Incorrect TS type is generated when involving Tuple

[yy9y]

To reproduce:

#types.py
from typing import List, Tuple
from pydantic import BaseModel


class Model(BaseModel):
    tuple: Tuple[str, int]
    list_of_tuples: List[Tuple[str, int]]

pydantic2ts --module ./types.py --output ./types.ts

yields:

//types.ts
export interface Model {
  tuple: [] | [string] | [string, number];
  list_of_tuples: [] | [string] | [string, number][];
}

Expected

The above Pydantic model should generate this:

export interface Model {
  tuple:  [string, number];
  list_of_tuples: [string, number][];
}

Versions:

python: 3.7.9
pydantic: 1.6 | 1.8.2
pydantic-to-typescript: 1.0.7

[phillipdupuis]

Hi, thanks for finding this. It's a bit of a funky one

It seems that the issue is resolved if the json schema has minItems: 2. If you run your test with this model, it passes

from typing import List, Tuple
from pydantic import BaseModel


class Model(BaseModel):
    tuple: Tuple[str, int]
    list_of_tuples: List[Tuple[str, int]]

    class Config:
        @staticmethod
        def schema_extra(schema: dict) -> None:
            """Customize the generated JSON schema via mutation"""
            schema["properties"]["tuple"]["minItems"] = 2
            schema["properties"]["list_of_tuples"]["items"]["minItems"] = 2

However, it appears that they already talked about adding this functionality into pydantic but decided against it:
pydantic/pydantic#718

We can certainly implement it ourselves by adding some behavior to the clean_schema function (https://github.com/phillipdupuis/pydantic-to-typescript/blob/master/pydantic2ts/cli/script.py#L114) , just need to take care not to break anything else in doing so