Initialize pydantic Literal from array of strings

Question

I would like to inizialize a pydantic Literal from an array of strings

from typing import Literal
from pydantic import BaseModel

CLASS_NAME_VALUES = ["primary", "secondary", "success", "danger", "warning", "info", "dark"]
ICON_NAME_VALUES = ["electricity.svg", "water.svg",
                    "internet.svg", "naturalGas.svg", "noCategory.svg"]

class CategoryRequest(BaseModel):
    class_name: Literal[CLASS_NAME_VALUES]
    icon_name: Literal[ICON_NAME_VALUES]

But I am getting the following error

  File "pydantic/main.py", line 252, in pydantic.main.ModelMetaclass.__new__
  File "pydantic/fields.py", line 309, in pydantic.fields.ModelField.infer
  File "pydantic/fields.py", line 271, in pydantic.fields.ModelField.__init__
  File "pydantic/fields.py", line 351, in pydantic.fields.ModelField.prepare
  File "pydantic/fields.py", line 529, in pydantic.fields.ModelField.populate_validators
  File "pydantic/validators.py", line 566, in find_validators
  File "pydantic/validators.py", line 410, in pydantic.validators.make_literal_validator
TypeError: unhashable type: 'list'

Any suggestion?

This question is similar to: Dynamically create Literal alias from list of valid values. If you believe it’s different, please edit the question, make it clear how it’s different and/or how the answers on that question are not helpful for your problem. — STerliakov
– STerliakov, Commented Dec 28, 2024 at 23:33

alex_noname · Accepted Answer · 2021-01-12 08:54:30Z

At runtime, an arbitrary value is allowed as type argument to Literal[...], but type checkers may impose restrictions. For example, mypy permits only one or more literal bool, int, str, bytes, enum values, None and aliases to other Literal types. For example, Literal[3 + 4] or List[(3, 4)] are disallowed.

As for pydantic, it permits uses values of hashable types in Literal, like tuple. For example, your sample could be rewritten using tuple as:

from typing import Literal
from pydantic import BaseModel

CLASS_NAME_VALUES = ("primary", "secondary", "success", "danger", "warning", "info", "dark")
ICON_NAME_VALUES = ("electricity.svg", "water.svg",
                    "internet.svg", "naturalGas.svg", "noCategory.svg")


class CategoryRequest(BaseModel):
    class_name: Literal[CLASS_NAME_VALUES]
    icon_name: Literal[ICON_NAME_VALUES]

But I would strongly advise against using this syntax, but sticking to the canonical one, which is also allowed by the type checkers:

CLASS_NAME_VALUES = Literal["primary", "secondary", "success", "danger", "warning", "info", "dark"]
ICON_NAME_VALUES = Literal["electricity.svg", "water.svg",
                    "internet.svg", "naturalGas.svg", "noCategory.svg"]


class CategoryRequest(BaseModel):
    class_name: CLASS_NAME_VALUES
    icon_name: ICON_NAME_VALUES

So if I need my ICON_NAME_VALUES as an iterable later in my code, I have to define two constants? Once the Literal["electricity.svg", "water.svg", ...] for type checking and once a list/tuple ["electricity.svg", "water.svg", ...] to use in the code? How cumbersome! Apparently, I cannot go vice versa and get a list from the Literal either?
@stefanbschneider it's possible other way around, and that's the canonical solution: stackoverflow.com/questions/64522040/…

Valerij · Accepted Answer · 2025-01-07 17:55:37Z

2

Starting with python 3.11 (see PEP-0646) you can just unpack tuple/array with a star operator:

class CategoryRequest(BaseModel):
    class_name: Literal[*CLASS_NAME_VALUES]
    icon_name: Literal[*ICON_NAME_VALUES]

Additional links:

How to unpack a list of colons and NumPy Nones to index an array? — similar question.
http://www.andy-pearce.com/blog/posts/2023/Jan/whats-new-in-python-311-type-hint-improvements/#variadic-generics — Variadic Generics explained.

edited Jan 7 at 17:55

answered Dec 28, 2024 at 17:25

Valerij

991 silver badge3 bronze badges

2 Comments

Yaakov Bressler Dec 30, 2024 at 0:16

Can you provide a citation for this?

user2357112 Jan 7 at 17:59

From a type-checker perspective, this is invalid, and you'll get a type-checking error. From a runtime perspective, this just does the same thing Literal[tuple(CLASS_NAME_VALUES)] already did in earlier versions.

Collectives™ on Stack Overflow

Initialize pydantic Literal from array of strings

2 Answers 2

2 Comments

2 Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

2 Comments

Linked

Related