Test Driven Development

[kb1lqc]

This project will use the style of test driven development (TDD). This website explains it well.

• Define tests of functions prior to writing functions
• All Pull Requests for new code additions must contain tests

[kb1lqc]

Ohh splinter helps automate website automated testing. Nice! Not terribly useful here unless we provide a statistics webpage with flask or something. Adding a note here.

https://splinter.readthedocs.io/en/latest/

[kb1lqc]

I noticed a few interesting things that must be done when developing in a virtual environment and using both python 2.7 and 3.5 on the same computer. These are:

• Install pip3 using this stackoverflow comment: https://stackoverflow.com/a/12262143
• Create a new Python 3 virtual environment using this stackoverflow comment: https://stackoverflow.com/a/23842752

@kb1lqd @hdkmike

[kb1lqc]

Also, when installing python-pytun on Ubuntu you will need to install the python3-dev package per: https://stackoverflow.com/a/21530768

[kb1lqc]

Alright. This might be easier to approach from the opposite direction and create a unittest.mock serial port which provides some test return_value. The issue I'm having is the same with the UDP end of defining these tests.. what data is being sent? I guess random UDP packets? Do I just generate them with dpkt or something? Any suggestions @reillyeon @hdkmike? I'll be around on Gitter Sunday if you want to try to catch me.

[kb1lqc]

⚠️ Need to add user to dailup group for good/easy linux serial port use

[kb1lqc]

Well then. Looks like pyserial can be used for loopback tests with serial compliant objects pointing to a local loopback port... I'll be damned.... nifty.

https://pythonhosted.org/pyserial/pyserial_api.html#serial.serial_for_url

@kb1lqd @hdkmike @reillyeon

[kb1lqc]

Oh wow, right on the front page of the documentation you can call a function with serial and see the available serial ports. This is really useful @kb1lqd !

https://pythonhosted.org/pyserial/shortintro.html#listing-ports

https://pythonhosted.org/pyserial/url_handlers.html

[kb1lqc]

Progress! Ended up deleting a bunch of my code but I like this approach much better. I am now starting to use pyserial's url based serial connection and using the loopback device to make a test serial port which I can then treat as any other serial port! Literally using python serial library to interact.

@kb1lqd @el-iso @reillyeon @hdkmike

https://github.com/kb1lqc/faradayio/blob/initial/tests/serialtestclass.py

class SerialTestClass(object):
    """A mock serial port test class"""
    def __init__(self):
        """Creates a mock serial port which is a loopback object"""
        self._port = "loop://"
        self._timeout = 0
        self._baudrate = 115200
        self.serialPort = \
            serial.serial_for_url(  url=self._port,
                                    timeout=self._timeout,
baudrate=self._baudrate)

A SUPER simple test to make sure this works is shown below

https://github.com/kb1lqc/faradayio/blob/initial/tests/test_serial.py

This simply writes "Hello World!" to the serial port and then reads it's back using serial for both read/write. I then check with assert that the string received was in-fact Hello World!. Super simple but it's a good foundation. I can actually start making progress defining the tests. I do need to figure out how this fits in with asyncio and whether that means I can use pyserial-asyncio too.

[kb1lqc]

Synchronous serial port send and receive are working with unit testing! At least as much as I can figure that it's unit testing. Not currently testing all possible inputs or exceptions but it's a start. I'd like to move to asynchronous now that this is working.

• https://github.com/kb1lqc/faradayio/blob/initial/tests/test_serial.py
• https://github.com/kb1lqc/faradayio/blob/initial/faradayio/faraday.py#L25

It turns out sliplib is not terribly well documented and the sliplib.Driver.receive() function actually sends messages to sliplib.Driver.messages instead of returning messages. This tripped me up a bit but I figured it out. Therefore I decided to make the faraday.receive() function a generator which yields each message. Let me know if this is a bad idea!

[kb1lqc]

@kb1lqd @reillyeon @hdkmike I sort of have a test working for asyncio based serial ports. I pretty much copied the pyserial-asyncio unit test and modified it for my own use.

https://github.com/kb1lqc/faradayio/blob/initial/tests/test_serial.py#L143

This uses the Input and Output classes in the faradayio module.

The problem I am having is that I can't seem to return a value from the test. I can only run the coroutine with run_until_complete(client). This test "Passes" but that's because it doesn't error. I am not asserting anything.

The pyserial-asyncio unit test checks that the receive list equals the bytes that were sent to it over the serial port (socket in this case). I can't do that since when the coroutine of aysncio completes, the list is destroyed. I am not sure how to go about and actually unit test a asyncio.Protocol.

@pyserial any help would be appreciated! @bufferoverflow @zsquareplusc

I really want to use asyncio as it appears to be everything I need but maybe I need to consider just doing something with synchronous pyserial... like we did before. https://stackoverflow.com/questions/17553543/pyserial-non-blocking-read-loop

[kb1lqc]

I removed the asyncio code for now and started building better unit tests for pyserial and sliplib implementations within faradayio. I need to test exceptions and stuff but this is a good step to making progress with serial and SLIP support with unit tests to back it up.

https://github.com/kb1lqc/faradayio/tree/initial

@kb1lqd @reillyeon @el-iso @hdkmike

[kb1lqc]

The documentation in CONTRIBUTIONS.md in our Athena Project needs to be updated to reflect a central location for this desired approach. Closing this ticket as I created FaradayRF/Athena#2 to reflect this need on that repository.