X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=04f38e0a751bf41747130472385c1febe6c8c96f;hb=8af10bb39253b97589c5f4b455b708c8fb9e233b;hp=0445597e635aa6673304d1fa734df0cc29371edc;hpb=463238cd3f894fc6fb5cebbcc7bb2b9584c640a8;p=pspp
diff --git a/dump.c b/dump.c
index 0445597e63..04f38e0a75 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
{
@@ -788,6 +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 0
+ 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)
{
@@ -874,8 +896,8 @@ dump_fonts(void)
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);
@@ -919,8 +941,105 @@ 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();
+
+ /* Skip first inner envelope. */
+ int inner_len = get_u32();
+ hex_dump(pos, inner_len);
+ pos += inner_len;
+
+ 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)
+ {
+ uint8_t *endp = memmem(&data[pos], outer_end - pos, "\5\0\0\0\4", 5);
+ assert(endp);
+
+ int end = endp - data;
+ if (pos != end)
+ {
+ printf("%s\n", get_string());
+ printf("%s\n", get_string());
+
+ if (pos != end)
+ {
+ match_u32_assert(0);
+ get_u32();
+ match_u32_assert(0);
+ }
+ }
+ //fprintf(stderr, "%d\n", end - pos);
+ //hex_dump(pos, end - pos);
+ assert(pos == end);
+ }
+
+ 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
+ pos = get_end();
}
int