Alaksiej Ščarbaty created NIFI-15175:
----------------------------------------
Summary: Escape json field names in JsonRecordSetWriter when
writing Avro schema
Key: NIFI-15175
URL: https://issues.apache.org/jira/browse/NIFI-15175
Project: Apache NiFi
Issue Type: Improvement
Affects Versions: 2.6.0
Reporter: Alaksiej Ščarbaty
When `JsonRecordSetWriter` with `Set 'avro.schema' Attribute` encounters a
record field with an invalid [Avro
name|https://avro.apache.org/docs/1.12.0/specification/#names] (e.g. one that
contains a hyphen) it fails with an error, even if the field name is valid from
a json perspective.
This complicates schema usage for json-encoded records.
Instead, the field name should be escaped before being written into an Avro
schema. The original field name can be stored in a field
[alias|https://avro.apache.org/docs/1.12.0/specification/#aliases].
During deserialization (e.g. by `JsonTreeReader`) the fields with escaped names
should use aliases when `RecordSchema` is created.
*Important: the written FlowFile content must not change.*
_Perhaps, instead of changing singular writers and readers, the escaping logic
should be introduced in
[AvroTypeUtil|https://github.com/apache/nifi/blob/main/nifi-extension-bundles/nifi-extension-utils/nifi-record-utils/nifi-avro-record-utils/src/main/java/org/apache/nifi/avro/AvroTypeUtil.java]._
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
