VARIABLE ROLE: New command.

[pspp] / src / data / sys-file-reader.c

diff --git a/src/data/sys-file-reader.c b/src/data/sys-file-reader.c
index c9d49129732062db02de437568f42cd27fdb0cfc..d38113e2883a47fb237548b856ab95cceb57e55a 100644 (file)
--- a/src/data/sys-file-reader.c
+++ b/src/data/sys-file-reader.c
@@ -83,7 +83,8 @@ enum
     EXT_VAR_ATTRS     = 18,     /* Variable attributes. */
     EXT_MRSETS2       = 19,     /* Multiple response sets (extended). */
     EXT_ENCODING      = 20,     /* Character encoding. */
-    EXT_LONG_LABELS   = 21      /* Value labels for long strings. */
+    EXT_LONG_LABELS   = 21,     /* Value labels for long strings. */
+    EXT_DATAVIEW      = 24      /* "Format properties in dataview table". */
   };
 
 /* Fields from the top-level header record. */
@@ -294,6 +295,7 @@ static void parse_data_file_attributes (struct sfm_reader *,
 static void parse_variable_attributes (struct sfm_reader *,
                                        const struct sfm_extension_record *,
                                        struct dictionary *);
+static void assign_variable_roles (struct sfm_reader *, struct dictionary *);
 static void parse_long_string_value_labels (struct sfm_reader *,
                                             const struct sfm_extension_record *,
                                             struct dictionary *);
@@ -523,7 +525,12 @@ sfm_open_reader (struct file_handle *fh, const char *volatile encoding,
 
   /* The following records use long names, so they need to follow renaming. */
   if (extensions[EXT_VAR_ATTRS] != NULL)
-    parse_variable_attributes (r, extensions[EXT_VAR_ATTRS], dict);
+    {
+      parse_variable_attributes (r, extensions[EXT_VAR_ATTRS], dict);
+
+      /* Roles use the $@Role attribute.  */
+      assign_variable_roles (r, dict);
+    }
 
   if (extensions[EXT_LONG_LABELS] != NULL)
     parse_long_string_value_labels (r, extensions[EXT_LONG_LABELS], dict);
@@ -887,6 +894,7 @@ read_extension_record (struct sfm_reader *r, int subtype)
       { EXT_VAR_SETS,     0, 0 },
       { EXT_DATE,         0, 0 },
       { EXT_DATA_ENTRY,   0, 0 },
+      { EXT_DATAVIEW,     0, 0 },
     };
 
   const struct extension_record_type *type;
@@ -1040,6 +1048,11 @@ parse_variable_records (struct sfm_reader *r, struct dictionary *dict,
                 {
                   double low = parse_float (r, rec->missing, 0);
                   double high = parse_float (r, rec->missing, 8);
+
+                  /* Deal with SPSS 21 change in representation. */
+                  if (low == SYSMIS)
+                    low = LOWEST;
+
                   mv_add_range (&mv, low, high);
                   ofs += 16;
                 }
@@ -1269,11 +1282,14 @@ parse_machine_float_info (struct sfm_reader *r,
                 "instead of %g (%a)."),
               highest, highest, "HIGHEST", HIGHEST, HIGHEST);
 
-  if (lowest != LOWEST)
+  /* SPSS before version 21 used a unique value just bigger than SYSMIS as
+     LOWEST.  SPSS 21 uses SYSMIS for LOWEST, which is OK because LOWEST only
+     appears in a context (missing values) where SYSMIS cannot. */
+  if (lowest != LOWEST && lowest != SYSMIS)
     sys_warn (r, record->pos,
               _("File specifies unexpected value %g (%a) as %s, "
-                "instead of %g (%a)."),
-              lowest, lowest, "LOWEST", LOWEST, LOWEST);
+                "instead of %g (%a) or %g (%a)."),
+              lowest, lowest, "LOWEST", LOWEST, LOWEST, SYSMIS, SYSMIS);
 }
 
 /* Parses record type 7, subtype 7 or 19. */
@@ -1504,9 +1520,8 @@ parse_display_parameters (struct sfm_reader *r,
       align = parse_int (r, record->data, ofs);
       ofs += 4;
 
-      /* SPSS 14 sometimes seems to set string variables' measure
-         to zero. */
-      if (0 == measure && var_is_alpha (v))
+      /* SPSS sometimes seems to set variables' measure to zero. */
+      if (0 == measure)
         measure = 1;
 
       if (measure < 1 || measure > 3 || align < 0 || align > 2)
@@ -1887,6 +1902,64 @@ parse_variable_attributes (struct sfm_reader *r,
   close_text_record (r, text);
 }
 
+static void
+assign_variable_roles (struct sfm_reader *r, struct dictionary *dict)
+{
+  size_t n_warnings = 0;
+  size_t i;
+
+  for (i = 0; i < dict_get_var_cnt (dict); i++)
+    {
+      struct variable *var = dict_get_var (dict, i);
+      struct attrset *attrs = var_get_attributes (var);
+      const struct attribute *attr = attrset_lookup (attrs, "$@Role");
+      if (attr != NULL)
+        {
+          int value = atoi (attribute_get_value (attr, 0));
+          enum var_role role;
+
+          switch (value)
+            {
+            case 0:
+              role = ROLE_NONE;
+              break;
+
+            case 1:
+              role = ROLE_INPUT;
+              break;
+
+            case 2:
+              role = ROLE_OUTPUT;
+              break;
+
+            case 3:
+              role = ROLE_BOTH;
+              break;
+
+            case 4:
+              role = ROLE_PARTITION;
+              break;
+
+            case 5:
+              role = ROLE_SPLIT;
+              break;
+
+            default:
+              role = ROLE_NONE;
+              if (n_warnings++ == 0)
+                sys_warn (r, 0, _("Invalid role for variable %s."),
+                          var_get_name (var));
+            }
+
+          var_set_role (var, role);
+        }
+    }
+
+  if (n_warnings > 1)
+    sys_warn (r, 0, _("%zu other variables had invalid roles."),
+              n_warnings - 1);
+}
+
 static void
 check_overflow (struct sfm_reader *r,
                 const struct sfm_extension_record *record,