se.py

"""
This module provides `serialize`, `is_serializable` `to_dict`, `to_tuple` and classes
and functions associated with serialization.
"""

from __future__ import annotations
import abc
import copy
import dataclasses
import functools
import typing
import itertools
import jinja2
from dataclasses import dataclass, is_dataclass
from typing import TypeVar, Literal, Generic, Optional, Any, Union
from collections.abc import Callable, Iterable, Iterator
from beartype import beartype, BeartypeConf
from beartype.door import is_bearable
from typing_extensions import dataclass_transform

from .compat import (
    SerdeError,
    SerdeSkip,
    T,
    get_origin,
    is_any,
    is_bare_dict,
    is_bare_list,
    is_bare_opt,
    is_bare_set,
    is_bare_tuple,
    is_class_var,
    is_datetime,
    is_datetime_instance,
    is_dict,
    is_enum,
    is_generic,
    is_list,
    is_literal,
    is_none,
    is_opt,
    is_primitive,
    is_set,
    is_str_serializable,
    is_str_serializable_instance,
    is_tuple,
    is_union,
    is_variable_tuple,
    iter_types,
    iter_unions,
    type_args,
    typename,
)
from .core import (
    ClassSerializer,
    CACHE,
    SERDE_SCOPE,
    TO_DICT,
    TO_ITER,
    UNION_SE_PREFIX,
    DefaultTagging,
    Field,
    Scope,
    Tagging,
    TypeCheck,
    add_func,
    coerce_object,
    disabled,
    strict,
    fields,
    is_instance,
    logger,
    raise_unsupported_type,
    union_func_name,
    GLOBAL_CLASS_SERIALIZER,
)
from .numpy import (
    is_numpy_array,
    is_numpy_jaxtyping,
    is_numpy_datetime,
    is_numpy_scalar,
    serialize_numpy_array,
    serialize_numpy_datetime,
    serialize_numpy_scalar,
)

__all__ = ["serialize", "is_serializable", "to_dict", "to_tuple"]


SerializeFunc = Callable[[type[Any], Any], Any]
""" Interface of Custom serialize function. """


def default_serializer(_cls: type[Any], obj: Any) -> Any:
    """
    Marker function to tell serde to use the default serializer. It's used when custom serializer
    is specified at the class but you want to override a field with the default serializer.
    """


def serde_legacy_custom_class_serializer(
    cls: type[Any], obj: Any, custom: SerializeFunc, default: Callable[[], Any]
) -> Any:
    try:
        return custom(cls, obj)
    except SerdeSkip:
        return default()


class Serializer(Generic[T], metaclass=abc.ABCMeta):
    """
    `Serializer` base class. Subclass this to customize serialize behaviour.

    See `serde.json.JsonSerializer` and `serde.msgpack.MsgPackSerializer` for example usage.
    """

    @classmethod
    @abc.abstractmethod
    def serialize(cls, obj: Any, **opts: Any) -> T:
        raise NotImplementedError


def _make_serialize(
    cls_name: str,
    fields: Iterable[Union[str, tuple[str, type[Any]], tuple[str, type[Any], Any]]],
    *args: Any,
    rename_all: Optional[str] = None,
    reuse_instances_default: bool = False,
    convert_sets_default: bool = False,
    serializer: Optional[SerializeFunc] = None,
    tagging: Tagging = DefaultTagging,
    type_check: TypeCheck = disabled,
    serialize_class_var: bool = False,
    class_serializer: Optional[ClassSerializer] = None,
    **kwargs: Any,
) -> type[Any]:
    """
    Create a serializable class programatically.
    """
    C = dataclasses.make_dataclass(cls_name, fields, *args, **kwargs)
    C = serialize(
        C,
        rename_all=rename_all,
        reuse_instances_default=reuse_instances_default,
        convert_sets_default=convert_sets_default,
        serializer=serializer,
        tagging=tagging,
        type_check=type_check,
        serialize_class_var=serialize_class_var,
        **kwargs,
    )
    return C


# The `serialize` function can call itself recursively when it needs to generate code for
# unmarked dataclasses. To avoid infinite recursion, this array remembers types for which code is
# currently being generated.
GENERATION_STACK = []


@dataclass_transform()
def serialize(
    _cls: Optional[type[T]] = None,
    rename_all: Optional[str] = None,
    reuse_instances_default: bool = False,
    convert_sets_default: bool = False,
    serializer: Optional[SerializeFunc] = None,
    tagging: Tagging = DefaultTagging,
    type_check: TypeCheck = strict,
    serialize_class_var: bool = False,
    class_serializer: Optional[ClassSerializer] = None,
    **kwargs: Any,
) -> type[T]:
    """
    A dataclass with this decorator is serializable into any of the data formats
    supported by pyserde.

    >>> from datetime import datetime
    >>> from serde import serialize
    >>> from serde.json import to_json
    >>>
    >>> @serialize
    ... class Foo:
    ...     i: int
    ...     s: str
    ...     f: float
    ...     b: bool
    >>>
    >>> to_json(Foo(i=10, s='foo', f=100.0, b=True))
    '{"i":10,"s":"foo","f":100.0,"b":true}'
    """

    def wrap(cls: type[T]) -> type[T]:
        tagging.check()

        # If no `dataclass` found in the class, dataclassify it automatically.
        if not is_dataclass(cls):
            dataclass(cls)

        if type_check.is_strict():
            serde_beartype = beartype(conf=BeartypeConf(violation_type=SerdeError))
            serde_beartype(cls)

        g: dict[str, Any] = {}

        # Create a scope storage used by serde.
        # Each class should get own scope. Child classes can not share scope with parent class.
        # That's why we need the "scope.cls is not cls" check.
        scope: Optional[Scope] = getattr(cls, SERDE_SCOPE, None)
        if scope is None or scope.cls is not cls:
            scope = Scope(
                cls,
                reuse_instances_default=reuse_instances_default,
                convert_sets_default=convert_sets_default,
            )
            setattr(cls, SERDE_SCOPE, scope)

        class_serializers: list[ClassSerializer] = list(
            itertools.chain(GLOBAL_CLASS_SERIALIZER, [class_serializer] if class_serializer else [])
        )

        # Set some globals for all generated functions
        g["cls"] = cls
        g["copy"] = copy
        g["serde_scope"] = scope
        g["SerdeError"] = SerdeError
        g["raise_unsupported_type"] = raise_unsupported_type
        g["enum_value"] = enum_value
        g["is_dataclass"] = is_dataclass
        g["typename"] = typename  # used in union functions
        g["is_instance"] = is_instance  # used in union functions
        g["to_obj"] = to_obj
        g["typing"] = typing
        g["Literal"] = Literal
        g["TypeCheck"] = TypeCheck
        g["disabled"] = disabled
        g["coerce_object"] = coerce_object
        g["class_serializers"] = class_serializers
        if serializer:
            g["serde_legacy_custom_class_serializer"] = functools.partial(
                serde_legacy_custom_class_serializer, custom=serializer
            )

        # Collect types used in the generated code.
        for typ in iter_types(cls):
            # When we encounter a dataclass not marked with serialize, then also generate serialize
            # functions for it.
            if is_dataclass_without_se(typ) and typ is not cls:
                # We call serialize and not wrap to make sure that we will use the default serde
                # configuration for generating the serialization function.
                serialize(typ)

            if is_primitive(typ) and not is_enum(typ):
                continue
            g[typename(typ)] = typ

        # render all union functions
        for union in iter_unions(cls):
            union_args = list(type_args(union))
            union_key = union_func_name(UNION_SE_PREFIX, union_args)
            add_func(scope, union_key, render_union_func(cls, union_args, tagging), g)
            scope.union_se_args[union_key] = union_args

        for f in sefields(cls, serialize_class_var):
            if f.skip_if:
                g[f.skip_if.name] = f.skip_if
            if f.serializer:
                g[f.serializer.name] = f.serializer

        add_func(
            scope,
            TO_ITER,
            render_to_tuple(cls, serializer, type_check, serialize_class_var, class_serializer),
            g,
        )
        add_func(
            scope,
            TO_DICT,
            render_to_dict(
                cls, rename_all, serializer, type_check, serialize_class_var, class_serializer
            ),
            g,
        )

        logger.debug(f"{typename(cls)}: {SERDE_SCOPE} {scope}")

        return cls

    if _cls is None:
        return wrap  # type: ignore

    if _cls in GENERATION_STACK:
        return _cls

    GENERATION_STACK.append(_cls)
    try:
        return wrap(_cls)
    finally:
        GENERATION_STACK.pop()


def is_serializable(instance_or_class: Any) -> bool:
    """
    Test if an instance or class is serializable.

    >>> @serialize
    ... class Foo:
    ...     pass

    Testing `Foo` class object returns `True`.
    >>> is_serializable(Foo)
    True

    Testing `Foo` object also returns `True`.
    >>> is_serializable(Foo())
    True
    """
    return hasattr(instance_or_class, SERDE_SCOPE)


def is_dataclass_without_se(cls: type[Any]) -> bool:
    if not dataclasses.is_dataclass(cls):
        return False
    if not hasattr(cls, SERDE_SCOPE):
        return True
    scope: Optional[Scope] = getattr(cls, SERDE_SCOPE)
    if not scope:
        return True
    return TO_DICT not in scope.funcs


def to_obj(
    o: Any,
    named: bool,
    reuse_instances: Optional[bool] = None,
    convert_sets: Optional[bool] = None,
    c: Optional[Any] = None,
) -> Any:
    def serializable_to_obj(object: Any) -> Any:
        serde_scope: Scope = getattr(object, SERDE_SCOPE)
        func_name = TO_DICT if named else TO_ITER
        return serde_scope.funcs[func_name](
            object, reuse_instances=reuse_instances, convert_sets=convert_sets
        )

    try:
        thisfunc = functools.partial(
            to_obj,
            named=named,
            reuse_instances=reuse_instances,
            convert_sets=convert_sets,
        )

        # If a class in the argument is a non-dataclass class e.g. Union[Foo, Bar],
        # pyserde generates a wrapper (de)serializable dataclass on the fly,
        # and use it to serialize the object.
        if c and is_union(c) and not is_opt(c):
            return CACHE.serialize_union(c, o)

        if o is None:
            return None
        if is_dataclass_without_se(o):
            # Do not automatically implement beartype if dataclass without serde decorator
            # is passed, because it is surprising for users
            # See https://github.com/yukinarit/pyserde/issues/480
            serialize(type(o), type_check=disabled)
            return serializable_to_obj(o)
        elif is_serializable(o):
            return serializable_to_obj(o)
        elif is_bearable(o, list):
            return [thisfunc(e) for e in o]
        elif is_bearable(o, tuple):
            return tuple(thisfunc(e) for e in o)
        elif is_bearable(o, set):
            return [thisfunc(e) for e in o]
        elif is_bearable(o, dict):
            return {k: thisfunc(v) for k, v in o.items()}
        elif is_str_serializable_instance(o) or is_datetime_instance(o):
            return CACHE.serialize(
                c or o.__class__, o, reuse_instances=reuse_instances, convert_sets=convert_sets
            )

        return o

    except Exception as e:
        raise SerdeError(e) from None


def astuple(v: Any) -> tuple[Any, ...]:
    """
    Serialize object into tuple.
    """
    return to_tuple(v, reuse_instances=False, convert_sets=False)


def to_tuple(
    o: Any,
    c: Optional[type[Any]] = None,
    reuse_instances: Optional[bool] = None,
    convert_sets: Optional[bool] = None,
) -> tuple[Any, ...]:
    """
    Serialize object into tuple.

    >>> @serialize
    ... class Foo:
    ...     i: int
    ...     s: str = 'foo'
    ...     f: float = 100.0
    ...     b: bool = True
    >>>
    >>> to_tuple(Foo(i=10))
    (10, 'foo', 100.0, True)

    You can pass any type supported by pyserde. For example,

    >>> lst = [Foo(i=10), Foo(i=20)]
    >>> to_tuple(lst)
    [(10, 'foo', 100.0, True), (20, 'foo', 100.0, True)]
    """
    return to_obj(  # type: ignore
        o, named=False, c=c, reuse_instances=reuse_instances, convert_sets=convert_sets
    )


def asdict(v: Any) -> dict[Any, Any]:
    """
    Serialize object into dictionary.
    """
    return to_dict(v, reuse_instances=False, convert_sets=False)


def to_dict(
    o: Any,
    c: Optional[type[Any]] = None,
    reuse_instances: Optional[bool] = None,
    convert_sets: Optional[bool] = None,
) -> dict[Any, Any]:
    """
    Serialize object into python dictionary. This function ensures that the dataclass's fields are
    accurately represented as key-value pairs in the resulting dictionary.

    * `o`: Any pyserde object that you want to convert to `dict`
    * `c`: Optional class argument
    * `reuse_instances`: pyserde will pass instances (e.g. Path, datetime) directly to serializer
    instead of converting them to serializable representation e.g. string. This behaviour allows
    to delegate serializtation to underlying data format packages e.g. `pyyaml` and potentially
    improve performance.
    * `convert_sets`: This option controls how sets are handled during serialization and
    deserialization. When `convert_sets` is set to True, pyserde will convert sets to lists during
    serialization and back to sets during deserialization. This is useful for data formats that
    do not natively support sets.

    >>> from serde import serde
    >>> @serde
    ... class Foo:
    ...     i: int
    ...     s: str = 'foo'
    ...     f: float = 100.0
    ...     b: bool = True
    >>>
    >>> to_dict(Foo(i=10))
    {'i': 10, 's': 'foo', 'f': 100.0, 'b': True}

    You can serialize not only pyserde objects but also objects of any supported types. For example,
    the following example serializes list of pyserde objects into dict.

    >>> lst = [Foo(i=10), Foo(i=20)]
    >>> to_dict(lst)
    [{'i': 10, 's': 'foo', 'f': 100.0, 'b': True}, {'i': 20, 's': 'foo', 'f': 100.0, 'b': True}]
    """
    return to_obj(  # type: ignore
        o, named=True, c=c, reuse_instances=reuse_instances, convert_sets=convert_sets
    )


@dataclass
class SeField(Field[T]):
    """
    Field class for serialization.
    """

    @property
    def varname(self) -> str:
        """
        Get variable name in the generated code e.g. obj.a.b
        """
        var = getattr(self.parent, "varname", None) if self.parent else None
        if var:
            return f"{var}.{self.name}"
        else:
            if self.name is None:
                raise SerdeError("Field name is None.")
            return self.name

    def __getitem__(self, n: int) -> SeField[Any]:
        typ = type_args(self.type)[n]
        opts: dict[str, Any] = {
            "kw_only": self.kw_only,
            "case": self.case,
            "alias": self.alias,
            "rename": self.rename,
            "skip": self.skip,
            "skip_if": self.skip_if,
            "skip_if_false": self.skip_if_false,
            "skip_if_default": self.skip_if_default,
            "serializer": self.serializer,
            "deserializer": self.deserializer,
            "flatten": self.flatten,
        }
        return SeField(typ, name=None, **opts)


def sefields(cls: type[Any], serialize_class_var: bool = False) -> Iterator[SeField[Any]]:
    """
    Iterate fields for serialization.
    """
    for f in fields(SeField, cls, serialize_class_var=serialize_class_var):
        f.parent = SeField(None, "obj")  # type: ignore
        yield f


jinja2_env = jinja2.Environment(
    loader=jinja2.DictLoader(
        {
            "dict": """
def {{func}}(obj, reuse_instances = None, convert_sets = None):
  if reuse_instances is None:
    reuse_instances = {{serde_scope.reuse_instances_default}}
  if convert_sets is None:
    convert_sets = {{serde_scope.convert_sets_default}}
  if not is_dataclass(obj):
    return copy.deepcopy(obj)

  res = {}
  {% for f in fields -%}
  {% if not f.skip -%}
    {% if f.skip_if -%}
  subres = {{rvalue(f)}}
  if not {{f.skip_if.name}}(subres):
    {{lvalue(f)}} = subres
    {% else -%}
  {{lvalue(f)}} = {{rvalue(f)}}
    {% endif -%}
  {% endif %}

  {% endfor -%}
  return res
""",
            "iter": """
def {{func}}(obj, reuse_instances=None, convert_sets=None):
  if reuse_instances is None:
    reuse_instances = {{serde_scope.reuse_instances_default}}
  if convert_sets is None:
    convert_sets = {{serde_scope.convert_sets_default}}
  if not is_dataclass(obj):
    return copy.deepcopy(obj)

  return (
  {% for f in fields -%}
  {% if not f.skip|default(False) %}
  {{rvalue(f)}},
  {% endif -%}
  {% endfor -%}
  )
""",
            "union": """
def {{func}}(obj, reuse_instances, convert_sets):
  union_args = serde_scope.union_se_args['{{func}}']

  {% for t in union_args %}
  if is_instance(obj, union_args[{{loop.index0}}]):
    {% if tagging.is_external() and is_taggable(t) %}
    return {"{{typename(t)}}": {{rvalue(arg(t))}}}

    {% elif tagging.is_internal() and is_taggable(t) %}
    res = {{rvalue(arg(t))}}
    res["{{tagging.tag}}"] = "{{typename(t)}}"
    return res

    {% elif tagging.is_adjacent() and is_taggable(t) %}
    res = {"{{tagging.content}}": {{rvalue(arg(t))}}}
    res["{{tagging.tag}}"] = "{{typename(t)}}"
    return res

    {% else %}
    return {{rvalue(arg(t))}}
    {% endif %}
  {% endfor %}
  raise SerdeError("Can not serialize " + \
                   repr(obj) + \
                   " of type " + \
                   typename(type(obj)) + \
                   " for {{union_name}}")
""",
        }
    )
)


def render_to_tuple(
    cls: type[Any],
    legacy_class_serializer: Optional[SerializeFunc] = None,
    type_check: TypeCheck = strict,
    serialize_class_var: bool = False,
    class_serializer: Optional[ClassSerializer] = None,
) -> str:
    renderer = Renderer(
        TO_ITER,
        legacy_class_serializer,
        suppress_coerce=(not type_check.is_coerce()),
        serialize_class_var=serialize_class_var,
        class_serializer=class_serializer,
        class_name=typename(cls),
    )
    return jinja2_env.get_template("iter").render(
        func=TO_ITER,
        serde_scope=getattr(cls, SERDE_SCOPE),
        fields=sefields(cls, serialize_class_var),
        type_check=type_check,
        rvalue=renderer.render,
    )


def render_to_dict(
    cls: type[Any],
    case: Optional[str] = None,
    legacy_class_serializer: Optional[SerializeFunc] = None,
    type_check: TypeCheck = strict,
    serialize_class_var: bool = False,
    class_serializer: Optional[ClassSerializer] = None,
) -> str:
    renderer = Renderer(
        TO_DICT,
        legacy_class_serializer,
        suppress_coerce=(not type_check.is_coerce()),
        class_serializer=class_serializer,
        class_name=typename(cls),
    )
    lrenderer = LRenderer(case, serialize_class_var)
    return jinja2_env.get_template("dict").render(
        func=TO_DICT,
        serde_scope=getattr(cls, SERDE_SCOPE),
        fields=sefields(cls, serialize_class_var),
        type_check=type_check,
        lvalue=lrenderer.render,
        rvalue=renderer.render,
    )


def render_union_func(
    cls: type[Any], union_args: list[type[Any]], tagging: Tagging = DefaultTagging
) -> str:
    """
    Render function that serializes a field with union type.
    """
    union_name = f"Union[{', '.join([typename(a) for a in union_args])}]"
    renderer = Renderer(TO_DICT, suppress_coerce=True, class_name=typename(cls))
    return jinja2_env.get_template("union").render(
        func=union_func_name(UNION_SE_PREFIX, union_args),
        serde_scope=getattr(cls, SERDE_SCOPE),
        union_args=union_args,
        union_name=union_name,
        tagging=tagging,
        is_taggable=Tagging.is_taggable,
        arg=lambda x: SeField(x, "obj"),
        rvalue=renderer.render,
        typename=typename,
    )


@dataclass
class LRenderer:
    """
    Render lvalue for various types.
    """

    case: Optional[str]
    serialize_class_var: bool = False

    def render(self, arg: SeField[Any]) -> str:
        """
        Render lvalue
        """
        if is_dataclass(arg.type) and arg.flatten:
            return self.flatten(arg)
        elif is_opt(arg.type) and arg.flatten:
            inner = arg[0]
            if is_dataclass(inner.type):
                return self.flatten(inner)

        return f'res["{arg.conv_name(self.case)}"]'

    def flatten(self, arg: SeField[Any]) -> str:
        """
        Render field with flatten attribute.
        """
        flattened = []
        for f in sefields(arg.type, self.serialize_class_var):
            flattened.append(self.render(f))
        return ", ".join(flattened)


@dataclass
class Renderer:
    """
    Render rvalue for code generation.
    """

    func: str
    legacy_class_serializer: Optional[SerializeFunc] = None
    suppress_coerce: bool = False
    """ Suppress type coercing because generated union serializer has its own type checking """
    serialize_class_var: bool = False
    class_serializer: Optional[ClassSerializer] = None
    class_name: Optional[str] = None

    def render(self, arg: SeField[Any]) -> str:
        """
        Render rvalue
        """
        implemented_methods: dict[type[Any], int] = {}
        class_serializers: Iterable[ClassSerializer] = itertools.chain(
            GLOBAL_CLASS_SERIALIZER, [self.class_serializer] if self.class_serializer else []
        )
        for n, class_serializer in enumerate(class_serializers):
            for sig in class_serializer.__class__.serialize.methods:  # type: ignore
                implemented_methods[sig.types[1]] = n

        custom_serializer_available = arg.type in implemented_methods
        if custom_serializer_available and not arg.serializer:
            res = f"class_serializers[{implemented_methods[arg.type]}].serialize({arg.varname})"
        elif arg.serializer and arg.serializer.inner is not default_serializer:
            res = self.custom_field_serializer(arg)
        elif is_dataclass(arg.type):
            res = self.dataclass(arg)
        elif is_opt(arg.type):
            res = self.opt(arg)
        elif is_list(arg.type):
            res = self.list(arg)
        elif is_set(arg.type):
            res = self.set(arg)
        elif is_dict(arg.type):
            res = self.dict(arg)
        elif is_tuple(arg.type):
            res = self.tuple(arg)
        elif is_enum(arg.type):
            res = self.enum(arg)
        elif is_numpy_datetime(arg.type):
            res = serialize_numpy_datetime(arg)
        elif is_numpy_scalar(arg.type):
            res = serialize_numpy_scalar(arg)
        elif is_numpy_array(arg.type):
            res = serialize_numpy_array(arg)
        elif is_numpy_jaxtyping(arg.type):
            res = serialize_numpy_array(arg)
        elif is_primitive(arg.type):
            res = self.primitive(arg)
        elif is_union(arg.type):
            res = self.union_func(arg)
        elif is_str_serializable(arg.type):
            res = f"{arg.varname} if reuse_instances else {self.string(arg)}"
        elif is_datetime(arg.type):
            res = f"{arg.varname} if reuse_instances else {arg.varname}.isoformat()"
        elif is_none(arg.type):
            res = "None"
        elif is_any(arg.type) or is_bearable(arg.type, TypeVar):
            res = f"to_obj({arg.varname}, True, False, False, c=typing.Any)"
        elif is_generic(arg.type):
            origin = get_origin(arg.type)
            assert origin
            arg.type = origin
            res = self.render(arg)
        elif is_literal(arg.type):
            res = self.literal(arg)
        elif is_class_var(arg.type):
            arg.type = type_args(arg.type)[0]
            res = self.render(arg)
        else:
            res = f"raise_unsupported_type({arg.varname})"

        # Custom field serializer overrides custom class serializer.
        if self.legacy_class_serializer and not arg.serializer and not custom_serializer_available:
            return (
                "serde_legacy_custom_class_serializer("
                f"{typename(arg.type)}, "
                f"{arg.varname}, "
                f"default=lambda: {res})"
            )
        else:
            return res

    def custom_field_serializer(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for the field with custom serializer.
        """
        assert arg.serializer
        return f"{arg.serializer.name}({arg.varname})"

    def dataclass(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for dataclass.
        """
        if arg.flatten:
            flattened = []
            for f in sefields(arg.type, self.serialize_class_var):
                f.parent = arg
                flattened.append(self.render(f))
            return ", ".join(flattened)
        else:
            return (
                f"{arg.varname}.{SERDE_SCOPE}.funcs['{self.func}']({arg.varname},"
                " reuse_instances=reuse_instances, convert_sets=convert_sets)"
            )

    def opt(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for optional.
        """
        if is_bare_opt(arg.type):
            return f"{arg.varname} if {arg.varname} is not None else None"

        inner = arg[0]
        inner.name = arg.varname
        if arg.flatten:
            return self.render(inner)
        else:
            return f"({self.render(inner)}) if {arg.varname} is not None else None"

    def list(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for list.
        """
        if is_bare_list(arg.type):
            return arg.varname
        else:
            earg = arg[0]
            earg.name = "v"
            return f"[{self.render(earg)} for v in {arg.varname}]"

    def set(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for set.
        """
        if is_bare_set(arg.type):
            return f"list({arg.varname}) if convert_sets else {arg.varname}"
        else:
            earg = arg[0]
            earg.name = "v"
            return (
                f"[{self.render(earg)} for v in {arg.varname}] "
                f"if convert_sets else set({self.render(earg)} for v in {arg.varname})"
            )

    def tuple(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for tuple.
        """
        if is_bare_tuple(arg.type):
            return arg.varname
        elif is_variable_tuple(arg.type):
            earg = arg[0]
            earg.name = "v"
            return f"tuple({self.render(earg)} for v in {arg.varname})"
        else:
            rvalues = []
            for i, _ in enumerate(type_args(arg.type)):
                r = arg[i]
                r.name = f"{arg.varname}[{i}]"
                rvalues.append(self.render(r))
            return f"({', '.join(rvalues)},)"  # trailing , is required for single element tuples

    def dict(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for dict.
        """
        if is_bare_dict(arg.type):
            return arg.varname
        else:
            karg = arg[0]
            karg.name = "k"
            varg = arg[1]
            varg.name = "v"
            return f"{{{self.render(karg)}: {self.render(varg)} for k, v in {arg.varname}.items()}}"

    def enum(self, arg: SeField[Any]) -> str:
        return f"enum_value({typename(arg.type)}, {arg.varname})"

    def primitive(self, arg: SeField[Any]) -> str:
        """
        Render rvalue for primitives.
        """
        typ = typename(arg.type)
        var = arg.varname
        if self.suppress_coerce:
            return var
        else:
            return f'coerce_object("{self.class_name}", "{arg.name}", {typ}, {var})'

    def string(self, arg: SeField[Any]) -> str:
        return f"str({arg.varname})"

    def union_func(self, arg: SeField[Any]) -> str:
        func_name = union_func_name(UNION_SE_PREFIX, list(type_args(arg.type)))
        return f"serde_scope.funcs['{func_name}']({arg.varname}, reuse_instances, convert_sets)"

    def literal(self, arg: SeField[Any]) -> str:
        return f"{arg.varname}"


def enum_value(cls: Any, e: Any) -> Any:
    """
    Helper function to get value from enum or enum compatible value.
    """
    if is_enum(e):
        v = e.value
        # Recursively get value of Nested enum.
        if is_enum(v):
            return enum_value(v.__class__, v)
        else:
            return v
    else:
        return cls(e).value