read_sql_* methods should accept dtype argument with same type as DataFrame.astype()

[Dr-Irv]

For the read_sql_* methods, the argument dtype is just passed to DataFrame.astype(), where we allow dtype: AstypeArg | Mapping[Any, Dtype] | Series, so the read_sql_*() methods should do the same.

In particular, doing something like dtype: {"col": "str"} is totally allowed, but not permitted in the stubs.

pandas-stubs version 2.0.1.230501

[Dr-Irv]

@ramvikrams can you fix this? You added the argument before, so maybe you can do this?

[ramvikrams]

Thank you @Dr-Irv I will surely do it

[Dr-Irv]

The solution here is actually simpler than I thought. Try changing

pandas-stubs/pandas-stubs/_typing.pyi

Line 79 in 6c0eed6

DtypeArg: TypeAlias = Dtype | dict[Any, Dtype]

to use Mapping instead of dict, and then just add a test that has a string in the value of the dict.

[ramvikrams]

The solution here is actually simpler than I thought. Try changing

pandas-stubs/pandas-stubs/_typing.pyi

Line 79 in 6c0eed6

DtypeArg: TypeAlias = Dtype | dict[Any, Dtype]

to use Mapping instead of dict, and then just add a test that has a string in the value of the dict.

i'll do it via this way then