X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=55ad4bc64ca55316839a244f0a7bb1692d023548;hb=eb5a888913207139985fcd85886b6b9c36aa01a0;hp=e1446edeb8189b3094742fdbc866013a4c054a11;hpb=016b5f18d03c3eafe8e0b14f0db2c9e23d6bfc03;p=pspp

diff --git a/dump.c b/dump.c
index e1446edeb8..55ad4bc64c 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)
 {
@@ -26,13 +28,6 @@ try_find(const char *target, size_t target_len)
   return pos ? pos - data : 0;
 }
 
-static size_t
-try_find_tail(const char *target, size_t target_len)
-{
-  size_t pos = try_find(target, target_len);
-  return pos ? pos + target_len : 0;
-}
-
 static size_t
 find(const char *target, size_t target_len)
 {
@@ -45,18 +40,6 @@ find(const char *target, size_t target_len)
   return pos;
 }
 
-static size_t
-find_tail(const char *target, size_t target_len)
-{
-  size_t pos = try_find_tail(target, target_len);
-  if (!pos)
-    {
-      fprintf (stderr, "not found\n");
-      exit(1);
-    }
-  return pos;
-}
-
 size_t pos;
 
 #define XSTR(x) #x
@@ -126,7 +109,13 @@ match_byte_assert(uint8_t b, const char *where)
 #define match_byte_assert(b) match_byte_assert(b, WHERE)
 
 static void
-dump_raw(FILE *stream, int start, int end, const char *separator)
+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; )
     {
@@ -138,7 +127,8 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           && i + 4 + data[i] + data[i + 1] * 256 <= end
           && all_ascii(&data[i + 4], data[i] + data[i + 1] * 256))
         {
-          fprintf(stream, "%s\"", separator);
+          newline(stream, i);
+          fprintf(stream, "\"");
           fwrite(&data[i + 4], 1, data[i] + data[i + 1] * 256, stream);
           fputs("\" ", stream);
 
@@ -152,8 +142,9 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           double d;
 
           memcpy (&d, &data[i + 4], 8);
-          fprintf (stream, "F40.%d(%.*f)%s", data[i], data[i], d, separator);
+          fprintf (stream, "F40.%d(%.*f)", data[i], data[i], d);
           i += 12;
+          newline (stream, i);
         }
       else if (i + 12 <= end
                && data[i + 1] == 40
@@ -163,8 +154,9 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
           double d;
 
           memcpy (&d, &data[i + 4], 8);
-          fprintf (stream, "PCT40.%d(%.*f)%s", data[i], data[i], d, separator);
+          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)
@@ -256,27 +248,47 @@ 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))
             {
               /* Only one example in the corpus. */
+              match_byte(1);
               match_byte(0);
               match_byte(0);
               match_byte(0);
@@ -284,11 +296,31 @@ dump_value_31(FILE *stream)
               get_string();     /* foreground */
               get_string();     /* background */
               get_string();     /* font */
-              match_byte_assert(12); /* size? */
+              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);
-          match_byte_assert(0x58);
           if (pos != outer_end)
             {
               fprintf(stderr, "outer end discrepancy\n");
@@ -303,10 +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
@@ -374,7 +407,7 @@ format_to_string (int type)
 }
 
 static void
-dump_value__(FILE *stream, int level, bool match1)
+dump_value(FILE *stream, int level, bool match1)
 {
   match_byte(0);
   match_byte(0);
@@ -419,7 +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 ();
@@ -485,7 +518,7 @@ 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, false);
           for (int j = 0; j <= level + 1; j++)
             fprintf (stream, "    ");
           fprintf (stream, "</substitution>\n");
@@ -522,9 +555,12 @@ 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)
 {
-  dump_value__ (stdout, level, true);
+  for (int i = 0; i <= level; i++)
+    fprintf (stdout, "    ");
+  printf ("<category>\n");
+  dump_value (stdout, level + 1, true);
   match_byte(0);
   match_byte(0);
   match_byte(0);
@@ -550,25 +586,35 @@ 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)
-    printf (", %d subcategories:", n_categories);
-  else
-    printf (", index %d", indx);
-  printf("\n");
+  if (n_categories == 0)
+    {
+      for (int i = 0; i <= level + 1; i++)
+        fprintf (stdout, "    ");
+      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");
 }
 
 static void
-dump_dim(void)
+dump_dim(int indx)
 {
   int n_categories;
-  printf("next dim\n");
-  dump_value__ (stdout, 0, false);
 
-  /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
+  printf ("<dimension index=\"%d\">\n", indx);
+  dump_value (stdout, 0, false);
+
+  /* This byte is usually 0x02 but many other values have been spotted. */
   pos++;
 
   if (!match_byte(0) && !match_byte(1))
@@ -580,16 +626,17 @@ dump_dim(void)
   if (!match_byte(0))
     match_byte_assert(1);
   match_byte_assert(1);
-  static int dim_indx = 0;
-  match_u32_assert(dim_indx++);
+  if (!match_u32(UINT32_MAX))
+    match_u32_assert(indx);
   n_categories = get_u32();
-  printf("%d nested categories\n", n_categories);
 
-  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");
 }
 
 int n_dims;
@@ -597,12 +644,8 @@ static void
 dump_dims(void)
 {
   n_dims = get_u32();
-  printf ("%u dimensions\n", n_dims);
   for (int i = 0; i < n_dims; i++)
-    {
-      printf("\n");
-      dump_dim ();
-    }
+    dump_dim (i);
 }
 
 static void
@@ -620,101 +663,88 @@ dump_data(void)
   check_permutation(a, n_dims, "dimensions");
 
   int x = get_u32();
-  printf ("%d data values, starting at %08x\n", x, pos);
+  printf ("<data>\n");
   for (int i = 0; i < x; i++)
     {
-      printf("%08x, index %d:\n", pos, get_u32());
+      printf ("    <datum index=\"%d\">\n", get_u32());
       match_u32_assert(0);
-      dump_value__(stdout, 0, false);
-      putchar('\n');
+      if (version == 1)
+        match_byte(0);
+      dump_value(stdout, 1, false);
+      fprintf (stdout, "    </datum>\n");
     }
+  printf ("</data>\n");
 }
 
 static void
 dump_title(void)
 {
   pos = 0x27;
-  printf("text:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
-  printf("subtype:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
+  printf ("<title-local>\n");
+  dump_value(stdout, 0, true);
+  printf ("</title-local>\n");
+
+  printf ("<subtype>\n");
+  dump_value(stdout, 0, true);
+  printf ("</subtype>\n");
+
   match_byte_assert(0x31);
-  printf("text_eng:\n");
-  dump_value__(stdout, 0, true); putchar('\n');
+
+  printf ("<title-c>\n");
+  dump_value(stdout, 0, true);
+  printf ("</title-c>\n");
+
   match_byte(0);
   match_byte_assert(0x58);
   if (match_byte(0x31))
     {
-      printf("caption:\n");
-      dump_value__(stdout, 0, false); putchar('\n');
+      printf ("<caption>\n");
+      dump_value(stdout, 0, false);
+      printf ("</caption>\n");
     }
   else
     match_byte_assert(0x58);
 
 
   int n_footnotes = get_u32();
-  if (n_footnotes >= 20)
+  for (int i = 0; i < n_footnotes; i++)
     {
-      fprintf(stderr, "%08x: %d footnotes\n", pos - 4, n_footnotes);
-      exit(1);
-    }
-
-  printf("------\n%d footnotes\n", n_footnotes);
-  if (n_footnotes < 20)
-    {
-      for (int i = 0; i < n_footnotes; i++)
+      printf ("<footnote index=\"%d\">\n", i);
+      dump_value(stdout, 0, false);
+      if (match_byte (0x31))
         {
-          printf("footnote %d:\n", i);
-          dump_value__(stdout, 0, false);
-          if (match_byte (0x31))
-            {
-              /* Custom footnote marker string. */
-              match_byte_assert(3);
-              get_string();
-              match_byte_assert(0x58);
-              match_u32_assert(0);
-              get_string();
-            }
-          else
-            match_byte_assert (0x58);
-          printf("(%d)\n", get_u32());
+          /* Custom footnote marker string. */
+          match_byte_assert(3);
+          get_string();
+          match_byte_assert(0x58);
+          match_u32_assert(0);
+          get_string();
         }
+      else
+        match_byte_assert (0x58);
+      printf("(%d)\n", get_u32());
+      printf ("</footnote>\n");
     }
 }
 
-static int
-find_dimensions(void)
-{
-  {
-    const char dimensions[] = "-,,,.\0";
-    int x = try_find_tail(dimensions, sizeof dimensions - 1);
-    if (x)
-      return x;
-  }
-
-  const char dimensions[] = "-,,, .\0";
-  return find_tail(dimensions, sizeof dimensions - 1);
-}
-
 static void
 dump_fonts(void)
 {
-  printf("fonts: offset=%08x\n", pos);
   match_byte(0);
   for (int i = 1; i <= 8; i++)
     {
-      printf("%08x: font %d, ", pos, i);
+      printf ("<style index=\"%d\"", i);
       match_byte_assert(i);
       match_byte_assert(0x31);
-      printf("%s, ", get_string());
+      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... */
@@ -723,31 +753,38 @@ dump_fonts(void)
 
       if (!match_u32(0) && !match_u32(1) && !match_u32(2))
         match_u32_assert(3);
-      printf ("%s, ", get_string());
-      printf ("%s, ", get_string());
+      printf (" fgcolor=\"%s\"", get_string());
+      printf (" bgcolor=\"%s\"", get_string());
       match_u32_assert(0);
       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");
     }
 
   match_u32_assert(240);
@@ -758,17 +795,22 @@ 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();
   pos += 4 * count;
 
-  char *encoding = get_string();
-  printf("encoding=%s\n", encoding);
+  printf ("<encoding>%s</encoding>\n", get_string ());
 
   if (!match_u32(0))
     match_u32_assert(UINT32_MAX);
@@ -777,25 +819,36 @@ dump_fonts(void)
   match_byte_assert(0);
   if (!match_byte(0))
     match_byte_assert(1);
-  if (!match_byte(0x99) && !match_byte(0x98))
-    match_byte_assert(0x97);
-  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);
+    }
+  if (match_u32(5))
+    {
+      for (int i = 0; i < 5; i++)
+        get_string();
     }
-  match_u32_assert(5);
-  for (int i = 0; i < 5; i++)
-    get_string();
-  pos += get_u32();
-  if (pos != find_dimensions())
-    fprintf (stderr, "%08x / %08x\n", pos, find_dimensions());
+  else
+    match_u32_assert(0);
+  int skip = get_u32();
+  pos += skip;
 }
 
 int
@@ -827,83 +880,95 @@ 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);
-          match_u32_assert(3);
-          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 ();
-          printf("\n\ndata:\n");
-          dump_data ();
-          match_byte (1);
-          if (pos != n)
-            {
-              fprintf (stderr, "%x / %x\n", pos, n);
-              exit(1);
-            }
-          exit(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"))
+    {
+      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);
+
+      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, "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, "\n");
+      dump_raw(stdout, start, n);
+    }
+  else
+    {
+      fprintf (stderr, "unknown section %s\n", argv[1]);
+      exit(1);
+    }
 
   return 0;
 }