[
https://issues.apache.org/jira/browse/CAMEL-26206?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Work on CAMEL-26206 started by Claus Ibsen.
-------------------------------------------
> camel-langchain4j-ingest - text mode ingests any body by its toString, for
> example a TextSegment
> ------------------------------------------------------------------------------------------------
>
> Key: CAMEL-26206
> URL: https://issues.apache.org/jira/browse/CAMEL-26206
> Project: Camel
> Issue Type: Bug
> Components: camel-langchain4j
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Minor
> Fix For: 4.24.0
>
>
> Found while adding the catalog body types of langchain4j-ingest (CAMEL-24961,
> PR #28401).
> h4. 1. Text mode ingests any body via toString without an error
> {{LangChain4jIngestProducer.textBody}} (main 2a465aa4814c, line 331) reads
> the body with {{getBody(String.class)}}, which for an object without a real
> text form gives its {{toString()}}. A {{TextSegment}} body is ingested as the
> text {{TextSegment \{ text = ...}}, a POJO or a Map as their {{toString()}},
> and nothing tells the route. Checked on 4.24.0-SNAPSHOT with deterministic
> embedding models and an {{InMemoryEmbeddingStore}}.
> Proposed: take the text of a {{TextSegment}} / {{Document}} body, and fail
> for bodies that are not text, bytes, a stream, a file or a Reader.
> _Claude Code on behalf of davsclaus_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
