[AVRO-3812] Handle null namespace properly for canonicalized schema representation

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.12.0
• Fix Version/s: 1.12.0, 1.11.3
• Component/s: rust
• Labels: pull-request-available
• Assignee: Martin Tzvetanov Grigorov
• Reporter: Kousuke Saruta
• Created: 23/Jul/23 13:26
• Updated: 26/Jul/23 21:51
• Resolved: 26/Jul/23 21:51

Description

Considering the following schema, which contains namespaces of "".

{
 "namespace": "",
 "type": "record",
 "name": "my_schema",
 "fields": [
   {
     "name": "a",
     "type": {
       "type": "enum",
       "name": "my_enum",
       "namespace": "",
       "symbols": ["a", "b"]
     }
   },  {
     "name": "b",
     "type": {
       "type": "fixed",
       "name": "my_fixed",
       "namespace": "",
       "size": 10
     }
   }
 ]
}

If we try to canonicalize this schema with the following code

let schema = Schema::parse_str(schema_str).unwrap().canonical_form();
println!("{schema}");

We get the following result.

{"name":".my_schema","type":"record","fields":[{"name":"a","type":{"name":".my_enum","type":"enum","symbols":["a","b"]}},{"name":"b","type":{"name":".my_fixed","type":"fixed","size":10}}]}

But .my_schema, .my_enum and .my_fixed should not starts with a dot.

Issue Links

• links to: GitHub Pull Request #2383