Stringify response before json.loads

[AroundTheLines]

This fixes the "the json object must be str, not 'bytes'" error that happens on line 145.

Sample error before fix:

Traceback (most recent call last):
  File "(...)/twitter.py", line 53, in <module>
    async_data = LineItem.async_stats_job_data(account, async_stats_job_result['url'])
  File "(...)/venv/lib/python3.5/site-packages/twitter_ads/resource.py", line 261, in async_stats_job_data
    raw_body=True, stream=True).perform()
  File "(...)/venv/lib/python3.5/site-packages/twitter_ads/http.py", line 73, in perform
    response = self.__oauth_request()
  File "(...)/venv/lib/python3.5/site-packages/twitter_ads/http.py", line 103, in __oauth_request
    body=response.raw, raw_body=raw_response_body)
  File "(...)/venv/lib/python3.5/site-packages/twitter_ads/http.py", line 145, in __init__
    self._body = json.loads(raw_response_body)
  File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/json/__init__.py", line 312, in loads
    s.__class__.__name__))
TypeError: the JSON object must be str, not 'bytes'

[frewsxcv]

This probably doesn't need to be in the try, unless str(..) raises a ValueError, which I don't think it does? Not that big of a deal though.

[AroundTheLines]

That was how I found the code, so I didn't change that. I did just change location of the stringify just now though.

[frewsxcv]

Under what scenario would raw_response_body not be a str?

[frewsxcv]

Just thinking out loud, if it's a str only when zlib.decompress gets called (a few lines above this), we should maybe move the decode() up in that conditional.

[AroundTheLines]

I'm not entirely sure if that's the case, I'll test it out and get back to you on that.

[itsamaik]

Please merge this fix into the latest version, currently I'm also getting this error when retrieving job results. After zlib.decompress it's returning b'{ .......... }' thus resulting into an error when it's put into json.loads().

Edit: This seems to be an issue with Python 3+, see https://bugs.python.org/issue13989. TextIOWrapper could solve this.

[jbabichjapan]

@AroundTheLines Hey could you check the latest feedback and let us know if you are ready to go with this change? I don't have time to help drive this one in right now but if you need more eyeballs please call out and hopefully some people will notice and help review

[juanshishido]

[Edited for clarity: changes only local]

As @frewsxcv suggested, I tested added adding .decode() to line 139:

raw_response_body = zlib.decompress(self._raw_body, 16 + zlib.MAX_WBITS).decode()

Async stats working locally for both Python 2.7.12 and Python 3.5.2.

Any thoughts on this, @AroundTheLines?

cc: @jbabichjapan, @tushdante

[juanshishido]

@AroundTheLines: We'd like to merge this fix. Please let us know if you're able to make the aforementioned changes. If not, we can do it, too. Thanks for bringing this up and proposing a solution!

[AroundTheLines]

@juanshishido: Hey, sorry I haven't been keeping up with this PR. I'd be glad to make the changes in a moment.

[AroundTheLines]

Like @juanshishido said, decompressing the response at the source solves python 3 compatibility issues, where bytes are returned instead of strings.

This should be ready for merging!
cc: @jbabichjapan, @tushdante

[juanshishido]

LGTM

[juanshishido]

Awesome! Thanks, @AroundTheLines! 👍

[madman-bob]

I'm having this issue sporadically in version 1.2.2.

Eyeballing the source, I'm guessing we also need a .decode() in http.py:97. ie. Changing
raw_response_body = response.raw.read() if stream else response.text
to
raw_response_body = response.raw.read().decode() if stream else response.text.

Can anyone more knowledgable than I confirm/deny?