X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=dump.c;h=e37202000c64f310a309f872ccbbd0ae2f46cc96;hb=9692b579384a142daf732706ad2f4401481a3077;hp=a7ccbdc88b1cbb1facd1b19015a1fdfdbf13f229;hpb=f7ab332bead0665caa4f3ccc7dcf28098f8b0406;p=pspp

diff --git a/dump.c b/dump.c
index a7ccbdc88b..e37202000c 100644
--- a/dump.c
+++ b/dump.c
@@ -1,3 +1,4 @@
+#include <assert.h>
 #include <float.h>
 #include <stdbool.h>
 #include <stdint.h>
@@ -6,10 +7,13 @@
 #include <string.h>
 #include <sys/stat.h>
 #include <unistd.h>
+#include "u8-mbtouc.h"
 
 static uint8_t *data;
 static size_t n;
 
+int version;
+
 static bool
 all_ascii(const uint8_t *p, size_t n)
 {
@@ -175,6 +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)
 {
@@ -232,7 +252,7 @@ dump_nested_string(void)
 }
 
 static void
-dump_value_31(FILE *stream)
+dump_optional_value(FILE *stream)
 {
   if (match_byte (0x31))
     {
@@ -246,22 +266,41 @@ 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))
@@ -280,7 +319,26 @@ dump_value_31(FILE *stream)
             }
           else
             match_byte_assert(0x58);
-          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);
           if (pos != outer_end)
             {
               fprintf(stderr, "outer end discrepancy\n");
@@ -295,10 +353,11 @@ 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
@@ -366,7 +425,7 @@ format_to_string (int type)
 }
 
 static void
-dump_value(FILE *stream, int level, bool match1)
+dump_value(FILE *stream, int level)
 {
   match_byte(0);
   match_byte(0);
@@ -379,7 +438,7 @@ dump_value(FILE *stream, int level, bool match1)
   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);
@@ -390,12 +449,10 @@ dump_value(FILE *stream, int level, bool match1)
       fprintf (stream, "/>\n");
       if (!match_byte (0))
         match_byte_assert(1);
-      if (match1)
-        match_byte (1);
     }
   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);
@@ -411,7 +468,7 @@ dump_value(FILE *stream, int level, bool match1)
       char *var, *vallab;
       double value;
 
-      match_byte_assert (0x58);
+      dump_optional_value (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -451,17 +508,15 @@ dump_value(FILE *stream, int level, bool match1)
       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",
                DBL_DIG, value, format_to_string(format >> 16), (format >> 8) & 0xff, format & 0xff);
-      if (match1)
-        match_byte (1);
     }
   else
     {
-      dump_value_31(stream);
+      dump_optional_value(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -477,7 +532,7 @@ 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);
           for (int j = 0; j <= level + 1; j++)
             fprintf (stream, "    ");
           fprintf (stream, "</substitution>\n");
@@ -514,27 +569,27 @@ 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)
 {
   for (int i = 0; i <= level; i++)
     fprintf (stdout, "    ");
   printf ("<category>\n");
-  dump_value (stdout, level + 1, true);
+  dump_value (stdout, level + 1);
+  match_byte(1);
   match_byte(0);
   match_byte(0);
   match_byte(0);
 
-  if (match_u32 (1))
-    match_byte (0);
-  else if (match_byte (1))
+  if (match_byte (1))
     {
-      match_byte (0);
-      if (!match_u32 (2))
-        match_u32_assert (1);
-      match_byte (0);
+      if (!match_byte (0) && !match_byte (1))
+        match_byte_assert (2);
     }
-  else if (!match_u32(2))
-    match_u32_assert (0);
+  else if (!match_byte (2))
+    match_byte_assert (0);
+  match_byte_assert (0);
+  match_byte_assert (0);
+  match_byte_assert (0);
 
   int indx = get_u32();
   int n_categories = get_u32();
@@ -545,6 +600,11 @@ 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)
@@ -554,7 +614,7 @@ dump_category(int level, int *indexes, int *n_indexes)
       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");
@@ -566,9 +626,9 @@ dump_dim(int indx)
   int n_categories;
 
   printf ("<dimension index=\"%d\">\n", indx);
-  dump_value (stdout, 0, false);
+  dump_value (stdout, 0);
 
-  /* This byte is usually 0x02 but 0x00 and 0x75 (!) have also been spotted. */
+  /* This byte is usually 0 but many other values have been spotted. */
   pos++;
 
   if (!match_byte(0) && !match_byte(1))
@@ -580,13 +640,14 @@ dump_dim(int indx)
   if (!match_byte(0))
     match_byte_assert(1);
   match_byte_assert(1);
-  match_u32_assert(indx);
+  if (!match_u32(UINT32_MAX))
+    match_u32_assert(indx);
   n_categories = get_u32();
 
-  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");
@@ -621,7 +682,9 @@ dump_data(void)
     {
       printf ("    <datum index=\"%d\">\n", get_u32());
       match_u32_assert(0);
-      dump_value(stdout, 1, false);
+      if (version == 1)
+        match_byte(0);
+      dump_value(stdout, 1);
       fprintf (stdout, "    </datum>\n");
     }
   printf ("</data>\n");
@@ -630,19 +693,21 @@ dump_data(void)
 static void
 dump_title(void)
 {
-  pos = 0x27;
   printf ("<title-local>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  match_byte(1);
   printf ("</title-local>\n");
 
   printf ("<subtype>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  match_byte(1);
   printf ("</subtype>\n");
 
   match_byte_assert(0x31);
 
   printf ("<title-c>\n");
-  dump_value(stdout, 0, true);
+  dump_value(stdout, 0);
+  match_byte(1);
   printf ("</title-c>\n");
 
   match_byte(0);
@@ -650,30 +715,23 @@ dump_title(void)
   if (match_byte(0x31))
     {
       printf ("<caption>\n");
-      dump_value(stdout, 0, false);
+      dump_value(stdout, 0);
       printf ("</caption>\n");
     }
   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, false);
+      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");
     }
 }
@@ -690,17 +748,21 @@ dump_fonts(void)
       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);
+      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);
@@ -710,26 +772,30 @@ dump_fonts(void)
       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)
         {
+          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");
     }
 
@@ -741,16 +807,23 @@ 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;
 
-  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);
@@ -759,23 +832,43 @@ dump_fonts(void)
   match_byte_assert(0);
   if (!match_byte(0))
     match_byte_assert(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);
+  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);
+
+  int decimal = data[pos];
+  int grouping = data[pos + 1];
   if (match_byte('.'))
-    match_byte_assert(',');
+    {
+      if (!match_byte(',') && !match_byte('\''))
+        match_byte_assert(' ');
+    }
   else
     {
       match_byte_assert(',');
-      if (!match_byte('.'))
-        match_byte_assert(' ');
+      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++)
+        printf("<CC%c>%s</CC%c>\n", 'A' + i, get_string(), 'A' + i);
     }
-  match_u32_assert(5);
-  for (int i = 0; i < 5; i++)
-    get_string();
-  pos += get_u32();
+  else
+    match_u32_assert(0);
+  int skip = get_u32();
+  pos += skip;
 }
 
 int
@@ -807,82 +900,137 @@ 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);
-          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, "%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"))
+    {
+      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;
 }