X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=1cdb749f81fc714f09876a9701391cb48416e2bb;hb=2aed65a53e0d5ae8d7abc77f6cbd7cf055b37ceb;hp=9b90fe651d62db98b82be547cb8618e99abeb384;hpb=6f432d8f62b7298cc4d6897a42e471774bd55938;p=pspp

diff --git a/dump.c b/dump.c
index 9b90fe651d..1cdb749f81 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)
 {
@@ -234,7 +252,7 @@ dump_nested_string(void)
 }
 
 static void
-dump_value_31(FILE *stream)
+dump_optional_value(FILE *stream)
 {
   if (match_byte (0x31))
     {
@@ -420,7 +438,7 @@ dump_value(FILE *stream, int level)
   if (match_byte (3))
     {
       char *text = get_string();
-      dump_value_31(stream);
+      dump_optional_value(stream);
       char *identifier = get_string();
       char *text_eng = get_string();
       fprintf (stream, "<string c=\"%s\"", text_eng);
@@ -434,7 +452,7 @@ dump_value(FILE *stream, int level)
     }
   else if (match_byte (5))
     {
-      dump_value_31(stream);
+      dump_optional_value(stream);
       char *name = get_string ();
       char *label = get_string ();
       fprintf (stream, "<variable name=\"%s\"", name);
@@ -450,7 +468,7 @@ 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 ();
@@ -490,7 +508,7 @@ dump_value(FILE *stream, int level)
       unsigned int format;
       double value;
 
-      dump_value_31(stream);
+      dump_optional_value(stream);
       format = get_u32 ();
       value = get_double ();
       fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
@@ -498,7 +516,7 @@ dump_value(FILE *stream, int level)
     }
   else
     {
-      dump_value_31(stream);
+      dump_optional_value(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -551,33 +569,51 @@ 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_byte(1);
-  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");
@@ -590,16 +626,25 @@ 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");
 }
 
@@ -611,7 +656,7 @@ 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))
@@ -630,7 +675,7 @@ 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");
@@ -649,14 +694,21 @@ 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();
@@ -676,7 +728,6 @@ dump_data(void)
 static void
 dump_title(void)
 {
-  pos = 0x27;
   printf ("<title-local>\n");
   dump_value(stdout, 0);
   match_byte(1);
@@ -705,24 +756,17 @@ 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");
     }
 }
@@ -741,15 +785,19 @@ 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);
@@ -761,7 +809,6 @@ dump_fonts(void)
 
       if (version > 1)
         {
-          /* These seem unlikely to be correct too. */
           if (i != 3)
             {
               if (!match_u32(8))
@@ -810,7 +857,8 @@ 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);
@@ -829,6 +877,9 @@ dump_fonts(void)
     }
   else
     match_u32_assert(UINT32_MAX);
+
+  int decimal = data[pos];
+  int grouping = data[pos + 1];
   if (match_byte('.'))
     {
       if (!match_byte(',') && !match_byte('\''))
@@ -840,10 +891,14 @@ 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);
@@ -898,6 +953,7 @@ main(int argc, char *argv[])
     }
   else if (!strcmp(argv[1], "title"))
     {
+      pos = 0x27;
       dump_title();
       exit(0);
     }
@@ -929,7 +985,7 @@ main(int argc, char *argv[])
 
       /* This might be a version number of some kind, because value 1 seems
          to only appear in an SPV file that also required its own weird
-         special cases in dump_value_31(). */
+         special cases in dump_optional_value(). */
       version = get_u32();
       pos -= 4;
       if (!match_u32(1))
@@ -938,14 +994,55 @@ main(int argc, char *argv[])
       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 ();