Dictionary Key and Fields

The <structure> clause describes the dictionary key and fields available for queries.

XML description:

  1. <dictionary>
  2. <structure>
  3. <id>
  4. <name>Id</name>
  5. </id>
  6. <attribute>
  7. <!-- Attribute parameters -->
  8. </attribute>
  9. ...
  10. </structure>
  11. </dictionary>

Attributes are described in the elements:

DDL query:

  1. CREATE DICTIONARY dict_name (
  2. Id UInt64,
  3. -- attributes
  4. )
  5. PRIMARY KEY Id
  6. ...

Attributes are described in the query body:

  • PRIMARY KEYKey column
  • AttrName AttrTypeData column. There can be a multiple number of attributes.

Key

ClickHouse supports the following types of keys:

  • Numeric key. UInt64. Defined in the <id> tag or using PRIMARY KEY keyword.
  • Composite key. Set of values of different types. Defined in the tag <key> or PRIMARY KEY keyword.

An xml structure can contain either <id> or <key>. DDL-query must contain single PRIMARY KEY.

Warning

You must not describe key as an attribute.

Numeric Key

Type: UInt64.

Configuration example:

  1. <id>
  2. <name>Id</name>
  3. </id>

Configuration fields:

  • name – The name of the column with keys.

For DDL-query:

  1. CREATE DICTIONARY (
  2. Id UInt64,
  3. ...
  4. )
  5. PRIMARY KEY Id
  6. ...
  • PRIMARY KEY – The name of the column with keys.

Composite Key

The key can be a tuple from any types of fields. The layout in this case must be complex_key_hashed or complex_key_cache.

Tip

A composite key can consist of a single element. This makes it possible to use a string as the key, for instance.

The key structure is set in the element <key>. Key fields are specified in the same format as the dictionary attributes. Example:

  1. <structure>
  2. <key>
  3. <attribute>
  4. <name>field1</name>
  5. <type>String</type>
  6. </attribute>
  7. <attribute>
  8. <name>field2</name>
  9. <type>UInt32</type>
  10. </attribute>
  11. ...
  12. </key>
  13. ...

or

  1. CREATE DICTIONARY (
  2. field1 String,
  3. field2 String
  4. ...
  5. )
  6. PRIMARY KEY field1, field2
  7. ...

For a query to the dictGet* function, a tuple is passed as the key. Example: dictGetString('dict_name', 'attr_name', tuple('string for field1', num_for_field2)).

Attributes

Configuration example:

  1. <structure>
  2. ...
  3. <attribute>
  4. <name>Name</name>
  5. <type>ClickHouseDataType</type>
  6. <null_value></null_value>
  7. <expression>rand64()</expression>
  8. <hierarchical>true</hierarchical>
  9. <injective>true</injective>
  10. <is_object_id>true</is_object_id>
  11. </attribute>
  12. </structure>

or

  1. CREATE DICTIONARY somename (
  2. Name ClickHouseDataType DEFAULT '' EXPRESSION rand64() HIERARCHICAL INJECTIVE IS_OBJECT_ID
  3. )

Configuration fields:

TagDescriptionRequired
nameColumn name.Yes
typeClickHouse data type: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64, UUID, Decimal32, Decimal64, Decimal128, Decimal256, String, Array.
ClickHouse tries to cast value from dictionary to the specified data type. For example, for MySQL, the field might be TEXT, VARCHAR, or BLOB in the MySQL source table, but it can be uploaded as String in ClickHouse.
Nullable is currently supported for Flat, Hashed, ComplexKeyHashed, Direct, ComplexKeyDirect, RangeHashed, Polygon, Cache, ComplexKeyCache, SSDCache, SSDComplexKeyCache dictionaries. In IPTrie dictionaries Nullable types are not supported.
Yes
null_valueDefault value for a non-existing element.
In the example, it is an empty string. NULL value can be used only for the Nullable types (see the previous line with types description).
Yes
expressionExpression that ClickHouse executes on the value.
The expression can be a column name in the remote SQL database. Thus, you can use it to create an alias for the remote column.

Default value: no expression.
No
hierarchicalIf true, the attribute contains the value of a parent key for the current key. See Hierarchical Dictionaries.

Default value: false.
No
injectiveFlag that shows whether the id -> attribute image is injective.
If true, ClickHouse can automatically place after the GROUP BY clause the requests to dictionaries with injection. Usually it significantly reduces the amount of such requests.

Default value: false.
No
is_object_idFlag that shows whether the query is executed for a MongoDB document by ObjectID.

Default value: false.
No

See Also