For exporting text nodes that are covering "real" (empty) tokens, the groupby feature does not work, since during following the coverage edges from the groupby nodes to the tokens the dfs never comes by the actual text nodes. Therefore, a bottom-up search of some kind needs to be added or put in place of the current version.
For exporting text nodes that are covering "real" (empty) tokens, the
groupbyfeature does not work, since during following the coverage edges from the groupby nodes to the tokens the dfs never comes by the actual text nodes. Therefore, a bottom-up search of some kind needs to be added or put in place of the current version.