Data integrity check

To verify that data isn't breaking the data structure we are implementing an API that scans neo4j for inconsistencies:

Check data for inconsistencies

• Check that all resources have CREATED_BY relations:

START resource=node(*)
WHERE has(resource.nodeType)
AND (resource.nodeType="kn_Post")
AND NOT (resource)-[:CREATED_BY]-()
RETURN resource

• Check that a connection is related to two resources: START connection=node(*)
WHERE has(connection.nodeType)
AND (connection.nodeType="kn_Edge")
AND NOT ()-[:RELATED_TO]-(connection)-[:RELATED_TO]-()
RETURN connection

• Check that all connection have CREATED_BY relations:
  START connection=node(*)
WHERE has(connection.nodeType)
AND (connection.nodeType="kn_Edge")
AND NOT (connection)-[:CREATED_BY]-()
RETURN connection

• Check that a comment is related to a connection: START comment=node(*)
WHERE has(comment.nodeType)
AND (comment.nodeType="kn_Comment")
AND NOT (comment)-[:COMMENT_OF]-()
RETURN comment

• Check that a comment is related to a user: START comment=node(*)
WHERE has(comment.nodeType)
AND (comment.nodeType="kn_Comment")
AND NOT (comment)-[:CREATED_BY]-()
RETURN comment

Deletion of corrupted data:

• DELETE resources that don't have CREATED_BY relations: START resource=node(*)
MATCH (resource)-[relation?]-()
WHERE has(resource.nodeType)
AND (resource.nodeType="kn_Post")
AND NOT (resource)-[:CREATED_BY]-()
DELETE resource, relation

• DELETE connections not related to two resources: START connection=node(*)
MATCH (connection)-[relation?]-()
WHERE has(connection.nodeType)
AND (connection.nodeType="kn_Edge")
AND NOT ()-[:RELATED_TO]-(connection)-[:RELATED_TO]-()
DELETE connection, relation

• DELETE all connection without CREATED_BY relations:
  START connection=node(*)
MATCH (connection)-[relation?]-()
WHERE has(connection.nodeType)
AND (connection.nodeType="kn_Edge")
AND NOT (connection)-[:CREATED_BY]-()
DELETE connection, relation

• DELETE comments without COMMENT_OF relation: START comment=node(*)
MATCH (comment)-[relation?]-()
WHERE has(comment.nodeType)
AND (comment.nodeType="kn_Comment")
AND NOT (comment)-[:COMMENT_OF]-()
DELETE comment, relation

• DELETE comment without CREATED_BY relation: START comment=node(*)
MATCH (comment)-[relation?]-()
WHERE has(comment.nodeType)
AND (comment.nodeType="kn_Comment")
AND NOT (comment)-[:CREATED_BY]-()
DELETE comment, relation