X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=73935a73374faf7df3695a23436ad8c2506385ba;hb=7674958d6669183799289f701e1148b6903b801a;hp=140050d6cba84956d5bd32ff594f2d7fe1baa664;hpb=a787bb1da9d64940bca6fa37209108addfbad54d;p=pspp

diff --git a/dump.c b/dump.c
index 140050d6cb..73935a7337 100644
--- a/dump.c
+++ b/dump.c
@@ -10,6 +10,8 @@
 static uint8_t *data;
 static size_t n;
 
+int version;
+
 static bool
 all_ascii(const uint8_t *p, size_t n)
 {
@@ -246,22 +248,41 @@ dump_value_31(FILE *stream)
           else
             match_u32_assert (0);
 
-          int outer_end = pos + get_u32();
-          int inner_end = pos + get_u32();
-          match_u32_assert(0);
-          if (match_byte(0x31))
+          if (version == 1)
             {
-              /* 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();
+              /* We only have one SPV file for this version (with many
+                 tables). */
+              match_byte(0);
+              if (!match_u32(1))
+                match_u32_assert(2);
+              match_byte(0);
+              match_byte(0);
+              if (!match_u32(0) && !match_u32(1) && !match_u32(2) && !match_u32(3) && !match_u32(4) && !match_u32(5) && !match_u32(6) && !match_u32(7) && !match_u32(8) && !match_u32(9))
+                match_u32_assert(10);
+              match_byte(0);
+              match_byte(0);
+              return;
             }
-          else
-            match_byte_assert(0x58);
+
+          int outer_end = pos + get_u32();
+          int inner_end = pos + get_u32();
           if (pos != inner_end)
             {
-              fprintf(stderr, "inner end discrepancy\n");
-              exit(1);
+              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);
+                }
             }
 
           if (match_byte(0x31))
@@ -282,15 +303,16 @@ dump_value_31(FILE *stream)
             match_byte_assert(0x58);
           if (match_byte(0x31))
             {
-              /* All the examples in the corpus, all from one SPV file, are
-                 exactly like this. */
+              /* Only two SPV files have anything like this, so it's hard to
+                 generalize. */
               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);
+              if (!match_byte(8) && !match_byte(1))
+                match_byte_assert(2);
               match_byte_assert(0);
               match_byte_assert(8);
               match_byte_assert(0);
@@ -313,10 +335,11 @@ dump_value_31(FILE *stream)
       else if (match_u32 (2))
         {
           fprintf(stream, "(special 2)");
+          if (!match_byte(0))
+            match_byte_assert(2);
           match_byte_assert(0);
-          match_byte_assert(0);
-          if (!match_u32 (2))
-            match_u32_assert(1);
+          if (!match_u32 (2) && !match_u32(1))
+            match_u32_assert(3);
           dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
         }
       else
@@ -429,7 +452,7 @@ dump_value(FILE *stream, int level, bool match1)
       char *var, *vallab;
       double value;
 
-      match_byte_assert (0x58);
+      dump_value_31 (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -532,7 +555,7 @@ check_permutation(int *a, int n, const char *name)
 }
 
 static void
-dump_category(int level, int *indexes, int *n_indexes)
+dump_category(int level, int *indexes, int *n_indexes, int max_indexes)
 {
   for (int i = 0; i <= level; i++)
     fprintf (stdout, "    ");
@@ -563,6 +586,11 @@ dump_category(int level, int *indexes, int *n_indexes)
           fprintf(stderr, "index not -1 but subcategories\n");
           exit(1);
         }
+      if (*n_indexes >= max_indexes)
+        {
+          fprintf(stderr, "too many categories (increase max_indexes)\n");
+          exit(1);
+        }
       indexes[(*n_indexes)++] = indx;
     }
   if (n_categories == 0)
@@ -572,7 +600,7 @@ dump_category(int level, int *indexes, int *n_indexes)
       fprintf (stdout, "<category-index>%d</category-index>\n", indx);
     }
   for (int i = 0; i < n_categories; i++)
-    dump_category (level + 1, indexes, n_indexes);
+    dump_category (level + 1, indexes, n_indexes, max_indexes);
   for (int i = 0; i <= level; i++)
     fprintf (stdout, "    ");
   printf ("</category>\n");
@@ -586,7 +614,7 @@ dump_dim(int indx)
   printf ("<dimension index=\"%d\">\n", indx);
   dump_value (stdout, 0, false);
 
-  /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
+  /* This byte is usually 0x02 but many other values have been spotted. */
   pos++;
 
   if (!match_byte(0) && !match_byte(1))
@@ -598,13 +626,14 @@ dump_dim(int indx)
   if (!match_byte(0))
     match_byte_assert(1);
   match_byte_assert(1);
-  match_u32_assert(indx);
+  if (!match_u32(UINT32_MAX))
+    match_u32_assert(indx);
   n_categories = get_u32();
 
-  int indexes[1024];
+  int indexes[2048];
   int n_indexes = 0;
   for (int i = 0; i < n_categories; i++)
-    dump_category (0, indexes, &n_indexes);
+    dump_category (0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
   check_permutation(indexes, n_indexes, "categories");
 
   fprintf (stdout, "</dimension>\n");
@@ -639,6 +668,8 @@ dump_data(void)
     {
       printf ("    <datum index=\"%d\">\n", get_u32());
       match_u32_assert(0);
+      if (version == 1)
+        match_byte(0);
       dump_value(stdout, 1, false);
       fprintf (stdout, "    </datum>\n");
     }
@@ -708,12 +739,12 @@ dump_fonts(void)
       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))
+      if (!match_byte(0x40) && !match_byte(0x20) && !match_byte(0x80) && !match_byte(0x10) && !match_byte(0x70))
         match_byte_assert(0x50);
       if (!match_byte(0x41))
         match_byte_assert(0x51);
-      if (!match_u32(0))
-        match_u32_assert(1);
+      if (!match_u32(0) && !match_u32(1))
+        match_u32_assert(2);
       match_byte_assert(0);
 
       /* OK, this seems really unlikely to be totally correct, but it matches my corpus... */
@@ -728,26 +759,31 @@ dump_fonts(void)
       match_u32_assert(0);
       match_byte_assert(0);
 
-      /* These seem unlikely to be correct too. */
-      if (i != 3)
-        {
-          match_u32_assert(8);
-          if (!match_u32(10))
-            match_u32_assert(11);
-          match_u32_assert(1);
-        }
-      else
+      if (version > 1)
         {
+          /* These seem unlikely to be correct too. */
+          if (i != 3)
+            {
+              if (!match_u32(8))
+                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);
+            }
+          else
+            {
+              get_u32();
+              if (!match_u32(-1) && !match_u32(8))
+                match_u32_assert(24);
+              if (!match_u32(-1) && !match_u32(2))
+                match_u32_assert(3);
+            }
+
+          /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
           get_u32();
-          if (!match_u32(-1) && !match_u32(8))
-            match_u32_assert(24);
-          if (!match_u32(-1) && !match_u32(2))
-            match_u32_assert(3);
         }
 
-      /* Who knows? Ranges from -1 to 8 with no obvious pattern. */
-      get_u32();
-
       printf ("/>\n");
     }
 
@@ -759,10 +795,16 @@ dump_fonts(void)
 
   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
     {
-      match_u32_assert(142);
-      pos += 142;
+      match_u32_assert(16);
+      pos += 16;
     }
 
   int count = get_u32();
@@ -777,23 +819,36 @@ dump_fonts(void)
   match_byte_assert(0);
   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 (version > 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);
+    }
+  else
+    match_u32_assert(UINT32_MAX);
   if (match_byte('.'))
-    match_byte_assert(',');
+    {
+      if (!match_byte(','))
+        match_byte_assert(' ');
+    }
   else
     {
       match_byte_assert(',');
-      if (!match_byte('.'))
-        match_byte_assert(' ');
+      if (!match_byte('.') && !match_byte(' '))
+        match_byte_assert(0);
     }
-  match_u32_assert(5);
-  for (int i = 0; i < 5; i++)
-    get_string();
-  pos += get_u32();
+  if (match_u32(5))
+    {
+      for (int i = 0; i < 5; i++)
+        get_string();
+    }
+  else
+    match_u32_assert(0);
+  int skip = get_u32();
+  pos += skip;
 }
 
 int
@@ -867,7 +922,15 @@ main(int argc, char *argv[])
           pos = 0;
           match_byte_assert(1);
           match_byte_assert(0);
-          match_u32_assert(3);
+
+          /* This might be a version number of some kind, because value 1 seems
+             to only appear in an SPV file that also required its own weird
+             special cases in dump_value_31(). */
+          version = get_u32();
+          pos -= 4;
+          if (!match_u32(1))
+            match_u32_assert(3);
+
           match_byte_assert(1);
           if (!match_byte(0))
             match_byte_assert(1);