- 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
- 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());
+ printf ("\">\n");
match_u32_assert(0);
- dump_data_value(0);
- putchar('\n');
+ if (version == 1)
+ match_byte(0);
+ dump_value(stdout, 1);
+ fprintf (stdout, " </datum>\n");
}
+ printf ("</data>\n");
}
static void
-dump_title_value(int level)
+dump_title(void)
{
- 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);
-}
+ printf ("<title-local>\n");
+ dump_value(stdout, 0);
+ match_byte(1);
+ printf ("</title-local>\n");
-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);
-}
+ printf ("<subtype>\n");
+ dump_value(stdout, 0);
+ match_byte(1);
+ printf ("</subtype>\n");
-static void
-dump_title(void)
-{
- pos = 0x27;
- dump_title_value(0); putchar('\n');
- dump_title_value(0); putchar('\n');
match_byte_assert(0x31);
- dump_title_value(0); putchar('\n');
- match_byte(0);
- match_byte_assert(0x58);
+
+ printf ("<title-c>\n");
+ dump_value(stdout, 0);
+ match_byte(1);
+ printf ("</title-c>\n");
+
if (match_byte(0x31))
{
- dump_footnote_value(0); putchar('\n');
+ printf ("<user-caption>\n");
+ dump_value(stdout, 0);
+ printf ("</user-caption>\n");
}
else
match_byte_assert(0x58);
-
-
- int n_footnotes = get_u32();
- if (n_footnotes >= 20)
+ if (match_byte(0x31))
{
- fprintf(stderr, "%08x: %d footnotes\n", pos - 4, n_footnotes);
- exit(1);
+ printf ("<caption>\n");
+ dump_value(stdout, 0);
+ printf ("</caption>\n");
}
+ else
+ match_byte_assert(0x58);
- printf("------\n%d footnotes\n", n_footnotes);
- if (n_footnotes < 20)
+ int n_footnotes = get_u32();
+ for (int i = 0; i < n_footnotes; i++)
{
- for (int i = 0; i < n_footnotes; i++)
- {
- printf("footnote %d:\n", i);
- dump_footnote_value(0);
- if (match_byte (0x31))
- {
- /* Custom footnote marker string. */
- match_byte_assert(3);
- get_string();
- match_byte_assert(0x58);
- match_u32_assert(0);
- get_string();
- }
- else
- match_byte_assert (0x58);
- printf("(%d)\n", get_u32());
- }
+ printf ("<footnote index=\"%d\">\n", i);
+ dump_value(stdout, 0);
+ /* Custom footnote marker string. */
+ if (match_byte (0x31))
+ dump_value(stdout, 0);
+ else
+ match_byte_assert (0x58);
+ get_u32 ();
+ printf ("</footnote>\n");
}
}
-static int
-find_dimensions(void)
-{
- {
- const char dimensions[] = "-,,,.\0";
- int x = try_find_tail(dimensions, sizeof dimensions - 1);
- if (x)
- return x;
- }
-
- const char dimensions[] = "-,,, .\0";
- return find_tail(dimensions, sizeof dimensions - 1);
-}
-
static void
dump_fonts(void)
{
- printf("fonts: offset=%08x\n", pos);
match_byte(0);
for (int i = 1; i <= 8; i++)
{
- printf("%08x: font %d, ", pos, i);
+ printf ("<style index=\"%d\"", i);
match_byte_assert(i);
match_byte_assert(0x31);
- printf("%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_assert(0x50);
- if (!match_byte(0x41))
- match_byte_assert(0x51);
- pos += 13;
- printf ("%s, ", get_string());
- printf ("%s, ", get_string());
- match_u32_assert(0);
+ printf(" font=\"%s\"", get_string());
+
+ 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);
+
+ printf (" fgcolor=\"%s\"", get_string());
+ printf (" bgcolor=\"%s\"", get_string());
+
+ if (!match_byte(0))
+ match_byte_assert(1);
match_u32_assert(0);
- pos++;
- get_u32();
- get_u32();
- get_u32();
- get_u32();
- putchar('\n');
- }
+ char *othercolor = get_string();
+ if (othercolor[0])
+ printf(" othercolor=\"%s\"", othercolor);
- match_u32_assert(240);
- pos += 240;
+ if (version > 1)
+ {
+ printf("\nfonts:");
+ for (int i = 0; i < 4; i++)
+ printf(" %2d", get_u32());
+ printf("\n");
+ }
- match_u32_assert(18);
- pos += 18;
+ printf ("/>\n");
+ }
- if (match_u32(117))
- pos += 117;
- 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++)
+ {
+ 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_u32_assert(142);
- pos += 142;
+ 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();
+ pos += get_be32();
+ get_string_be();
+ char *look = get_string_be();
+ if (look[0])
+ printf(" look=\"%s\"", look);
+ printf(">\n");
}
+ pos = x3_end;
int count = get_u32();
pos += 4 * count;
- char *encoding = get_string();
- printf("encoding=%s\n", encoding);
+ 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 (!match_byte(0x99) && !match_byte(0x98))
- match_byte_assert(0x97);
- match_byte_assert(7);
- match_byte_assert(0);
- match_byte_assert(0);
+ printf("<epoch>%d</epoch>\n", get_u32());
+
+ int decimal = data[pos];
+ int grouping = data[pos + 1];
if (match_byte('.'))
- match_byte_assert(',');
+ {
+ if (!match_byte(',') && !match_byte('\''))
+ match_byte_assert(' ');
+ }
else
{
match_byte_assert(',');
- if (!match_byte('.'))
- match_byte_assert(' ');
+ if (!match_byte('.') && !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++)
+ printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
+ }
+ else
+ match_u32_assert(0);
+
+ /* 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);
+
+ printf("<epoch2>%d</epoch2>\n", get_u32());
+
+ if (match_byte('.'))
+ {
+ if (!match_byte(',') && !match_byte('\''))
+ match_byte_assert(' ');
+ }
+ else
+ {
+ match_byte_assert(',');
+ if (!match_byte('.') && !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)
+ {
+ printf("<seed>%d</seed>\n", get_u32());
+ match_u32_assert(0);
+ }
+ assert(pos == outer_end);
+
+ pos = outer_end;
+ }
+ else
+ {
+ pos = get_end();
}
- match_u32_assert(5);
- for (int i = 0; i < 5; i++)
- get_string();
- pos += get_u32();
- if (pos != find_dimensions())
- fprintf (stderr, "%08x / %08x\n", pos, find_dimensions());
}
int
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);
- match_u32_assert(3);
- 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 ();
- printf("\n\ndata:\n");
- dump_data ();
- match_byte (1);
- if (pos != n)
- {
- fprintf (stderr, "%x / %x\n", pos, n);
- exit(1);
- }
- exit(0);
- }
- else
- {
- fprintf (stderr, "unknown section %s\n", argv[1]);
- exit(1);
- }
+ fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
+ exit (1);
}
- else
- start = 0x27;
- for (size_t i = start; i < n; )
+ if (!strcmp(argv[1], "title0"))
{
- 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);
+ 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);
- 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;
+ /* 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_modifier(). */
+ version = get_u32();
+ pos -= 4;
+ if (!match_u32(1))
+ match_u32_assert(3);
- 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;
+ match_byte_assert(1);
+ if (!match_byte(0))
+ match_byte_assert(1);
- 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
+ /* Offset 8. */
+ match_byte_assert(0);
+ if (!match_byte(0))
+ match_byte_assert(1);
+ 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++;
+ pos++;
+ pos++;
+
+ /* 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)
{
- printf("%02x ", data[i]);
- i++;
+ fprintf (stderr, "%x / %x\n", pos, n);
+ exit(1);
}
+ exit(0);
+ }
+ else if (!strcmp(argv[1], "raw"))
+ {
+ start = 0x27;
+
+ dump_raw(stdout, start, n);
+ }
+ else
+ {
+ fprintf (stderr, "unknown section %s\n", argv[1]);
+ exit(1);
}
return 0;
projects / pspp / blobdiff