Python

See also

Using Python

Example Output

personinfo.py

Overview

The Python generator will generate a Python object model from a LinkML schema

To run:

gen-python personinfo.yaml > personinfo.py

This object model can now be used to create, load, and serialize objects. For example:

from personinfo import Person
p1 = Person('P1', name='John Smith')

The linkml-runtime framework can be used to serialize these objects to JSON, YAML, RDF, and TSV, as well as to load them

Dataclasses

We use the standard Python dataclasses framework

Note that there is an alternative generator for using Pydantic, see Pydantic Generator

Inheritance

The linkml is_a slot is mapped to superclasses in the generated classes. Note that mixins are not currently handled in the same way - we avoid multiple inheritance. Instead, mixins are “rolled down” to their implementing classes.

The root term for all classes is the class YAMLRoot in the linkml-runtime.

Object Creation

Generated class include specific initialization code in post_init that is permissive in how objects are created. For example, when adding inlined objects any of the following can be used:

event = EmploymentEvent(...)
# method 1: creation with objects
p1 = Person('P1', name='John Smith', employment_history=[event])

# method 2: creation with dicts
p1 = Person('P1', name='John Smith', employment_history=[{...}])

# method 3: implicit lists
p1 = Person('P1', name='John Smith', employment_history={...})

Non-inlined references

In LinkML schemas, references to other objects can be declared as inlined. If a reference is not inlined, and the referenced object has an identifier, then the value of the reference will be that identifier.

In the generated python, class are created for these foreign keys

For example, in the personinfo schema, the employment event class references organization via a non-inlined reference, so the following code is generated:

class OrganizationId(NamedThingId):

...

@dataclass
class EmploymentEvent(Event):

    employed_at: Optional[Union[str, OrganizationId]] = None

Types

In LinkML types represent terminal scalar values and are mapped to built in Python objects such integers.

For user-defined types, a class is created which inherits from that Python base type.

Code Docs

Command Line

gen-python

Generate python classes to represent a LinkML model

gen-python [OPTIONS] YAMLFILE

Options

--head, --no-head

Emit metadata heading

Default:

True

--genmeta, --no-genmeta

Generating metamodel. Only use this for generating meta.py

Default:

False

--classvars, --no-classvars

Generate CLASSVAR info

Default:

True

--slots, --no-slots

Generate Slot information

Default:

True

--validate, --no-validate

Validate generated code by compiling it

Default:

False

-V, --version

Show the version and exit.

-f, --format <format>

Output format

Default:

'py'

Options:

py

--metadata, --no-metadata

Include metadata in output

Default:

True

--useuris, --metauris

Use class and slot URIs over model uris

Default:

True

-im, --importmap <importmap>

Import mapping file

--log_level <log_level>

Logging level

Default:

'WARNING'

Options:

CRITICAL | ERROR | WARNING | INFO | DEBUG

-v, --verbose

Verbosity. Takes precedence over –log_level.

--mergeimports, --no-mergeimports

Merge imports into source file (default=mergeimports)

--stacktrace, --no-stacktrace

Print a stack trace when an error occurs

Default:

False

Arguments

YAMLFILE

Required argument

Code

class linkml.generators.pythongen.PythonGenerator(schema: str | ~typing.TextIO | ~linkml_runtime.linkml_model.meta.SchemaDefinition | ~linkml.utils.generator.Generator | ~pathlib.Path, schemaview: ~linkml_runtime.utils.schemaview.SchemaView | None = None, format: str | None = None, metadata: bool = True, useuris: bool | None = None, log_level: int | None = 30, mergeimports: bool | None = True, source_file_date: str | None = None, source_file_size: int | None = None, logger: ~logging.Logger | None = None, verbose: bool | None = None, output: str | None = None, namespaces: ~linkml_runtime.utils.namespaces.Namespaces | None = None, directory_output: bool = False, base_dir: str = None, metamodel_name_map: ~typing.Dict[str, str] = None, importmap: str | ~typing.Mapping[str, str] | None = None, emit_prefixes: ~typing.Set[str] = <factory>, metamodel: ~linkml.utils.schemaloader.SchemaLoader = None, stacktrace: bool = False, include: str | ~pathlib.Path | ~linkml_runtime.linkml_model.meta.SchemaDefinition | None = None, gen_classvars: bool = True, gen_slots: bool = True, genmeta: bool = False, emit_metadata: bool = True, dataclass_repr: bool = False, **_kwargs)

Generates Python dataclasses from a LinkML model

See Python Generator Docs

serialize(**kwargs) → str

Generate output in the required format

Parameters:

kwargs – Generator specific parameters

Returns:

Generated output

Additional Notes

Code generation for LinkML slots

The generation of slot entries in class definitions is split into two parts:

• 1. The declaration of the slot type

• 2) The __post_init__ processing transforms several allowable input forms into single, consistent internal representation.

The elements that control the python for slot generation include

• Whether range references an instance of a LinkML type or a LinkML class

• If the slot range is a class:

  • Whether the class has a key (key: <type>), an identifier (identifier: <type>), or neither

  • If the class has a key or identifier, whether the

    instances are inlined as a dictionary (inlined: true), inlined as a list (inlined_as_list: true) or are referenced elsewhere in the model (Default).

• Whether the slot is concrete (abstract: false) or abstract (abstract: true or mixin: true) (Default: false)

• Whether a slot is required (required: true) or optional (required: false) (Default: false)

• Whether a slot is single-valued (multivalued: false) or multi-valued (multivalued: true) (Default: false)

• The ifabsent attribute (not covered in this document)

• The default value (not covered in this document)

These various situations are described more detail below:

1) Slot range is a LinkML type definition

LinkML type definitions can take one of three forms:

• 1. Builtin python type

yaml     types:        <type>:           base: <builtin python type> (e.g. 'str', 'int', 'float', etc)           ...

• 2. Defined type

yaml     types:        <type>:           base: <type defined in linkml/linkml_runtime/utils/metamodelcore.py> (e.g. URIorCURIE, Date, NCName, etc)           ...

3. Inherited type

   yaml  types:     <type>:        typeof: <parent type> Each of these are outlined below

2) Python generation for basic python Type.

For this example, we show basic permutations on the python int type as a base YAML:

types:
  integer:
    uri: xsd:integer
    base: int
    description: An integer

classes:
    Integers:
      description: various permutations of the integer type
      attributes:
        opt_integer:
          range: integer
        mand_integer:
          range: integer
          required: true
        opt_multi_integer:
          range: integer
          multivalued: true
        mand_multi_integer:
          range: integer
          multivalued: true
          required: true

Generated Python:

class Integer(int):
    """ An integer """
    type_class_uri = XSD.integer
    type_class_curie = "xsd:integer"
    type_name = "integer"
    type_model_uri = METATYPE.Integer

@dataclass
class Integers(YAMLRoot):
    """
    various permutations of the integer type
    """

    mand_integer: int = None
    mand_multi_integer: Union[int, List[int]] = None
    opt_integer: Optional[int] = None
    opt_multi_integer: Optional[Union[int, List[int]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.mand_integer is None:
            raise ValueError("mand_integer must be supplied")
        if not isinstance(self.mand_integer, int):
            self.mand_integer = int(self.mand_integer)

        if self.mand_multi_integer is None:
            raise ValueError("mand_multi_integer must be supplied")
        elif not isinstance(self.mand_multi_integer, list):
            self.mand_multi_integer = [self.mand_multi_integer]
        elif len(self.mand_multi_integer) == 0:
            raise ValueError(f"mand_multi_integer must be a non-empty list")
        self.mand_multi_integer = [v if isinstance(v, int) else int(v) for v in self.mand_multi_integer]

        if self.opt_integer is not None and not isinstance(self.opt_integer, int):
            self.opt_integer = int(self.opt_integer)

        if self.opt_multi_integer is None:
            self.opt_multi_integer = []
        if not isinstance(self.opt_multi_integer, list):
            self.opt_multi_integer = [self.opt_multi_integer]
        self.opt_multi_integer = [v if isinstance(v, int) else int(v) for v in self.opt_multi_integer]

        super().__post_init__(**kwargs)

The python first emits a class declaration for the base class – in this case, the class Integer which derives from the builtin type int. Four class variables are included in the generation: * type_class_uri - the URIRef for the type itself * type_class_curie - the string CURIE representation of the type * type_name - the non-mangled name assigned to the type in the original definition * type_model_uri - the URIRef of the type definition in the default LinkML namespace

The python then emits a class definition for the Integers class, where we have defined four slot type permutations:

1. mand_integer - a single valued required type:

   • The type, int, is from the base attribute. The python generator does make use of strictly positional parameters, so all type declarations will have an appropriate default, which is None in this case:

     mand_integer: int = None

   • Check for the presence of the value itself: if self.mand_integer is None:       raise ValueError("mand_integer must be supplied")

   • Coerce the input value to int if it isn’t one already: if not isinstance(self.mand_integer, int):       self.mand_integer = int(self.mand_integer)

2. mand_multi_integer - a mandatory list of integers

   • Accept either a list of integers or a single integer, which will be converted to a list:

     mand_multi_integer: Union[int, List[int]] = None

   • Check that the value is present. Note: LinkML treats an empty list as the same as an absent value.

     if self.mand_multi_integer is None:
         raise ValueError("mand_multi_integer must be supplied")
     

   • Convert a non-list value into a list. This allows constructors to be lazy when it comes to single valued entries. Note, however, that this does require that list-of-list constructs to be explicitly declared because, while the code treats x = C(1) as the same as x = C([1]) in the multivalued case, it does not attempt to differentiate x = C([1]) and x = C([[1]]). This could be done if there is a compelling reason to, however:

     elif not isinstance(self.mand_multi_integer, list):
         self.mand_multi_integer = [self.mand_multi_integer]
     

   • Check that the mandatory list has at least one element (our definition of mandatory list: elif len(self.mand_multi_integer) == 0:      raise ValueError(f"mand_multi_integer must be a non-empty list")

   • Coerce all of the member of the list into the target type (int):

     self.mand_multi_integer = [v if isinstance(v, int) else int(v) for v in self.mand_multi_integer]

3. opt_integer - Optional non-list.

   • Accept either an int or None, the latter being entailed in the Optional type: opt_integer: Optional[int] = None

   • Coerce to an integer if the value is present, otherwise let it be: if self.opt_integer is not None and not isinstance(self.opt_integer, int):       self.opt_integer = int(self.opt_integer)

   • Type is Optional[<type>]

   • __post_init__ adds type coercion if necessary

4. opt_multi_integer: list:

   • Accept one of None, int, list of int, or anything that can be coerced into List[int]:

     opt_multi_integer: Optional[Union[int, List[int]]] = empty_list()

   • The absence of a list property is always represented as an empty list: if self.opt_multi_integer is None:       self.opt_multi_integer = []

   • Coerce any other non-list into a list: if not isinstance(self.opt_multi_integer, list):       self.opt_multi_integer = [self.opt_multi_integer]

   • Coerce all the members of the list into the range type self.opt_multi_integer = [v if isinstance(v, int) else int(v) for v in self.opt_multi_integer]

3) Python generation for a Type defined in metamodelcore.py – XSDDate in this example

YAML:

types:
  date:
    uri: xsd:date
    base: XSDDate
    repr: str
    description: a date (year, month and day) in an idealized calendar

classes:
  Dates:
    description: various permutations of the date type
    attributes:
      opt_date:
        range: date
      mand_date:
        range: date
        required: true
      opt_multi_date:
        range: date
        multivalued: true
      mand_multi_date:
        range: date
        multivalued: true
        required: true

python:

class Date(XSDDate):
    """ a date (year, month and day) in an idealized calendar """
    type_class_uri = XSD.date
    type_class_curie = "xsd:date"
    type_name = "date"
    type_model_uri = METATYPE.Date

@dataclass
class Dates(YAMLRoot):
    """
    various permutations of the date type
    """
    ...

    mand_date: Union[str, XSDDate] = None
    mand_multi_date: Union[Union[str, XSDDate], List[Union[str, XSDDate]]] = None
    opt_date: Optional[Union[str, XSDDate]] = None
    opt_multi_date: Optional[Union[Union[str, XSDDate], List[Union[str, XSDDate]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.mand_date is None:
            raise ValueError("mand_date must be supplied")
        if not isinstance(self.mand_date, XSDDate):
            self.mand_date = XSDDate(self.mand_date)

        if self.mand_multi_date is None:
            raise ValueError("mand_multi_date must be supplied")
        elif not isinstance(self.mand_multi_date, list):
            self.mand_multi_date = [self.mand_multi_date]
        elif len(self.mand_multi_date) == 0:
            raise ValueError(f"mand_multi_date must be a non-empty list")
        self.mand_multi_date = [v if isinstance(v, XSDDate) else XSDDate(v) for v in self.mand_multi_date]

        if self.opt_date is not None and not isinstance(self.opt_date, XSDDate):
            self.opt_date = XSDDate(self.opt_date)

        if self.opt_multi_date is None:
            self.opt_multi_date = []
        if not isinstance(self.opt_multi_date, list):
            self.opt_multi_date = [self.opt_multi_date]
        self.opt_multi_date = [v if isinstance(v, XSDDate) else XSDDate(v) for v in self.opt_multi_date]

        super().__post_init__(**kwargs)

The generated python for this type is identical to that of the builtin type with the exception that type declarations are the Union of the class representing the element and the repr element (e.g. XSDDate and str):

int type:

mand_multi_integer: Union[int, List[int]] = None

XSDDate type:

mand_multi_date: Union[Union[str, XSDDate], List[Union[str, XSDDate]]] = None

4) Python generation for an inherited Type

The following YAML shows three inherited types - one that inherits from a builtin type, a second from a metamodelcore type and a third from a the inherited type.

types:
  InheritedType:
    typeof: integer

  InheritedType2:
    typeof: uriorcurie

  InheritedType3:
    typeof: InheritedType2

Python:

# Types
class InheritedType(Integer):
    type_class_uri = XSD.integer
    type_class_curie = "xsd:integer"
    type_name = "InheritedType"
    type_model_uri = PTYPES.InheritedType


class InheritedType2(Uriorcurie):
    type_class_uri = XSD.anyURI
    type_class_curie = "xsd:anyURI"
    type_name = "InheritedType2"
    type_model_uri = PTYPES.InheritedType2


class InheritedType3(InheritedType2):
    type_class_uri = XSD.anyURI
    type_class_curie = "xsd:anyURI"
    type_name = "InheritedType3"
    type_model_uri = PTYPES.InheritedType3

     ...

@dataclass
class InheritedTypes(YAMLRoot):
    """
    various permutations of a typeof referencing a builtin
    """
    ...

    mand_InheritedType: Union[int, InheritedType] = None
    mand_multi_InheritedType: Union[Union[int, InheritedType], List[Union[int, InheritedType]]] = None
    opt_InheritedType: Optional[Union[int, InheritedType]] = None
    opt_multi_InheritedType: Optional[Union[Union[int, InheritedType], List[Union[int, InheritedType]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.mand_InheritedType is None:
            raise ValueError("mand_InheritedType must be supplied")
        if not isinstance(self.mand_InheritedType, InheritedType):
            self.mand_InheritedType = InheritedType(self.mand_InheritedType)

        if self.mand_multi_InheritedType is None:
            raise ValueError("mand_multi_InheritedType must be supplied")
        elif not isinstance(self.mand_multi_InheritedType, list):
            self.mand_multi_InheritedType = [self.mand_multi_InheritedType]
        elif len(self.mand_multi_InheritedType) == 0:
            raise ValueError(f"mand_multi_InheritedType must be a non-empty list")
        self.mand_multi_InheritedType = [v if isinstance(v, InheritedType) else InheritedType(v) for v in self.mand_multi_InheritedType]

        if self.opt_InheritedType is not None and not isinstance(self.opt_InheritedType, InheritedType):
            self.opt_InheritedType = InheritedType(self.opt_InheritedType)

        if self.opt_multi_InheritedType is None:
            self.opt_multi_InheritedType = []
        if not isinstance(self.opt_multi_InheritedType, list):
            self.opt_multi_InheritedType = [self.opt_multi_InheritedType]
        self.opt_multi_InheritedType = [v if isinstance(v, InheritedType) else InheritedType(v) for v in self.opt_multi_InheritedType]

        super().__post_init__(**kwargs)

The Python patterns for each of the inherited types (Inherited from basic (e.g. int), inherited from metamodelcore.py (e.g. uriorcurie) or from another type are identical to those generated above for XSDDate

5) Slot range is a LinkML class definition

The previous section described the Python that was generated for slot ranges that reference LinkML Type definitions. This section addresses the second use case, where the slot range references another LinkML Class. Classes are a more complex example, because, as opposed to types, classes can either be represented by value or, if they have been declared to have unique identifiers, by reference. In addition, classes that are represented by value can either be represented as lists of values or dictionaries.

The following attributes control how the python is generated for a LinkML class definition: * The slot abstract setting * The slot required setting * The slot multivalued setting * Whether a class includes a slot identified as a “key” ("key": true), an “identifier” ("identifier": true) or neither. Note that at the moment, a class may have at most one key or identifier slot. Two keys, two identifiers or a key and an identifier are all considered errors. It is anticipated that a future version of LinkML will allow multiple key slots, directly implementing the notion of a “compound key”. * The slot inlined and inlined_as_list properties * The slot ifabsent attribute * The slot default value

Various permutations of these cases are described below:

6) Non-keyed, non-identified classes

Classes without keys or identifiers cannot only be realized inline. The following code shows examples of various permutations of this type of class:

classes:
  OptionalOneElementRange:
    description: Range is a optional class that contains one non-key/non-identifier element
    attributes:
      v1:
        range: OneElementClass

  RequiredOneElementRange:
    description: Range is a required class that contains one non-key/non-identifier element
    is_a: OptionalOneElementRange
    attributes:
      v1:
        range: OneElementClass
        required: true

  OptionalOneElementRangeList:
    description: Range is a optional list of a class that contain one non-key/non-identifier element
    attributes:
      v1:
        range: OneElementClass
        multivalued: true

  RequiredThreeElementRangeList:
    description: Range is a required list of a class that contain one non-key/non-identifier element
    attributes:
      v1:
        range: ThreeElementClass
        multivalued: true
        required: true

An optional, single element class generates the following code:

@dataclass
    class OptionalOneElementRange(YAMLRoot):
        """
        Range is a optional class that contains one non-key/non-identifier element
        """
        ...

        v1: Optional[Union[dict, OneElementClass]] = None

        def __post_init__(self, **kwargs: Dict[str, Any]):
            if self.v1 is not None and not isinstance(self.v1, OneElementClass):
                self.v1 = OneElementClass(**self.v1)

            super().__post_init__(**kwargs)

v1 can be one of: * None - as in indicated by the Optional type * any python dict, which is the type that underlies all class representations * an instance of OneElementClass

The only additional initialization code coerces the input to a OneElementClass if it exists (is not None) and it isn’t already an instance.

Switching from optional to required produces:

@dataclass
class RequiredOneElementRange(YAMLRoot):
    """
    Range is a required class that contains one non-key/non-identifier element
    """
    ...

    v1: Union[dict, OneElementClass] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, OneElementClass):
            self.v1 = OneElementClass(**self.v1)

        super().__post_init__(**kwargs)

Which is the same as the previous example, with the exception that the value, v1 must be supplied.

The third example, an optional list of OneElementClass instances:

@dataclass
class OptionalOneElementRangeList(YAMLRoot):
    """
    Range is a optional list of a class that contain one non-key/non-identifier element
    """
    ...

    v1: Optional[Union[Union[dict, OneElementClass], List[Union[dict, OneElementClass]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, list):
            self.v1 = [self.v1]
        self.v1 = [v if isinstance(v, OneElementClass) else OneElementClass(**v) for v in self.v1]

        super().__post_init__(**kwargs)

Accepts any of: * None - the list is not present * An empty list ([]) - same semantics as None * A dictionary * An instance of OneElementClass * A list of dictionaries and/or OneElementClass instances

The code: 1) translates the None value into the equivalent empty list 2) converts a naked dict or OneElementClass instance into a single element list 3) coerces the list of zero or more items into a list of OneElementClass instances

In the final case, we’ve attempted to show both a) what a required inline list looks like and b) show that the target type is irrelevant as long as it is a LinkML Class and has no key or identifier slots

@dataclass
class RequiredThreeElementRangeList(YAMLRoot):
    """
    Range is a required list of a class that contain two non-key/non-identifier elements
    """
    ...

    v1: Union[Union[dict, ThreeElementClass], List[Union[dict, ThreeElementClass]]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        elif not isinstance(self.v1, list):
            self.v1 = [self.v1]
        elif len(self.v1) == 0:
            raise ValueError(f"v1 must be a non-empty list")
        self.v1 = [v if isinstance(v, ThreeElementClass) else ThreeElementClass(**v) for v in self.v1]

        super().__post_init__(**kwargs)

The structural differences between the above class and its predecessor are: 1) The v1 signature does not include None as an option (i.e. does not begin with Optional[]) 2) A None value is flagged as an error rather than converted to an empty list 3) An empty list is flagged as an error.

Keyed or Identified Classes

When the range of a slot is a LinkML class that has a key or an identifier, additional options present themselves: 1) Should elements be represented inline or as references? 2) If inline, should the elements be represented as lists or as dictionaries, whose key is the key or identifier value.

We discuss each of these options below

Key or identified class references

Using the following classes as building blocks

KeyedThreeElementClass:
  description: A keyed class with an additional integer and date
  attributes:
    name:
      range: string
      key: true
    value:
      range: integer
    modifier:
      range: date

IdentifiedThreeElementClass:
  description: A identified class with an additional integer and date
  attributes:
    name:
      range: string
      identifier: true
    value:
      range: integer
    modifier:
      range: date

Python:

class KeyedThreeElementClassName(extended_str):
    pass

class IdentifiedThreeElementClassName(extended_str):
    pass

@dataclass
class KeyedThreeElementClass(YAMLRoot):
    """
    A keyed class with an additional integer and date
    """
    ...

    name: Union[str, KeyedThreeElementClassName] = None
    value: Optional[int] = None
    modifier: Optional[Union[str, XSDDate]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.name is None:
            raise ValueError("name must be supplied")
        if not isinstance(self.name, KeyedThreeElementClassName):
            self.name = KeyedThreeElementClassName(self.name)

        if self.value is not None and not isinstance(self.value, int):
            self.value = int(self.value)

        if self.modifier is not None and not isinstance(self.modifier, XSDDate):
            self.modifier = XSDDate(self.modifier)

        super().__post_init__(**kwargs)

@dataclass
class IdentifiedThreeElementClass(YAMLRoot):
    """
    A identified class with an additional integer and date
    """
    ...

    name: Union[str, IdentifiedThreeElementClassName] = None
    value: Optional[int] = None
    modifier: Optional[Union[str, XSDDate]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.name is None:
            raise ValueError("name must be supplied")
        if not isinstance(self.name, IdentifiedThreeElementClassName):
            self.name = IdentifiedThreeElementClassName(self.name)

        if self.value is not None and not isinstance(self.value, int):
            self.value = int(self.value)

        if self.modifier is not None and not isinstance(self.modifier, XSDDate):
            self.modifier = XSDDate(self.modifier)

        super().__post_init__(**kwargs)

We can now begin with the following LinkML class definitions:

OptionalKeyedThreeElementRange:
  description: Range is a optional class that contains one key and two regular elements
  attributes:
    v1:
      range: KeyedThreeElementClass

OptionalKeyedThreeElementRangeList:
  description: Range is a optional list of a class that contains one key and two regular elements
  attributes:
    v1:
      range: KeyedThreeElementClass
      multivalued: true

RequiredIdentifiedThreeElementRange:
  description: Range is a required class that contains one identifier and two regular elements
  attributes:
    v1:
      range: IdentifiedThreeElementClass
      required: true

RequiredIdentifiedThreeElementRangeList:
  description: Range is a optional list of a class that contains one identifier and two regular elements
  attributes:
    v1:
      range: IdentifiedThreeElementClass
      multivalued: true
      required: true

Python equivalents

@dataclass
class OptionalKeyedThreeElementRange(YAMLRoot):
    """
    Range is a optional class that contains one key and two regular elements
    """
    ...

    v1: Optional[Union[str, KeyedThreeElementClassName]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is not None and not isinstance(self.v1, KeyedThreeElementClassName):
            self.v1 = KeyedThreeElementClassName(self.v1)

        super().__post_init__(**kwargs)

In the above example, v1 is a single, optional reference to a KeyedThreeElementClass instance. The constructor can take one of: 1) None - because the link is optional 2) str - the type of the key to the KeyedThreeElementClass 3) ThreeElementClassName - an instance of the actual key. 4) Any python object that can be coerced into a KeyedThreeElementClassName

The key is recorded in the class. Note that the Python code, as it exists today, does NOT check that there actually exists an instance of KeyedThreeElementClass with the name in v1 and, as such, there is currently no way to get directly from an instance of the class, OptionalKeyedThreeElementRange to an instance of KeyedThreeElementClass. This check and capability may be added to a later version of the LinkML tooling. The second example, below, shows an optional list

@dataclass
class OptionalKeyedThreeElementRangeList(YAMLRoot):
    """
    Range is a optional list of a class that contains one key and two regular elements
    """
    ...

    v1: Optional[Union[Union[str, KeyedThreeElementClassName], List[Union[str, KeyedThreeElementClassName]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, (list, dict)):
            self.v1 = [self.v1]
        self.v1 = [v if isinstance(v, KeyedThreeElementClassName) else KeyedThreeElementClassName(v) for v in self.v1]

        super().__post_init__(**kwargs)

The above signature is a bit confusing. It allows: 1) None - no list has been supplied 2) str - the type of the key to a single KeyedThreeElementClass 3) ThreeElementClassName - an instance of an actual key 4) Any python object that can be coerced into a KeyedThreeElementClass (not explicit in the signature) 4) A list consisting of any combination of str, ThreeElementClassNames, or other objects that can be coerced into KeyedThreeElementClassName

The remaining two non-line examples are similar enough to the above that they are included below without comment. Note, in particular, that there is no difference between keyed and identified class representation.

@dataclass
class RequiredIdentifiedThreeElementRange(YAMLRoot):
    """
    Range is a required class that contains one identifier and two regular elements
    """
    ...

    v1: Union[str, IdentifiedThreeElementClassName] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, IdentifiedThreeElementClassName):
            self.v1 = IdentifiedThreeElementClassName(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class RequiredIdentifiedThreeElementRangeList(YAMLRoot):
    """
    Range is a optional list of a class that contains one identifier and two regular elements
    """
    ...

    v1: Union[Union[str, IdentifiedThreeElementClassName], List[Union[str, IdentifiedThreeElementClassName]]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        elif not isinstance(self.v1, (list, dict)):
            self.v1 = [self.v1]
        if len(self.v1) == 0:
            raise ValueError(f"v1 must be a non-empty list dictionary or class")
        self.v1 = [v if isinstance(v, IdentifiedThreeElementClassName) else IdentifiedThreeElementClassName(v) for v in self.v1]

        super().__post_init__(**kwargs)

Python generation options for single-value (non-list) slots

The table below describes the various python generation options for single-valued slots ("multivalued": false)

ca se #	multiv alued	inlined	inlined_as_list	range class has key or identif ier	res ult
1. 1	false (defau lt)			N	slo t val ue in of con tai nin g cla ss (i. e. inl ine )
1. 2	false	N	N	Y	slo t key or ide nti fie r is in con tai nin g cla ss
1. 3	false	Y		Y	slo t val ue in of con tai nin g cla ss (i. e. inl ine )
1. 4	false		Y	Y	slo t val ue in of con tai nin g cla ss (i. e. inl ine )

The YAML examples below the above options for both optional and required slots.

OptionalThreeElementRange:
  description: Case 1.1(o) -- single values optional slot - range has no keys or identifiers
  attributes:
    v1:
      range: ThreeElementClass

RequiredThreeElementRange:
  description: Case 1.1(r) -- single values optional slot - range has no keys or identifiers
  attributes:
    v1:
      range: ThreeElementClass
      required: true

OptionalIdentifiedThreeElementRange:
  description: Case 1.2(o) -- single values optional slot - range has an identifier
  attributes:
    v1:
      range: IdentifiedThreeElementClass

RequiredKeyedThreeElementRange:
  description: Case 1.2(r) -- single values optional slot - range has a key
  attributes:
    v1:
      range: KeyedThreeElementClass
      required: true

OptionalInlinedKeyedThreeElementRange:
  description: Case 1.3(o) -- single values optional slot - range has an identifier
  attributes:
    v1:
      range: KeyedThreeElementClass
      inlined: true

RequiredInlinedIdentifiedThreeElementRange:
  description: Case 1.3(r) -- single values optional slot - range has a key
  attributes:
    v1:
      range: IdentifiedThreeElementClass
      required: true
      inlined: true

OptionalInlinedAsListKeyedThreeElementRange:
  description: Case 1.4(o) -- single values optional slot - range has an identifier
  attributes:
    v1:
      range: KeyedThreeElementClass
      inlined_as_list: true

RequiredInlinedAsListIdentifiedThreeElementRange:
  description: Case 1.4(r) -- single values optional slot - range has a key
  attributes:
    v1:
      range: IdentifiedThreeElementClass
      required: true
      inlined_as_list: true

The emitted python is as described for types with the exception of case 1.2 – non-inlined keyed or identified class, where the identifier or key of the class is the range of the v1 slot in Python:

@dataclass
class OptionalThreeElementRange(YAMLRoot):
    """
    Case 1.1(o) -- single values optional slot - range has no keys or identifiers
    """
    ...
    v1: Optional[Union[dict, ThreeElementClass]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is not None and not isinstance(self.v1, ThreeElementClass):
            self.v1 = ThreeElementClass(**self.v1)

        super().__post_init__(**kwargs)


@dataclass
class RequiredThreeElementRange(YAMLRoot):
    """
    Case 1.1(r) -- single values optional slot - range has no keys or identifiers
    """
    ...
    v1: Union[dict, ThreeElementClass] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, ThreeElementClass):
            self.v1 = ThreeElementClass(**self.v1)

        super().__post_init__(**kwargs)


@dataclass
class OptionalIdentifiedThreeElementRange(YAMLRoot):
    """
    Case 1.2(o) -- single values optional slot - range has an identifier
    """
    ...
    v1: Optional[Union[str, IdentifiedThreeElementClassName]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is not None and not isinstance(self.v1, IdentifiedThreeElementClassName):
            self.v1 = IdentifiedThreeElementClassName(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class RequiredKeyedThreeElementRange(YAMLRoot):
    """
    Case 1.2(r) -- single values optional slot - range has a key
    """
    ...
    v1: Union[str, KeyedThreeElementClassName] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, KeyedThreeElementClassName):
            self.v1 = KeyedThreeElementClassName(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class OptionalInlinedKeyedThreeElementRange(YAMLRoot):
    """
    Case 1.3(o) -- single values optional slot - range has an identifier
    """
    ...
    v1: Optional[Union[dict, KeyedThreeElementClass]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is not None and not isinstance(self.v1, KeyedThreeElementClass):
            self.v1 = KeyedThreeElementClass(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class RequiredInlinedIdentifiedThreeElementRange(YAMLRoot):
    """
    Case 1.3(r) -- single values optional slot - range has a key
    """
    ...
    v1: Union[dict, IdentifiedThreeElementClass] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, IdentifiedThreeElementClass):
            self.v1 = IdentifiedThreeElementClass(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class OptionalInlinedAsListKeyedThreeElementRange(YAMLRoot):
    """
    Case 1.4(o) -- single values optional slot - range has an identifier
    """
    ...
    v1: Optional[Union[dict, KeyedThreeElementClass]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is not None and not isinstance(self.v1, KeyedThreeElementClass):
            self.v1 = KeyedThreeElementClass(self.v1)

        super().__post_init__(**kwargs)


@dataclass
class RequiredInlinedAsListIdentifiedThreeElementRange(YAMLRoot):
    """
    Case 1.4(r) -- single values optional slot - range has a key
    """
    ...
    v1: Union[dict, IdentifiedThreeElementClass] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        if not isinstance(self.v1, IdentifiedThreeElementClass):
            self.v1 = IdentifiedThreeElementClass(self.v1)

        super().__post_init__(**kwargs)

Python generation options for multivalued slots

The python generated for multivalued slots is a more interesting situation.

ca se #	multiv alued	inlined	inlined_as_list	range class has key or identif ier	res ult
2. 1	true			N	Slo t is inl ine d as a lis t
2. 2	true	false (default)	false (default)	Y	Slo t is a lis t of ran ge key s or ide nti fie rs
2. 3	true		true		Slo t is inl ine d as a lis t
2. 4	true	true	false	Y	Slo t is inl ine d as a dic tio nar y, key ed by the slo t key or ide nti fie r

The first two cases (2.1) and (2.2) do not differ significantly from what was described earlier

OptionalMultivaluedThreeElementRange:
    description: Case 2.1(o) -- multivalued optional slot - range has no key or identifier
    attributes:
      v1:
        range: ThreeElementClass
        multivalued: true

  RequiredMultivaluedThreeElementRange:
    description: Case 2.1(r) -- multivalued optional slot - range has no key or identifier
    attributes:
      v1:
        range: ThreeElementClass
        multivalued: true
        required: true

  OptionalMultivaluedKeyedThreeElementRange:
    description: Case 2.2(o) -- multivalued optional slot - range has a key
    attributes:
      v1:
        range: KeyedThreeElementClass
        multivalued: true

  RequiredMultivaluedIdentifiedThreeElementRange:
    description: Case 2.2(r) -- multivalued optional slot - range has an identifier
    attributes:
      v1:
        range: IdentifiedThreeElementClass
        multivalued: true
        required: true

@dataclass
class OptionalMultivaluedThreeElementRange(YAMLRoot):
    """
    Case 2.1(o) -- multivalued optional slot - range has no key or identifier
    """
    ...
    v1: Optional[Union[Union[dict, ThreeElementClass], List[Union[dict, ThreeElementClass]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, list):
            self.v1 = [self.v1]
        self.v1 = [v if isinstance(v, ThreeElementClass) else ThreeElementClass(**v) for v in self.v1]

        super().__post_init__(**kwargs)


@dataclass
class RequiredMultivaluedThreeElementRange(YAMLRoot):
    """
    Case 2.1(r) -- multivalued optional slot - range has no key or identifier
    """
    ...
    v1: Union[Union[dict, ThreeElementClass], List[Union[dict, ThreeElementClass]]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        elif not isinstance(self.v1, list):
            self.v1 = [self.v1]
        elif len(self.v1) == 0:
            raise ValueError(f"v1 must be a non-empty list")
        self.v1 = [v if isinstance(v, ThreeElementClass) else ThreeElementClass(**v) for v in self.v1]

        super().__post_init__(**kwargs)

@dataclass
class OptionalMultivaluedKeyedThreeElementRange(YAMLRoot):
    """
    Case 2.2(o) -- multivalued optional slot - range has a key
    """
    ...
    v1: Optional[Union[Union[str, KeyedThreeElementClassName], List[Union[str, KeyedThreeElementClassName]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, list):
            self.v1 = [self.v1]
        self.v1 = [v if isinstance(v, KeyedThreeElementClassName) else KeyedThreeElementClassName(v) for v in self.v1]

        super().__post_init__(**kwargs)


@dataclass
class RequiredMultivaluedIdentifiedThreeElementRange(YAMLRoot):
    """
    Case 2.2(r) -- multivalued required slot - range has an identifier
    """
    ...
    v1: Union[Union[str, IdentifiedThreeElementClassName], List[Union[str, IdentifiedThreeElementClassName]]] = None

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        elif not isinstance(self.v1, list):
            self.v1 = [self.v1]
        elif len(self.v1) == 0:
            raise ValueError(f"v1 must be a non-empty list")
        self.v1 = [v if isinstance(v, IdentifiedThreeElementClassName) else IdentifiedThreeElementClassName(v) for v in self.v1]

        super().__post_init__(**kwargs)

Case 2.3 (Inlined as list) shows a new pattern:

OptionalMultivaluedInlinedListIdentifiedThreeElementRange:
  description: 2.3(o) Range is an optional identified three element class that is represented as an inlined list
  attributes:
    v1:
      range: IdentifiedThreeElementClass
      multivalued: true
      inlined_as_list: true


RequiredMultivaluedInlinedListKeyedThreeElementRangeList:
  description: 2.3(r) Range is a required keyed three element class that is represented as an inlined list
  attributes:
    v1:
      range: KeyedThreeElementClass
      multivalued: true
      inlined_as_list: true
      required: true

The Python that is generated from the above entries will require some more description:

@dataclass
class OptionalMultivaluedInlinedListIdentifiedThreeElementRange(YAMLRoot):
    """
    2.3(o) Range is an optional identified three element class that is represented as an inlined list
    """
    ...
    v1: Optional[Union[Dict[Union[str, IdentifiedThreeElementClassName], Union[dict, IdentifiedThreeElementClass]], List[Union[dict, IdentifiedThreeElementClass]]]] = empty_list()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, (list, dict)):
            self.v1 = [self.v1]
        self._normalize_inlined_slot(slot_name="v1", slot_type=IdentifiedThreeElementClass, key_name="name", inlined_as_list=True, keyed=True)

        super().__post_init__(**kwargs)

Starting with the signature:

v1: Optional[Union[Dict[Union[str, IdentifiedThreeElementClassName], Union[dict, IdentifiedThreeElementClass]], List[Union[dict, IdentifiedThreeElementClass]]]] = empty_list()

The outermost declaration: Optional[...] = empty_list() reflects: 1) v1 is not required: None is a valid value 2) the internal storage structure is going to be a list. Note that None, [] and {} (Missing scalar, empty list and empty dictionary all indicate absence of a value)

Moving inwards, Union[Dict[...], List[...]] means that the inputs can either be in the form of a dictionary or a list.

Input in the form of a dictionary

If the input is in the form of a dictionary, its signature is “Dict[Union[str, IdentifiedThreeElementClassName], Union[dict, IdentifiedThreeElementClass]]”, that is a dictionary whose keys are valid IdentifiedThreeElementClassNames and whose values are instances of IdentifiedThreeElementClass. The str part of the key signature and the dict component of the value indicate that, raw strings can be substituted for class names and plain dictionaries for class values.

Noting the definition for KeyedThreeElementClass is:

KeyedThreeElementClass:
  description: A keyed class with an additional integer and date
  attributes:
    name:
      range: string
      key: true
    value:
      range: integer
    modifier:
      range: date

One can represent instances of OptionalMultivaluedInlinedListIdentifiedThreeElementRange in YAML as:

element1:
  value: 17
  modifier: 2012-03-11
element2:
element3:
  name: element3
  value: 42

or in python as:

ktec_examples = OptionalMultivaluedInlinedListIdentifiedThreeElementRange(
    {
        element1: {value: 17, modifier: "2012-03-11"},
        KeyedThreeElementClassKey('element2'): {},
        element3: {name: 'element3', value: 42}
    }
)

Note that the following is not valid because, while the name attribute (the key) is not required, as it has already been established by the key, if present it must match the key itself

element3:
  name: test
  value: 42

One of the reasons for the explicit key types (e.g. KeyedThreeElementClassKey) is to provide some level of checking for potential link mismatches. The following code snippet:

itc = IdentifiedThreeElementClass('entry1', 17)
ktec_examples: OptionalMultivaluedInlinedListIdentifiedThreeElementRange = {
    itc.name: {value: 42}
}

Would result in an error, as itc.name is an instance of IdentifiedThreeElementClassKey while KeyedThreeElementClass expects a KeyedThreeElementClassKey

Input in the form of a list

Starting with the skeleton we visited earlier, Moving inwards, Union[Dict[...], List[...]] the second option, List[Union[dict, IdentifiedThreeElementClass]] indicates that the input to the constructor can also be a list of dictionaries that conform to the IdentifiedThreeElementClass model. Referencing the examples that we used above, we could also use:

- name: element1
  value: 17
  modifier: 2012-03-11
- name: element2
- name: element3
  value: 42

And the Python equivalent would be:

ktec_examples = OptionalMultivaluedInlinedListIdentifiedThreeElementRange(
    [ {name: 'element1', value: 17, modifier: "2012-03-11"},
      {name: KeyedThreeElementClassKey('element2')},
      {name: 'element3', value: 42}
    ]
)

We then move on to the generated code:

def __post_init__(self, **kwargs: Dict[str, Any]):
    if self.v1 is None:
        self.v1 = []
    if not isinstance(self.v1, (list, dict)):
        self.v1 = [self.v1]

The first two tests are pretty close to what we saw earlier, with the exception that you can’t pass an isolated dictionary. You can construct a list of integers with a single value:

my_entries = ListOfIntegers(17)

Which would, if the class ListOfIntegers’s first variable was a multivalued slot with a range of integers, result in my_entries.v == [17]. When dealing with classes, however, the following:

my_entries = ListOfClasses(dict(name='element1', value=17))

would fail, as it would try to convert this into a list with two IdentifiedThreeElementClass entries.

The final line in the __post_init__ section:

self._normalize_inlined_slot(slot_name="v1", slot_type=IdentifiedThreeElementClass, key_name="name", inlined_as_list=True, keyed=True)

passes the remainder of the post initialization normalization to a function defined in YAMLRoot.

The final case, Case 2.4 (Inlined as dictionary) is essentially the same, with the exception that the target type is a dictionary rather than a list.

OptionalMultivaluedInlinedIdentifiedThreeElementRangeList:
    description: 2.4(o) Range is an optional identified three element class that is represented as an inlined dictionary
    attributes:
      v1:
        range: IdentifiedThreeElementClass
        multivalued: true
        inlined: true


  RequiredMultivaluedInlinedKeyedThreeElementRange:
    description: 2.4(r) Range is a required keyed three element class that is represented as an inlined dictionary
    attributes:
      v1:
        range: KeyedThreeElementClass
        multivalued: true
        inlined: true
        required: true

@dataclass
class OptionalMultivaluedInlinedIdentifiedThreeElementRangeList(YAMLRoot):
    """
    2.4(o) Range is an optional identified three element class that is represented as an inlined dictionary
    """
    ...
    v1: Optional[Union[Dict[Union[str, IdentifiedThreeElementClassName], Union[dict, IdentifiedThreeElementClass]], List[Union[dict, IdentifiedThreeElementClass]]]] = empty_dict()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            self.v1 = []
        if not isinstance(self.v1, (list, dict)):
            self.v1 = [self.v1]
        self._normalize_inlined_slot(slot_name="v1", slot_type=IdentifiedThreeElementClass, key_name="name", inlined_as_list=None, keyed=True)

        super().__post_init__(**kwargs)


@dataclass
class RequiredMultivaluedInlinedKeyedThreeElementRange(YAMLRoot):
    """
    2.4(r) Range is a required keyed three element class that is represented as an inlined dictionary
    """
    ...
    v1: Union[Dict[Union[str, KeyedThreeElementClassName], Union[dict, KeyedThreeElementClass]], List[Union[dict, KeyedThreeElementClass]]] = empty_dict()

    def __post_init__(self, **kwargs: Dict[str, Any]):
        if self.v1 is None:
            raise ValueError("v1 must be supplied")
        elif not isinstance(self.v1, (list, dict)):
            self.v1 = [self.v1]
        if len(self.v1) == 0:
            raise ValueError(f"v1 must be a non-empty list, dictionary, or class")
        self._normalize_inlined_slot(slot_name="v1", slot_type=KeyedThreeElementClass, key_name="name", inlined_as_list=None, keyed=True)

        super().__post_init__(**kwargs)

The generated code is identical to the previous case (2.3) with the exception that the default initializer is empty_dict() instead of empty_list(), and the _normalized_inlined_slot function is passed a false inlined_as_list value (well, ‘None’ but…).

As a final note on the inlined_as_list vs. just plain inlined, while the internal storage structure is different, the initialization code is exactly the same – YAML emitted from a list structure can be loaded as a dictionary and visa-versa.

Slots

One bit that we haven’t touched on here is the slots section of the emitted Python.

At the moment, the following code is emitted:

# Slots
class slots:
    pass

    ...

slots.RequiredInlinedKeyedTwoElementRange_v1 = Slot(uri=LISTS_AND_KEYS.v1, name="RequiredInlinedKeyedTwoElementRange_v1", curie=LISTS_AND_KEYS.curie('v1'),
                   model_uri=LISTS_AND_KEYS.RequiredInlinedKeyedTwoElementRange_v1, domain=RequiredInlinedKeyedTwoElementRange, range=Union[dict, KeyedTwoElementClass])

slots.RequiredInlinedKeyedTwoElementRangeList_v1 = Slot(uri=LISTS_AND_KEYS.v1, name="RequiredInlinedKeyedTwoElementRangeList_v1", curie=LISTS_AND_KEYS.curie('v1'),
                   model_uri=LISTS_AND_KEYS.RequiredInlinedKeyedTwoElementRangeList_v1, domain=RequiredInlinedKeyedTwoElementRangeList, range=Union[Dict[Union[str, KeyedTwoElementClassName], Union[dict, KeyedTwoElementClass]], List[Union[dict, KeyedTwoElementClass]]])

slots.RequiredInlinedKeyedThreeElementRange_v1 = Slot(uri=LISTS_AND_KEYS.v1, name="RequiredInlinedKeyedThreeElementRange_v1", curie=LISTS_AND_KEYS.curie('v1'),
                   model_uri=LISTS_AND_KEYS.RequiredInlinedKeyedThreeElementRange_v1, domain=RequiredInlinedKeyedThreeElementRange, range=Union[dict, KeyedThreeElementClass])

slots.RequiredInlinedKeyedThreeElementRangeList_v1 = Slot(uri=LISTS_AND_KEYS.v1, name="RequiredInlinedKeyedThreeElementRangeList_v1", curie=LISTS_AND_KEYS.curie('v1'),
                   model_uri=LISTS_AND_KEYS.RequiredInlinedKeyedThreeElementRangeList_v1, domain=RequiredInlinedKeyedThreeElementRangeList, range=Union[Dict[Union[str, KeyedThreeElementClassName], Union[dict, KeyedThreeElementClass]], List[Union[dict, KeyedThreeElementClass]]])

The slots class is here to keep the module namespace from getting too confusing, the idea that, if you want to reference a slot, you use slots.RequiredInlinedTwoElementRange_v1 (which is the mangling of the owner class name and the actual slot)

At the moment, you have the following elements:

• uri - the SLOT URI

• name - the mangled name

• curie - the SLOT CURIE

• model_uri - the slot URI where the base is the Model URI vs. one assigned via slot_uri

• domain - the slot domain

• range - the range signature

Note, however, that the slots area is still being enhanced, so keep an eye on entries like Issue #272