Figured out more details, documented most values, work on substitutions.

[pspp] / dump.c

diff --git a/dump.c b/dump.c
index b383f4359083cd6e1bef39c075947e02c324dfb6..a52cafddf0d414cb06ddd8464c9795949b91e2d9 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -265,22 +265,25 @@ hex_dump(int ofs, int n)
 }
 
 static void
-dump_style(void)
+dump_style(FILE *stream)
 {
   match_byte(1);
   match_byte(0);
   match_byte(0);
   match_byte(0);
   match_byte_assert(1);
-  get_string();     /* foreground */
-  get_string();     /* background */
-  get_string();     /* font */
+  char *fg = get_string();     /* foreground */
+  char *bg = get_string();     /* background */
+  char *font = get_string();     /* font */
+  int size = data[pos];
   if (!match_byte(14))
     match_byte_assert(12); /* size? */
+  fprintf(stream, " fgcolor=\"%s\" bgcolor=\"%s\" font=\"%s\" size=\"%d\"",
+          fg, bg, font, size);
 }
 
 static char *
-dump_nested_string(void)
+dump_nested_string(FILE *stream)
 {
   char *s = NULL;
 
@@ -288,8 +291,10 @@ dump_nested_string(void)
   match_byte_assert (0);
   int outer_end = get_end();
   s = dump_counted_string();
+  if (s)
+    fprintf(stream, " \"%s\"", s);
   if (match_byte(0x31))
-    dump_style();
+    dump_style(stream);
   else
     match_byte_assert(0x58);
   match_byte_assert(0x58);
@@ -303,20 +308,23 @@ dump_nested_string(void)
 }
 
 static void
-dump_optional_value(FILE *stream)
+dump_value_modifier(FILE *stream)
 {
   if (match_byte (0x31))
     {
       if (match_u32 (0))
         {
+          fprintf(stream, "<special0");
           if (match_u32 (1))
             {
               /* Corpus frequencies:
                  124 "a"
                  12 "b"
                  8 "a, b"
+
+                 The given text is appended to the cell in a subscript font.
               */
-              get_string();
+              fprintf(stream, " subscript=\"%s\"", get_string());
             }
           else
             match_u32_assert (0);
@@ -334,6 +342,7 @@ dump_optional_value(FILE *stream)
                 match_u32_assert(10);
               match_byte(0);
               match_byte(0);
+              fprintf(stream, "/>\n");
               return;
             }
 
@@ -341,10 +350,12 @@ dump_optional_value(FILE *stream)
           
           /* This counted-string appears to be a template string,
              e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
-          dump_counted_string();
+          char *template = dump_counted_string();
+          if (template)
+            fprintf(stream, " template=\"%s\"", template);
 
           if (match_byte(0x31))
-            dump_style();
+            dump_style(stream);
           else
             match_byte_assert(0x58);
           if (match_byte(0x31))
@@ -372,32 +383,36 @@ dump_optional_value(FILE *stream)
               fprintf(stderr, "outer end discrepancy\n");
               exit(1);
             }
+          fprintf(stream, "/>\n");
         }
       else if (match_u32 (1))
         {
-          fprintf(stream, "(footnote %d) ", get_u32());
-          dump_nested_string();
+          fprintf(stream, "<footnote-ref index=%d", get_u32());
+          dump_nested_string(stream);
+          fprintf(stream, "/>\n");
         }
       else if (match_u32 (2))
         {
-          fprintf(stream, "(special 2)");
+          fprintf(stream, "<special2 %d %d", data[pos], data[pos + 2]);
           if (!match_byte(0) && !match_byte(1))
             match_byte_assert(2);
           match_byte_assert(0);
           if (!match_u32 (2) && !match_u32(1))
             match_u32_assert(3);
-          dump_nested_string();
+          dump_nested_string(stream);
+          fprintf(stream, "/>\n");
         }
       else
         {
+          fprintf(stream, "<special3");
           match_u32_assert(3);
-          fprintf(stream, "(special 3)");
           match_byte_assert(0);
           match_byte_assert(0);
           match_byte_assert(1);
           match_byte_assert(0);
           match_u32_assert(2);
-          dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+          dump_nested_string(stream); /* Our corpus doesn't contain any examples with strings though. */
+          fprintf(stream, "/>\n");
         }
     }
   else
@@ -468,7 +483,7 @@ dump_value(FILE *stream, int level)
       unsigned int format;
       double value;
 
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       format = get_u32 ();
       value = get_double ();
       fprintf (stream, "<number value=\"%.*g\" format=\"%s%d.%d\"/>\n",
@@ -480,7 +495,7 @@ dump_value(FILE *stream, int level)
       char *var, *vallab;
       double value;
 
-      dump_optional_value (stream);
+      dump_value_modifier (stream);
       format = get_u32 ();
       value = get_double ();
       var = get_string ();
@@ -498,7 +513,7 @@ dump_value(FILE *stream, int level)
   else if (match_byte (3))
     {
       char *text =  get_string();
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       char *identifier = get_string();
       char *text_eng = get_string();
       fprintf (stream, "<string c=\"%s\"", text_eng);
@@ -515,7 +530,7 @@ dump_value(FILE *stream, int level)
       unsigned int format;
       char *var, *vallab, *value;
 
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       format = get_u32 ();
       vallab = get_string ();
       var = get_string ();
@@ -532,7 +547,7 @@ dump_value(FILE *stream, int level)
     }
   else if (match_byte (5))
     {
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
       char *name = get_string ();
       char *label = get_string ();
       fprintf (stream, "<variable name=\"%s\"", name);
@@ -544,7 +559,7 @@ dump_value(FILE *stream, int level)
     }
   else
     {
-      dump_optional_value(stream);
+      dump_value_modifier(stream);
 
       char *base = get_string();
       int x = get_u32();
@@ -1130,7 +1145,7 @@ 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_optional_value(). */
+         special cases in dump_value_modifier(). */
       version = get_u32();
       pos -= 4;
       if (!match_u32(1))