Ignore whitespace when checking for content in fields with idrefs.

Fixes #50.

Author: Bryan Worrell

sdv/utils.py

@@ -295,3 +295,31 @@ def has_tzinfo(timestamp):
     """
     ts = parse_timestamp(timestamp)
     return ts and bool(ts.tzinfo)
+
+
+def strip_whitespace(string):
+    """Returns a copy of `string` with all whitespace removed.
+
+    """
+    if string is None:
+        return None
+
+    return ''.join(string.split())
+
+
+def has_content(node):
+    """Returns ``True`` if the `node` has children or text nodes.
+
+    Note:
+        This will ignore whitespace and XML comments.
+
+    """
+    if node is None:
+        return False
+
+    if len(node.findall('*')) > 0:
+        return True
+
+    stripped = strip_whitespace(node.text)
+    return bool(stripped)
+

sdv/validators/stix/best_practice.py

@@ -425,7 +425,8 @@ def _check_idref_with_content(self, root, namespaces, version):  # noqa
         def is_invalid(node):
             if common.is_idref_content_exception(node):
                 return False
-            return bool(node.text) or len(node.findall('*')) > 0
+
+            return utils.has_content(node)
 
         nodes = root.xpath("//*[@idref]")
         warnings = [BestPracticeWarning(x) for x in nodes if is_invalid(x)]
@@ -575,8 +576,6 @@ def _check_timestamp_usage(self, root, namespaces, **kwargs):  # noqa
                     warning['timestamp'] = timestamp
                     results.append(warning)
 
-            warning = None  # overwritten below
-
             if id_ and not timestamp:
                 warning = BestPracticeWarning(
                     node=node,