How to search file geodatabase for feature classes containing attributes with names that meet a regular expression?

I need to scan an existing but unpopulated (i.e. no data) file geodatabase (GDB) and determine which feature classes have attributes whose names match a given regular expression. I then need essentially to extract the pairs of feature_class_name/matching_attribute_name and in some way to use those pairs subsequently in the translation. So:

1. What transformer would I use to scan an unpopulated GDB for attribute names matching a regex?
2. How best to configure and to capture information from that transformer to use subsequently in translation?
3. [Related to the beginnings of a solution idea in my head ...] If I use, say, a FeatureReader to extract schema features from an unpopulated GDB with Ignore Relationship Info unchecked, will the schema features carry format attributes (similar to geodb_feature_has_relationships on data features) indicating feature classes and attributes that participate in relationships?

Thanks, all, for your ideas on this.