Figure out a few more details of one of the chunks in the fonts.

[pspp] / dump2.c

diff --git a/dump2.c b/dump2.c
index 4fd7531ab530df5860a01cb27195008f47aef06d..d7f8ad3d68abbcff43539cae1f56ef4515aa2e06 100644 (file)
--- a/dump2.c
+++ b/dump2.c
@@ -7,6 +7,7 @@
 #include <string.h>
 #include <sys/stat.h>
 #include <unistd.h>
+#include "u8-mbtouc.h"
 
 static uint8_t *data;
 static size_t n, pos;
@@ -86,10 +87,25 @@ all_ascii(const uint8_t *p)
   return true;
 }
 
+static bool
+all_utf8(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 == 127 || uc == 0xfffd)
+        return false;
+    }
+  return true;
+}
+
 static char *
 get_fixed_string(int len, const char *where)
 {
-  if (pos + len > n || !memchr(&data[pos], 0, len) || !all_ascii(&data[pos]))
+  if (pos + len > n || !memchr(&data[pos], 0, len) || !all_utf8(&data[pos]))
     {
       fprintf(stderr, "%s: 0x%x: bad fixed-width string\n", where, pos);
       exit(1);
@@ -190,7 +206,7 @@ dump_raw(FILE *stream, int start, int end, const char *separator)
 }
 
 static void
-dump_source(int count, int n_series)
+dump_source(int end, int count, int n_series)
 {
   const union
     {
@@ -216,10 +232,10 @@ dump_source(int count, int n_series)
       printf ("\n");
     }
 
-  if (!n_sysmis)
+  if (pos >= end)
     return;
 
-  printf ("\n    %08x:", pos);
+  printf ("\n    %08x: (%d sysmis)", pos, n_sysmis);
   printf (" %d", get_u32());
   printf (", \"%s\"\n", get_string());
 
@@ -284,6 +300,7 @@ main(int argc, char **argv)
 
   pos = 0;
   match_byte_assert(0);
+  int version = data[pos];
   if (!match_byte(0xaf))
     match_byte_assert(0xb0);
   int n_sources = data[pos++];
@@ -291,19 +308,36 @@ main(int argc, char **argv)
 
   match_u32_assert(s.st_size);
   printf ("%d sources\n", n_sources);
+
+  struct source
+    {
+      int offset, count, n_series;
+    }
+  sources[n_sources];
   for (int i = 0; i < n_sources; i++)
     {
-      pos = 8 + 80 * i;
       int count = get_u32();
       int n_series = get_u32();
       int offset = get_u32();
-      char *name = get_fixed_string(64);
-      int dunno = get_u32();
+      char *name = get_fixed_string(version == 0xb0 ? 64 : 28);
+      int dunno = version == 0xb0 ? get_u32() : 0;
       printf ("source %d: %d series, %d observations per series, offset %08x, \"%s\", %x\n",
               i, n_series, count, offset, name, dunno);
-      pos = offset;
-      dump_source(count, n_series);
+      sources[i].offset = offset;
+      sources[i].count = count;
+      sources[i].n_series = n_series;
+    }
+
+  for (int i = 0; i < n_sources; i++)
+    {
+      if (pos != sources[i].offset)
+        {
+          fprintf (stderr, "pos=0x%x expected=0x%x reading source %d\n", pos, sources[i].offset, i);
+          exit(1);
+        }
+      dump_source(i + 1 >= n_sources ? n : sources[i + 1].offset, sources[i].count, sources[i].n_series);
     }
+  assert(pos == n);
 
   return 0;
 }