Some minor refinements from expanding the corpus.

[pspp] / dump.c

diff --git a/dump.c b/dump.c
index 04f38e0a751bf41747130472385c1febe6c8c96f..be5139c6045a5eb2620242d9b2ca4212243e724d 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -246,6 +246,39 @@ dump_counted_string(void)
   return s;
 }
 
+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_style(void)
+{
+  match_byte(1);
+  match_byte(0);
+  match_byte(0);
+  match_byte(0);
+  match_byte_assert(1);
+  get_string();     /* foreground */
+  get_string();     /* background */
+  get_string();     /* font */
+  if (!match_byte(14))
+    match_byte_assert(12); /* size? */
+}
+
 static char *
 dump_nested_string(void)
 {
@@ -255,7 +288,10 @@ dump_nested_string(void)
   match_byte_assert (0);
   int outer_end = get_end();
   s = dump_counted_string();
-  match_byte_assert(0x58);
+  if (match_byte(0x31))
+    dump_style();
+  else
+    match_byte_assert(0x58);
   match_byte_assert(0x58);
   if (pos != outer_end)
     {
@@ -308,19 +344,7 @@ dump_optional_value(FILE *stream)
           dump_counted_string();
 
           if (match_byte(0x31))
-            {
-              /* Only one example in the corpus. */
-              match_byte(1);
-              match_byte(0);
-              match_byte(0);
-              match_byte(0);
-              match_byte_assert(1);
-              get_string();     /* foreground */
-              get_string();     /* background */
-              get_string();     /* font */
-              if (!match_byte(14))
-                match_byte_assert(12); /* size? */
-            }
+            dump_style();
           else
             match_byte_assert(0x58);
           if (match_byte(0x31))
@@ -357,7 +381,7 @@ dump_optional_value(FILE *stream)
       else if (match_u32 (2))
         {
           fprintf(stream, "(special 2)");
-          if (!match_byte(0))
+          if (!match_byte(0) && !match_byte(1))
             match_byte_assert(2);
           match_byte_assert(0);
           if (!match_u32 (2) && !match_u32(1))
@@ -792,24 +816,6 @@ 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)
 {
@@ -854,8 +860,8 @@ dump_fonts(void)
                 match_u32_assert(5);
               if (!match_u32(10) && !match_u32(11) && !match_u32(5))
                 match_u32_assert(9);
-              if (!match_u32(0))
-                match_u32_assert(1);
+              if (!match_u32(0) && !match_u32(1))
+                match_u32_assert(2);
             }
           else
             {
@@ -879,19 +885,15 @@ 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)
+        printf("<tablelook>%.*s</tablelook>\n", len, &data[pos + 0x35]);
     }
+  pos += x3;
 
   int count = get_u32();
   pos += 4 * count;
@@ -908,8 +910,9 @@ dump_fonts(void)
     match_byte_assert(1);
   if (version > 1)
     {
-      if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
-        match_byte_assert(0x9a);
+      if (!match_byte(0x97) && !match_byte(0x98)
+          && !match_byte(0x99) && !match_byte(0x9a))
+        match_byte_assert(0x9b);
       match_byte_assert(7);
       match_byte_assert(0);
       match_byte_assert(0);
@@ -949,17 +952,16 @@ dump_fonts(void)
     {
       int outer_end = get_end();
 
-      /* Skip first inner envelope. */
-      int inner_len = get_u32();
-      hex_dump(pos, inner_len);
-      pos += inner_len;
+      /* 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);
+      if (!match_byte(3) && !match_byte(4))
+        match_byte_assert(5);
       match_byte_assert(0);
       match_byte_assert(0);
       match_byte_assert(0);
@@ -978,8 +980,9 @@ dump_fonts(void)
       if (!match_byte(0))
         match_byte_assert(1);
 
-      if (!match_byte(0x97) && !match_byte(0x98) && !match_byte(0x99))
-        match_byte_assert(0x9a);
+      if (!match_byte(0x97) && !match_byte(0x98)
+          && !match_byte(0x99) && !match_byte(0x9a))
+        match_byte_assert(0x9b);
       match_byte_assert(7);
       match_byte_assert(0);
       match_byte_assert(0);
@@ -1001,25 +1004,26 @@ dump_fonts(void)
 
       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)
+          /* 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());
 
-              if (pos != end)
-                {
-                  match_u32_assert(0);
-                  get_u32();
-                  match_u32_assert(0);
-                }
+              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))
@@ -1034,12 +1038,17 @@ dump_fonts(void)
       if (!match_byte(0))
         match_byte_assert(1);
 
+      if (pos < outer_end)
+        {
+          match_u32_assert(2000000);
+          match_u32_assert(0);
+        }
       assert(pos == outer_end);
 
       pos = outer_end;
     }
   else
-    pos = get_end();
+    match_u32_assert(0);
 }
 
 int