F.42. rum

F.42.1. Introduction
F.42.2. Common operators and functions
F.42.3. Operator classes

F.42.1. Introduction

The rum module provides access method to work with RUM index. It is based on the GIN access methods code.

GIN index allows to perform fast full text search using tsvector and tsquery types. But full text search with GIN index has several problems:

  • Slow ranking. It is need position information about lexems to ranking. GIN index doesn’t store positions of lexems. So after index scan we need additional heap scan to retrieve lexems positions.

  • Slow phrase search with GIN index. This problem relates with previous problem. It is need position information to perform phrase search.

  • Slow ordering by timestamp. GIN index can’t store some related information in index with lexemes. So it is necessary to perform additional heap scan.

RUM solves this problems by storing additional information in posting tree. For example, positional information of lexemes or timestamps.

Drawback of RUM is that it has slower build and insert time than GIN. It is because we need to store additional information besides keys and because RUM uses generic WAL records.

F.42.2. Common operators and functions

rum module provides next operators.

Operator Returns Description
tsvector <=> tsquery float4 Returns distance between tsvector and tsquery.
timestamp <=> timestamp float8 Returns distance between two timestamps.
timestamp <=| timestamp float8 Returns distance only for left timestamps.
timestamp |=> timestamp float8 Returns distance only for right timestamps.

Last three operations also works for types timestamptz, int2, int4, int8, float4, float8 and oid.

F.42.3. Operator classes

rum provides next operator classes.

F.42.3.1. rum_tsvector_ops

For type: tsvector

This operator class stores tsvector lexemes with positional information. Supports ordering by <=> operator and prefix search. There is the example.

Let us assume we have the table:

CREATE TABLE test_rum(t text, a tsvector);

CREATE TRIGGER tsvectorupdate
BEFORE UPDATE OR INSERT ON test_rum
FOR EACH ROW EXECUTE PROCEDURE tsvector_update_trigger('a', 'pg_catalog.english', 't');

INSERT INTO test_rum(t) VALUES ('The situation is most beautiful');
INSERT INTO test_rum(t) VALUES ('It is a beautiful');
INSERT INTO test_rum(t) VALUES ('It looks like a beautiful place');

To create the rum index we need create an extension:

CREATE EXTENSION rum;

Then we can create new index:

CREATE INDEX rumidx ON test_rum USING rum (a rum_tsvector_ops);

And we can execute the following queries:

SELECT t, a <=> to_tsquery('english', 'beautiful | place') AS rank
    FROM test_rum
    WHERE a @@ to_tsquery('english', 'beautiful | place')
    ORDER BY a <=> to_tsquery('english', 'beautiful | place');
                t                |  rank
---------------------------------+---------
 It looks like a beautiful place | 8.22467
 The situation is most beautiful | 16.4493
 It is a beautiful               | 16.4493
(3 rows)

SELECT t, a <=> to_tsquery('english', 'place | situation') AS rank
    FROM test_rum
    WHERE a @@ to_tsquery('english', 'place | situation')
    ORDER BY a <=> to_tsquery('english', 'place | situation');
                t                |  rank
---------------------------------+---------
 The situation is most beautiful | 16.4493
 It looks like a beautiful place | 16.4493
(2 rows)

F.42.3.2. rum_tsvector_hash_ops

For type: tsvector

This operator class stores hash of tsvector lexemes with positional information. Supports ordering by <=> operator. But doesn’t support prefix search.

F.42.3.3. rum_TYPE_ops

For types: int2, int4, int8, float4, float8, oid, time, timetz, date, interval, macaddr, inet, cidr, text, varchar, char, bytea, bit, varbit, numeric, timestamp, timestamptz

Supported operations: <, <=, =, >=, > for all types and <=>, <=| and |=> for int2, int4, int8, float4, float8, oid, timestamp and timestamptz types.

Supports ordering by <=>, <=| and |=> operators. Can be used with rum_tsvector_addon_ops, rum_tsvector_hash_addon_ops' andrum_anyarray_addon_ops` operator classes.

F.42.3.4. rum_tsvector_addon_ops

For type: tsvector

This operator class stores tsvector lexems with any supported by module field. There is the example.

Let us assume we have the table:

CREATE TABLE tsts (id int, t tsvector, d timestamp);

\copy tsts from 'rum/data/tsts.data'

CREATE INDEX tsts_idx ON tsts USING rum (t rum_tsvector_addon_ops, d)
    WITH (attach = 'd', to = 't');

Now we can execute the following queries:

EXPLAIN (costs off)
    SELECT id, d, d <=> '2016-05-16 14:21:25' FROM tsts WHERE t @@ 'wr&qh' ORDER BY d <=> '2016-05-16 14:21:25' LIMIT 5;
                                    QUERY PLAN
-----------------------------------------------------------------------------------
 Limit
   ->  Index Scan using tsts_idx on tsts
         Index Cond: (t @@ '''wr'' & ''qh'''::tsquery)
         Order By: (d <=> 'Mon May 16 14:21:25 2016'::timestamp without time zone)
(4 rows)

SELECT id, d, d <=> '2016-05-16 14:21:25' FROM tsts WHERE t @@ 'wr&qh' ORDER BY d <=> '2016-05-16 14:21:25' LIMIT 5;
 id  |                d                |   ?column?
-----+---------------------------------+---------------
 355 | Mon May 16 14:21:22.326724 2016 |      2.673276
 354 | Mon May 16 13:21:22.326724 2016 |   3602.673276
 371 | Tue May 17 06:21:22.326724 2016 |  57597.326724
 406 | Wed May 18 17:21:22.326724 2016 | 183597.326724
 415 | Thu May 19 02:21:22.326724 2016 | 215997.326724
(5 rows)

Warning: Currently RUM has bogus behaviour when one creates an index using ordering over pass-by-reference additional information. This is due to the fact that posting trees have fixed length right bound and fixed length non-leaf posting items. It isn’t allowed to create such indexes.

F.42.3.5. rum_tsvector_hash_addon_ops

For type: tsvector

This operator class stores hash of tsvector lexems with any supported by module field.

Doesn’t support prefix search.

F.42.3.6. rum_tsquery_ops

For type: tsquery

Stores branches of query tree in additional information. For example we have the table:

CREATE TABLE query (q tsquery, tag text);

INSERT INTO query VALUES ('supernova & star', 'sn'),
    ('black', 'color'),
    ('big & bang & black & hole', 'bang'),
    ('spiral & galaxy', 'shape'),
    ('black & hole', 'color');

CREATE INDEX query_idx ON query USING rum(q);

Now we can execute the following fast query:

SELECT * FROM query
    WHERE to_tsvector('black holes never exists before we think about them') @@ q;
        q         |  tag
------------------+-------
 'black'          | color
 'black' & 'hole' | color
(2 rows)

F.42.3.7. rum_anyarray_ops

For type: anyarray

This operator class stores anyarray elements with length of the array. Supports operators &&, @>, <@, =, % operators. Supports ordering by <=> operator. For example we have the table:

CREATE TABLE test_array (i int2[]);

INSERT INTO test_array VALUES ('{}'), ('{0}'), ('{1,2,3,4}'), ('{1,2,3}'), ('{1,2}'), ('{1}');

CREATE INDEX idx_array ON test_array USING rum (i rum_anyarray_ops);

Now we can execute the query using index scan:

SET enable_seqscan TO off;

EXPLAIN (COSTS OFF) SELECT * FROM test_array WHERE i && '{1}' ORDER BY i <=> '{1}' ASC;
                QUERY PLAN
------------------------------------------
 Index Scan using idx_array on test_array
   Index Cond: (i && '{1}'::smallint[])
   Order By: (i <=> '{1}'::smallint[])
(3 rows

SELECT * FROM test_array WHERE i && '{1}' ORDER BY i <=> '{1}' ASC;
     i
-----------
 {1}
 {1,2}
 {1,2,3}
 {1,2,3,4}
(4 rows)

F.42.3.8. rum_anyarray_addon_ops

For type: anyarray

This operator class stores anyarray elements with any supported by module field.