[
https://issues.apache.org/jira/browse/CAMEL-26228?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Work on CAMEL-26228 started by Andrea Cosentino.
------------------------------------------------
> camel-neo4j - CREATE_NODE with no body and a similarity search with a List or
> text vector fail with a NullPointerException, and a Cypher retrieve needs a
> node first column
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: CAMEL-26228
> URL: https://issues.apache.org/jira/browse/CAMEL-26228
> Project: Camel
> Issue Type: Bug
> Components: camel-neo4j
> Reporter: Claus Ibsen
> Assignee: Andrea Cosentino
> Priority: Minor
> Fix For: 4.24.0
>
>
> Found while adding the catalog body types of neo4j (CAMEL-24961, PR #28398).
> h4. 1. CREATE_NODE with no body: bare NullPointerException
> {{Neo4jProducer.createNode}} (main 2a465aa4814c, line 137) builds
> {{Map.of("props", body)}}, which throws a NullPointerException when the body
> is null. Checked on 4.24.0-SNAPSHOT with Neo4j from test-infra (docker).
> Proposed: fail with a clear message that CREATE_NODE needs a Map or JSON body.
> h4. 2. VECTOR_SIMILARITY_SEARCH with a List<Double> or text vector
> {{similaritySearch}} (line 417) converts the body with
> {{getBody(float[].class)}}. A {{List<Double>}} gives a misleading conversion
> error, and text such as {{0.75,0.65}} converts to null, so {{Map.of(...,
> "embeddingValue", vectors, ...)}} (line 433) throws a NullPointerException.
> Checked on 4.24.0-SNAPSHOT with Neo4j from test-infra (docker).
> Proposed: accept a List of numbers, and fail with a clear message when the
> body is not a vector.
> h4. 3. A Cypher retrieve fails when the first column is not a node
> {{nodeProperties}} (line 251) maps every record with
> {{record.get(0).asNode()}}, so
> {{RETRIEVE_NODES_AND_UPDATE_WITH_CYPHER_QUERY}} with a query that returns a
> property, a count or a path first fails with "Cannot coerce STRING to Node".
> Checked on 4.24.0-SNAPSHOT with Neo4j from test-infra (docker).
> Proposed: return the record as a Map when the first column is not a node.
> h4. 4. The similarity search drops the score
> The query (lines 426-431) yields {{node, score}}, but {{nodeProperties}}
> keeps only the node properties, so the score of each hit is lost. Checked on
> 4.24.0-SNAPSHOT with Neo4j from test-infra (docker).
> Proposed: add the score to each result Map (or to a header).
> h4. 5. Header metadata is wrong
> In {{Neo4jHeaders}}: {{CamelNeo4jQueryResultListNeo4jNodes}} (line 43) is
> described as "Query execution time in Milliseconds" but holds the list of
> records (Neo4jProducer line 242); {{CamelNeo4jQueryResultRetrieveSize}} (line
> 41) is declared Long but the producer sets an Integer (line 241); the
> {{CamelNeo4jLabel}} header (line 47) and the {{query}} endpoint option
> ({{Neo4jConfiguration}} line 242) are never read. Read in the code, not run.
> Proposed: fix the descriptions and types, and use (or remove) the unused
> header and option.
> _Claude Code on behalf of davsclaus_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
