[THRIFT-1679] NodeJS: protocol readString() should treat string as utf8, not binary

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Implemented
• Affects Version/s: 0.8
• Fix Version/s: 0.9
• Component/s: Node.js - Library

Environment

All

• Assignee: Unassigned
• Reporter: Michael Fleming
• Created: 24/Aug/12 22:39
• Updated: 20/May/13 03:13
• Resolved: 11/May/13 19:11

Description

Using Node.js's Thrift library to talk to Hive:

a) Read back a Hive result that contains utf8 characters

Expected:

• Get a Javascript string object that contains the correct utf8 sequence

Actual:

• Get a Javascript sequence that is indicative of UTF8 double-encoding

Underlying:

In protocol.js, change "readString()"

var r = this.readBinary().toString('binary');

to this:

var r = this.readBinary().toString('utf8');

Per http://thrift.apache.org/docs/types/

"string: A text string encoded using UTF-8 encoding"