dump: Relax a constraint to accept a few more spv files.

[pspp] / dump.c

diff --git a/dump.c b/dump.c
index dab02a412079dd5487ba0fa772556f8bedb41c71..7994db13c19517bb33a66c5a71586f48d8722b2e 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -19,15 +19,41 @@ all_ascii(const uint8_t *p, size_t n)
 }
 
 static size_t
-find(const char *target, size_t target_len)
+try_find(const char *target, size_t target_len)
 {
   const uint8_t *pos = (const uint8_t *) memmem (data, n, target, 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)
+{
+  size_t pos = try_find(target, target_len);
   if (!pos)
     {
       fprintf (stderr, "not found\n");
       exit(1);
     }
-  return pos - data;
+  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;
@@ -99,10 +125,10 @@ match_byte_assert(uint8_t b, const char *where)
 #define match_byte_assert(b) match_byte_assert(b, WHERE)
 
 static char *
-get_string(void)
+get_string(const char *where)
 {
   if (data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0
-      && all_ascii(&data[pos + 4], data[pos]))
+      /*&& all_ascii(&data[pos + 4], data[pos])*/)
     {
       int len = data[pos];
       char *s = malloc(len + 1);
@@ -114,25 +140,51 @@ get_string(void)
     }
   else
     {
-      fprintf(stderr, "0x%x: expected string\n", pos);
+      fprintf(stderr, "%s: 0x%x: expected string\n", where, pos);
       exit(1);
     }
 }
+#define get_string() get_string(WHERE)
 
 static void
-dump_category(int level)
+dump_value(int level)
 {
   for (int i = 0; i <= level; i++)
     printf ("    ");
 
-  match_byte (0);
   if (match_byte (3))
     {
       get_string();
-      match_byte_assert (0x58);
+      if (match_byte (0x31))
+        {
+          if (match_u32 (1))
+            {
+              printf("(footnote %d) ", get_u32());
+              match_byte_assert (0);
+              match_byte_assert (0);
+            }
+          else
+            {
+              match_u32_assert (2);
+              printf("(special 2)");
+              match_byte_assert(0);
+              match_byte_assert(0);
+              match_u32_assert(1);
+              match_byte_assert(0);
+              match_byte_assert(0);
+            }
+          int subn = get_u32 ();
+          printf ("nested %d bytes", subn);
+          pos += subn;
+        }
+      else
+        match_byte_assert (0x58);
       get_string();
       printf("string \"%s\"", get_string());
-      match_byte_assert (1);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (1);
       match_byte (0);
       match_byte (0);
       match_byte (0);
@@ -148,6 +200,7 @@ dump_category(int level)
       match_byte (0);
       match_byte (0);
       match_byte (0);
+      match_byte (0);
     }
   else if (match_byte (2))
     {
@@ -164,13 +217,133 @@ dump_category(int level)
               value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
       if (!match_u32 (3))
         match_u32_assert (2);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
     }
-  else
+  else if (match_byte (4))
+    {
+      unsigned int format;
+      char *var, *vallab, *value;
+
+      match_byte_assert (0x58);
+      format = get_u32 ();
+      vallab = get_string ();
+      var = get_string ();
+      match_byte_assert (2);
+      value = get_string ();
+      printf ("value \"%s\" format %d(%d.%d) var \"%s\" vallab \"%s\"",
+              value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+    }
+  else if (match_byte (1))
     {
       unsigned int format;
       double value;
 
+      match_byte_assert (0x58);
+      format = get_u32 ();
+      value = get_double ();
+      printf ("value %g format %d(%d.%d)", value, format >> 16, (format >> 8) & 0xff, format & 0xff);
+      match_byte (1);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (1);
+    }
+  else
+    {
+      int subn;
+      int total_subs = 1;
+
+      match_byte (0);
+      match_byte_assert (0x31);
+      match_u32_assert (0);
+      match_u32_assert (0);
+      subn = get_u32 ();
+      printf ("nested %d bytes", subn);
+      pos += subn;
+      printf ("; \"%s\", substitutions:", get_string());
+      for (;;)
+        {
+          int n_subst = get_u32();
+          if (!n_subst)
+            break;
+          printf (" %d", n_subst);
+          total_subs *= n_subst;
+        }
+
+      for (int i = 0; i < total_subs; i++)
+        {
+          putc ('\n', stdout);
+          dump_value (level + 1);
+        }
+    }
+}
+
+static void
+dump_dim_value(int level)
+{
+  for (int i = 0; i <= level; i++)
+    printf ("    ");
+
+  if (match_byte (3))
+    {
+      get_string();
+      if (match_byte (0x31))
+        {
+          match_u32 (1);
+          printf("(footnote %d) ", get_u32());
+          match_byte_assert (0);
+          match_byte_assert (0);
+          int subn = get_u32 ();
+          printf ("nested %d bytes", subn);
+          pos += subn;
+        }
+      else
+        match_byte_assert (0x58);
+      get_string();
+      printf("string \"%s\"", get_string());
+      match_byte (0);
       match_byte_assert (1);
+      match_byte (0);
+      match_byte (0);
+      match_byte (0);
+      match_byte (1);
+    }
+  else if (match_byte (5))
+    {
+      match_byte_assert (0x58);
+      printf ("variable \"%s\"", get_string());
+      get_string();
+      match_byte_assert (2);
+    }
+  else if (match_byte (2))
+    {
+      unsigned int format;
+      char *var, *vallab;
+      double value;
+
+      match_byte_assert (0x58);
+      format = get_u32 ();
+      value = get_double ();
+      var = get_string ();
+      vallab = get_string ();
+      printf ("value %g format %d(%d.%d) var \"%s\" vallab \"%s\"",
+              value, format >> 16, (format >> 8) & 0xff, format & 0xff, var, vallab);
+      if (!match_u32 (3))
+        match_u32_assert (2);
+      match_byte (0);
+    }
+  else if (match_byte (1))
+    {
+      unsigned int format;
+      double value;
+
       match_byte_assert (0x58);
       format = get_u32 ();
       value = get_double ();
@@ -181,6 +354,44 @@ dump_category(int level)
       match_byte (0);
       match_byte (1);
     }
+  else
+    {
+      int subn;
+      int total_subs = 1;
+
+      match_byte (0);
+      match_byte_assert (0x31);
+      match_u32_assert (0);
+      match_u32_assert (0);
+      subn = get_u32 ();
+      printf ("nested %d bytes", subn);
+      pos += subn;
+      printf ("; \"%s\", substitutions:", get_string());
+      for (;;)
+        {
+          int n_subst = get_u32();
+          if (!n_subst)
+            break;
+          printf (" %d", n_subst);
+          total_subs *= n_subst;
+        }
+
+      for (int i = 0; i < total_subs; i++)
+        {
+          putc ('\n', stdout);
+          dump_value (level + 1);
+        }
+    }
+}
+
+static void
+dump_category(int level)
+{
+  match_byte (0);
+  match_byte (0);
+  match_byte (0);
+  match_byte (0);
+  dump_value (level);
 
   if (match_u32 (2))
     get_u32 ();
@@ -191,11 +402,19 @@ dump_category(int level)
       match_byte (0);
       get_u32 ();
     }
+  else if (match_byte (1))
+    {
+      match_byte (0);
+      match_u32_assert (1);
+      match_byte (0);
+      get_u32();
+    }
   else
     {
       match_u32_assert (0);
       get_u32 ();
     }
+
   int n_categories = get_u32();
   if (n_categories > 0)
     printf (", %d subcategories:", n_categories);
@@ -208,6 +427,8 @@ static void
 dump_dim(void)
 {
   int n_categories;
+  printf("next dim\n");
+  match_byte(0);
   if (match_byte(3))
     {
       get_string();
@@ -216,7 +437,7 @@ dump_dim(void)
       printf("string \"%s\": ", get_string());
       match_byte_assert(1);
     }
-  else if (match_byte(5))
+  else if (match_byte(5)) 
     {
       match_byte_assert(0x58);
       printf("variable \"%s\": ", get_string());
@@ -226,8 +447,30 @@ dump_dim(void)
     }
   else
     {
-      fprintf(stderr, "%08x: unexpected byte\n", pos);
-      exit(1);
+      int subn;
+      int total_subs = 1;
+
+      match_byte_assert(0x31);
+      match_u32_assert (0);
+      match_u32_assert (0);
+      subn = get_u32 ();
+      printf ("nested %d bytes", subn);
+      pos += subn;
+      printf ("; \"%s\", substitutions:", get_string());
+      for (;;)
+        {
+          int n_subst = get_u32();
+          if (!n_subst)
+            break;
+          printf (" %d", n_subst);
+          total_subs *= n_subst;
+        }
+
+      for (int i = 0; i < total_subs; i++)
+        {
+          putc ('\n', stdout);
+          dump_dim_value (0);
+        }
     }
 
   match_byte_assert(0);
@@ -295,6 +538,16 @@ main(int argc, char *argv[])
 
   if (argc > 1)
     {
+      if (!strcmp(argv[1], "title0"))
+        {
+          pos = 0x27;
+          if (match_byte (0x03)
+              || (match_byte (0x05) && match_byte (0x58)))
+            printf ("%s\n", get_string());
+          else
+            printf ("<unknown>\n");
+          return 0;
+        }
       if (!strcmp(argv[1], "title"))
         {
           const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
@@ -317,8 +570,17 @@ main(int argc, char *argv[])
         }
       else if (!strcmp(argv[1], "dimensions"))
         {
-          const char dimensions[] = "-,,,.\0";
-          start = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
+          {
+            const char dimensions[] = "-,,,.\0";
+            start = try_find_tail(dimensions, sizeof dimensions - 1);
+          }
+
+          if (!start)
+            {
+              const char dimensions[] = "-,,, .\0";
+              start = find_tail(dimensions, sizeof dimensions - 1);
+            }
+
           pos = start;
           dump_dims ();
           return 0;