CODEWITHSUNDEEP
lucenestratiocassandra-lucene-index
Keith Nordstrom
Keith Nordstrom

Reputation: 354

Stratio Lucene Index 3.0.9: Purpose of INet mapper

From the docs for the Stratio Lucene Index Plugin, the INet mapper looks interesting but there's not a lot of motivation behind it: ie. https://github.com/Stratio/cassandra-lucene-index/blob/branch-3.0.9/doc/documentation.rst#inet-mapper.

Given that it represents an IP address (typed) and is therefore distinct from being simple a string, what sort of queries can I apply to it? In particular, is it possible to do a ranged query on it?

Upvotes: 0

Views: 37

Answers (1)

Andrés de la Peña
Andrés de la Peña

Reputation: 331

Lucene index inet mapper is intended to map CQL inet data type. The only advantage on it is to make parsing more flexible. For example, given the following data:

CREATE TABLE t (
    pk int PRIMARY KEY,
    address inet
) ;

CREATE CUSTOM INDEX i ON t () 
USING 'com.stratio.cassandra.lucene.Index' 
WITH OPTIONS = {
  'refresh_seconds': '1',
  'schema': '{
       fields : {
          address : {type: "inet"},
          address_s : {type: "string", column: "address"}
       }
    }'};

INSERT INTO t(pk, address) VALUES (0, '::FFFF:8:8:8');

All these queries will found the indexed row:

SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address", value: "::ffFF:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address", value: "::0:ffff:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address", value: "0:0:0:0:ffff:8:8:8"}}');

However, the same queries wouldn't work with a string mapper:

SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address_s", value: "::ffFF:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address_s", value: "::0:ffff:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"match", field:"address_s", value: "0:0:0:0:ffff:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"wildcard", field:"address_s", value: "*:8:8:8"}}');

Aside from this, the generated Lucene field is a string field, so there is no special treatment for wildcard or range queries, that should use the expanded format of the IP address and will have a lexicographical behaviour:

SELECT * FROM t WHERE expr(i, '{filter:{type:"wildcard", field:"address", value: "*:8:8:8"}}');
SELECT * FROM t WHERE expr(i, '{filter:{type:"range", field:"address", lower: "::FFFF:8:8:7"}}');

Upvotes: 1

Related Questions