diff --git a/src/caoscrawler/sync_graph.py b/src/caoscrawler/sync_graph.py index 71b1f7eb8b654245a3794395ebfbf69a67270deb..72bd3bd23c05aa41d707a6f3b62020ee5dc5f29a 100644 --- a/src/caoscrawler/sync_graph.py +++ b/src/caoscrawler/sync_graph.py @@ -158,12 +158,14 @@ class SyncGraph(): # initialize reference mappings (see _create_reference_mapping) ( - self.forward_references, # full set of reference properties of the entity - self.backward_references, # full set of reference properties (of other entities) that reference the entity - self.forward_references_id_props, # subset of the reference properties that are part of identifiables - self.backward_references_id_props, # subset of the backward references that are part of identifiables - self.forward_references_backref, # subset of the reference properties that are part of identifiables as "referenced_by" properties - self.backward_references_backref, # same for backward references + self.forward_references, # id(node) -> full set of nodes referenced by the given node + self.backward_references, # id(node) -> full set of nodes referencing the given node + # as above, subset where the reference properties are part of identifiables + self.forward_references_id_props, + self.backward_references_id_props, + # as above, subset where references are part of identifiables due to "referenced_by" + self.forward_references_backref, + self.backward_references_backref, ) = self._create_reference_mapping(self.nodes) # remove entities with path or ID from unchecked list