Claus Ibsen created CAMEL-26553:
-----------------------------------
Summary: camel-spring-ai-embeddings and
camel-langchain4j-embeddings - embed bodies that have a real converter to
String, such as char[] and ByteBuf
Key: CAMEL-26553
URL: https://issues.apache.org/jira/browse/CAMEL-26553
Project: Camel
Issue Type: Improvement
Reporter: Claus Ibsen
Found in the review of CAMEL-26512 (PR #29108).
camel-spring-ai-embeddings (with CAMEL-26512) and camel-langchain4j-embeddings
({{LangChain4jEmbeddingsProducer.isReadableAsText}}) only embed a body of a
fixed list of text types, and fail other bodies. Bodies that have a real
converter to String fail too, while before CAMEL-26512 spring-ai embedded them:
* {{char[]}}
* a Netty {{ByteBuf}} (netty-http with {{disableStreamCache=true}})
* a custom type with its own converter to String
* a List with such an element
Asking the type converter does not tell them apart from a Map or POJO: after
one fallback conversion, {{lookup(String.class, X.class)}} returns the promoted
{{ToStringTypeConverter}}. {{BodyFacts.kindOf}} does not help either (ByteBuf
is DATA, char[] is COLLECTION, Integer is PAYLOAD).
Today the workaround is {{convertBodyTo(String.class)}} before the endpoint (in
the 4.24 upgrade guide). Proposed: widen the accepted types in both components
the same way, e.g. CharSequence, char[], and bodies whose converter to String
is not the toString fallback.
_Claude Code on behalf of Claus Ibsen (davsclaus)_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
