[PARQUET-357] Parquet-thrift generates wrong schema for Thrift binary fields

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.5.0, 1.6.0, 1.7.0, 1.8.0
• Fix Version/s: 1.10.0
• Component/s: parquet-mr
• Assignee: Nándor Kollár
• Reporter: Cheng Lian
• Created: 13/Aug/15 11:03
• Updated: 23/Jun/24 03:27
• Resolved: 04/Jan/18 16:52

Description

Thrift doesn't have true BINARY type. The BINARY type is actually just an unencoded STRING. Quoted from Thrift Types section of official Thrift documentation:

binary: a sequence of unencoded bytes

N.B.: This is currently a specialized form of the string type above, added to provide better interoperability with Java. The current plan-of-record is to elevate this to a base type at some point.

The consequence is that, Thrift BINARY and STRING are both passed to parquet-thrift as STRING, and are always encoded as BINARY (UTF8).

This is really a problem on Thrift side. One possible workaround is to inspect binary fields in the actual generated Java classes to see whether the type is ByteBuffer.

Issue Links

• links to: PR