A newer version of this documentation is available.

View Latest

NEST Clause

  • reference
    +
    The NEST clause creates an input object by producing a single result of nesting keyspaces.

    Purpose

    The NEST clause is used within the FROM clause. It enables you to create an input object by producing a single result of nesting keyspaces via ANSI NEST, Lookup NEST, or Index NEST.

    Prerequisites

    For you to select data from keyspace or expression, you must have the query_select privilege on that keyspace. For more details about user roles, see Authorization.

    Syntax

    nest clause

    FROM Term

    The FROM term is a keyspace reference or expression which defines the input object(s) for the query, representing the left-hand side of the NEST clause.

    The FROM term is recursive, which enables you to chain the NEST clause with any of the terms which are permitted in the FROM clause, including other NEST clauses. For more information, see the page on the FROM clause.

    There are restrictions on what types of FROM terms may be chained and in what order — see the descriptions on this page for more details.

    A summary of the different types of FROM term is given in the following table.

    Type Example

    keyspace identifier

    `travel-sample`.inventory.hotel

    generic expression

    20+10 AS Total

    subquery

    SELECT ARRAY_AGG(t1.city) AS cities,
      SUM(t1.city_cnt) AS apnum
    FROM (
      SELECT city, city_cnt, country,
        ARRAY_AGG(airportname) AS apnames
      FROM `travel-sample`.inventory.airport
      GROUP BY city, country
      LETTING city_cnt = COUNT(city)
    ) AS t1
    WHERE t1.city_cnt > 5;

    previous join, nest, or unnest

    SELECT *
    FROM `travel-sample`.inventory.route AS rte
    JOIN `travel-sample`.inventory.airport AS apt
      ON rte.destinationairport = apt.faa
    NEST `travel-sample`.inventory.landmark AS lmk
      ON apt.city = lmk.city
    LIMIT 5;

    ANSI NEST Clause

    ANSI JOIN and ANSI NEST clauses have much more flexible functionality than their earlier INDEX and LOOKUP equivalents. Since these are standard compliant and more flexible, we recommend you to use ANSI JOIN and ANSI NEST exclusively, where possible.

    ANSI NEST supports more nest types than Lookup NEST or Index NEXT. ANSI NEST can nest arbitrary fields of the documents, and can be chained together.

    The key difference between ANSI NEST and other supported NEST types is the replacement of the ON KEYS or ON KEY … FOR clauses with a simple ON clause. The ON KEYS or ON KEY … FOR clauses dictate that those nests can only be done on a document key (primary key for a document). The ON clause can contain any expression, and thus it opens up many more nest possibilities.

    Syntax

    ansi-nest-clause ::= [ ansi-nest-type ] NEST ansi-nest-rhs ansi-nest-predicate
    ansi-nest-type? 'NEST' ansi-nest-rhs ansi-nest-predicate

    Nest Type

    ansi-nest-type ::= INNER | ( LEFT [ OUTER ] )
    ansi nest type

    This clause represents the type of ANSI nest.

    INNER

    For each nested object produced, both the left-hand and right-hand source objects must be non-MISSING and non-NULL.

    LEFT [OUTER]

    [Query Service interprets LEFT as LEFT OUTER]

    For each nested object produced, only the left-hand source objects must be non-MISSING and non-NULL.

    This clause is optional. If omitted, the default is INNER.

    Nest Right-Hand Side

    ansi-nest-rhs ::= keyspace-ref [ [ AS ] alias ]
    keyspace-ref ( 'AS'? alias )?
    Keyspace Reference

    Keyspace reference or expression representing the right-hand side of the NEST clause. For details, see Keyspace Reference.

    AS Alias

    Assigns another name to the right-hand side of the NEST clause. For details, see AS Clause.

    Assigning an alias to the keyspace reference is optional. If you assign an alias to the keyspace reference, the AS keyword may be omitted.

    Nest Predicate

    ansi-nest-predicate ::= ON expr
    ansi nest predicate
    expr

    Boolean expression representing the nest condition between the left-hand side FROM Term and the right-hand side Keyspace Reference. This expression may contain fields, constant expressions, or any complex N1QL expression.

    Limitations

    The following nest types are currently not supported:

    • Full OUTER NEST.

    • Cross NEST.

    • No mixing of new ANSI NEST syntax with NEST syntax in the same FROM clause.

    • The right-hand-side of any nest must be a keyspace. Expressions, subqueries, or other join combinations cannot be on the right-hand-side of a nest.

    • A nest can only be executed when appropriate index exists on the inner side of the ANSI NEST (similar to current NEST support).

    • Adaptive indexes are not considered when selecting indexes on inner side of the nest.

    Examples

    Example 1. Inner ANSI NEST

    List only airports in Toulouse which have routes starting from them, and nest details of the routes.

    Query
    SELECT *
    FROM `travel-sample`.inventory.airport a
      INNER NEST `travel-sample`.inventory.route r
      ON a.faa = r.sourceairport
    WHERE a.city = "Toulouse"
    ORDER BY a.airportname;
    Results
    [
      {
        "a": {
          "airportname": "Blagnac",
          "city": "Toulouse",
          "country": "France",
          "faa": "TLS",
          "geo": {
            "alt": 499,
            "lat": 43.629075,
            "lon": 1.363819
          },
          "icao": "LFBO",
          "id": 1273,
          "type": "airport",
          "tz": "Europe/Paris"
        },
        "r": [
          {
            "airline": "AH",
            "airlineid": "airline_794",
            "destinationairport": "ALG",
            "distance": 787.299015326995,
            "equipment": "736",
            "id": 10265,
    // ...
          },
          {
            "airline": "AH",
            "airlineid": "airline_794",
            "destinationairport": "ORN",
            "distance": 906.1483088609814,
            "equipment": "736",
            "id": 10266,
    // ...
        ]
      }
    ]
    Example 2. Left Outer ANSI NEST

    List all airports in Toulouse, and nest details of any routes that start from each airport.

    Query
    SELECT *
    FROM `travel-sample`.inventory.airport a
      LEFT NEST `travel-sample`.inventory.route r
      ON a.faa = r.sourceairport
    WHERE a.city = "Toulouse"
    ORDER BY a.airportname;
    Results
    [
      {
        "a": {
          "airportname": "Blagnac",
          "city": "Toulouse",
          "country": "France",
          "faa": "TLS",
          "geo": {
            "alt": 499,
            "lat": 43.629075,
            "lon": 1.363819
          },
          "icao": "LFBO",
          "id": 1273,
          "type": "airport",
          "tz": "Europe/Paris"
        },
        "r": [
          {
            "airline": "AH",
            "airlineid": "airline_794",
            "destinationairport": "ALG",
            "distance": 787.299015326995,
            "equipment": "736",
            "id": 10265,
    // ...
          }
        ]
      },
      {
        "a": {
          "airportname": "Francazal",
          "city": "Toulouse",
          "country": "France",
          "faa": null,
          "geo": {
            "alt": 535,
            "lat": 43.545555,
            "lon": 1.3675
          },
          "icao": "LFBF",
          "id": 1266,
          "type": "airport",
          "tz": "Europe/Paris"
        },
        "r": [] (1)
      },
      {
        "a": {
          "airportname": "Lasbordes",
          "city": "Toulouse",
          "country": "France",
          "faa": null,
          "geo": {
            "alt": 459,
            "lat": 43.586113,
            "lon": 1.499167
          },
          "icao": "LFCL",
          "id": 1286,
          "type": "airport",
          "tz": "Europe/Paris"
        },
        "r": []
      }
    ]
    1 The LEFT OUTER NEST lists all the left-side results, even if there are no matching right-side documents, as indicated by the results in which the fields from the route keyspace are null or missing.

    Lookup NEST Clause

    Nesting is conceptually the inverse of unnesting. Nesting performs a join across two keyspaces. But instead of producing a cross-product of the left and right inputs, a single result is produced for each left input, while the corresponding right inputs are collected into an array and nested as a single array-valued field in the result object.

    Syntax

    lookup-nest-clause ::= [ lookup-nest-type ] NEST lookup-nest-rhs lookup-nest-predicate
    lookup-nest-type? 'NEST' lookup-nest-rhs lookup-nest-predicate

    Nest Type

    lookup-nest-type ::= INNER | ( LEFT [ OUTER ] )
    lookup nest type

    This clause represents the type of lookup nest.

    INNER

    For each result object produced, both the left-hand and right-hand source objects must be non-MISSING and non-NULL.

    LEFT [OUTER]

    [Query Service interprets LEFT as LEFT OUTER]

    A left-outer unnest is performed, and at least one result object is produced for each left source object.

    For each joined object produced, only the left-hand source objects must be non-MISSING and non-NULL.

    This clause is optional. If omitted, the default is INNER.

    Nest Right-Hand Side

    lookup-nest-rhs ::= keyspace-ref [ [ AS ] alias ]
    keyspace-ref ( 'AS'? alias )?
    Keyspace Reference

    Keyspace reference for the right-hand side of the lookup nest. For details, see Keyspace Reference.

    AS Alias

    Assigns another name to the right-hand side of the lookup nest. For details, see AS Clause.

    Assigning an alias to the keyspace reference is optional. If you assign an alias to the keyspace reference, the AS keyword may be omitted.

    Nest Predicate

    lookup-nest-predicate ::= ON KEYS expr
    lookup nest predicate

    The ON KEYS expression produces a document key or array of document keys for the right-hand side of the lookup nest.

    expr

    [Required] String or expression representing the primary keys of the documents for the right-hand side keyspace.

    Return Values

    If the right-hand source object is NULL, MISSING, empty, or a non-array value, then the result object’s right-side value is MISSING (omitted).

    Nests can be chained with other NEST, JOIN, and UNNEST clauses. By default, an INNER NEST is performed. This means that for each result object produced, both the left and right source objects must be non-missing and non-null. The right-hand side result of NEST is always an array or MISSING. If there is no matching right source object, then the right source object is as follows:

    If the ON KEYS expression evaluates to Then the right-side value is

    MISSING

    MISSING

    NULL

    MISSING

    an array

    an empty array

    a non-array value

    an empty array

    Examples

    Example 3. Join two keyspaces producing an output for each left input

    Show one set of routes for one airline in the airline keyspace.

    Query
    SELECT *
    FROM `travel-sample`.inventory.route
      INNER NEST `travel-sample`.inventory.airline
      ON KEYS route.airlineid
    LIMIT 1;
    Results
    [
      {
        "airline": [
          {
            "callsign": "AIRFRANS",
            "country": "France",
            "iata": "AF",
            "icao": "AFR",
            "id": 137,
            "name": "Air France",
            "type": "airline"
          }
        ],
        "route": {
          "airline": "AF",
          "airlineid": "airline_137",
          "destinationairport": "MRS",
          "distance": 2881.617376098415,
          "equipment": "320",
          "id": 10000,
          "schedule": [
    // ...
          ],
          "sourceairport": "TLV",
          "stops": 0,
          "type": "route"
        }
      }
    ]

    Index NEST Clause

    Index NESTs allow you to flip the direction of a Lookup NEST clause. Index NESTs can be used efficiently when Lookup NESTs cannot efficiently nest left-hand side documents with right-to-left nests, and your situation cannot be flipped because your predicate needs to be on the left-hand side, such as Example 3 above where airline documents have no reference to route documents.

    For index nests, the syntax uses ON KEY (singular) instead of ON KEYS (plural). This is because an Index NEST’s ON KEY expression must produce a scalar value; whereas a Lookup NEST’s ON KEYS expression can produce either a scalar or an array value.

    Syntax

    index-nest-clause ::= [ index-nest-type ] NEST index-nest-rhs index-nest-predicate
    index-nest-type? 'NEST' index-nest-rhs index-nest-predicate

    Nest Type

    index-nest-type ::= INNER | ( LEFT [ OUTER ] )
    index nest type

    This clause represents the type of index nest.

    INNER

    For each nested object produced, both the left-hand and right-hand source objects must be non-MISSING and non-NULL.

    LEFT [OUTER]

    [Query Service interprets LEFT as LEFT OUTER]

    For each nested object produced, only the left-hand source objects must be non-MISSING and non-NULL.

    This clause is optional. If omitted, the default is INNER.

    Nest Right-Hand Side

    index-nest-rhs ::= keyspace-ref [ [ AS ] alias ]
    keyspace-ref ( 'AS'? alias )?
    Keyspace Reference

    Keyspace reference or expression representing the right-hand side of the NEST clause. For details, see Keyspace Reference.

    AS Alias

    Assigns another name to the right-hand side of the NEST clause. For details, see AS Clause.

    Assigning an alias to the keyspace reference is optional. If you assign an alias to the keyspace reference, the AS keyword may be omitted.

    Nest Predicate

    index-nest-predicate ::= ON KEY expr FOR alias
    index nest predicate
    expr

    Expression in the form rhs-expression.lhs-expression-key:

    rhs-expression

    Keyspace reference for the right-hand side of the index nest.

    lhs-expression-key

    String or expression representing the attribute in rhs-expression and referencing the document key for alias.

    alias

    Keyspace reference for the left-hand side of the index nest.

    Examples

    Example 4. Use INDEX nest to flip the direction of Example 3 above

    This example nests the airline routes for each airline after creating the following index. (Note that the index will not match if it contains a WHERE clause.)

    Index
    CREATE INDEX route_airlineid ON `travel-sample`.inventory.route(airlineid);
    Query
    SELECT *
    FROM `travel-sample`.inventory.airline aline
      INNER NEST `travel-sample`.inventory.route rte
      ON KEY rte.airlineid FOR aline
    LIMIT 1;
    Results
    [
      {
        "aline": {
          "callsign": "MILE-AIR",
          "country": "United States",
          "iata": "Q5",
          "icao": "MLA",
          "id": 10,
          "name": "40-Mile Air",
          "type": "airline"
        },
        "rte": [
          {
            "airline": "Q5",
            "airlineid": "airline_10",
            "destinationairport": "FAI",
            "distance": 118.20183585107631,
            "equipment": "CNA",
            "id": 46587,
            "schedule": [
    // ...
            ],
            "sourceairport": "HKB",
            "stops": 0,
            "type": "route"
          },
          {
            "airline": "Q5",
            "airlineid": "airline_10",
            "destinationairport": "HKB",
            "distance": 118.20183585107631,
            "equipment": "CNA",
            "id": 46586,
            "schedule": [
    // ...
            ],
            "sourceairport": "FAI",
            "stops": 0,
            "type": "route"
          }
        ]
      }
    ]

    If you generalize the same query, it looks like the following:

    CREATE INDEX on-key-for-index-name rhs-expression (lhs-expression-key);
    SELECT projection-list
    FROM lhs-expression
      NEST rhs-expression
      ON KEY rhs-expression.lhs-expression-key FOR lhs-expression
    [ WHERE predicates ] ;

    There are three important changes in the index scan syntax example above:

    • CREATE INDEX on the ON KEY expression rte.airlineid to access route documents using airlineid (which are produced on the left-hand side).

    • The ON KEY rte.airlineid FOR aline enables N1QL to use the index route_airlineid.

    • Create any optional index, such as route_airline, that can be used on airline (left-hand side).

    Appendix: Summary of NEST Types

    ANSI

    Left-Hand Side (lhs)

    Any field or expression that produces a value that will be matched on the right-hand side.

    Right-Hand Side (rhs)

    Anything that can have a proper index on the join expression.

    Syntax

    lhs-expr
    NEST rhs-keyspace
    ON any nest condition

    Example

    SELECT *
    FROM `travel-sample`.inventory.route r
    NEST `travel-sample`.inventory.airline a
    ON r.airlineid = META(a).id
    LIMIT 4;

    Lookup

    Left-Hand Side (lhs)

    Must produce a Document Key for the right-hand side.

    Right-Hand Side (rhs)

    Must have a Document Key.

    Syntax

    lhs-expr
    NEST rhs-keyspace
    ON KEYS lhs-expr.foreign_key

    Example

    SELECT *
    FROM `travel-sample`.inventory.route r
    NEST `travel-sample`.inventory.airline a
    ON KEYS r.airlineid
    LIMIT 4;

    Index

    Left-Hand Side (lhs)

    Must produce a key for the right-hand side index.

    Right-Hand Side (rhs)

    Must have a proper index on the field or expression that maps to the Document Key of the left-hand side.

    Syntax

    lhs-keyspace
    NEST rhs-keyspace
    ON KEY rhs-kspace.idx_key
    FOR lhs-keyspace

    Example

    SELECT *
    FROM `travel-sample`.inventory.airline a
    NEST `travel-sample`.inventory.route r
    ON KEY r.airlineid FOR a
    LIMIT 4;