Start work on documenting values.

[pspp] / dump.c

diff --git a/dump.c b/dump.c
index 0445597e635aa6673304d1fa734df0cc29371edc..c247c4af661be8c50ef6312aa18f05c524088b68 100644 (file)
--- a/dump.c
+++ b/dump.c
@@ -219,13 +219,9 @@ get_string(const char *where)
 #define get_string() get_string(WHERE)
 
 static char *
-dump_nested_string(void)
+dump_counted_string(void)
 {
   char *s = NULL;
-
-  match_byte_assert (0);
-  match_byte_assert (0);
-  int outer_end = pos + get_u32();
   int inner_end = pos + get_u32();
   if (pos != inner_end)
     {
@@ -240,6 +236,18 @@ dump_nested_string(void)
           exit(1);
         }
     }
+  return s;
+}
+
+static char *
+dump_nested_string(void)
+{
+  char *s = NULL;
+
+  match_byte_assert (0);
+  match_byte_assert (0);
+  int outer_end = pos + get_u32();
+  s = dump_counted_string();
   match_byte_assert(0x58);
   match_byte_assert(0x58);
   if (pos != outer_end)
@@ -260,7 +268,11 @@ dump_optional_value(FILE *stream)
         {
           if (match_u32 (1))
             {
-              /* Only "a" observed as a sample value (although it appears 44 times in the corpus). */
+              /* Corpus frequencies:
+                 124 "a"
+                 12 "b"
+                 8 "a, b"
+              */
               get_string();
             }
           else
@@ -283,25 +295,10 @@ dump_optional_value(FILE *stream)
             }
 
           int outer_end = pos + get_u32();
-          int inner_end = pos + get_u32();
-          if (pos != inner_end)
-            {
-              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);
-                }
-            }
+
+          /* This counted-string appears to be a template string,
+             e.g. "Design\: [:^1:]1 Within Subjects Design\: [:^1:]2". */
+          dump_counted_string();
 
           if (match_byte(0x31))
             {
@@ -358,7 +355,7 @@ dump_optional_value(FILE *stream)
           match_byte_assert(0);
           if (!match_u32 (2) && !match_u32(1))
             match_u32_assert(3);
-          dump_nested_string(); /* Our corpus doesn't contain any examples with strings though. */
+          dump_nested_string();
         }
       else
         {
@@ -435,32 +432,16 @@ dump_value(FILE *stream, int level)
   for (int i = 0; i <= level; i++)
     fprintf (stream, "    ");
 
-  if (match_byte (3))
-    {
-      char *text = get_string();
-      dump_optional_value(stream);
-      char *identifier = get_string();
-      char *text_eng = get_string();
-      fprintf (stream, "<string c=\"%s\"", text_eng);
-      if (identifier[0])
-        fprintf (stream, " identifier=\"%s\"", identifier);
-      if (strcmp(text_eng, text))
-        fprintf (stream, " local=\"%s\"", text);
-      fprintf (stream, "/>\n");
-      if (!match_byte (0))
-        match_byte_assert(1);
-    }
-  else if (match_byte (5))
+  if (match_byte (1))
     {
+      unsigned int format;
+      double value;
+
       dump_optional_value(stream);
-      char *name = get_string ();
-      char *label = get_string ();
-      fprintf (stream, "<variable name=\"%s\"", name);
-      if (label[0])
-        fprintf (stream, " label=\"%s\"", label);
-      fprintf (stream, "/>\n");
-      if (!match_byte(1) && !match_byte(2))
-        match_byte_assert(3);
+      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);
     }
   else if (match_byte (2))
     {
@@ -483,12 +464,27 @@ dump_value(FILE *stream, int level)
       if (!match_byte (1) && !match_byte(2))
         match_byte_assert (3);
     }
+  else if (match_byte (3))
+    {
+      char *text =  get_string();
+      dump_optional_value(stream);
+      char *identifier = get_string();
+      char *text_eng = get_string();
+      fprintf (stream, "<string c=\"%s\"", text_eng);
+      if (identifier[0])
+        fprintf (stream, " identifier=\"%s\"", identifier);
+      if (strcmp(text_eng, text))
+        fprintf (stream, " local=\"%s\"", text);
+      fprintf (stream, "/>\n");
+      if (!match_byte (0))
+        match_byte_assert(1);
+    }
   else if (match_byte (4))
     {
       unsigned int format;
       char *var, *vallab, *value;
 
-      match_byte_assert (0x58);
+      dump_optional_value(stream);
       format = get_u32 ();
       vallab = get_string ();
       var = get_string ();
@@ -503,16 +499,17 @@ dump_value(FILE *stream, int level)
         fprintf (stream, " label=\"%s\"/>\n", vallab);
       fprintf (stream, "/>\n");
     }
-  else if (match_byte (1))
+  else if (match_byte (5))
     {
-      unsigned int format;
-      double value;
-
       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);
+      char *name = get_string ();
+      char *label = get_string ();
+      fprintf (stream, "<variable name=\"%s\"", name);
+      if (label[0])
+        fprintf (stream, " label=\"%s\"", label);
+      fprintf (stream, "/>\n");
+      if (!match_byte(1) && !match_byte(2))
+        match_byte_assert(3);
     }
   else
     {