Removed encoding parameter from bytes call

[jimbobbennett]

As dicsussed in #20, the encoding parameter passed to the byes call gives a TypeError: wrong number of arguments

[FoamyGuy]

Do you know of public API that accepts bytearray data via POST like this that you can share? If so I can test this out with a pyportal.

[jimbobbennett]

I've been testing with the Azure Custom Vision service at customvision.ai. I created a model and posted to the prediction service. Source for how to do it is here: https://github.com/jimbobbennett/mandmcounter

This code has my fixed version of adafruit_requests in requests.py. You'd need to create an object detection model in Custom Vision and set the values in the secrets.py file for your WiFi and keys for the model.

I'm on a plane at the moment, but tomorrow I can provide detailed instructions on how to do it if that would help.

[FoamyGuy]

Thank you! I think I should be able to get it tested using what you've posted. I will give it a try either tonight or tomorrow.

[tannewt]

I'm ok removing this but would love to see an issue for adding support for the encoding parameter to CircuitPython. In general, we should ensure CircuitPython matches CPython.

[jimbobbennett]

@tannewt - if you look at #20, @FoamyGuy has compared the behavior to CPython and CircuitPython, and the behavior is roughly the same.

If you call bytes() against a string, you can pass the encoding parameter and it works correctly on both platforms.
If you call bytes() against a bytearray, you get an error passing the encoding parameter, albeit a different, more descriptive error on CPython

The difference here is the error given by CircuitPython is different, and less instructive than the CPython one. Where is the right place to raise an issue to get the error message improved? Is it in https://github.com/adafruit/circuitpython?

[FoamyGuy]

@tannewt do you mean adding encoding parameter to the request() method in this library? bytes() seems to be supporting the encoding parameter already, it just doesn't work with a bytearray as the data, that fails in CPython as well, but with a slightly different error.

[FoamyGuy]

Looks good to me, and works as expected. I ended up testing with a basic Flask server locally, my msoft account was not allowing me to use Azure without signing up for some more stuff. I was able to verify with these changes to the library we can now use this Circuit Python code:

# ... imports ... 
# ... connect to esp and AP etc ...
requests.set_socket(socket, esp)

f = open("small_image.bmp", "rb")
file_data = f.read()
f.close()

copy_data = bytearray(file_data)

response = requests.post("http://192.168.1.117:5000/test", data=copy_data)
print(response)

to POST some image data to a server and we now get the the same (successful) results as this CPython code:

import requests

f = open("small_image.bmp", "rb")
file_data = f.read()
f.close()

copy_data = bytearray(file_data)

r = requests.post("http://localhost:5000/test", data=copy_data)

Whereas prior to these changes it was the error noted in the issue:

  File "/lib/adafruit_requests.py", line 285, in post
  File "/lib/adafruit_requests.py", line 242, in request
  File "/lib/adafruit_requests.py", line 224, in request
TypeError: wrong number of arguments

[tannewt]

@tannewt - if you look at #20, @FoamyGuy has compared the behavior to CPython and CircuitPython, and the behavior is roughly the same.

If you call bytes() against a string, you can pass the encoding parameter and it works correctly on both platforms.
If you call bytes() against a bytearray, you get an error passing the encoding parameter, albeit a different, more descriptive error on CPython

The difference here is the error given by CircuitPython is different, and less instructive than the CPython one. Where is the right place to raise an issue to get the error message improved? Is it in https://github.com/adafruit/circuitpython?

Ah! Ok. I misunderstood that. Sounds like it is close enough. It would be nice to ensure it is the same exception thrown but it's much less important.

This change looks good to me.

[FoamyGuy]

Thank you @jimbobbennett!