This does some very simple inference by assuming that a parent node spans all of its children, and if the children come from multiple {@link InputSource source}s chooses the region spanning the source from which the most children come.
We could do more advanced propagation by using a constraint solver to capture constraints like child-follows-sibling and parent-contains-children but the current approach is sufficient to do useful inference on
___.readPub(tmp___, 'y')where {@code 'y'} has a proper FilePosition, perhaps because it wasderived from {@link com.google.caja.parser.quasiliteral.Rule#toStringLiteral}which propagates position info. @author mikesamuel@gmail.com
|
|