X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump2.c;h=15670ecf4bb7c8de1ba8986290fa0fde8f53046f;hb=f926879f4be2b8eca78860e22f00d1001b91c175;hp=af01b408cdc5f77c1ae6021458d1001e1e6473cd;hpb=324edb7f2fc042792ae418618d6b6d04d16a54a3;p=pspp

diff --git a/dump2.c b/dump2.c
index af01b408cd..15670ecf4b 100644
--- 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 end, int count, int n_series)
+dump_source(int end, int count, int n_series, const char *name)
 {
   const union
     {
@@ -219,14 +235,19 @@ dump_source(int end, int count, int n_series)
   if (pos >= end)
     return;
 
-  printf ("\n    %08x: (%d sysmis)", pos, n_sysmis);
-  printf (" %d", get_u32());
-  printf (", \"%s\"\n", get_string());
+  match_u32_assert(1);
+  char *name2 = get_string();
+  assert(!strcmp(name, name2));
 
   printf ("\n    %08x:", pos);
   int n_more_series = get_u32();
+  if (n_series != n_more_series)
+    printf("different series counts: %d %d\n", n_series, n_more_series);
+  assert(n_more_series <= n_series);
   printf (" %d series to come\n", n_more_series);
 
+  int max1 = -1;
+  int ofs = pos;
   for (int i = 0; i < n_more_series; i++)
     {
       printf ("%08x:", pos);
@@ -236,21 +257,45 @@ dump_source(int end, int count, int n_series)
         {
           int x = get_u32();
           int y = get_u32();
-          printf (" (%d,%d)", x, y);
+          printf (" (%d, %d)", x, y);
+          if (y > max1)
+            max1 = y;
         }
       printf ("\n");
     }
 
   printf ("\n%08x:", pos);
   int n_strings = get_u32();
+  assert(n_strings == max1 + 1);
   printf (" %d strings\n", n_strings);
+
+  char **strings = malloc(n_strings * sizeof *strings);
   for (int i = 0; i < n_strings; i++)
     {
-      int x = get_u32();
+      int frequency = get_u32();
       char *s = get_string();
-      printf ("%d: \"%s\" (%d)\n", i, s, x);
+      printf ("%d: \"%s\" (%d)\n", i, s, frequency);
+      strings[i] = s;
     }
   printf ("\n");
+
+  assert (pos == end);
+  pos = ofs;
+  printf("Strings:\n");
+  for (int i = 0; i < n_more_series; i++)
+    {
+      printf ("  \"%s\"\n", get_string());
+      int n_pairs = get_u32();
+      for (int j = 0; j < n_pairs; j++)
+        {
+          int x = get_u32();
+          //assert (x == j);
+          int y = get_u32();
+          printf ("    %d: \"%s\"\n", x, strings[y]);
+        }
+      printf ("\n");
+    }
+  pos = end;
 }
 
 int
@@ -296,6 +341,7 @@ main(int argc, char **argv)
   struct source
     {
       int offset, count, n_series;
+      char *name;
     }
   sources[n_sources];
   for (int i = 0; i < n_sources; i++)
@@ -310,6 +356,7 @@ main(int argc, char **argv)
       sources[i].offset = offset;
       sources[i].count = count;
       sources[i].n_series = n_series;
+      sources[i].name = name;
     }
 
   for (int i = 0; i < n_sources; i++)
@@ -319,7 +366,7 @@ main(int argc, char **argv)
           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);
+      dump_source(i + 1 >= n_sources ? n : sources[i + 1].offset, sources[i].count, sources[i].n_series, sources[i].name);
     }
   assert(pos == n);