dump: Title section sometimes has 0x31 plus fixed block in place of 0x58.

[pspp] / dump.c

diff --git a/dump.c b/dump.c
index 64ddcfdda85c16d6c978b89d5bfa73d3f3cb62de..140050d6cba84956d5bd32ff594f2d7fe1baa664 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -107,7 +107,13 @@ match_byte_assert(uint8_t b, const char *where)
 #define match_byte_assert(b) match_byte_assert(b, WHERE)
 
 static void
-dump_raw(FILE *stream, int start, int end, const char *separator)
+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; )
     {
@@ -119,7 +125,8 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           && i + 4 + data[i] + data[i + 1] * 256 <= end
           && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
         {
-          fprintf(stream, "%s\"", separator);
+          newline(stream, i);
+          fprintf(stream, "\"");
           fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
           fputs("\" ", stream);
 
@@ -133,8 +140,9 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           double d;
 
           memcpy (&d, &data[i + 4], 8);
-          fprintf (stream, "F40.%d(%.*f)%s", data[i], data[i], d, separator);
+          fprintf (stream, "F40.%d(%.*f)", data[i], data[i], d);
           i += 12;
+          newline (stream, i);
         }
       else if (i + 12 <= end
                && data[i + 1] == 40
@@ -144,8 +152,9 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           double d;
 
           memcpy (&d, &data[i + 4], 8);
-          fprintf (stream, "PCT40.%d(%.*f)%s", data[i], data[i], d, separator);
+          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)
@@ -258,6 +267,7 @@ dump_value_31(FILE *stream)
           if (match_byte(0x31))
             {
               /* Only one example in the corpus. */
+              match_byte(1);
               match_byte(0);
               match_byte(0);
               match_byte(0);
@@ -265,11 +275,30 @@ dump_value_31(FILE *stream)
               get_string();     /* foreground */
               get_string();     /* background */
               get_string();     /* font */
-              match_byte_assert(12); /* size? */
+              if (!match_byte(14))
+                match_byte_assert(12); /* size? */
+            }
+          else
+            match_byte_assert(0x58);
+          if (match_byte(0x31))
+            {
+              /* All the examples in the corpus, all from one SPV file, are
+                 exactly like this. */
+              match_u32_assert(0);
+              match_u32_assert(0);
+              match_u32_assert(0);
+              match_u32_assert(0);
+              match_byte_assert(1);
+              match_byte_assert(0);
+              match_byte_assert(8);
+              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);
-          match_byte_assert(0x58);
           if (pos != outer_end)
             {
               fprintf(stderr, "outer end discrepancy\n");
@@ -355,7 +384,7 @@ format_to_string (int type)
 }
 
 static void
-dump_value__(FILE *stream, int level, bool match1)
+dump_value(FILE *stream, int level, bool match1)
 {
   match_byte(0);
   match_byte(0);
@@ -466,7 +495,7 @@ dump_value__(FILE *stream, int level, bool match1)
             fprintf (stream, "    ");
           fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
           for (int j = 0; j < y; j++)
-            dump_value__ (stream, level + 2, false);
+            dump_value (stream, level + 2, false);
           for (int j = 0; j <= level + 1; j++)
             fprintf (stream, "    ");
           fprintf (stream, "</substitution>\n");
@@ -508,7 +537,7 @@ dump_category(int level, int *indexes, int *n_indexes)
   for (int i = 0; i <= level; i++)
     fprintf (stdout, "    ");
   printf ("<category>\n");
-  dump_value__ (stdout, level + 1, true);
+  dump_value (stdout, level + 1, true);
   match_byte(0);
   match_byte(0);
   match_byte(0);
@@ -555,7 +584,7 @@ dump_dim(int indx)
   int n_categories;
 
   printf ("<dimension index=\"%d\">\n", indx);
-  dump_value__ (stdout, 0, false);
+  dump_value (stdout, 0, false);
 
   /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
   pos++;
@@ -610,7 +639,7 @@ dump_data(void)
     {
       printf ("    <datum index=\"%d\">\n", get_u32());
       match_u32_assert(0);
-      dump_value__(stdout, 1, false);
+      dump_value(stdout, 1, false);
       fprintf (stdout, "    </datum>\n");
     }
   printf ("</data>\n");
@@ -621,17 +650,17 @@ dump_title(void)
 {
   pos = 0x27;
   printf ("<title-local>\n");
-  dump_value__(stdout, 0, true);
+  dump_value(stdout, 0, true);
   printf ("</title-local>\n");
 
   printf ("<subtype>\n");
-  dump_value__(stdout, 0, true);
+  dump_value(stdout, 0, true);
   printf ("</subtype>\n");
 
   match_byte_assert(0x31);
 
   printf ("<title-c>\n");
-  dump_value__(stdout, 0, true);
+  dump_value(stdout, 0, true);
   printf ("</title-c>\n");
 
   match_byte(0);
@@ -639,7 +668,7 @@ dump_title(void)
   if (match_byte(0x31))
     {
       printf ("<caption>\n");
-      dump_value__(stdout, 0, false);
+      dump_value(stdout, 0, false);
       printf ("</caption>\n");
     }
   else
@@ -650,7 +679,7 @@ dump_title(void)
   for (int i = 0; i < n_footnotes; i++)
     {
       printf ("<footnote index=\"%d\">\n", i);
-      dump_value__(stdout, 0, false);
+      dump_value(stdout, 0, false);
       if (match_byte (0x31))
         {
           /* Custom footnote marker string. */
@@ -748,8 +777,8 @@ dump_fonts(void)
   match_byte_assert(0);
   if (!match_byte(0))
     match_byte_assert(1);
-  if (!match_byte(0x99) && !match_byte(0x98))
-    match_byte_assert(0x97);
+  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);
@@ -871,7 +900,7 @@ main(int argc, char *argv[])
   else
     start = 0x27;
 
-  dump_raw(stdout, start, n, "\n");
+  dump_raw(stdout, start, n);
 
   return 0;
 }