dump: All output into pseudo-XML.

diff --git a/dump.c b/dump.c
index e1446edeb8189b3094742fdbc866013a4c054a11..64ddcfdda85c16d6c978b89d5bfa73d3f3cb62de 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -26,13 +26,6 @@ try_find(const char *target, size_t target_len)
   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)
 {
@@ -45,18 +38,6 @@ 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
@@ -524,7 +505,10 @@ check_permutation(int *a, int n, const char *name)
 static void
 dump_category(int level, int *indexes, int *n_indexes)
 {
-  dump_value__ (stdout, level, true);
+  for (int i = 0; i <= level; i++)
+    fprintf (stdout, "    ");
+  printf ("<category>\n");
+  dump_value__ (stdout, level + 1, true);
   match_byte(0);
   match_byte(0);
   match_byte(0);
@@ -552,20 +536,25 @@ dump_category(int level, int *indexes, int *n_indexes)
         }
       indexes[(*n_indexes)++] = indx;
     }
-  if (n_categories > 0)
-    printf (", %d subcategories:", n_categories);
-  else
-    printf (", index %d", indx);
-  printf("\n");
+  if (n_categories == 0)
+    {
+      for (int i = 0; i <= level + 1; i++)
+        fprintf (stdout, "    ");
+      fprintf (stdout, "<category-index>%d</category-index>\n", indx);
+    }
   for (int i = 0; i < n_categories; i++)
     dump_category (level + 1, indexes, n_indexes);
+  for (int i = 0; i <= level; i++)
+    fprintf (stdout, "    ");
+  printf ("</category>\n");
 }
 
 static void
-dump_dim(void)
+dump_dim(int indx)
 {
   int n_categories;
-  printf("next dim\n");
+
+  printf ("<dimension index=\"%d\">\n", indx);
   dump_value__ (stdout, 0, false);
 
   /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
@@ -580,16 +569,16 @@ dump_dim(void)
   if (!match_byte(0))
     match_byte_assert(1);
   match_byte_assert(1);
-  static int dim_indx = 0;
-  match_u32_assert(dim_indx++);
+  match_u32_assert(indx);
   n_categories = get_u32();
-  printf("%d nested categories\n", n_categories);
 
   int indexes[1024];
   int n_indexes = 0;
   for (int i = 0; i < n_categories; i++)
     dump_category (0, indexes, &n_indexes);
   check_permutation(indexes, n_indexes, "categories");
+
+  fprintf (stdout, "</dimension>\n");
 }
 
 int n_dims;
@@ -597,12 +586,8 @@ static void
 dump_dims(void)
 {
   n_dims = get_u32();
-  printf ("%u dimensions\n", n_dims);
   for (int i = 0; i < n_dims; i++)
-    {
-      printf("\n");
-      dump_dim ();
-    }
+    dump_dim (i);
 }
 
 static void
@@ -620,93 +605,78 @@ dump_data(void)
   check_permutation(a, n_dims, "dimensions");
 
   int x = get_u32();
-  printf ("%d data values, starting at %08x\n", x, pos);
+  printf ("<data>\n");
   for (int i = 0; i < x; i++)
     {
-      printf("%08x, index %d:\n", pos, get_u32());
+      printf ("    <datum index=\"%d\">\n", get_u32());
       match_u32_assert(0);
-      dump_value__(stdout, 0, false);
-      putchar('\n');
+      dump_value__(stdout, 1, false);
+      fprintf (stdout, "    </datum>\n");
     }
+  printf ("</data>\n");
 }
 
 static void
 dump_title(void)
 {
   pos = 0x27;
-  printf("text:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
-  printf("subtype:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
+  printf ("<title-local>\n");
+  dump_value__(stdout, 0, true);
+  printf ("</title-local>\n");
+
+  printf ("<subtype>\n");
+  dump_value__(stdout, 0, true);
+  printf ("</subtype>\n");
+
   match_byte_assert(0x31);
-  printf("text_eng:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
+
+  printf ("<title-c>\n");
+  dump_value__(stdout, 0, true);
+  printf ("</title-c>\n");
+
   match_byte(0);
   match_byte_assert(0x58);
   if (match_byte(0x31))
     {
-      printf("caption:\n");
-      dump_value__(stdout, 0, false); putchar('\n');
+      printf ("<caption>\n");
+      dump_value__(stdout, 0, false);
+      printf ("</caption>\n");
     }
   else
     match_byte_assert(0x58);
 
 
   int n_footnotes = get_u32();
-  if (n_footnotes >= 20)
+  for (int i = 0; i < n_footnotes; i++)
     {
-      fprintf(stderr, "%08x: %d footnotes\n", pos - 4, n_footnotes);
-      exit(1);
-    }
-
-  printf("------\n%d footnotes\n", n_footnotes);
-  if (n_footnotes < 20)
-    {
-      for (int i = 0; i < n_footnotes; i++)
+      printf ("<footnote index=\"%d\">\n", i);
+      dump_value__(stdout, 0, false);
+      if (match_byte (0x31))
         {
-          printf("footnote %d:\n", i);
-          dump_value__(stdout, 0, false);
-          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());
+          /* 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>\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());
+      printf(" font=\"%s\"", get_string());
       match_byte_assert(0);
       match_byte_assert(0);
       if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10))
@@ -723,8 +693,8 @@ dump_fonts(void)
 
       if (!match_u32(0) && !match_u32(1) && !match_u32(2))
         match_u32_assert(3);
-      printf ("%s, ", get_string());
-      printf ("%s, ", get_string());
+      printf (" fgcolor=\"%s\"", get_string());
+      printf (" bgcolor=\"%s\"", get_string());
       match_u32_assert(0);
       match_u32_assert(0);
       match_byte_assert(0);
@@ -748,6 +718,8 @@ dump_fonts(void)
 
       /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
       get_u32();
+
+      printf ("/>\n");
     }
 
   match_u32_assert(240);
@@ -767,8 +739,7 @@ dump_fonts(void)
   int count = get_u32();
   pos += 4 * count;
 
-  char *encoding = get_string();
-  printf("encoding=%s\n", encoding);
+  printf ("<encoding>%s</encoding>\n", get_string ());
 
   if (!match_u32(0))
     match_u32_assert(UINT32_MAX);
@@ -794,8 +765,6 @@ dump_fonts(void)
   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
@@ -884,7 +853,6 @@ main(int argc, char *argv[])
           dump_title ();
           dump_fonts();
           dump_dims ();
-          printf("\n\ndata:\n");
           dump_data ();
           match_byte (1);
           if (pos != n)