infi.clickhouse_orm/docs/field_options.md
Itai Shirav 113ac7ad4a docs
2020-05-28 19:38:41 +03:00

5.5 KiB

Field Options

All field types accept the following arguments:

  • default
  • alias
  • materialized
  • readonly
  • codec

Note that default, alias and materialized are mutually exclusive - you cannot use more than one of them in a single field.

default

Specifies a default value to use for the field. If not given, the field will have a default value based on its type: empty string for string fields, zero for numeric fields, etc. The default value can be a Python value suitable for the field type, or an expression. For example:

class Event(Model):

    name = StringField(default="EVENT")
    repeated = UInt32Field(default=1)
    created = DateTimeField(default=F.now())

    engine = Memory()
    ...

When creating a model instance, any fields you do not specify get their default value. Fields that use a default expression are assigned a sentinel value of infi.clickhouse_orm.utils.NO_VALUE instead. For example:

>>> event = Event()
>>> print(event.to_dict())
{'name': 'EVENT', 'repeated': 1, 'created': <NO_VALUE>}

⚠️ Due to a bug in ClickHouse versions prior to 20.1.2.4, insertion of records with expressions for default values may fail.

alias / materialized

The alias and materialized attributes expect an expression that gets calculated by the database. The difference is that alias fields are calculated on the fly, while materialized fields are calculated when the record is inserted, and are stored on disk. You can use any expression, and can refer to other model fields. For example:

class Event(Model):

    created = DateTimeField()
    created_date = DateTimeField(materialized=F.toDate(created))
    name = StringField()
    normalized_name = StringField(alias=F.upper(F.trim(name)))

    engine = Memory()

For backwards compatibility with older versions of the ORM, you can pass the expression as an SQL string:

    created_date = DateTimeField(materialized="toDate(created)")

Both field types can't be inserted into the database directly, so they are ignored when using the Database.insert() method. ClickHouse does not return the field values if you use "SELECT * FROM ..." - you have to list these field names explicitly in the query.

Usage:

obj = Event(created=datetime.now(), name='MyEvent')
db = Database('my_test_db')
db.insert([obj])
# All values will be retrieved from database
db.select('SELECT created, created_date, username, name FROM $db.event', model_class=Event)
# created_date and username will contain a default value
db.select('SELECT * FROM $db.event', model_class=Event)

When creating a model instance, any alias or materialized fields are assigned a sentinel value of infi.clickhouse_orm.utils.NO_VALUE since their real values can only be known after insertion to the database.

codec

This attribute specifies the compression algorithm to use for the field (instead of the default data compression algorithm defined in server settings).

Supported compression algorithms:

Codec Argument Comment
NONE None No compression.
LZ4 None LZ4 compression.
LZ4HC(level) Possible level range: [3, 12]. Default value: 9. Greater values stands for better compression and higher CPU usage. Recommended value range: [4,9].
ZSTD(level) Possible levelrange: [1, 22]. Default value: 1. Greater values stands for better compression and higher CPU usage. Levels >= 20, should be used with caution, as they require more memory.
Delta(delta_bytes) Possible delta_bytes range: 1, 2, 4 , 8. Default value for delta_bytes is sizeof(type) if it is equal to 1, 2,4 or 8 and equals to 1 otherwise.

Codecs can be combined by separating their names with commas. The default database codec is not included into pipeline (if it should be applied to a field, you have to specify it explicitly in pipeline).

Recommended usage for codecs:

  • When values for particular metric do not differ significantly from point to point, delta-encoding allows to reduce disk space usage significantly.
  • DateTime works great with pipeline of Delta, ZSTD and the column size can be compressed to 2-3% of its original size (given a smooth datetime data)
  • Numeric types usually enjoy best compression rates with ZSTD
  • String types enjoy good compression rates with LZ4HC

Example:

class Stats(Model):

    id                  = UInt64Field(codec='ZSTD(10)')
    timestamp           = DateTimeField(codec='Delta,ZSTD')
    timestamp_date      = DateField(codec='Delta(4),ZSTD(22)')
    metadata_id         = Int64Field(codec='LZ4')
    status              = StringField(codec='LZ4HC(10)')
    calculation         = NullableField(Float32Field(), codec='ZSTD')
    alerts              = ArrayField(FixedStringField(length=15), codec='Delta(2),LZ4HC')

    engine = MergeTree('timestamp_date', ('id', 'timestamp'))

Note: This feature is supported on ClickHouse version 19.1.16 and above. Codec arguments will be ignored by the ORM for older versions of ClickHouse.

readonly

This attribute is set automatically for fields with alias or materialized attributes, you do not need to pass it yourself.


<< Querysets | Table of Contents | Field Types >>