GraphQL mutation apis

[ljeub-pometry]

What changes were proposed in this pull request?

Add support for creating and mutating graphs using GraphQL

Graph Functions

• graph.add_node()
• graph.add_edge()
• graph.delete_edge()
• graph.add_properties()
• graph.add_constant_properties()
• graph.window.subgraph.etc.export_to("path/for/other/graph/")

Node functions

• graph.node.add_updates()
• graph.node.add_constant_properties()
• graph.node.update_constant_properties()

Edge functions

• graph.edge.add_updates()
• graph.edge.delete()
• graph.edge.add_constant_properties()
• graph.edge.update_constant_properties()

Why are the changes needed?

Does this PR introduce any user-facing change? If yes is this documented?

How was this patch tested?

Issues

If this resolves any issues, please link to them here, the format is a KEYWORD followed by @_
KEYWORDS available are close, closes, closed, fix, fixes, fixed, resolve, resolves, resolved.
Please delete this text before creating your PR

Are there any further changes required?

[miratepuffin]

Random stuff I have found

• If there is an invalid graph/file within your working directory on startup the sever won't start - perhaps should just skip the file and log a warning
• new_graph function gives "IO operation failed" when graph already exists - should be 'Graph already exists'
• new_graph function currently doesn't handle parent dirs correctly.
• If we delete all graphs in a namespace, do we want to also delete the namespace?
• When creating a new RaphtoryClient, it doesn't check the given URL, meaning an error is only thrown when you run your first query
• props needs to be properties to keep it inline with the python apis
• props in bulk updates need to be optional
• "Field "addNode" of type "GqlMutableNode" must have a selection of subfields"

[ljeub-pometry]

[ ] "Field "addNode" of type "GqlMutableNode" must have a selection of subfields"

that is just how GraphQL works

[miratepuffin]

LGTM

[ljeub-pometry]

Note that the nicer error message check is not atomic. There is still a small chance of getting the IO error instead.

[ljeub-pometry]

this can still fail with the mysterious IO error here if someone manages to create the file between the previous check and this one. Probably better to only use this to check and map the error for a better message.