11/07 graph-explorer

[jackson-millard]

This merges in the workbench updates as of 11/07/22 from Bitbucket to Github.

[srondelli]

Hello, thanks for sharing the code.

1. The repository seems missing the actual code. i.e. the neptune-graph-explorer/ folder.
2. Is it necessary to upload the code for all the node_modules?

[srondelli]

Gave some comments to the README. Going to take a look to the code. Note: I haven;t actually tried to follow it yet (i.e. checkout the code and build and run), once I do that i may have more comments.

[srondelli]

I got just at the beginning of SPARQL. Will continue later this week.

[gopuneet]

Reviewed the proxy-server code and added comments.

[saikiranboga]

This query is fine for a single hop and fetching smaller set of results, but a more efficient query to fetch 1 hop vertices and edges could be:

g.V("1")
.bothE("knows").otherV().hasLabel("person").dedup().range(0,10)
.path()

This would produce the below path results for each solution. Each solution is a path containing the starting vertex, an edge and the 1-hop vertex.

path[v[1], e[7][1-knows->2], v[2]]
path[v[1], e[8][1-knows->4], v[4]]

[joywa]

@nestoralvarezd could you please take a look?

[nestoralvarezd]

I've reviewed the above query. The .path() chain in the query give me the information of the relation and the id of the neighbor. However, in the original query, I'm getting the properties of each edge and node which is related to the searched node. So, this change will imply doing 1 extra query by each resulting node.

[saikiranboga]

The original query might be returning properties because of the serializer being used. It is best to always query for the exact data needed for each vertex/edge instead of relying of serialization.

In case properties, id and label info is needed for vertices and edges on the path, you could get them as maps in the path using by modulation on the last path step:

g.V("1").bothE("knows").otherV().hasLabel("person").dedup().range(0,10)
.path().by(valueMap().with(WithOptions.tokens,WithOptions.all))

This would return results as below:

path[{<T.label: 4>: 'person', 'name': ['marko'], <T.id: 1>: '1', 'age': [29]}, {<T.label: 4>: 'knows', 'weight': 0.5, <T.id: 1>: '7'}, {<T.label: 4>: 'person', 'name': ['vadas'], <T.id: 1>: '2', 'age': [27]}]

path[{<T.label: 4>: 'person', 'name': ['marko'], <T.id: 1>: '1', 'age': [29]}, {<T.label: 4>: 'knows', 'weight': 1.0, <T.id: 1>: '8'}, {<T.label: 4>: 'person', 'name': ['josh'], <T.id: 1>: '4', 'age': [32]}]

[nestoralvarezd]

I've reviewed this version and I believe that it'll work as expected. Do you want to update the queries in a new PR?

[saikiranboga]

This could also be written similar to the above query, but should be ok for a smaller set of results.

[joywa]

@nestoralvarezd could you please take a look?

[nestoralvarezd]

Same answer here, .path() works if I want to only know the relation and its attributes. However, I'm fetching the properties as well

[saikiranboga]

Reviewed Gremlin queries and mostly look good to me. Added couple of optional suggestions.

[srondelli]

LGTM, thanks for all the changes.

1. Change sparQL to sparql or SPARQL.
2. SPARQL queries should be signed off by Charles.

[joywa]

@jackson-millard-experoinc thanks, this commit looks great. Can you make the small modifications I just left?

[joywa]

Approved. Will address bnode support + Gremlin optimizations in later PRs.