DOC: Fix to lzma.LZMAFile

[ShaharNaveh]

• closes #xxxx
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[jbrockmendel]

possibly type(LZMAFile)? usually this would indicate that we're getting an instance of that class

[WillAyd]

Should annotate the function return here as well

[ShaharNaveh]

Should annotate the function return here as well

I thought of that. but the problem is that the function tests for the imports of the lzma module. If I import it beforehand it kinda losing it's point (I think)

[ShaharNaveh]

Should annotate the function return here as well

@WillAyd I tried it now, I imported the lzma module at the beginning of the file and annotated _get_lzma_file with -> lzma.LZMAFile, and mypy complained:

pandas/io/common.py:468: error: "LZMAFile" not callable

Any thoughts?

[WillAyd]

Should annotate the function return here as well

[WillAyd]

It would need to be Type[lzma.LZMAFile], though not sure if we can do that given this is technically an optional dependency from the stdlib

[WillAyd]

Does Type[IO] work? May be the best bet

[topper-123]

LZMA has been part of the standard library since python 3.3. We should just assume it’s installed, that will simplify this a bit.

[jbrockmendel]

LZMA has been part of the standard library since python 3.3. We should just remove assume it’s installed, that will simplify this a bit.

I'm pretty sure that just recently i had to do an apt-get install to get this working on an ubuntu 18 machine

[WillAyd]

We had an issue previously with that assumption, hence the existence of this compat to begin with. See #27575

[topper-123]

Ok, that’s surprising. Yeah, changing it isn’t possible then, I guess...

[ShaharNaveh]

It would need to be Type[lzma.LZMAFile], though not sure if we can do that given this is technically an optional dependency from the stdlib

mypy's log:

Success: no issues found in 808 source files

Does Type[IO] work? May be the best bet

mypy's log:

pandas/io/common.py:468: error: Too many arguments for "IO"
Found 1 error in 1 file (checked 808 source files)

[ShaharNaveh]

It would need to be Type[lzma.LZMAFile], though not sure if we can do that given this is technically an optional dependency from the stdlib

mypy's log:

Success: no issues found in 808 source files

Does Type[IO] work? May be the best bet

mypy's log:

pandas/io/common.py:468: error: Too many arguments for "IO"
Found 1 error in 1 file (checked 808 source files)

I just found a bug (I think), unless I'm missing something obvious.

When _get_lzma_file is annotated with -> Type[IO], normally mypy will say:

pandas/io/common.py:468: error: Too many arguments for "IO"
Found 1 error in 1 file (checked 808 source files)

But if the lzma module is imported at the beginning of the file you mypy will say:

Success: no issues found in 808 source files

---

So two questions actually:

• Is this a bug?
• What option do we prefer? (if any)

[WillAyd]

Thanks @MomIsBestFriend

Can come back to types later for the whole module if something you are interested in