Fixed chunk_size values in iter_content() and iter_lines(). False default instead of None.

[mponton]

Following discussion about issue #539 Kenneth agreed that a default
chunk_size of 1 for iter_lines() would make more sense, but
apparently changed chunk_size for iter_content() by mistake
instead, which did solve the iter_lines() issue but also changed the
default iter_content() chunk to 1 bytes which is not desirable.

Also changed default for decode_unicode to False instead of None in iter_lines().

[travisbot]

This pull request fails (merged 85e83c4 into c84012e).

[mponton]

Hmmm, I think the logic in

if pending is not None:
    chunk = pending + chunk
lines = chunk.splitlines()

if lines and lines[-1] and chunk and lines[-1][-1] == chunk[-1]:
    pending = lines.pop()
else:
    pending = None

Does not handle "\r\n" as splitlines() does when using a chunk_size of 1. The quote:

quote = (
            '''Agamemnon  \n'''
            '''\tWhy will he not upon our fair request\r\n'''
            '''\tUntent his person and share the air with us?'''
        )

used in test_requests.py contains \r\n which ends up becoming a line in iter_lines() when the iterator hits the \r character, next the \n character is read and an empty line is spit by chunk.splitlines(). Thus the test fails as 4 lines are returned instead of 3. If the chunk_size is larger and \r\n are read in a single chunk, then it works as expected.

[kennethreitz]

decode_unicode is None for a reason :)

[kennethreitz]

iter_content also has a proper size now, so closing. Thanks though!

[mponton]

OK, not sure why decode_unicode should be None but you may then want to change it to None in iter_content() too...? :-)