X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;ds=sidebyside;f=dump.c;h=9e1bfd9f960b2f97379db046204193bcd6cdfd6a;hb=77b70801d56d19a531c970c57bc585cd69395edb;hp=65955d81995b16e50c631e430ee33fe95cf485d1;hpb=db2c3800433514d5de7fb4fc9372eabaa30c2857;p=pspp
diff --git a/dump.c b/dump.c
index 65955d8199..9e1bfd9f96 100644
--- a/dump.c
+++ b/dump.c
@@ -218,15 +218,18 @@ get_string(const char *where)
}
#define get_string() get_string(WHERE)
+static int
+get_end(void)
+{
+ int len = get_u32();
+ return pos + len;
+}
+
static char *
-dump_nested_string(void)
+dump_counted_string(void)
{
char *s = NULL;
-
- match_byte_assert (0);
- match_byte_assert (0);
- int outer_end = pos + get_u32();
- int inner_end = pos + get_u32();
+ int inner_end = get_end();
if (pos != inner_end)
{
match_u32_assert(0);
@@ -240,6 +243,18 @@ dump_nested_string(void)
exit(1);
}
}
+ return s;
+}
+
+static char *
+dump_nested_string(void)
+{
+ char *s = NULL;
+
+ match_byte_assert (0);
+ match_byte_assert (0);
+ int outer_end = get_end();
+ s = dump_counted_string();
match_byte_assert(0x58);
match_byte_assert(0x58);
if (pos != outer_end)
@@ -260,7 +275,11 @@ dump_optional_value(FILE *stream)
{
if (match_u32 (1))
{
- /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
+ /* Corpus frequencies:
+ 124 "a"
+ 12 "b"
+ 8 "a, b"
+ */
get_string();
}
else
@@ -282,26 +301,11 @@ dump_optional_value(FILE *stream)
return;
}
- int outer_end = pos + get_u32();
- int inner_end = pos + get_u32();
- if (pos != inner_end)
- {
- match_u32_assert(0);
- if (match_byte(0x31))
- {
- /* Appears to be a template string, e.g. '^1 cells (^2) expf < 5. Min exp = ^3...'.
- Probably doesn't actually appear in output because many examples look unpolished,
- e.g. 'partial list cases value ^1 shown upper...' */
- get_string();
- }
- else
- match_byte_assert(0x58);
- if (pos != inner_end)
- {
- fprintf(stderr, "inner end discrepancy\n");
- exit(1);
- }
- }
+ 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))
{
@@ -358,7 +362,7 @@ dump_optional_value(FILE *stream)
match_byte_assert(0);
if (!match_u32 (2) && !match_u32(1))
match_u32_assert(3);
- dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+ dump_nested_string();
}
else
{
@@ -435,32 +439,16 @@ dump_value(FILE *stream, int level)
for (int i = 0; i <= level; i++)
fprintf (stream, " ");
- if (match_byte (3))
- {
- char *text = get_string();
- dump_optional_value(stream);
- char *identifier = get_string();
- char *text_eng = get_string();
- fprintf (stream, "\n");
- if (!match_byte (0))
- match_byte_assert(1);
- }
- else if (match_byte (5))
+ if (match_byte (1))
{
+ unsigned int format;
+ double value;
+
dump_optional_value(stream);
- char *name = get_string ();
- char *label = get_string ();
- fprintf (stream, "\n");
- if (!match_byte(1) && !match_byte(2))
- match_byte_assert(3);
+ format = get_u32 ();
+ value = get_double ();
+ fprintf (stream, "\n",
+ DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
}
else if (match_byte (2))
{
@@ -483,12 +471,27 @@ dump_value(FILE *stream, int level)
if (!match_byte (1) && !match_byte(2))
match_byte_assert (3);
}
+ else if (match_byte (3))
+ {
+ char *text = get_string();
+ dump_optional_value(stream);
+ char *identifier = get_string();
+ char *text_eng = get_string();
+ 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_optional_value(stream);
format = get_u32 ();
vallab = get_string ();
var = get_string ();
@@ -503,16 +506,17 @@ dump_value(FILE *stream, int level)
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_optional_value(stream);
- format = get_u32 ();
- value = get_double ();
- fprintf (stream, "\n",
- DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
+ char *name = get_string ();
+ char *label = get_string ();
+ fprintf (stream, "\n");
+ if (!match_byte(1) && !match_byte(2))
+ match_byte_assert(3);
}
else
{
@@ -569,38 +573,51 @@ check_permutation(int *a, int n, const char *name)
}
static void
-dump_category(int level, int *indexes, int *n_indexes, int max_indexes)
+dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_indexes)
{
for (int i = 0; i <= level; i++)
- fprintf (stdout, " ");
+ fprintf (stream, " ");
printf ("\n");
- dump_value (stdout, level + 1);
+ dump_value (stream, level + 1);
- int start = pos;
- int d0 = data[pos];
- match_byte(1);
- int c = 0;
- c += match_byte(0);
- c += match_byte(0);
- c += match_byte(0);
+ int merge = data[pos];
+ if (!match_byte(0))
+ match_byte_assert (1);
- int d1 = data[pos];
- int d2 = 9;
- if (match_byte (1))
- {
- d2 = data[pos];
- if (!match_u32 (0) && !match_u32 (1))
- match_u32_assert (2);
- }
- else
+ match_byte_assert (0);
+
+ 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 len = pos - start;
int indx = get_u32();
int n_categories = get_u32();
- if (indx != -1)
+ if (indx == -1)
+ {
+ if (merge)
+ {
+ for (int i = 0; i <= level + 1; i++)
+ fprintf (stream, " ");
+ fprintf (stream, "\n");
+ }
+ }
+ else
{
- fprintf (stderr, "category %d %d %d %d %d:", d0, c, d1, d2, len);
+ 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");
@@ -613,26 +630,29 @@ dump_category(int level, int *indexes, int *n_indexes, int max_indexes)
}
indexes[(*n_indexes)++] = indx;
}
- else
- fprintf (stderr, "group %d %d %d %d %d:", d0, c, d1, d2, len);
- for (int i = 0; i < len; i++)
- fprintf (stderr, " %02x", data[start + i]);
- putc ('\n', stderr);
+
+ int expected_unindexed = indx == -1;
+ if (unindexed != expected_unindexed)
+ {
+ fprintf(stderr, "unindexed (%d) mismatch with indx (%d)\n",
+ unindexed, indx);
+ exit(1);
+ }
if (n_categories == 0)
{
for (int i = 0; i <= level + 1; i++)
- fprintf (stdout, " ");
- fprintf (stdout, "%d\n", indx);
+ fprintf (stream, " ");
+ fprintf (stream, "%d\n", indx);
}
for (int i = 0; i < n_categories; i++)
- dump_category (level + 1, indexes, n_indexes, max_indexes);
+ dump_category (stream, level + 1, indexes, n_indexes, max_indexes);
for (int i = 0; i <= level; i++)
- fprintf (stdout, " ");
+ fprintf (stream, " ");
printf ("\n");
}
-static void
+static int
dump_dim(int indx)
{
int n_categories;
@@ -659,40 +679,64 @@ dump_dim(int indx)
int indexes[2048];
int n_indexes = 0;
for (int i = 0; i < n_categories; i++)
- dump_category (0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
+ dump_category (stdout, 0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
check_permutation(indexes, n_indexes, "categories");
fprintf (stdout, "\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();
+ assert(n_dims < MAX_DIMS);
for (int i = 0; i < n_dims; i++)
- dump_dim (i);
+ dim_n_cats[i] = dump_dim (i);
}
static void
dump_data(void)
{
/* The first three numbers add to the number of dimensions. */
- int t = get_u32();
- t += get_u32();
- match_u32_assert(n_dims - t);
+ 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++)
- a[i] = get_u32();
+ {
+ 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");
int x = get_u32();
printf ("\n");
for (int i = 0; i < x; i++)
{
- printf (" \n", get_u32());
+ unsigned int indx = get_u32();
+ printf (" \n");
match_u32_assert(0);
if (version == 1)
match_byte(0);
@@ -748,6 +792,24 @@ dump_title(void)
}
}
+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)
{
@@ -817,25 +879,21 @@ dump_fonts(void)
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)
+ fprintf(stderr, "%.*s\n", len, &data[pos + 0x35]);
}
+ pos += x3;
int count = get_u32();
pos += 4 * count;
- const char *encoding = get_string();
- printf ("%s\n", encoding);
+ const char *locale = get_string();
+ printf ("%s\n", locale);
if (!match_u32(0))
match_u32_assert(UINT32_MAX);
@@ -879,8 +937,105 @@ dump_fonts(void)
}
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_assert(4);
+ match_byte_assert(0);
+ match_byte_assert(0);
+ match_byte_assert(0);
+
+ printf("%s\n", get_string());
+ printf("%s\n", get_string());
+ printf("%s\n", get_string());
+ printf("%s\n", get_string());
+ printf("%s\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_assert(0x9a);
+ 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("%s\n", get_string());
+ printf("%s\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("%s\n", 'A' + i, get_string(), 'A' + i);
+ }
+ else
+ match_u32_assert(0);
+
+ match_byte_assert(0x2e);
+ if (!match_byte(0))
+ match_byte_assert(1);
+
+ assert(pos == outer_end);
+
+ pos = outer_end;
+ }
+ else
+ match_u32_assert(0);
}
int