jahnavi created SPARK-56654:
-------------------------------

             Summary: Enforce strict Unicode validation in JSON parsing
                 Key: SPARK-56654
                 URL: https://issues.apache.org/jira/browse/SPARK-56654
             Project: Spark
          Issue Type: Improvement
          Components: Spark Core
    Affects Versions: 4.0.0
            Reporter: jahnavi


The JVM execution path uses a Jackson-based parser (e.g., 
ReaderBasedJsonParser), which is permissive. It does not strictly validate 
Unicode surrogate pairs and will accept invalid sequences such as a lone high 
surrogate (\uD835). When this happens, it silently replaces the invalid 
character with a placeholder (?) and returns a parsed result.

 

The alternative execution like simdjson uses a stricter JSON parser this is 
used in photon in Databricks. It explicitly checks for correct surrogate 
pairing and rejects malformed sequences. As a result, it returns NULL for 
try_parse_json or throws an error for parse_json.

 

Because of this difference:

 

The same input JSON can be rejected in one execution path and accepted (with 
data corruption) in another.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to