add support for streaming uploads #8
Conversation
|
Hi @fabian-paul, Sorry! I only just noticed this PR (end-of-year was pretty crazy). Supporting streaming uploads sounds like a great idea. David |
| elif isinstance(self.prepared.body, (io.RawIOBase, io.BufferedIOBase)): | ||
| self.curl.setopt(pycurl.READFUNCTION, self.prepared.body.read) | ||
| self.curl.setopt(pycurl.TRANSFER_ENCODING, 1) | ||
| elif hasattr(self.prepared.body, "__iter__"): # TODO: call iter instead of checking (e.g. to support delegates) |
There was a problem hiding this comment.
It might be better to use isinstance(obj, Collection).
collections.abc.Collection guarantees that a type implements __contains__, __iter__ and __len__. You can also use collections.abc.Iterable to specifically test for __iter__.
| else: | ||
| self.curl.setopt(pycurl.TRANSFER_ENCODING, 0) | ||
| self.curl.setopt(pycurl.INFILESIZE_LARGE, n_bytes) | ||
| reader = ChunkIterableReader(iter(self.prepared.body)) |
There was a problem hiding this comment.
I wonder if we can use the two argument form of iter here:
iter(self.prepared.body, "") # for string
iter(self.prepared.body, b"") # for bytes| def close(self): # TODO | ||
| try: | ||
| self._iterator.close() | ||
| except AttributeError: | ||
| pass |
There was a problem hiding this comment.
I don't think you should be closing the iterator from within the library. Typically I would expect a user to be using a with block or calling close manually.
|
|
||
| #: Is this _really_ PyCurl-Requests? | ||
| #: Should be used when testing for PyCurl-Requests extensions. | ||
| IS_PYCURL_REQUESTS = requests.__name__ == 'pycurl_requests' | ||
|
|
||
|
|
||
| test_data = bytes(random.getrandbits(8) for _ in range(123456)) |
There was a problem hiding this comment.
I'd recommend using os.urandom here.
| if not allow_chunked: | ||
| self.response('This endpoint has chunked transfer deactivated.', status=(400, "Bad Request")) | ||
| return | ||
| body = b"" |
There was a problem hiding this comment.
Joining immutable bytes objects can be potentially expensive. bytearray is a better choice if you need something mutable.
Hi all,
I happened to use streaming uploads with requests and curl a lot (using both individually but not in combination). So I felt like adding support for the data access protocols that requests supports to this cool project would a useful contribution.
The PR isn't ready yet and I hope to finish it soon. There are still a few sections that I have marked with TODO comments. I'm also not completely sure how the Content-Length header and libcurl's
INFILESIZE_LARGEinteract. Would be nice to have your opinion on this PR.Fabian