How to Splice a Linked Lists in Cypher

Question

I have a Linked-List of LogEntries. Each entry has a date. Every user has exactly 1 Log.

What's the best way to splice in a new log entry if I receive a date in the past?

(log:Log)
(e1:LogEntry {date: Date('2020-03-19')})
(e2:LogEntry {date: Date('2020-03-17')})
…

CREATE (log)-[:PREV_ENTRY]->(e1)
CREATE (e1)-[:PREV_ENTRY]->(e2)
CREATE (e2)-[:PREV_ENTRY]->(e3)
CREATE (e3)-[:PREV_ENTRY]->(e4)

I'm building a public graphql API to trace contact of COVID-19 case: https://github.com/mmmoli/contacttracing.app-graphql-api/blob/master/src/appConfigs/schema.ts#L55

Any help would be awesome!

Inspiration from Max's Fraud post: https://maxdemarzi.com/2019/08/19/finding-fraud/

Answer 1

Is this when you're first creating the database (i.e. the entries weren't sorted to start) or is this entry appearing later on and you're trying to update/insert with a new entry? If I'm understanding correctly, you want to insert a node in the chain for the log. How about

EDIT: Included additional constraint to make sure matching the chain finds one set of nodes between which to insert the new entry

CREATE (eInsert:LogEntry:NewEntry {date: Date('2020-03-18')})
WITH eInsert

MATCH (log:Log {id: "log1"})
WITH log, eInsert

MATCH (log)-[:PREV_ENTRY*]->(e1)-[oldLink:PREV_ENTRY]->(e2:LogEntry)
WHERE e2.date < eInsert.date
  AND e1.date >= eInsert.date 
WITH e1, e2, eInsert, oldLink

CREATE (e1)-[:PREV_ENTRY]->(eInsert)-[:PREV_ENTRY]->(e2)
DELETE oldLink

That should work, but for completeness, here's the script I used to create the sample chain

CREATE (log:Log {id: 'log1'})
CREATE (e1:LogEntry {date: Date('2020-03-30')})
CREATE (e2:LogEntry {date: Date('2020-03-28')})
CREATE (e3:LogEntry {date: Date('2020-03-23')})
CREATE (e4:LogEntry {date: Date('2020-03-17')})
CREATE (e5:LogEntry {date: Date('2020-03-09')})

CREATE (log)-[:PREV_ENTRY]->(e1)
CREATE (e1)-[:PREV_ENTRY]->(e2)
CREATE (e2)-[:PREV_ENTRY]->(e3)
CREATE (e3)-[:PREV_ENTRY]->(e4)
CREATE (e4)-[:PREV_ENTRY]->(e5)

Answer 2

I've come up with this so far:

MATCH p=(log:Log {id: "log1"})-[:PREV_ENTRY*]->(e:LogEntry)
FOREACH (rel IN relationships(p) | 
  DELETE rel
)
WITH e, log
ORDER BY e.id DESC
WITH collect(e) AS entries, log, e
CALL apoc.nodes.link(entries, 'PREV_ENTRY')
WITH apoc.agg.first(e) AS latest, log
CREATE (log)-[:PREV_ENTRY]->(latest)
RETURN latest

But it doesn't work quite yet.