I've changed almost all of this - this PR now adds tests for compresslevel (these were lacking before) and documents preset (this was undocumented before).

This PR fixes tarfile.open() when using the w:xz mode and a compresslevel parameter.

TarFile.xzopen() seems to want to do what TarFile.gzopen() and TarFile.bzopen() do (see here), but it uses the keyword preset instead of compresslevel.

This means the preset never gets passed to lzma.LZMAFile(), the compresslevel keyword is thus picked up by **kwargs and passed to TarFile.__init__(), which doesn't recognize it.

>>> import tarfile
>>> good = tarfile.open('/tmp/dummy.tar.gz', 'w:gz', compresslevel=1)
>>> bad = tarfile.open('/tmp/dummy.tar.xz', 'w:xz', compresslevel=1)
Traceback (most recent call last):
  File "<pyshell#3>", line 1, in <module>
    bad = tarfile.open('/tmp/dummy.tar.xz', 'w:xz', compresslevel=1)
  File "/home/bo/Code/cpython/Lib/tarfile.py", line 1588, in open
    return func(name, filemode, fileobj, **kwargs)
  File "/home/bo/Code/cpython/Lib/tarfile.py", line 1699, in xzopen
    t = cls.taropen(name, mode, fileobj, **kwargs)
  File "/home/bo/Code/cpython/Lib/tarfile.py", line 1618, in taropen
    return cls(name, mode, fileobj, **kwargs)
TypeError: __init__() got an unexpected keyword argument 'compresslevel'

In this PR I change the signature of TarFile.xzopen() to match its neighbors. I also change TarFile.taropen() and document that compresslevel is meaningless when not using one of the w:<some compression> modes. (compresslevel is already ignored when using a read mode; you can do tarfile.open('/tmp/dummy.tar.gz', 'r:gz', compresslevel='kittens') and not get an exception).

I also add tests for compresslevel with the various modes; these were lacking previously.

https://bugs.python.org/issue32695