F.17. intarray

The intarray module provides a number of useful functions and operators for manipulating null-free arrays of integers. There is also support for indexed searches using some of the operators.

All of these operations will throw an error if a supplied array contains any NULL elements.

Many of these operations are only sensible for one-dimensional arrays. Although they will accept input arrays of more dimensions, the data is treated as though it were a linear array in storage order.

F.17.1. intarray Functions and Operators

The functions provided by the intarray module are shown in Table F.10, the operators in Table F.11.

Table F.10. intarray Functions

FunctionReturn TypeDescriptionExampleResult
icount(int[])intnumber of elements in arrayicount('{1,2,3}'::int[])3
sort(int[], text dir)int[]sort array — dir must be asc or descsort('{1,2,3}'::int[], 'desc'){3,2,1}
sort(int[])int[]sort in ascending ordersort(array[11,77,44]){11,44,77}
sort_asc(int[])int[]sort in ascending order
sort_desc(int[])int[]sort in descending order
uniq(int[])int[]remove adjacent duplicatesuniq(sort('{1,2,3,2,1}'::int[])){1,2,3}
idx(int[], int item)intindex of first element matching item (0 if none)idx(array[11,22,33,22,11], 22)2
subarray(int[], int start, int len)int[]portion of array starting at position start, len elementssubarray('{1,2,3,2,1}'::int[], 2, 3){2,3,2}
subarray(int[], int start)int[]portion of array starting at position startsubarray('{1,2,3,2,1}'::int[], 2){2,3,2,1}
intset(int)int[]make single-element arrayintset(42){42}

Table F.11. intarray Operators

OperatorReturnsDescription
int[] && int[]booleanoverlap — true if arrays have at least one common element
int[] @> int[]booleancontains — true if left array contains right array
int[] <@ int[]booleancontained — true if left array is contained in right array
# int[]intnumber of elements in array
int[] # intintindex (same as idx function)
int[] + intint[]push element onto array (add it to end of array)
int[] + int[] int[]array concatenation (right array added to the end of left one)
int[] - intint[]remove entries matching right argument from array
int[] - int[]int[]remove elements of right array from left
int[] | intint[]union of arguments
int[] | int[]int[]union of arrays
int[] & int[]int[]intersection of arrays
int[] @@ query_intbooleantrue if array satisfies query (see below)
query_int ~~ int[]booleantrue if array satisfies query (commutator of @@)

(Before PostgreSQL 8.2, the containment operators @> and <@ were respectively called @ and ~. These names are still available, but are deprecated and will eventually be retired. Notice that the old names are reversed from the convention formerly followed by the core geometric data types!)

The operators &&, @> and <@ are equivalent to PostgreSQL's built-in operators of the same names, except that they work only on integer arrays that do not contain nulls, while the built-in operators work for any array type. This restriction makes them faster than the built-in operators in many cases.

The @@ and ~~ operators test whether an array satisfies a query, which is expressed as a value of a specialized data type query_int. A query consists of integer values that are checked against the elements of the array, possibly combined using the operators & (AND), | (OR), and ! (NOT). Parentheses can be used as needed. For example, the query 1&(2|3) matches arrays that contain 1 and also contain either 2 or 3.

F.17.2. Index Support

intarray provides index support for the &&, @>, <@, and @@ operators, as well as regular array equality.

Two GiST index operator classes are provided: gist__int_ops (used by default) is suitable for small- to medium-size data sets, while gist__intbig_ops uses a larger signature and is more suitable for indexing large data sets (i.e., columns containing a large number of distinct array values). The implementation uses an RD-tree data structure with built-in lossy compression.

There is also a non-default GIN operator class gin__int_ops supporting the same operators.

The choice between GiST and GIN indexing depends on the relative performance characteristics of GiST and GIN, which are discussed elsewhere. As a rule of thumb, a GIN index is faster to search than a GiST index, but slower to build or update; so GIN is better suited for static data and GiST for often-updated data.

F.17.3. Example

-- a message can be in one or more sections
CREATE TABLE message (mid INT PRIMARY KEY, sections INT[], ...);

-- create specialized index
CREATE INDEX message_rdtree_idx ON message USING GIST (sections gist__int_ops);

-- select messages in section 1 OR 2 - OVERLAP operator
SELECT message.mid FROM message WHERE message.sections && '{1,2}';

-- select messages in sections 1 AND 2 - CONTAINS operator
SELECT message.mid FROM message WHERE message.sections @> '{1,2}';

-- the same, using QUERY operator
SELECT message.mid FROM message WHERE message.sections @@ '1&2'::query_int;

F.17.4. Benchmark

The source directory contrib/intarray/bench contains a benchmark test suite. To run:

cd .../bench
createdb TEST
psql TEST < ../_int.sql
./create_test.pl | psql TEST
./bench.pl

The bench.pl script has numerous options, which are displayed when it is run without any arguments.

F.17.5. Authors

All work was done by Teodor Sigaev () and Oleg Bartunov (). See http://www.sai.msu.su/~megera/postgres/gist/ for additional information. Andrey Oktyabrski did a great work on adding new functions and operations.