X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=c8019cad9e338618a6add67bdc2d7fa487192990;hb=38a126b38f4963ba033120754e7a4f026e9fa01e;hp=30e34facaf00efcd2fe03470bbeae77df4ba1b28;hpb=b8ec4632f2e4d5f7e638af92e1c2dcc2a339d7de;p=pspp

diff --git a/dump.c b/dump.c
index 30e34facaf..c247c4af66 100644
--- a/dump.c
+++ b/dump.c
@@ -1,3 +1,5 @@
+#include <assert.h>
+#include <float.h>
 #include <stdbool.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -5,10 +7,13 @@
 #include <string.h>
 #include <sys/stat.h>
 #include <unistd.h>
+#include "u8-mbtouc.h"
 
 static uint8_t *data;
 static size_t n;
 
+int version;
+
 static bool
 all_ascii(const uint8_t *p, size_t n)
 {
@@ -19,15 +24,22 @@ 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
+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;
 }
 
 size_t pos;
@@ -78,7 +90,7 @@ match_u32_assert(uint32_t x, const char *where)
 static bool
 match_byte(uint8_t b)
 {
-  if (data[pos] == b)
+  if (pos < n && data[pos] == b)
     {
       pos++;
       return true;
@@ -98,187 +110,814 @@ match_byte_assert(uint8_t b, const char *where)
 }
 #define match_byte_assert(b) match_byte_assert(b, WHERE)
 
+static void
+newline(FILE *stream, int pos)
+{
+  fprintf(stream, "\n%08x: ", pos);
+}
+
+static void
+dump_raw(FILE *stream, int start, int end)
+{
+  for (size_t i = start; i < end; )
+    {
+      if (i + 5 <= n
+          && data[i]
+          //&& !data[i + 1]
+          && !data[i + 2]
+          && !data[i + 3]
+          && i + 4 + data[i] + data[i + 1] * 256 <= end
+          && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
+        {
+          newline(stream, i);
+          fprintf(stream, "\"");
+          fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
+          fputs("\" ", stream);
+
+          i += 4 + data[i] + data[i + 1] * 256;
+        }
+      else if (i + 12 <= end
+               && data[i + 1] == 40
+               && data[i + 2] == 5
+               && data[i + 3] == 0)
+        {
+          double d;
+
+          memcpy (&d, &data[i + 4], 8);
+          fprintf (stream, "F40.%d(%.*f)", data[i], data[i], d);
+          i += 12;
+          newline (stream, i);
+        }
+      else if (i + 12 <= end
+               && data[i + 1] == 40
+               && data[i + 2] == 31
+               && data[i + 3] == 0)
+        {
+          double d;
+
+          memcpy (&d, &data[i + 4], 8);
+          fprintf (stream, "PCT40.%d(%.*f)", data[i], data[i], d);
+          i += 12;
+          newline(stream, i);
+        }
+      else if (i + 4 <= end
+               && (data[i] && data[i] != 88 && data[i] != 0x41)
+               && !data[i + 1]
+               && !data[i + 2]
+               && !data[i + 3])
+        {
+          fprintf (stream, "i%d ", data[i]);
+          i += 4;
+        }
+      else
+        {
+          fprintf(stream, "%02x ", data[i]);
+          i++;
+        }
+    }
+
+
+}
+
+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(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]))
+  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];
+      int len = data[pos] + data[pos + 1] * 256;
       char *s = malloc(len + 1);
 
       memcpy(s, &data[pos + 4], len);
       s[len] = 0;
-      pos += 4 + data[pos];
+      pos += 4 + len;
       return s;
     }
   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 char *
+dump_counted_string(void)
+{
+  char *s = NULL;
+  int inner_end = pos + get_u32();
+  if (pos != inner_end)
+    {
+      match_u32_assert(0);
+      if (match_byte(0x31))
+        s = get_string();
+      else
+        match_byte_assert(0x58);
+      if (pos != inner_end)
+        {
+          fprintf(stderr, "inner end discrepancy\n");
+          exit(1);
+        }
+    }
+  return s;
+}
+
+static char *
+dump_nested_string(void)
+{
+  char *s = NULL;
+
+  match_byte_assert (0);
+  match_byte_assert (0);
+  int outer_end = pos + get_u32();
+  s = dump_counted_string();
+  match_byte_assert(0x58);
+  match_byte_assert(0x58);
+  if (pos != outer_end)
+    {
+      fprintf(stderr, "outer end discrepancy\n");
+      exit(1);
+    }
+
+  return s;
+}
 
 static void
-dump_category(int level)
+dump_optional_value(FILE *stream)
 {
-  for (int i = 0; i <= level; i++)
-    printf ("    ");
-
-  match_byte (0);
-  match_byte (0);
-  match_byte (0);
-  match_byte (0);
-  if (match_byte (3))
-    {
-      get_string();
-      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);
+  if (match_byte (0x31))
+    {
+      if (match_u32 (0))
+        {
+          if (match_u32 (1))
+            {
+              /* Corpus frequencies:
+                 124 "a"
+                 12 "b"
+                 8 "a, b"
+              */
+              get_string();
+            }
+          else
+            match_u32_assert (0);
+
+          if (version == 1)
+            {
+              /* 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;
+            }
+
+          int outer_end = pos + get_u32();
+
+          /* This counted-string appears to be a template string,
+             e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
+          dump_counted_string();
+
+          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? */
+            }
+          else
+            match_byte_assert(0x58);
+          if (match_byte(0x31))
+            {
+              /* 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);
+              if (!match_byte(8) && !match_byte(1))
+                match_byte_assert(2);
+              match_byte_assert(0);
+              match_byte_assert(8);
+              match_byte_assert(0);
+              match_byte_assert(10);
+              match_byte_assert(0);
+            }
+          else
+            match_byte_assert(0x58);
+          if (pos != outer_end)
+            {
+              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();
+        }
+      else
+        {
+          match_u32_assert(3);
+          fprintf(stream, "(special 3)");
+          match_byte_assert(0);
+          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. */
+        }
     }
-  else if (match_byte (5))
+  else
+    match_byte_assert (0x58);
+}
+
+static const char *
+format_to_string (int type)
+{
+  static char tmp[16];
+  switch (type)
     {
-      match_byte_assert (0x58);
-      printf ("variable \"%s\"", get_string());
-      get_string();
-      if (!match_byte (3))
-        match_byte_assert (2);
-      match_byte (0);
-      match_byte (0);
-      match_byte (0);
+    case 1: return "A";
+    case 2: return "AHEX";
+    case 3: return "COMMA";
+    case 4: return "DOLLAR";
+    case 5: case 40: return "F";
+    case 6: return "IB";
+    case 7: return "PIBHEX";
+    case 8: return "P";
+    case 9: return "PIB";
+    case 10: return "PK";
+    case 11: return "RB";
+    case 12: return "RBHEX";
+    case 15: return "Z";
+    case 16: return "N";
+    case 17: return "E";
+    case 20: return "DATE";
+    case 21: return "TIME";
+    case 22: return "DATETIME";
+    case 23: return "ADATE";
+    case 24: return "JDATE";
+    case 25: return "DTIME";
+    case 26: return "WKDAY";
+    case 27: return "MONTH";
+    case 28: return "MOYR";
+    case 29: return "QYR";
+    case 30: return "WKYR";
+    case 31: return "PCT";
+    case 32: return "DOT";
+    case 33: return "CCA";
+    case 34: return "CCB";
+    case 35: return "CCC";
+    case 36: return "CCD";
+    case 37: return "CCE";
+    case 38: return "EDATE";
+    case 39: return "SDATE";
+    default:
+      abort();
+      sprintf(tmp, "<%d>", type);
+      return tmp;
     }
-  else if (match_byte (2))
+}
+
+static void
+dump_value(FILE *stream, int level)
+{
+  match_byte(0);
+  match_byte(0);
+  match_byte(0);
+  match_byte(0);
+
+  for (int i = 0; i <= level; i++)
+    fprintf (stream, "    ");
+
+  if (match_byte (1))
     {
       unsigned int format;
-      char *var, *vallab;
       double value;
 
-      match_byte_assert (0x58);
+      dump_optional_value(stream);
       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);
+      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 (1))
+  else if (match_byte (2))
     {
       unsigned int format;
+      char *var, *vallab;
       double value;
 
-      match_byte_assert (0x58);
+      dump_optional_value (stream);
       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);
+      var = get_string ();
+      vallab = get_string ();
+      fprintf (stream, "<numeric-datum value=\"%.*g\" format=\"%s%d.%d\"",
+              DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
+      if (var[0])
+        fprintf (stream, " variable=\"%s\"", var);
+      if (vallab[0])
+        fprintf (stream, " label=\"%s\"/>\n", vallab);
+      fprintf (stream, "/>\n");
+      if (!match_byte (1) && !match_byte(2))
+        match_byte_assert (3);
     }
-  else
+  else if (match_byte (3))
     {
-      int subn;
-
-      match_byte_assert (0x31);
-      get_u32 ();
-      get_u32 ();
-      subn = get_u32 ();
-      printf ("nested %d bytes", subn);
-      pos += subn;
-      printf ("; \"%s\"", get_string());
-      goto next;
+      char *text =  get_string();
+      dump_optional_value(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;
 
-  if (match_u32 (2))
-    get_u32 ();
-  else if (match_u32 (1))
+      dump_optional_value(stream);
+      format = get_u32 ();
+      vallab = get_string ();
+      var = get_string ();
+      if (!match_byte(1) && !match_byte(2))
+        match_byte_assert (3);
+      value = get_string ();
+      fprintf (stream, "<string-datum value=\"%s\" format=\"%s%d.%d\"",
+              value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
+      if (var[0])
+        fprintf (stream, " variable=\"%s\"", var);
+      if (vallab[0])
+        fprintf (stream, " label=\"%s\"/>\n", vallab);
+      fprintf (stream, "/>\n");
+    }
+  else if (match_byte (5))
     {
-      match_byte (0);
-      match_byte (0);
-      match_byte (0);
-      get_u32 ();
+      dump_optional_value(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
     {
-      match_u32_assert (0);
-      get_u32 ();
+      dump_optional_value(stream);
+
+      char *base = get_string();
+      int x = get_u32();
+      fprintf (stream, "<template format=\"%s\">\n", base);
+      for (int i = 0; i < x; i++)
+        {
+          int y = get_u32();
+          if (!y)
+            y = 1;
+          else
+            match_u32_assert(0);
+          for (int j = 0; j <= level + 1; j++)
+            fprintf (stream, "    ");
+          fprintf (stream, "<substitution index=\"%d\">\n", i + 1);
+          for (int j = 0; j < y; j++)
+            dump_value (stream, level + 2);
+          for (int j = 0; j <= level + 1; j++)
+            fprintf (stream, "    ");
+          fprintf (stream, "</substitution>\n");
+        }
+      for (int j = 0; j <= level; j++)
+        fprintf (stream, "    ");
+      fprintf (stream, "</template>\n");
     }
-next:;
-  int n_categories = get_u32();
-  if (n_categories > 0)
-    printf (", %d subcategories:", n_categories);
-  printf("\n");
-  for (int i = 0; i < n_categories; i++)
-    dump_category (level + 1);
+}
+
+static int
+compare_int(const void *a_, const void *b_)
+{
+  const int *a = a_;
+  const int *b = b_;
+  return *a < *b ? -1 : *a > *b;
 }
 
 static void
-dump_dim(void)
+check_permutation(int *a, int n, const char *name)
 {
-  int n_categories;
-  if (match_byte(3))
+  int b[n];
+  memcpy(b, a, n * sizeof *a);
+  qsort(b, n, sizeof *b, compare_int);
+  for (int i = 0; i < n; i++)
+    if (b[i] != i)
+      {
+        fprintf(stderr, "bad %s permutation:", name);
+        for (int i = 0; i < n; i++)
+          fprintf(stderr, " %d", a[i]);
+        putc('\n', stderr);
+        exit(1);
+      }
+}
+
+static void
+dump_category(FILE *stream, int level, int *indexes, int *n_indexes, int max_indexes)
+{
+  for (int i = 0; i <= level; i++)
+    fprintf (stream, "    ");
+  printf ("<category>\n");
+  dump_value (stream, level + 1);
+
+  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)
     {
-      get_string();
-      match_byte_assert(0x58);
-      get_string();
-      printf("string \"%s\": ", get_string());
-      match_byte_assert(1);
+      if (merge)
+        {
+          for (int i = 0; i <= level + 1; i++)
+            fprintf (stream, "    ");
+          fprintf (stream, "<merge/>\n");
+        }
     }
-  else if (match_byte(5))
+  else
     {
-      match_byte_assert(0x58);
-      printf("variable \"%s\": ", get_string());
-      get_string();
-      if (!match_byte(2))
-        match_byte_assert(3);
+      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)
+        {
+          fprintf(stderr, "too many categories (increase max_indexes)\n");
+          exit(1);
+        }
+      indexes[(*n_indexes)++] = indx;
     }
-  else
+
+  int expected_unindexed = indx == -1;
+  if (unindexed != expected_unindexed)
     {
-      fprintf(stderr, "%08x: unexpected byte\n", pos);
+      fprintf(stderr, "unindexed (%d) mismatch with indx (%d)\n",
+              unindexed, indx);
       exit(1);
     }
 
-  match_byte_assert(0);
+  if (n_categories == 0)
+    {
+      for (int i = 0; i <= level + 1; i++)
+        fprintf (stream, "    ");
+      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);
+  for (int i = 0; i <= level; i++)
+    fprintf (stream, "    ");
+  printf ("</category>\n");
+}
+
+static int
+dump_dim(int indx)
+{
+  int n_categories;
+
+  printf ("<dimension index=\"%d\">\n", indx);
+  dump_value (stdout, 0);
+
+  /* This byte is usually 0 but many other values have been spotted. */
+  pos++;
+
   if (!match_byte(0) && !match_byte(1))
     match_byte_assert(2);
-  match_u32_assert(2);
+  if (!match_u32(0))
+    match_u32_assert(2);
   if (!match_byte(0))
     match_byte_assert(1);
-  match_byte(0);
-  match_byte(0);
-  match_byte(0);
-  match_byte(0);
-  get_u32();
-  match_byte(0);
-  match_byte(0);
-  match_byte(0);
-  match_byte(0);
+  if (!match_byte(0))
+    match_byte_assert(1);
+  match_byte_assert(1);
+  if (!match_u32(UINT32_MAX))
+    match_u32_assert(indx);
   n_categories = get_u32();
-  printf("%d nested categories\n", n_categories);
+
+  int indexes[2048];
+  int n_indexes = 0;
   for (int i = 0; i < n_categories; i++)
-    dump_category (0);
+    dump_category (stdout, 0, indexes, &n_indexes, sizeof indexes / sizeof *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)
 {
-  int n_dims = get_u32();
+  n_dims = get_u32();
+  assert(n_dims < MAX_DIMS);
+  for (int i = 0; i < n_dims; i++)
+    dim_n_cats[i] = dump_dim (i);
+}
 
-  printf ("%u dimensions\n", n_dims);
+static void
+dump_data(void)
+{
+  /* The first three numbers add to the number of dimensions. */
+  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++)
     {
-      printf("\n");
-      dump_dim ();
+      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++)
+    {
+      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);
+      if (version == 1)
+        match_byte(0);
+      dump_value(stdout, 1);
+      fprintf (stdout, "    </datum>\n");
+    }
+  printf ("</data>\n");
+}
+
+static void
+dump_title(void)
+{
+  printf ("<title-local>\n");
+  dump_value(stdout, 0);
+  match_byte(1);
+  printf ("</title-local>\n");
+
+  printf ("<subtype>\n");
+  dump_value(stdout, 0);
+  match_byte(1);
+  printf ("</subtype>\n");
+
+  match_byte_assert(0x31);
+
+  printf ("<title-c>\n");
+  dump_value(stdout, 0);
+  match_byte(1);
+  printf ("</title-c>\n");
+
+  match_byte(0);
+  match_byte_assert(0x58);
+  if (match_byte(0x31))
+    {
+      printf ("<caption>\n");
+      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);
+      /* Custom footnote marker string. */
+      if (match_byte (0x31))
+        dump_value(stdout, 0);
+      else
+        match_byte_assert (0x58);
+      get_u32 ();
+      printf ("</footnote>\n");
+    }
+}
+
+static void
+dump_fonts(void)
+{
+  match_byte(0);
+  for (int i = 1; i <= 8; i++)
+    {
+      printf ("<style index=\"%d\"", i);
+      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);
+        }
+
+      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 (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 ("/>\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
+    {
+      match_u32_assert(16);
+      pos += 16;
+    }
+
+  int count = get_u32();
+  pos += 4 * count;
+
+  const char *encoding = get_string();
+  printf ("<encoding>%s</encoding>\n", encoding);
+
+  if (!match_u32(0))
+    match_u32_assert(UINT32_MAX);
+  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);
+
+  int decimal = data[pos];
+  int grouping = data[pos + 1];
+  if (match_byte('.'))
+    {
+      if (!match_byte(',') && !match_byte('\''))
+        match_byte_assert(' ');
+    }
+  else
+    {
+      match_byte_assert(',');
+      if (!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++)
+        printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
     }
+  else
+    match_u32_assert(0);
+  int skip = get_u32();
+  pos += skip;
 }
 
 int
@@ -310,97 +949,136 @@ main(int argc, char *argv[])
       exit(1);
     }
 
-  if (argc > 1)
+  if (argc != 2)
     {
-      if (!strcmp(argv[1], "title"))
-        {
-          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"))
-        {
-          const char dimensions[] = "-,,,.\0";
-          start = find(dimensions, sizeof dimensions - 1) + sizeof dimensions - 1;
-          pos = start;
-          dump_dims ();
-          return 0;
-        }
-      else
-        {
-          fprintf (stderr, "unknown section %s\n", argv[1]);
-          exit(1);
-        }
+      fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
+      exit (1);
     }
-  else
-    start = 0x27;
 
-  for (size_t i = start; i < n; )
+  if (!strcmp(argv[1], "title0"))
     {
-      if (i + 5 <= n
-          && data[i]
-          && !data[i + 1]
-          && !data[i + 2]
-          && !data[i + 3]
-          && i + 4 + data[i] <= n
-          && all_ascii(&data[i + 4], data[i]))
-        {
-          fputs("\n\"", stdout);
-          fwrite(&data[i + 4], 1, data[i], stdout);
-          fputs("\" ", stdout);
+      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);
 
-          i += 4 + data[i];
-        }
-      else if (i + 12 <= n
-               && data[i + 1] == 40
-               && data[i + 2] == 5
-               && data[i + 3] == 0)
-        {
-          double d;
+      /* 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(). */
+      version = get_u32();
+      pos -= 4;
+      if (!match_u32(1))
+        match_u32_assert(3);
 
-          memcpy (&d, &data[i + 4], 8);
-          printf ("F40.%d(%.*f)\n", data[i], data[i], d);
-          i += 12;
-        }
-      else if (i + 12 <= n
-               && data[i + 1] == 40
-               && data[i + 2] == 31
-               && data[i + 3] == 0)
-        {
-          double d;
+      match_byte_assert(1);
+      if (!match_byte(0))
+        match_byte_assert(1);
 
-          memcpy (&d, &data[i + 4], 8);
-          printf ("PCT40.%d(%.*f)\n", data[i], data[i], d);
-          i += 12;
-        }
-      else if (i + 4 <= n
-               && (data[i] && data[i] != 88 && data[i] != 0x41)
-               && !data[i + 1]
-               && !data[i + 2]
-               && !data[i + 3])
-        {
-          printf ("i%d ", data[i]);
-          i += 4;
-        }
-      else
+      /* Offset 8. */
+      match_byte_assert(0);
+      match_byte_assert(0);
+      if (!match_byte(0))
+        match_byte_assert(1);
+
+      /* Offset 11. */
+      pos++;
+      match_byte_assert(0);
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      /* Offset 15. */
+      pos++;
+      if (!match_byte(0))
+        match_byte_assert(1);
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      /* Offset 19. */
+      pos++;
+      if (!match_byte(0))
+        match_byte_assert(1);
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      /* Offset 23. */
+      pos++;
+      if (!match_byte(0))
+        match_byte_assert(1);
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      /* Offset 27. */
+      pos++;
+      pos++;
+      match_byte_assert(0);
+      match_byte_assert(0);
+
+      /* Offset 31.
+
+         This is the tableId, e.g. -4154297861994971133 would be 0xdca00003.
+         We don't have enough context to validate it. */
+      pos += 4;
+
+      /* Offset 35. */
+      pos += 4;
+
+      dump_title ();
+      dump_fonts();
+      dump_dims ();
+      dump_data ();
+      match_byte (1);
+      if (pos != n)
         {
-          printf("%02x ", data[i]);
-          i++;
+          fprintf (stderr, "%x / %x\n", pos, n);
+          exit(1);
         }
+      exit(0);
+    }
+  else if (!strcmp(argv[1], "raw"))
+    {
+      start = 0x27;
+
+      dump_raw(stdout, start, n);
+    }
+  else
+    {
+      fprintf (stderr, "unknown section %s\n", argv[1]);
+      exit(1);
     }
 
   return 0;