X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=c247c4af661be8c50ef6312aa18f05c524088b68;hb=38a126b38f4963ba033120754e7a4f026e9fa01e;hp=cba4509cd41ebdecdff224aba3c46e5a7ba374b3;hpb=92a04dcbe26bed7d6c044f06414bc54e0dbb1ad1;p=pspp

diff --git a/dump.c b/dump.c
index cba4509cd4..c247c4af66 100644
--- a/dump.c
+++ b/dump.c
@@ -1,3 +1,4 @@
+#include <assert.h>
 #include <float.h>
 #include <stdbool.h>
 #include <stdint.h>
@@ -6,6 +7,7 @@
 #include <string.h>
 #include <sys/stat.h>
 #include <unistd.h>
+#include "u8-mbtouc.h"
 
 static uint8_t *data;
 static size_t n;
@@ -177,6 +179,22 @@ 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)
 {
@@ -201,13 +219,9 @@ get_string(const char *where)
 #define get_string() get_string(WHERE)
 
 static char *
-dump_nested_string(void)
+dump_counted_string(void)
 {
   char *s = NULL;
-
-  match_byte_assert (0);
-  match_byte_assert (0);
-  int outer_end = pos + get_u32();
   int inner_end = pos + get_u32();
   if (pos != inner_end)
     {
@@ -222,6 +236,18 @@ dump_nested_string(void)
           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)
@@ -234,7 +260,7 @@ dump_nested_string(void)
 }
 
 static void
-dump_value_31(FILE *stream)
+dump_optional_value(FILE *stream)
 {
   if (match_byte (0x31))
     {
@@ -242,7 +268,11 @@ dump_value_31(FILE *stream)
         {
           if (match_u32 (1))
             {
-              /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
+              /* Corpus frequencies:
+                 124 "a"
+                 12 "b"
+                 8 "a, b"
+              */
               get_string();
             }
           else
@@ -265,25 +295,10 @@ dump_value_31(FILE *stream)
             }
 
           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);
-                }
-            }
+
+          /* 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))
             {
@@ -340,7 +355,7 @@ dump_value_31(FILE *stream)
           match_byte_assert(0);
           if (!match_u32 (2) && !match_u32(1))
             match_u32_assert(3);
-          dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+          dump_nested_string();
         }
       else
         {
@@ -417,33 +432,16 @@ dump_value(FILE *stream, int level)
   for (int i = 0; i <= level; i++)
     fprintf (stream, "    ");
 
-  if (match_byte (3))
+  if (match_byte (1))
     {
-      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);
+      unsigned int format;
+      double value;
 
-    }
-  else if (match_byte (5))
-    {
-      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);
+      dump_optional_value(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))
     {
@@ -451,7 +449,7 @@ dump_value(FILE *stream, int level)
       char *var, *vallab;
       double value;
 
-      dump_value_31 (stream);
+      dump_optional_value (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -466,12 +464,27 @@ dump_value(FILE *stream, int level)
       if (!match_byte (1) && !match_byte(2))
         match_byte_assert (3);
     }
+  else if (match_byte (3))
+    {
+      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;
 
-      match_byte_assert (0x58);
+      dump_optional_value(stream);
       format = get_u32 ();
       vallab = get_string ();
       var = get_string ();
@@ -486,20 +499,21 @@ dump_value(FILE *stream, int level)
         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);
+      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
     {
-      dump_value_31(stream);
+      dump_optional_value(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -552,30 +566,51 @@ 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 *n_indexes, int max_indexes)
 {
   for (int i = 0; i <= level; i++)
-    fprintf (stdout, "    ");
+    fprintf (stream, "    ");
   printf ("<category>\n");
-  dump_value (stdout, level + 1);
-  match_u32(1);
+  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");
@@ -588,20 +623,29 @@ dump_category(int level, int *indexes, int *n_indexes, int max_indexes)
         }
       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, n_indexes, max_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;
@@ -609,7 +653,7 @@ dump_dim(int indx)
   printf ("<dimension index=\"%d\">\n", indx);
   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))
@@ -628,40 +672,64 @@ dump_dim(int indx)
   int indexes[2048];
   int n_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, &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)
 {
   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);
       if (version == 1)
         match_byte(0);
@@ -674,7 +742,6 @@ dump_data(void)
 static void
 dump_title(void)
 {
-  pos = 0x27;
   printf ("<title-local>\n");
   dump_value(stdout, 0);
   match_byte(1);
@@ -703,24 +770,17 @@ dump_title(void)
   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))
-        {
-          /* 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");
     }
 }
@@ -739,15 +799,19 @@ dump_fonts(void)
       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);
+      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))
-        match_u32_assert(0xfaad);
+        {
+          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);
@@ -759,7 +823,6 @@ dump_fonts(void)
 
       if (version > 1)
         {
-          /* These seem unlikely to be correct too. */
           if (i != 3)
             {
               if (!match_u32(8))
@@ -808,7 +871,8 @@ dump_fonts(void)
   int count = get_u32();
   pos += 4 * count;
 
-  printf ("<encoding>%s</encoding>\n", get_string ());
+  const char *encoding = get_string();
+  printf ("<encoding>%s</encoding>\n", encoding);
 
   if (!match_u32(0))
     match_u32_assert(UINT32_MAX);
@@ -827,9 +891,12 @@ dump_fonts(void)
     }
   else
     match_u32_assert(UINT32_MAX);
+
+  int decimal = data[pos];
+  int grouping = data[pos + 1];
   if (match_byte('.'))
     {
-      if (!match_byte(','))
+      if (!match_byte(',') && !match_byte('\''))
         match_byte_assert(' ');
     }
   else
@@ -838,10 +905,14 @@ dump_fonts(void)
       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++)
-        get_string();
+        printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
     }
   else
     match_u32_assert(0);
@@ -878,90 +949,137 @@ 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);
-
-          /* 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);
+      fprintf (stderr, "usage: %s TYPE < .bin", argv[0]);
+      exit (1);
+    }
 
-          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);
-        }
+  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_optional_value(). */
+      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);
+
+      /* 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)
         {
-          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;
 }