X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=205167951ca20dab22cf99d3b8143111d5d59abb;hb=55514d1999dd218a25028cb8bd6a7f608b52c2ef;hp=697ace697badba8c220db73de40d6c0b38bd7c35;hpb=1170cd2e9215a87aaaad36796fd9661691a4e29a;p=pspp

diff --git a/dump.c b/dump.c
index 697ace697b..205167951c 100644
--- a/dump.c
+++ b/dump.c
@@ -1,3 +1,4 @@
+#include <assert.h>
 #include <float.h>
 #include <stdbool.h>
 #include <stdint.h>
@@ -5,7 +6,9 @@
 #include <stdlib.h>
 #include <string.h>
 #include <sys/stat.h>
+#include <time.h>
 #include <unistd.h>
+#include "u8-mbtouc.h"
 
 static uint8_t *data;
 static size_t n;
@@ -46,6 +49,12 @@ 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)
 {
@@ -55,6 +64,33 @@ get_u32(void)
   return x;
 }
 
+static unsigned long long int
+get_u64(void)
+{
+  uint64_t x;
+  memcpy(&x, &data[pos], 8);
+  pos += 8;
+  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)
 {
@@ -64,6 +100,15 @@ 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)
 {
@@ -85,6 +130,48 @@ match_u32_assert(uint32_t x, const char *where)
 }
 #define match_u32_assert(x) match_u32_assert(x, WHERE)
 
+static bool __attribute__((unused))
+match_u64(uint64_t x)
+{
+  if (get_u64() == x)
+    return true;
+  pos -= 8;
+  return false;
+}
+
+static void __attribute__((unused))
+match_u64_assert(uint64_t x, const char *where)
+{
+  unsigned long long int y = get_u64();
+  if (x != y)
+    {
+      fprintf(stderr, "%s: 0x%x: expected u64:%llu, got u64:%llu\n", where, pos - 8, x, y);
+      exit(1);
+    }
+}
+#define match_u64_assert(x) match_u64_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)
 {
@@ -177,6 +264,22 @@ dump_raw(FILE *stream, int start, int end)
 
 }
 
+static bool __attribute__((unused))
+all_utf8(const char *p_)
+{
+  const uint8_t *p = (const uint8_t *) p_;
+  size_t len = strlen ((char *) p);
+  for (size_t ofs = 0, mblen; ofs < len; ofs += mblen)
+    {
+      ucs4_t uc;
+
+      mblen = u8_mbtouc (&uc, p + ofs, len - ofs);
+      if ((uc < 32 && uc != '\n') || uc == 127 || uc == 0xfffd)
+        return false;
+    }
+  return true;
+}
+
 static char *
 get_string(const char *where)
 {
@@ -201,17 +304,70 @@ get_string(const char *where)
 #define get_string() get_string(WHERE)
 
 static char *
-dump_nested_string(void)
+get_string_be(const char *where)
 {
-  char *s = NULL;
+  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);
 
-  match_byte_assert (0);
-  match_byte_assert (0);
-  int outer_end = pos + get_u32();
-  int inner_end = pos + get_u32();
+      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)
+{
+  int len = get_u32();
+  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))
+        printf("-");
+      else
+        printf(" ");
+#endif
+      printf("%02x", c);
+    }
+  for (int i = 0; i < n; i++)
+    {
+      int c = data[ofs + i];
+      printf("%c", c >= 32 && c < 127 ? c : '.');
+    }
+}
+
+static char *
+dump_counted_string(void)
+{
+  char *s = NULL;
+  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
@@ -222,7 +378,42 @@ dump_nested_string(void)
           exit(1);
         }
     }
-  match_byte_assert(0x58);
+  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(FILE *stream)
+{
+  char *s = NULL;
+
+  match_byte_assert (0);
+  match_byte_assert (0);
+  int outer_end = get_end();
+  s = dump_counted_string();
+  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)
     {
@@ -234,16 +425,23 @@ dump_nested_string(void)
 }
 
 static void
-dump_value_31(FILE *stream)
+dump_value_modifier(FILE *stream)
 {
   if (match_byte (0x31))
     {
       if (match_u32 (0))
         {
+          fprintf(stream, "<special0");
           if (match_u32 (1))
             {
-              /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
-              get_string();
+              /* Corpus frequencies:
+                 124 "a"
+                 12 "b"
+                 8 "a, b"
+
+                 The given text is appended to the cell in a subscript font.
+              */
+              fprintf(stream, " subscript=\"%s\"", get_string());
             }
           else
             match_u32_assert (0);
@@ -252,48 +450,29 @@ dump_value_31(FILE *stream)
             {
               /* We only have one SPV file for this version (with many
                  tables). */
-              match_u32_assert(0x200);
-              match_u32_assert(0x1000000);
-              match_u32_assert(0);
-              match_byte_assert(0);
+              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);
+              fprintf(stream, "/>\n");
               return;
             }
 
-          int outer_end = pos + get_u32();
-          int inner_end = pos + get_u32();
-          if (pos != inner_end)
-            {
-              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);
-                }
-            }
+          int outer_end = get_end();
+          
+          /* This counted-string appears to be a template string,
+             e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
+          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))
@@ -321,31 +500,23 @@ dump_value_31(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)");
-          match_byte_assert(0);
-          match_byte_assert(0);
-          if (!match_u32 (2))
-            match_u32_assert(1);
-          dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+          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
@@ -401,7 +572,7 @@ format_to_string (int type)
 }
 
 static void
-dump_value(FILE *stream, int level, bool match1)
+dump_value(FILE *stream, int level)
 {
   match_byte(0);
   match_byte(0);
@@ -411,34 +582,16 @@ dump_value(FILE *stream, int level, bool match1)
   for (int i = 0; i <= level; i++)
     fprintf (stream, "    ");
 
-  if (match_byte (3))
-    {
-      char *text = get_string();
-      dump_value_31(stream);
-      char *identifier = get_string();
-      char *text_eng = get_string();
-      fprintf (stream, "<string c=\"%s\"", text_eng);
-      if (identifier[0])
-        fprintf (stream, " identifier=\"%s\"", identifier);
-      if (strcmp(text_eng, text))
-        fprintf (stream, " local=\"%s\"", text);
-      fprintf (stream, "/>\n");
-      if (!match_byte (0))
-        match_byte_assert(1);
-      if (match1)
-        match_byte (1);
-    }
-  else if (match_byte (5))
+  if (match_byte (1))
     {
-      dump_value_31(stream);
-      char *name = get_string ();
-      char *label = get_string ();
-      fprintf (stream, "<variable name=\"%s\"", name);
-      if (label[0])
-        fprintf (stream, " label=\"%s\"", label);
-      fprintf (stream, "/>\n");
-      if (!match_byte(1) && !match_byte(2))
-        match_byte_assert(3);
+      unsigned int format;
+      double value;
+
+      dump_value_modifier(stream);
+      format = get_u32 ();
+      value = get_double ();
+      fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
+               DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
     }
   else if (match_byte (2))
     {
@@ -446,7 +599,7 @@ dump_value(FILE *stream, int level, bool match1)
       char *var, *vallab;
       double value;
 
-      dump_value_31 (stream);
+      dump_value_modifier (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -456,17 +609,32 @@ dump_value(FILE *stream, int level, bool match1)
       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);
     }
+  else if (match_byte (3))
+    {
+      char *text =  get_string();
+      dump_value_modifier(stream);
+      char *identifier = get_string();
+      char *text_eng = get_string();
+      fprintf (stream, "<string c=\"%s\"", text_eng);
+      if (identifier[0])
+        fprintf (stream, " identifier=\"%s\"", identifier);
+      if (strcmp(text_eng, text))
+        fprintf (stream, " local=\"%s\"", text);
+      fprintf (stream, "/>\n");
+      if (!match_byte (0))
+        match_byte_assert(1);
+    }
   else if (match_byte (4))
     {
       unsigned int format;
       char *var, *vallab, *value;
 
-      match_byte_assert (0x58);
+      dump_value_modifier(stream);
       format = get_u32 ();
       vallab = get_string ();
       var = get_string ();
@@ -481,22 +649,21 @@ dump_value(FILE *stream, int level, bool match1)
         fprintf (stream, " label=\"%s\"/>\n", vallab);
       fprintf (stream, "/>\n");
     }
-  else if (match_byte (1))
+  else if (match_byte (5))
     {
-      unsigned int format;
-      double value;
-
-      dump_value_31(stream);
-      format = get_u32 ();
-      value = get_double ();
-      fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
-               DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
-      if (match1)
-        match_byte (1);
+      dump_value_modifier(stream);
+      char *name = get_string ();
+      char *label = get_string ();
+      fprintf (stream, "<variable name=\"%s\"", name);
+      if (label[0])
+        fprintf (stream, " label=\"%s\"", label);
+      fprintf (stream, "/>\n");
+      if (!match_byte(1) && !match_byte(2))
+        match_byte_assert(3);
     }
   else
     {
-      dump_value_31(stream);
+      dump_value_modifier(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -512,7 +679,7 @@ dump_value(FILE *stream, int level, bool match1)
             fprintf (stream, "    ");
           fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
           for (int j = 0; j < y; j++)
-            dump_value (stream, level + 2, false);
+            dump_value (stream, level + 2);
           for (int j = 0; j <= level + 1; j++)
             fprintf (stream, "    ");
           fprintf (stream, "</substitution>\n");
@@ -549,66 +716,95 @@ check_permutation(int *a, int n, const char *name)
 }
 
 static void
-dump_category(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 (stdout, "    ");
+    fprintf (stream, "    ");
   printf ("<category>\n");
-  dump_value (stdout, level + 1, true);
-  match_byte(0);
-  match_byte(0);
-  match_byte(0);
+  dump_value (stream, level + 1);
 
-  if (match_u32 (1))
-    match_byte (0);
-  else if (match_byte (1))
-    {
-      match_byte (0);
-      if (!match_u32 (2))
-        match_u32_assert (1);
-      match_byte (0);
-    }
-  else if (!match_u32(2))
-    match_u32_assert (0);
+  int merge = data[pos];
+  if (!match_byte(0))
+    match_byte_assert (1);
+
+  match_byte_assert (0);
+
+  int unindexed = data[pos];
+  if (!match_byte(0))
+    match_byte_assert (1);
+
+  int x = get_u32 ();
+  pos -= 4;
+  if (!match_u32 (0))
+    match_u32_assert (2);
 
   int indx = get_u32();
   int n_categories = get_u32();
-  if (indx != -1)
+  if (indx == -1)
+    {
+      if (merge)
+        {
+          for (int i = 0; i <= level + 1; i++)
+            fprintf (stream, "    ");
+          fprintf (stream, "<merge/>\n");
+        }
+    }
+  else
     {
+      if (merge)
+        {
+          fprintf(stderr, "index not -1 but merged\n");
+          exit(1);
+        }
+      if (x != 2)
+        {
+          fprintf(stderr, "index not -1 but x != 2\n");
+          exit(1);
+        }
       if (n_categories != 0)
         {
           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;
+  if (unindexed != expected_unindexed)
+    {
+      fprintf(stderr, "unindexed (%d) mismatch with indx (%d)\n",
+              unindexed, indx);
+      exit(1);
+    }
+
   if (n_categories == 0)
     {
       for (int i = 0; i <= level + 1; i++)
-        fprintf (stdout, "    ");
-      fprintf (stdout, "<category-index>%d</category-index>\n", indx);
+        fprintf (stream, "    ");
+      fprintf (stream, "<category-index>%d</category-index>\n", indx);
     }
   for (int i = 0; i < n_categories; i++)
-    dump_category (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 (stdout, "    ");
+    fprintf (stream, "    ");
   printf ("</category>\n");
 }
 
-static void
+static int
 dump_dim(int indx)
 {
   int n_categories;
 
   printf ("<dimension index=\"%d\">\n", indx);
-  dump_value (stdout, 0, false);
+  dump_value (stdout, 0);
 
-  /* This byte is usually 0x02 but many other values have been spotted. */
+  /* This byte is usually 0 but many other values have been spotted. */
   pos++;
 
   if (!match_byte(0) && !match_byte(1))
@@ -624,45 +820,72 @@ 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 (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");
+  return n_indexes;
 }
 
 int n_dims;
+static int dim_n_cats[64];
+#define MAX_DIMS (sizeof dim_n_cats / sizeof *dim_n_cats)
+
 static void
 dump_dims(void)
 {
   n_dims = get_u32();
+  assert(n_dims < MAX_DIMS);
   for (int i = 0; i < n_dims; i++)
-    dump_dim (i);
+    dim_n_cats[i] = dump_dim (i);
 }
 
 static void
 dump_data(void)
 {
   /* The first three numbers add to the number of dimensions. */
-  int t = get_u32();
-  t += get_u32();
-  match_u32_assert(n_dims - t);
+  int l = get_u32();
+  int r = get_u32();
+  int c = n_dims - l - r;
+  match_u32_assert(c);
 
   /* The next n_dims numbers are a permutation of the dimension numbers. */
   int a[n_dims];
   for (int i = 0; i < n_dims; i++)
-    a[i] = get_u32();
+    {
+      int dim = get_u32();
+      a[i] = dim;
+
+      const char *name = i < l ? "layer" : i < l + r ? "row" : "column";
+      printf ("<%s dimension=\"%d\"/>\n", name, dim);
+    }
   check_permutation(a, n_dims, "dimensions");
 
   int x = get_u32();
   printf ("<data>\n");
   for (int i = 0; i < x; i++)
     {
-      printf ("    <datum index=\"%d\">\n", get_u32());
+      unsigned int indx = get_u32();
+      printf ("    <datum index=\"%d\" coords=", indx);
+
+      int coords[MAX_DIMS];
+      for (int i = n_dims; i-- > 0; )
+        {
+          coords[i] = indx % dim_n_cats[i];
+          indx /= dim_n_cats[i];
+        }
+      for (int i = 0; i < n_dims; i++)
+        printf("%c%d", i ? ',' : '"', coords[i]);
+
+      printf ("\">\n");
       match_u32_assert(0);
-      dump_value(stdout, 1, false);
+      if (version == 1)
+        match_byte(0);
+      dump_value(stdout, 1);
       fprintf (stdout, "    </datum>\n");
     }
   printf ("</data>\n");
@@ -671,50 +894,51 @@ dump_data(void)
 static void
 dump_title(void)
 {
-  pos = 0x27;
   printf ("<title-local>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  match_byte(1);
   printf ("</title-local>\n");
 
   printf ("<subtype>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  match_byte(1);
   printf ("</subtype>\n");
 
   match_byte_assert(0x31);
 
   printf ("<title-c>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  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");
-      dump_value(stdout, 0, false);
+      dump_value(stdout, 0);
       printf ("</caption>\n");
     }
   else
     match_byte_assert(0x58);
 
-
   int n_footnotes = get_u32();
   for (int i = 0; i < n_footnotes; i++)
     {
       printf ("<footnote index=\"%d\">\n", i);
-      dump_value(stdout, 0, false);
+      dump_value(stdout, 0);
+      /* Custom footnote marker string. */
       if (match_byte (0x31))
-        {
-          /* Custom footnote marker string. */
-          match_byte_assert(3);
-          get_string();
-          match_byte_assert(0x58);
-          match_u32_assert(0);
-          get_string();
-        }
+        dump_value(stdout, 0);
       else
         match_byte_assert (0x58);
-      printf("(%d)\n", get_u32());
+      get_u32 ();
       printf ("</footnote>\n");
     }
 }
@@ -729,118 +953,300 @@ 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);
-      if (!match_byte(0x41))
-        match_byte_assert(0x51);
-      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))
-        match_u32_assert(0xfaad);
+      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(" halign=%d", halign);
+
+      int valign = get_u32();
+      printf(" valign=%d", 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());
-      match_u32_assert(0);
-      match_u32_assert(0);
-      match_byte_assert(0);
+
+      if (!match_byte(0))
+        match_byte_assert(1);
+
+      char *alt_fgcolor = get_string();
+      if (alt_fgcolor[0])
+        printf (" altfg=\"%s\"", alt_fgcolor);
+      char *alt_bgcolor = get_string();
+      if (alt_bgcolor[0])
+        printf (" altbg=\"%s\"", alt_bgcolor);
 
       if (version > 1)
         {
-          /* These seem unlikely to be correct too. */
-          if (i != 3)
+          printf(" margins=\"");
+          for (int i = 0; i < 4; i++)
             {
-              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);
+              if (i)
+                putchar(' ');
+              printf("%d", get_u32());
             }
-          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();
+          putchar('"');
         }
 
       printf ("/>\n");
     }
 
-  match_u32_assert(240);
-  pos += 240;
+  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++)
+    {
+      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);
 
-  match_u32_assert(18);
-  pos += 18;
+  int skip = get_u32();
+  assert(skip == 18 || skip == 25);
+  pos += skip;
 
-  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();
+  int x3_end = pos + x3;
+  if (version == 3)
     {
-      match_u32_assert(16);
-      pos += 16;
+      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();
+      int nbytes = get_be32();
+      printf("\n");
+      hex_dump(pos, nbytes);
+      printf("\n");
+      pos += nbytes;
+      char *notes = get_string_be();
+      if (notes[0])
+        printf(" notes=\"%s\"", notes);
+      char *look = get_string_be();
+      if (look[0])
+        printf(" look=\"%s\"", look);
+      printf(">\n");
     }
+  pos = x3_end;
 
+  /* Manual column widths, if present. */
   int count = get_u32();
-  pos += 4 * count;
+  if (count > 0)
+    {
+      printf("<columnwidths>");
+      for (int i = 0; i < count; i++)
+        {
+          if (i)
+            putchar(' ');
+          printf("%d", get_u32());
+        }
+      printf("</columnwidths>\n");
+    }
 
-  printf ("<encoding>%s</encoding>\n", get_string ());
+  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];
   if (match_byte('.'))
     {
-      if (!match_byte(','))
+      if (!match_byte(',') && !match_byte('\''))
         match_byte_assert(' ');
     }
   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);
+  if (grouping)
+    putchar(grouping);
+  printf("\"/>\n");
   if (match_u32(5))
     {
       for (int i = 0; i < 5; i++)
-        get_string();
+        printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
     }
   else
     match_u32_assert(0);
-  int skip = get_u32();
-  pos += skip;
+
+  /* The last chunk is an outer envelope that contains two inner envelopes.
+     The second inner envelope has some interesting data like the encoding and
+     the locale. */
+  if (version == 3)
+    {
+      int outer_end = get_end();
+
+      /* First inner envelope: byte*33 int[n] int*[n]. */
+      int inner_len = get_u32();
+      int inner_end = pos + inner_len;
+      int array_start = pos + 33;
+      match_byte_assert(0);
+      pos++;                    /* 0, 1, 10 seen. */
+      match_byte_assert(0);
+      pos++;                    /* 0...11 seen. */
+      if (!match_byte(0) && !match_byte(1) && !match_byte(2))
+        match_byte_assert(3);
+      if (!match_byte(0) && !match_byte(2))
+        match_byte_assert(3);
+      if (!match_u64(0))
+        match_u64_assert(UINT64_MAX);
+      match_u32_assert(0);
+      match_u32_assert(0);
+      match_u32_assert(0);
+      match_u32_assert(0);
+      match_byte_assert(0);
+      if (!match_byte(0))
+        match_byte_assert(1);
+      match_byte_assert(1);
+      pos = array_start;
+#if 1
+      printf("widths:");
+      while (pos < inner_end)
+        printf(" %d", get_u32());
+      printf("\n");
+#endif
+      pos = inner_end;;
+
+      /* Second inner envelope. */
+      assert(get_end() == outer_end);
+
+      match_byte_assert(1);
+      match_byte_assert(0);
+      if (!match_byte(3) && !match_byte(4))
+        match_byte_assert(5);
+      match_byte_assert(0);
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      printf("<command>%s</command>\n", get_string());
+      printf("<subcommand>%s</subcommand>\n", get_string());
+      printf("<language>%s</language>\n", get_string());
+      printf("<charset>%s</charset>\n", get_string());
+      printf("<locale>%s</locale>\n", get_string());
+
+      if (!match_byte(0))
+        match_byte_assert(1);
+      match_byte_assert(0);
+      if (!match_byte(0))
+        match_byte_assert(1);
+      if (!match_byte(0))
+        match_byte_assert(1);
+
+      printf("<epoch2>%d</epoch2>\n", get_u32());
+
+      if (match_byte('.'))
+        {
+          if (!match_byte(',') && !match_byte('\''))
+            match_byte_assert(' ');
+        }
+      else
+        {
+          match_byte_assert(',');
+          if (!match_byte('.') && !match_byte(' ') && !match_byte(','))
+            match_byte_assert(0);
+        }
+
+      printf ("small: %g\n", get_double());
+
+      match_byte_assert(1);
+      if (outer_end - pos > 6)
+        {
+          /* 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());
+
+              match_u32_assert(0);
+
+              time_t date = get_u32();
+              struct tm tm = *localtime(&date);
+              char s[128];
+              strftime(s, sizeof s, "%a, %d %b %Y %H:%M:%S %z", &tm);
+              printf("<date>%s</date>\n", s);
+
+              match_u32_assert(0);
+            }
+        }
+
+      if (match_u32(5))
+        {
+          for (int i = 0; i < 5; i++)
+            printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
+        }
+      else
+        match_u32_assert(0);
+
+      match_byte_assert('.');
+      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
+    {
+      pos = get_end();
+    }
 }
 
 int
@@ -872,90 +1278,103 @@ main(int argc, char *argv[])
       exit(1);
     }
 
-  if (argc > 1)
+  if (argc != 2)
     {
-      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;
-        }
-      else if (!strcmp(argv[1], "title"))
-        {
-          dump_title();
-          exit(0);
-        }
-      else if (!strcmp(argv[1], "titleraw"))
-        {
-          const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
-          start = 0x27;
-          n = find(fonts, sizeof fonts - 1);
-        }
-      else if (!strcmp(argv[1], "fonts"))
-        {
-          const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
-          const char styles[] = "\xf0\0\0\0";
-          start = find(fonts, sizeof fonts - 1);
-          n = find(styles, sizeof styles - 1);
-        }
-      else if (!strcmp(argv[1], "styles"))
-        {
-          const char styles[] = "\xf0\0\0\0";
-          const char dimensions[] = "-,,,.\0";
-          start = find(styles, sizeof styles - 1);
-          n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
-        }
-      else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
-        {
-          pos = 0;
-          match_byte_assert(1);
-          match_byte_assert(0);
+      fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
+      exit (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;
+    }
+  else if (!strcmp(argv[1], "title"))
+    {
+      pos = 0x27;
+      dump_title();
+      exit(0);
+    }
+  else if (!strcmp(argv[1], "titleraw"))
+    {
+      const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
+      start = 0x27;
+      n = find(fonts, sizeof fonts - 1);
+    }
+  else if (!strcmp(argv[1], "fonts"))
+    {
+      const char fonts[] = "\x01\x31\x09\0\0\0SansSerif";
+      const char styles[] = "\xf0\0\0\0";
+      start = find(fonts, sizeof fonts - 1);
+      n = find(styles, sizeof styles - 1);
+    }
+  else if (!strcmp(argv[1], "styles"))
+    {
+      const char styles[] = "\xf0\0\0\0";
+      const char dimensions[] = "-,,,.\0";
+      start = find(styles, sizeof styles - 1);
+      n = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
+    }
+  else if (!strcmp(argv[1], "dimensions") || !strcmp(argv[1], "all"))
+    {
+      pos = 0;
+      match_byte_assert(1);
+      match_byte_assert(0);
 
-          /* 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);
+      /* 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_modifier(). */
+      version = get_u32();
+      pos -= 4;
+      if (!match_u32(1))
+        match_u32_assert(3);
 
+      match_byte_assert(1);
+      for (int i = 0; i < 4; i++)
+        if (!match_byte(0))
           match_byte_assert(1);
-          if (!match_byte(0))
-            match_byte_assert(1);
-          match_byte_assert(0);
-          match_byte_assert(0);
-          if (!match_byte(0))
-            match_byte_assert(1);
-          pos++;
-          match_byte_assert(0);
-          match_byte_assert(0);
-          match_byte_assert(0);
-          dump_title ();
-          dump_fonts();
-          dump_dims ();
-          dump_data ();
-          match_byte (1);
-          if (pos != n)
-            {
-              fprintf (stderr, "%x / %x\n", pos, n);
-              exit(1);
-            }
-          exit(0);
-        }
-      else
+      get_u32();
+
+      int min_col_width = get_u32();
+      int max_col_width = get_u32();
+      int min_row_width = get_u32();
+      int max_row_width = get_u32();
+      printf("<label-width min-col=\"%d\" max-col=\"%d\" min-row=\"%d\" "
+             "max-row=\"%d\"/>\n",
+             min_col_width, max_col_width,
+             min_row_width, max_row_width);
+
+      /* Offset 31. */
+      printf("<tableid>%lld</tableid>", get_u64());
+
+      dump_title ();
+      dump_fonts();
+      dump_dims ();
+      dump_data ();
+      match_byte (1);
+      if (pos != n)
         {
-          fprintf (stderr, "unknown section %s\n", argv[1]);
+          fprintf (stderr, "%x / %x\n", pos, n);
           exit(1);
         }
+      exit(0);
     }
-  else
-    start = 0x27;
+  else if (!strcmp(argv[1], "raw"))
+    {
+      start = 0x27;
 
-  dump_raw(stdout, start, n);
+      dump_raw(stdout, start, n);
+    }
+  else
+    {
+      fprintf (stderr, "unknown section %s\n", argv[1]);
+      exit(1);
+    }
 
   return 0;
 }