VersionedGraphCollection "Redo" Doesn't Bring Changes Back to Same Graph Name

[rileywong311]

It appears that redo() for the VersionedGraphCollection, although indeed adding the changeset back to the collection, it doesn't do it under the same graph name. So if I serialize the entire graph collection after undoing and redoing, it appears in the graph/file, but if I serialize only the specific graph I specified originally, it no longer appears in the file even though it is part of the graph collection.

from brickschema.persistent import VersionedGraphCollection, Changeset
from brickschema.namespaces import A, BRICK
from rdflib import Namespace

vg = VersionedGraphCollection("sqlite://")

BLDG = Namespace("test:") 

with vg.new_changeset("my-building") as cs:
    cs.add((BLDG.zone1, A, BRICK.HVAC_Zone))

with vg.new_changeset("my-building") as cs:
    cs.add((BLDG.zone2, A, BRICK.HVAC_Zone))

vg.undo()
vg.redo()

vg.serialize("UndoRedo1.ttl") # serialize the entire graph collection
vg.graph_at(graph="my-building").serialize("UndoRedo2.ttl") # serialize only the graph "my-building"

A potential fix that seems to work might be this (persistent.py on line 171):

# persistent.py
# redo():
triple = pickle.loads(row["triple"])
graph = self.get_context(redo_record["graph"])
if row["is_insertion"]:
  graph.remove((triple[0], triple[1], triple[2]))
else:
  graph.add((triple[0], triple[1], triple[2]))

[gtfierro]

Yes, I think this the right solution. Can you do a PR? Thanks for finding this!