Add ZstandardFile class (and "open" method) that supports 'r' and 'w' modes, just like in Python's stdlib

[d33tah]

Consider the following script that runs fine on my PC (with no output):

for lname in ['LZMA', 'BZ2', 'Gzip']:
    lib = __import__(lname.lower())
    cls = getattr(lib, lname + 'File')
    #  cls = getattr(lib, 'open')  # uncomment to see that this also works
    example_payload = b'a'
    with cls('a.bin', 'w') as f:
        f.write(example_payload)
    with cls('a.bin', 'r') as f:
        if f.read() != example_payload:
            print('WTF!')

All three compression libraries in the Python standard library follow a pattern: there's an SomethingFile object that supports reading and writing just as if it was a regular file. Perhaps it makes sense to add such a class to python-zstandard as well?

[d33tah]

Here's my lazy attempt at addressing this:

import zstandard
import contextlib

old_open = open

@contextlib.contextmanager
def open(*, fileobj=None, file=None, mode='r'):
    if mode not in ('r', 'w'):
        raise ValueError('Mode should be "r" or "w")')
    if not fileobj and not file:
        raise ValueError('either fileobj or file must be provided')
    if mode == "r":
        cctx = zstandard.ZstdDecompressor()
        if file:
            with old_open(file, 'rb') as fh, cctx.stream_reader(fh) as reader:
                yield reader
        else:
            yield cctx.stream_reader(fileobj)
    elif mode == "w":
        cctx = zstandard.ZstdCompressor()
        if file:
            with old_open(file, 'wb') as fh, cctx.stream_writer(fh) as writer:
                yield writer
        else:
            yield cctx.stream_writer(fileobj)

[indygreg]

I'm closing this issue because it is a duplicate of #23 and #64.

I do agree that having these APIs would be a good idea. I may add these to the next release. We'll see.