[AVRO-2520] Schema parsing does not work for non-record types with logicalType

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Not A Problem
• Affects Version/s: 1.9.0
• Component/s: java
• Assignee: Unassigned
• Reporter: Teodor Kostov
• Created: 23/Aug/19 11:59
• Updated: 17/Jan/20 21:31
• Resolved: 05/Nov/19 17:28

Description

Parsing the following schema fails:

new Schema.Parser().parse("{ \"name\": \"test\", \"type\": { \"type\": \"long\", \"logicalType\": \"MY_TIMESTAMP\" }}")

with the following exception:

No type: {"name":"test","type":{"type":"long","logicalType":"MY_TIMESTAMP"}}
org.apache.avro.SchemaParseException: No type: {"name":"test","type":{"type":"long","logicalType":"MY_TIMESTAMP"}}
	at org.apache.avro.Schema.getRequiredText(Schema.java:1725)
	at org.apache.avro.Schema.parse(Schema.java:1576)
	at org.apache.avro.Schema$Parser.parse(Schema.java:1366)
	at org.apache.avro.Schema$Parser.parse(Schema.java:1354)

The complex type cannot be properly parsed because it is not a simple string but a JSON object. Hence when calling `jsonNode.textValue()` the result is null. Check `getOptionalText(...)` for more information.

When supplying logical types it should be possible to parse a simple schema.