Awesome

Node Bindings for Hadoop Hive

Installation

npm install node-hive

Usage

hive = require('node-hive').for({ server:"hive.myserver" });

hive.fetch("SELECT * FROM my_table", function(err, data) {
  data.each(function(record) {
    console.log(record);
  });
});

Hive instances currently support the following functions

hive.fetch(query, callback);
hive.fetchInBatch(batchSize, query, callback);
hive.execute(query, [callback]);

Query callbacks receive two arguments...

• error which is true if there was an error
• result which is either a ResultSet or an error message depending on the state of error

The result of a query is returned as a ResultSet which wraps the results with some convenience functions...

• result.rows - The original string based rows returned by thrift.
• result.schema - The schema returned from hive.
• result.each(callback) - Iterate through rows converting them to friendly JS objects.
• result.headers() - An Array of the column headers.
• result.toTSV(headers=false) - produce a TSV version of the whole ResultSet.

See the examples folder for some more usage hints.

Connections

The Hive Thrift Server currently only supports one blocking query at a time. Due to the Async model of node we want to be able to run several queries at once, for this to work we create a new connection for each query to run in and then close it when the query is completed. There is currently no support for connection pooling as most users run a small number of long running hive queries but pooling should be possible if and when it's needed.

Notes

Thrift module has been cloned from https://github.com/wadey/node-thrift/commit/25c0eb4eb85aa63cfb49a8e8c815bd57e2b8043a

nodejs bindinga are generated from hive 0.6.1 CDH3B4