X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=c737c01d1dd4f8252967d8e8b66785f992cf9a70;hb=b84bc4577a96fdafefda66b414fa0eae35e81331;hp=4d7d7f468e694bfa11a59900f45c46920dd84146;hpb=6a45803579c7f58e62668306240784beec7a611a;p=pspp

diff --git a/dump.c b/dump.c
index 4d7d7f468e..c737c01d1d 100644
--- a/dump.c
+++ b/dump.c
@@ -48,6 +48,12 @@ size_t pos;
 #define STR(x) XSTR(x)
 #define WHERE __FILE__":" STR(__LINE__)
 
+static uint8_t
+get_byte(void)
+{
+  return data[pos++];
+}
+
 static unsigned int
 get_u32(void)
 {
@@ -57,6 +63,24 @@ get_u32(void)
   return x;
 }
 
+static unsigned int
+get_be32(void)
+{
+  uint32_t x;
+  x = (data[pos] << 24) | (data[pos + 1] << 16) | (data[pos + 2] << 8) | data[pos + 3];
+  pos += 4;
+  return x;
+}
+
+static unsigned int
+get_u16(void)
+{
+  uint16_t x;
+  memcpy(&x, &data[pos], 2);
+  pos += 2;
+  return x;
+}
+
 static double
 get_double(void)
 {
@@ -66,6 +90,15 @@ get_double(void)
   return x;
 }
 
+static double __attribute__((unused))
+get_float(void)
+{
+  float x;
+  memcpy(&x, &data[pos], 4);
+  pos += 4;
+  return x;
+}
+
 static bool
 match_u32(uint32_t x)
 {
@@ -87,6 +120,27 @@ match_u32_assert(uint32_t x, const char *where)
 }
 #define match_u32_assert(x) match_u32_assert(x, WHERE)
 
+static bool __attribute__((unused))
+match_be32(uint32_t x)
+{
+  if (get_be32() == x)
+    return true;
+  pos -= 4;
+  return false;
+}
+
+static void
+match_be32_assert(uint32_t x, const char *where)
+{
+  unsigned int y = get_be32();
+  if (x != y)
+    {
+      fprintf(stderr, "%s: 0x%x: expected be%u, got be%u\n", where, pos - 4, x, y);
+      exit(1);
+    }
+}
+#define match_be32_assert(x) match_be32_assert(x, WHERE)
+
 static bool
 match_byte(uint8_t b)
 {
@@ -218,6 +272,29 @@ get_string(const char *where)
 }
 #define get_string() get_string(WHERE)
 
+static char *
+get_string_be(const char *where)
+{
+  if (1
+      /*data[pos + 1] == 0 && data[pos + 2] == 0 && data[pos + 3] == 0*/
+      /*&& all_ascii(&data[pos + 4], data[pos])*/)
+    {
+      int len = data[pos + 2] * 256 + data[pos + 3];
+      char *s = malloc(len + 1);
+
+      memcpy(s, &data[pos + 4], len);
+      s[len] = 0;
+      pos += 4 + len;
+      return s;
+    }
+  else
+    {
+      fprintf(stderr, "%s: 0x%x: expected string\n", where, pos);
+      exit(1);
+    }
+}
+#define get_string_be() get_string_be(WHERE)
+
 static int
 get_end(void)
 {
@@ -225,6 +302,28 @@ get_end(void)
   return pos + len;
 }
 
+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);
+    }
+  for (int i = 0; i < n; i++)
+    {
+      int c = data[ofs + i];
+      fprintf(stderr, "%c", c >= 32 && c < 127 ? c : '.');
+    }
+  fprintf(stderr, "\n");
+}
+
 static char *
 dump_counted_string(void)
 {
@@ -232,7 +331,13 @@ dump_counted_string(void)
   int inner_end = get_end();
   if (pos != inner_end)
     {
-      match_u32_assert(0);
+      if (match_u32(5))
+        {
+          match_u32_assert(0);
+          match_byte_assert(0x58);
+        }
+      else
+        match_u32_assert(0);
       if (match_byte(0x31))
         s = get_string();
       else
@@ -246,8 +351,26 @@ dump_counted_string(void)
   return s;
 }
 
+static void
+dump_style(FILE *stream)
+{
+  match_byte(1);
+  match_byte(0);
+  match_byte(0);
+  match_byte(0);
+  match_byte_assert(1);
+  char *fg = get_string();     /* foreground */
+  char *bg = get_string();     /* background */
+  char *font = get_string();     /* font */
+  int size = data[pos];
+  if (!match_byte(14))
+    match_byte_assert(12); /* size? */
+  fprintf(stream, " fgcolor=\"%s\" bgcolor=\"%s\" font=\"%s\" size=\"%d\"",
+          fg, bg, font, size);
+}
+
 static char *
-dump_nested_string(void)
+dump_nested_string(FILE *stream)
 {
   char *s = NULL;
 
@@ -255,7 +378,12 @@ dump_nested_string(void)
   match_byte_assert (0);
   int outer_end = get_end();
   s = dump_counted_string();
-  match_byte_assert(0x58);
+  if (s)
+    fprintf(stream, " \"%s\"", s);
+  if (match_byte(0x31))
+    dump_style(stream);
+  else
+    match_byte_assert(0x58);
   match_byte_assert(0x58);
   if (pos != outer_end)
     {
@@ -267,20 +395,23 @@ dump_nested_string(void)
 }
 
 static void
-dump_optional_value(FILE *stream)
+dump_value_modifier(FILE *stream)
 {
   if (match_byte (0x31))
     {
       if (match_u32 (0))
         {
+          fprintf(stream, "<special0");
           if (match_u32 (1))
             {
               /* Corpus frequencies:
                  124 "a"
                  12 "b"
                  8 "a, b"
+
+                 The given text is appended to the cell in a subscript font.
               */
-              get_string();
+              fprintf(stream, " subscript=\"%s\"", get_string());
             }
           else
             match_u32_assert (0);
@@ -298,6 +429,7 @@ dump_optional_value(FILE *stream)
                 match_u32_assert(10);
               match_byte(0);
               match_byte(0);
+              fprintf(stream, "/>\n");
               return;
             }
 
@@ -305,22 +437,12 @@ dump_optional_value(FILE *stream)
           
           /* This counted-string appears to be a template string,
              e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
-          dump_counted_string();
+          char *template = dump_counted_string();
+          if (template)
+            fprintf(stream, " template=\"%s\"", template);
 
           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(stream);
           else
             match_byte_assert(0x58);
           if (match_byte(0x31))
@@ -348,32 +470,23 @@ dump_optional_value(FILE *stream)
               fprintf(stderr, "outer end discrepancy\n");
               exit(1);
             }
-        }
-      else if (match_u32 (1))
-        {
-          fprintf(stream, "(footnote %d) ", get_u32());
-          dump_nested_string();
-        }
-      else if (match_u32 (2))
-        {
-          fprintf(stream, "(special 2)");
-          if (!match_byte(0))
-            match_byte_assert(2);
-          match_byte_assert(0);
-          if (!match_u32 (2) && !match_u32(1))
-            match_u32_assert(3);
-          dump_nested_string();
+          fprintf(stream, "/>\n");
         }
       else
         {
-          match_u32_assert(3);
-          fprintf(stream, "(special 3)");
-          match_byte_assert(0);
+          int count = get_u32();
+          fprintf(stream, "<footnote-ref indexes=\"");
+          for (int i = 0; i < count; i++)
+            {
+              if (i)
+                putc(' ', stream);
+              fprintf(stream, "%d", get_u16());
+            }
+          putc('"', stream);
           match_byte_assert(0);
-          match_byte_assert(1);
           match_byte_assert(0);
-          match_u32_assert(2);
-          dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+          dump_nested_string(stream);
+          fprintf(stream, "/>\n");
         }
     }
   else
@@ -444,7 +557,7 @@ dump_value(FILE *stream, int level)
       unsigned int format;
       double value;
 
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       format = get_u32 ();
       value = get_double ();
       fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
@@ -456,7 +569,7 @@ dump_value(FILE *stream, int level)
       char *var, *vallab;
       double value;
 
-      dump_optional_value (stream);
+      dump_value_modifier (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -466,7 +579,7 @@ dump_value(FILE *stream, int level)
       if (var[0])
         fprintf (stream, " variable=\"%s\"", var);
       if (vallab[0])
-        fprintf (stream, " label=\"%s\"/>\n", vallab);
+        fprintf (stream, " label=\"%s\"", vallab);
       fprintf (stream, "/>\n");
       if (!match_byte (1) && !match_byte(2))
         match_byte_assert (3);
@@ -474,7 +587,7 @@ dump_value(FILE *stream, int level)
   else if (match_byte (3))
     {
       char *text =  get_string();
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       char *identifier = get_string();
       char *text_eng = get_string();
       fprintf (stream, "<string c=\"%s\"", text_eng);
@@ -491,7 +604,7 @@ dump_value(FILE *stream, int level)
       unsigned int format;
       char *var, *vallab, *value;
 
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       format = get_u32 ();
       vallab = get_string ();
       var = get_string ();
@@ -508,7 +621,7 @@ dump_value(FILE *stream, int level)
     }
   else if (match_byte (5))
     {
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       char *name = get_string ();
       char *label = get_string ();
       fprintf (stream, "<variable name=\"%s\"", name);
@@ -520,7 +633,7 @@ dump_value(FILE *stream, int level)
     }
   else
     {
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -573,7 +686,8 @@ check_permutation(int *a, int n, const char *name)
 }
 
 static void
-dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_indexes)
+dump_category(FILE *stream, int level, int **indexes, int *allocated_indexes,
+              int *n_indexes)
 {
   for (int i = 0; i <= level; i++)
     fprintf (stream, "    ");
@@ -623,12 +737,12 @@ dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_ind
           fprintf(stderr, "index not -1 but subcategories\n");
           exit(1);
         }
-      if (*n_indexes >= max_indexes)
+      if (*n_indexes >= *allocated_indexes)
         {
-          fprintf(stderr, "too many categories (increase max_indexes)\n");
-          exit(1);
+          *allocated_indexes = *allocated_indexes ? 2 * *allocated_indexes : 16;
+          *indexes = realloc(*indexes, *allocated_indexes * sizeof **indexes);
         }
-      indexes[(*n_indexes)++] = indx;
+      (*indexes)[(*n_indexes)++] = indx;
     }
 
   int expected_unindexed = indx == -1;
@@ -646,7 +760,7 @@ dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_ind
       fprintf (stream, "<category-index>%d</category-index>\n", indx);
     }
   for (int i = 0; i < n_categories; i++)
-    dump_category (stream, level + 1, indexes, n_indexes, max_indexes);
+    dump_category (stream, level + 1, indexes, allocated_indexes, n_indexes);
   for (int i = 0; i <= level; i++)
     fprintf (stream, "    ");
   printf ("</category>\n");
@@ -676,10 +790,11 @@ dump_dim(int indx)
     match_u32_assert(indx);
   n_categories = get_u32();
 
-  int indexes[2048];
+  int *indexes = NULL;
   int n_indexes = 0;
+  int allocated_indexes = 0;
   for (int i = 0; i < n_categories; i++)
-    dump_category (stdout, 0, indexes, &n_indexes, sizeof indexes / sizeof *indexes);
+    dump_category (stdout, 0, &indexes, &allocated_indexes, &n_indexes);
   check_permutation(indexes, n_indexes, "categories");
 
   fprintf (stdout, "</dimension>\n");
@@ -766,8 +881,14 @@ dump_title(void)
   match_byte(1);
   printf ("</title-c>\n");
 
-  match_byte(0);
-  match_byte_assert(0x58);
+  if (match_byte(0x31))
+    {
+      printf ("<user-caption>\n");
+      dump_value(stdout, 0);
+      printf ("</user-caption>\n");
+    }
+  else
+    match_byte_assert(0x58);
   if (match_byte(0x31))
     {
       printf ("<caption>\n");
@@ -792,24 +913,6 @@ 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 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_fonts(void)
 {
@@ -820,78 +923,101 @@ dump_fonts(void)
       match_byte_assert(i);
       match_byte_assert(0x31);
       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) && !match_byte(0x70))
-        match_byte_assert(0x50);
-      match_byte_assert(0x41);
-      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... */
-      if (!match_u32(0) && !match_u32(2))
-        {
-          if (i == 7)
-            match_u32_assert(0xfaad);
-          else
-            match_u32_assert(0);
-        }
+      printf(" size=\"%gpt\"", get_float());
+
+      int style = get_u32();
+      if (style & 1)
+        printf(" bold=\"true\"");
+      if (style & 2)
+        printf(" italic=\"true\"");
+
+      bool underline = data[pos++];
+      if (underline)
+        printf(" underline=\"true\"");
+
+      int halign = get_u32();
+      printf("\nhalign=%d\n", halign);
+
+      int valign = get_u32();
+      printf("\nvalign=%d\n", valign);
 
-      if (!match_u32(0) && !match_u32(1) && !match_u32(2))
-        match_u32_assert(3);
       printf (" fgcolor=\"%s\"", get_string());
       printf (" bgcolor=\"%s\"", get_string());
+
+      if (!match_byte(0))
+        match_byte_assert(1);
       match_u32_assert(0);
-      match_u32_assert(0);
-      match_byte_assert(0);
+      char *othercolor = get_string();
+      if (othercolor[0])
+        printf(" othercolor=\"%s\"", othercolor);
 
       if (version > 1)
         {
-          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();
+          printf("\nfonts:");
+          for (int i = 0; i < 4; i++)
+            printf(" %2d", get_u32());
+          printf("\n");
         }
 
       printf ("/>\n");
     }
 
-  match_u32_assert(240);
-  pos += 240;
-
-  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 x1 = get_u32();
+  int x1_end = pos + x1;
+  printf("<borders>\n");
+  match_be32_assert(1);
+  int n_borders = get_be32();
+  for (int i = 0; i < n_borders; i++)
     {
-      match_u32_assert(16);
-      pos += 16;
+      int type = get_be32();
+      int stroke = get_be32();
+      int color = get_be32();
+      printf("  <border type=\"%d\" stroke=\"%s\" color=\"#%06x\"/>\n",
+             type,
+             (stroke == 0 ? "none"
+              : stroke == 1 ? "solid"
+              : stroke == 2 ? "dashed"
+              : stroke == 3 ? "thick"
+              : stroke == 4 ? "thin"
+              : stroke == 5 ? "double"
+              : "<error>"),
+             color);
     }
+  bool grid = get_byte();
+  pos += 3;
+  printf("  <grid show=\"%s\">\n", grid ? "yes" : "no");
+  printf("</borders>\n");
+  assert(pos == x1_end);
+
+  int skip = get_u32();
+  assert(skip == 18 || skip == 25);
+  pos += skip;
+
+  int x3 = get_u32();
+  int x3_end = pos + x3;
+  if (version == 3)
+    {
+      match_be32_assert(1);
+      get_be32();
+      printf("<settings layer=\"%d\"", get_be32());
+      if (!get_byte())
+        printf(" skipempty=\"false\"");
+      if (!get_byte())
+        printf(" showdimensionincorner=\"false\"");
+      if (!get_byte())
+        printf(" markers=\"numeric\"");
+      if (!get_byte())
+        printf(" footnoteposition=\"subscript\"");
+      get_byte();
+      pos += get_be32();
+      get_string_be();
+      char *look = get_string_be();
+      if (look[0])
+        printf(" look=\"%s\"", look);
+      printf(">\n");
+    }
+  pos = x3_end;
 
   int count = get_u32();
   pos += 4 * count;
@@ -899,23 +1025,13 @@ dump_fonts(void)
   const char *locale = get_string();
   printf ("<locale>%s</locale>\n", locale);
 
-  if (!match_u32(0))
-    match_u32_assert(UINT32_MAX);
+  get_u32();            /* Seen: 0, UINT32_MAX, 2, 3, 4, 5, 6, 8, 9, 21, 24. */
   if (!match_byte(0))
     match_byte_assert(1);
   match_byte_assert(0);
   if (!match_byte(0))
     match_byte_assert(1);
-  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);
+  printf("<epoch>%d</epoch>\n", get_u32());
 
   int decimal = data[pos];
   int grouping = data[pos + 1];
@@ -927,7 +1043,7 @@ dump_fonts(void)
   else
     {
       match_byte_assert(',');
-      if (!match_byte('.') && !match_byte(' '))
+      if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
         match_byte_assert(0);
     }
   printf("<format decimal=\"%c\" grouping=\"", decimal);
@@ -957,8 +1073,8 @@ dump_fonts(void)
 
       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);
@@ -977,11 +1093,7 @@ 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);
-      match_byte_assert(7);
-      match_byte_assert(0);
-      match_byte_assert(0);
+      printf("<epoch2>%d</epoch2>\n", get_u32());
 
       if (match_byte('.'))
         {
@@ -991,7 +1103,7 @@ dump_fonts(void)
       else
         {
           match_byte_assert(',');
-          if (!match_byte('.') && !match_byte(' '))
+          if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
             match_byte_assert(0);
         }
 
@@ -1034,12 +1146,19 @@ dump_fonts(void)
       if (!match_byte(0))
         match_byte_assert(1);
 
+      if (pos < outer_end)
+        {
+          printf("<seed>%d</seed>\n", get_u32());
+          match_u32_assert(0);
+        }
       assert(pos == outer_end);
 
       pos = outer_end;
     }
   else
-    match_u32_assert(0);
+    {
+      pos = get_end();
+    }
 }
 
 int
@@ -1121,7 +1240,7 @@ main(int argc, char *argv[])
 
       /* 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_optional_value(). */
+         special cases in dump_value_modifier(). */
       version = get_u32();
       pos -= 4;
       if (!match_u32(1))
@@ -1133,7 +1252,8 @@ main(int argc, char *argv[])
 
       /* Offset 8. */
       match_byte_assert(0);
-      match_byte_assert(0);
+      if (!match_byte(0))
+        match_byte_assert(1);
       if (!match_byte(0))
         match_byte_assert(1);
 
@@ -1167,8 +1287,8 @@ main(int argc, char *argv[])
       /* Offset 27. */
       pos++;
       pos++;
-      match_byte_assert(0);
-      match_byte_assert(0);
+      pos++;
+      pos++;
 
       /* Offset 31.