lab02 : FtoC and CtoF with test cases

num ready? description assigned due
lab02 true FtoC and CtoF with test cases Thu 08/17 09:30AM Wed 08/23 04:45PM

In this lab, you’ll practice:

Step 0: Install pytest for your account (or on your machine)

This lab is one that you may find you need to do on the CSIL machines.

It is probably the case that pytest is not installed for your version of Python3. You can check by typing python3 at the Terminal prompt to get to the Python Shell Prompt >>>, and then typing import pytest.

If you get an error message like this one, then pytest is not installed.

[cgaucho@csil-12 ~]$ python3
Python 3.4.3 (default, Aug 9 2016, 15:36:17) [GCC 5.3.1 20160406 (Red Hat 5.3.1-6)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pytest
Traceback (most recent call last):
File "", line 1, in ImportError: No module named 'pytest'
>>>
[cgaucho@csil-12 ~]$ pip3 install --user pytest

To install it, type this command into the terminal session (the Unix Terminal prompt) to install pytest for your CSIL account:

pip3 install --user pytest

You can also try this command on Mac or Windows. It might work, or it might not. If it does—great, then you can do this lab on your own machine. If not, then you’ll need to do it on CSIL.

The output should look something like this:

[cgaucho@csil-12 ~]$ pip3 install --user pytest
You are using pip version 7.1.0, however version 9.0.1 is available.
You should consider upgrading via the 'pip install --upgrade pip' command.
Collecting pytest
  Downloading pytest-3.2.1-py2.py3-none-any.whl (186kB) 100% 188kB 1.5MB/s
Collecting py>=1.4.33 (from pytest)
  Downloading py-1.4.34-py2.py3-none-any.whl (84kB) 100% 86kB 2.0MB/s
Requirement already satisfied (use --upgrade to upgrade): setuptools in /usr/lib/python3.4/site-packages (from pytest)
Installing collected packages: py, pytest
Successfully installed py pytest
[cgaucho@csil-12 ~]$

To tell whether it worked or not, try the import pytest command again:

[cgaucho@csil-12 ~]$ python3
Python 3.4.3 (default, Aug 9 2016, 15:36:17) [GCC 5.3.1 20160406 (Red Hat 5.3.1-6)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pytest
>>>

The lack of an error message (just another >>> prompt) means “it worked!”. We are going to use the Python prompt in the next step anyway, so just stay at the Python prompt. (Or if you need to get out of Python, use CTRL-D to return to the Unix shell prompt.)

Step 1: Warmup–experiencing floating point inaccuracy

If you are not already at the Python prompt, bring up a terminal window, and just type python3. This should give you the Python Shell Prompt (>>>) where you can type in some expressions and see the resulting values.

Type in the import math, followed by math.sqrt(2). It should look like this:

>>> import math
>>> math.sqrt(2)
1.4142135623730951
>>> 

The value we get back is the square root of 2, which is an irrational number—its decimal representation goes on forever. Unfortunately, real world computing devices typically store numbers with a finite number of decimal places†. So, the representation we see for the square root of two, is, in fact an approximation.

(† Technically, “binary places”, or “binary digits” rather than “decimal places”. For purposes of this discussion it amounts to the same thing. Also, some computer systems do work with “symbolic” representations of numeric quantities e.g retaning or as symbolic values rather than as numerical approximations. On those systems, you can exact results, without losing precision, at the expense of speed. We won’t discuss that kind of software in this course.)

We can see this if we multiply math.sqrt(2) by itself. Try it:

>>> math.sqrt(2) * math.sqrt(2)
2.0000000000000004
>>> 

See that pesky 4 digit in the ten quadrillionths place? My goodness, we are really, really close to 2.0, but if we ask whether the values are equal, Python says no:

>>> math.sqrt(2) * math.sqrt(2)== 2.0
False

In fact, Python is very clear about the difference between 2.0 and math.sqrt(2) * math.sqrt(2), and can even tell us how big that difference is. The 4 digit is only the tip of the very, very, very small iceberg:

>>> math.sqrt(2) * math.sqrt(2)- 2.0
4.440892098500626e-16
>>> 

This fact is going to be annoying to us many times. One consequence is that when we test software involving floating point numbers, we must allow for some inaccuracy. This “allowable inaccuracy” is sometimes called the tolerance, and it might be a small value such as 0.001 (1x10-3, or 0.000001 (1x10-6).

In Python, we can write 0.001 as 1e-03, and 0.000001 as 1e-06. (The lowercase e is the way that Python represents scientific notation.)

You can see that Python, by default, formats numbers in this notation once the fifth decimal place is reached:

>>> 0.01
0.01
>>> 0.0001
0.0001
>>> 0.00001
1e-05
>>> 0.000001
1e-06
>>>

We’ll come back to that idea later in this lab.

As a reminder, CTRL-D gets you out of the Python Shell Prompt (>>>) and returns you to the Unix shell prompt.

Step 1: Make a ~/cs8/lab02 folder

In previous labs, you should already have created folders ~/cs8/lab00 and ~/cs8/lab01. You are now going to create folder ~/cs8/lab02 for the files for this lab.

In general, its probably a good idea to keep your work for this class under ~/cs8, in folders called lab00, lab01, lab02 etc.

This isn’t exactly required (no-one is going to check), but it’s probably a good habit to develop. Here’s why: all the rest of the instructions will be written based on the assumption that you did things this way. And if you continue to take CS courses at UCSB, you’ll often find that’s the case from one course to the next.

So, I’d strongly encourage you to do it.

As a reminder:

With that information, you should be able to determine how to create a ~/cs8/lab02 directory, and make that directory your current working directory (the one that comes up when you type pwd.) Do that now.

Then, you are ready for the next step.

Step 2: Create a file called convert.py in your ~/cs8/lab02 directory

The contents of convert.py should be as shown below. This contains two Python function definitions that, at the moment are incorrect.

Choose “File => New File” in IDLE to bring up an “untitled” window, then copy and paste this code into that window.

Note that the formulas for converting between Celsius and Fahrenheit are incorrect. That’s deliberate, so just go with it for now. We’ll fix those at a later step.

def fToC(fTemp):
    return fTemp - 32  # TODO: Fix this line of code

def cToF(cTemp):
    return cTemp + 32 # TODO: Fix this line of code

Step 3: Test your code by hand

To test this code, we’ll first do what many programmers do: test the code by hand.

That is, we’ll run the file in IDLE, and then enter some function calls in the Python shell to see what results we get. These two functions are supposed to convert between Fahrenheit where water freezes at 32 degrees, and Celsius where it freezes at 0 degrees. Let’s see if they work properly.

Use the Run Module command to run the code, and then try entering these function calls at the Python Shell prompt. You should see output similar to what is shown below:

>>> fToC(32)
0
>>> cToF(0)
32
>>> 

As you can see, for those two particular values, the function appears to return the correct answer—32 degrees fahrenheit is indeed 0 degrees celsius, and 0 degrees celsius is indeed 32 degrees fahrenheit.

So clearly, testing with a single value is not enough. Indeed, if we test with another well known value, 212 Fahrenheit and 100 Celsius (the boiling point of water), we see that the output is incorrect:

>>> fToC(212)
180
>>> cToF(100)
132
>>> 

One of the problems with testing by hand is that it is tedious, and folks have a tendency to skip it. So, experienced programmers have learned that its generally a better idea to automate the process of testing. We’ll learn how to do that next. We’ll see that when we set up these four tests, two of them will pass, and two of them will fail.

Step 4: Setting up automated tests

Add this line at the top of your convert.py file:

import pytest

Then, add the following code to your convert.py file after the function definitions for ftoC and cToF that defines four automated tests:

def test_fToC_freezing():
   assert fToC(32.0)==pytest.approx(0.0) 

def test_cToF_freezing():
   assert cToF(0.0)==pytest.approx(32.0) 

def test_fToC_boiling():
   assert fToC(212.0)==pytest.approx(100.0) 

def test_cToF_boiling():
   assert cToF(100.0)==pytest.approx(212.0) 

These are automated tests that use a module known as pytest. When defining tests using the pytest module, we typically define functions that:

If the expresssion after assert is true, the test passes, otherwise it fails

We are using pytext.approx() here because any time you are testing with floating point numbers, we have to be aware that there may be some inaccuracy, as we discussed earlier.

(Recall our discussion of what happens when you multiple math.sqrt(2.0) by itself. Here, its probably overkill since we aren’t using any irrational numbers, but it is still safer to always use some way of approximating equality when dealing with floating point.)

You can click the plus to open a dropdown showing what your entire file should look like at this point:

Contents of convert.py so far

import pytest

def fToC(fTemp):
    return fTemp - 32

def cToF(cTemp):
    return cTemp + 32
    
def test_fToC_freezing():
   assert fToC(32.0)==pytest.approx(0.0) 

def test_cToF_freezing():
   assert cToF(0.0)==pytest.approx(32.0) 

def test_fToC_boiling():
   assert fToC(212.0)==pytest.approx(100.0) 

def test_cToF_boiling():
   assert cToF(100.0)==pytest.approx(212.0) 

After entering this, save the file and use “Run Module” to make sure there are no error messages (red output in the Python Shell Window.). If not, then you are ready for the next step.

Step 5: Running the test cases

Running the test cases requires us to go outside of IDLE back to the terminal shell prompt.

Your current directory needs to be the same one that your convert.py file is stored in. That should be ~/cs8/lab02, but if it isn’t, then fix things so that the convert.py file is in that directory, and you are in that directory. If you need help, ask for assistance.

You should be able to type the ls command (or on Windows, dir) at the terminal shell prompt and see the convert.py file listed:

your-prompt-here $ ls
convert.py
your-prompt-here $ 

When that’s the case, try this command:

python3 -m pytest convert.py

You should see output like this. It may be a little overwhelming at first, but don’t let it intimidate you. Once you know what you are looking for, it is very easy to read. After the output, there is a guide to understanding it.

169-231-175-204:lab02 pconrad$ python3 -m pytest convert.py
==================================== test session starts ====================================
platform darwin -- Python 3.6.2, pytest-3.2.1, py-1.4.34, pluggy-0.4.0
rootdir: /Users/pconrad/github/ucsb-cs8/Lecture5_0816/lab02, inifile:
collected 4 items                                                                            

convert.py ..FF

========================================= FAILURES ==========================================
_____________________________________ test_fToC_boiling _____________________________________

    def test_fToC_boiling():
>      assert fToC(212.0)==pytest.approx(100.0)
E      assert 180.0 == 100.0 ± 1.0e-04
E       +  where 180.0 = fToC(212.0)
E       +  and   100.0 ± 1.0e-04 = <function approx at 0x1026c40d0>(100.0)
E       +    where <function approx at 0x1026c40d0> = pytest.approx

convert.py:16: AssertionError
_____________________________________ test_cToF_boiling _____________________________________

    def test_cToF_boiling():
>      assert cToF(100.0)==pytest.approx(212.0)
E      assert 132.0 == 212.0 ± 2.1e-04
E       +  where 132.0 = cToF(100.0)
E       +  and   212.0 ± 2.1e-04 = <function approx at 0x1026c40d0>(212.0)
E       +    where <function approx at 0x1026c40d0> = pytest.approx

convert.py:19: AssertionError
============================ 2 failed, 2 passed in 0.03 seconds =============================
169-231-175-204:lab02 pconrad$ 

Ok, let’s now break down this output.

Step 6: Understanding the output of pytest

Here’s how to understand pytest output.

  1. Get the big picture first from the convert.py ..FF line.

    Look for a line near the beginning that has the name of your file, followed by a list of either dots, letter ‘E’ or letter ‘F’ characters, like this one:

    convert.py ..FF
    

    In this case, the . characters are tests that passed. If you have four tests, the ideal thing you’d want to see is convert.py …. which means that four tests for convert.py all passed.

    Instead, here, we see ..FF, which means we had two test failures. Later in the output, we’ll see more detail about each of those failures.

  2. Understand the overall structure of the output.

    The rest of the output will be divided into sections, one for each failure. Here is what it look like with the details taken out so that you can see the big picture more easily:

    ==================================== test session starts ====================================
    (blah blah here that you can ignore)
       
    convert.py ..FF
    ========================================= FAILURES ==========================================
    _____________________________________ test_fToC_boiling _____________________________________
    
    ...
    (details about the first test case failure are here)
    ...
       
    _____________________________________ test_cToF_boiling _____________________________________
    
    ...
    (details about the second test case failure are here)
    ...
       
    ========================= 2 failed, 2 passed in 0.03 seconds =============================
    

    Note that the last line gives us a nice summary: 2 failed, 2 passed in 0.03 seconds. We now know that we need to focus on the two failures. And the headers tell us which test cases failed, namely test_fToC_boiling and test_cToF_boiling. So let’s focus on those next, by first looking in detail at the first one:

  3. Focus in on the first test case failure.

    Let’s focus just on the detailed output for the first test case failure, test_fToC_boiling. What does all of the detailed output mean?

    In general, its a breakdown of why the assertion turned out to be false, showing every step in the calculation. Let’s break it down one line at a time:

    line of output meaning
    def test_fToC_boiling(): first line of the failing test case
    E assert 180.0 == 100.0 ± 1.0e-04 This is the assertion that turned out not to be true
    E + where 180.0 = fToC(212.0) This tells us why 180.0 was on the left hand side of the ==. It was the result of computing ftoC(212.0)
    E + and 100.0 ± 1.0e-04 = <function approx at 0x1026c40d0>(100.0) This tells us why 100.0 ± 1.0e-04 was on the right hand side of the ==. It shows the expected value (100.0) and the tolerance ( ± 1.0e-04).
    E + where <function approx at 0x1026c40d0> = pytest.approx This tells us that pytest.approx was used to calculate the tolerance.
       
    convert.py:16: AssertionError This shows which line in the file convert.py had the failed assertion, namely, line 16. That helps us find the error faster if we are dealing with a large file of code.

    If we look at this, we can see that amidst all of the clutter is the crucial fact that fToC(212.0) returned 180.0 when what we were expecting was 100.0, with a tolerance of ± 1.0e-04.

Step 7: Fixing the code

So, if you have failing test cases, the thing to do is fix the code so that the test cases pass.

To do that you’ll need to correct the forumla.

Keep in mind that in Python:

Also, the order of operations in Python is that multiplication and division are done before addition and subtraction. Some examples:

When you replace return ftemp - 32.0 with the correct formula for converting a Fahrenheit temperature to Celsius, you should leave out the comment that says # TODO: Fix this line of code .

You’ll also want to replace the similar line in the cToF function.

When you have the test cases passing, try running the pytest command again—remembering that:

python3 -m pytest convert.py

When you see four passing tests, your output will look like this. Also, the last line will be a pleasant shade of green instead of an angry looking red.

169-231-175-204:lab02 pconrad$ python3 -m pytest convert.py
======================== test session starts =========================
platform darwin -- Python 3.6.2, pytest-3.2.1, py-1.4.34, pluggy-0.4.0
rootdir: /Users/pconrad/github/ucsb-cs8/Lecture5_0816/lab02, inifile:
collected 4 items                                                     

convert.py ....

====================== 4 passed in 0.01 seconds ======================
169-231-175-204:lab02 pconrad$ 

At that point, you are ready to submit your work to submit.cs

Step 8: Submit your convert.py file to submit.cs

To submit your file to submit.cs, you can visit this page:

https://submit.cs.ucsb.edu/form/project/770/submission

Navigate to that page, and upload your convert.py file.

Or, if you are working on the ECI/CSIL/lab linux systems, you can also submit at the command line with this command, provided you are in the correct folder/diretory:

~submit/submit -p 770 convert.py

The first time you use this method, it will ask for your email address: use your full umail address (e.g. cgaucho@umail.ucsb.edu). For password, use the password that you enter for the submit.cs system. You may save these credentials if you don’t want to have to type them in every time.