Combined queries

Combine query expressions

A combined query is a combination of several query types, such as:

You can use logical query operators to combine query expressions for numeric, tag, and text fields. For vector fields, you can combine a KNN query with a pre-filter.

Note:
The operators are interpreted slightly differently depending on the query dialect used. The default dialect is DIALECT 1; see this article for information on how to change the dialect version. This article uses the second version of the query dialect, DIALECT 2, and uses additional brackets ((...)) to help clarify the examples. Further details can be found in the query syntax documentation.

The examples in this article use the following schema:

Field name Field type
description TEXT
condition TAG
price NUMERIC
vector VECTOR

AND

The binary operator (space) is used to intersect the results of two or more expressions.

FT.SEARCH index "(expr1) (expr2)"

If you want to perform an intersection based on multiple values within a specific text field, then you should use the following simplified notion:

FT.SEARCH index "@text_field:( value1 value2 ... )"

The following example shows you a query that finds bicycles in new condition and in a price range from 500 USD to 1000 USD:

You might also be interested in bicycles for kids. The query below shows you how to combine a full-text search with the criteria from the previous query:

OR

You can use the binary operator | (vertical bar) to perform a union.

FT.SEARCH index "(expr1) | (expr2)"
Note:
The logical AND takes precedence over OR when using dialect version two. The expression expr1 expr2 | expr3 expr4 means (expr1 expr2) | (expr3 expr4). Version one of the query dialect behaves differently. Using parentheses in query strings is advised to ensure the order is clear.

If you want to perform the union based on multiple values within a single tag or text field, then you should use the following simplified notion:

FT.SEARCH index "@text_field:( value1 | value2 | ... )"
FT.SEARCH index "@tag_field:{ value1 | value2 | ... }"

The following query shows you how to find used bicycles that contain either the word 'kids' or 'small':

The previous query searches across all text fields. The following example shows you how to limit the search to the description field:

If you want to extend the search to new bicycles, then the below example shows you how to do that:

NOT

A minus (-) in front of a query expression negates the expression.

FT.SEARCH index "-(expr)"

If you want to exclude new bicycles from the search within the previous price range, you can use this query:

Numeric filter

The FT.SEARCH command allows you to combine any query expression with a numeric filter.

FT.SEARCH index "expr" FILTER numeric_field start end

Please see the range query article to learn more about numeric range queries and such filters.

Pre-filter for a KNN vector query

You can use a simple or more complex query expression with logical operators as a pre-filter in a KNN vector query.

FT.SEARCH index "(filter_expr)=>[KNN num_neighbours @field $vector]" PARAMS 2 vector "binary_data" DIALECT 2

Here is an example:

The vector search article provides further details about vector queries in general.

RATE THIS PAGE
Back to top ↑