Issue 5383 - Add tests for the Chrome and Firefox packagers

tox.ini

 [tox]
 envlist = py27
 skipsdist = true
 
 [flake8]
 putty-ignore =
     build.py : A102,A104,A107,A201,A206,A302,E501,E711,F401,N802,N803,N806
     chainedconfigparser.py : A107,E501
     ensure_dependencies.py : A102,A107,A108,A302,E129,E501,E713,E721,F821
     localeTools.py : A103,A104,A107,A108,A206,A301,A302,E129,E501,E711,E713,F401,F812,N802,N803,N806
     packager.py : A102,A107,A206,A302,E501,E711,N802,N803,N806
     packagerChrome.py : A101,A104,A105,A107,A108,A111,A112,A302,E129,E501,E711,E713,E731,F401,F841,N802,N803,N806
     packagerGecko.py : A101,A104,A105,A107,A108,A201,A302,E501,E711,E713,F401,N802,N803,N806
     packagerSafari.py : A101,A107,A301,A302,E501,N802,N803,N806
     publicSuffixListUpdater.py : A108,E501,F821,N802,N803
     releaseAutomation.py : A102,A107,A108,A302,E501,F401,N803,N806
 
 [testenv]
 setenv =
     PYTHONPATH = {toxinidir}/..
 deps =
+    pycrypto
+    pillow

[Sebastian Noack, 2017/08/14 15:20:12]

The README (in adblockpluschrome) lists PIL (not Pillow) as a dependency:
https://hg.adblockplus.org/adblockpluschrome/file/552ac1922795/README.md#l22

Pillow is meant to be a compatible drop-in replacement for PIL. So in theory
buildtools should work with both, and since in some environments only either is
available, it would be a good idea to ensure compatiblity with PIL and Pillow:

  [tox]
  envlist=py27-{PIL,pillow}

  [testenv]
  deps=
      PIL: PIL
      pillow: pillow
      ...

[tlucas, 2017/09/06 09:04:52]

Got rid of PIL / pillow dependency completely (by adding pre-generated images)

[Sebastian Noack, 2017/09/06 18:00:00]

Well, the make_icons() function in packagerChrome.py is using PIL/Pillow as
well. But since this seems to be the only use of PIL/Pillow, we still have, and
getting the size of a PNG image is simple enough without an image processing
library, we might want to just do that and get rid of that dependency.

https://stackoverflow.com/questions/8032642/how-to-obtain-image-size-using-st...

[Vasily Kuznetsov, 2017/09/07 20:27:50]

Guys! Writing your own image format parser is rather radical NIHilism, given
that Pillow library is available and is quite stable and standard. However, if
it's only about PNG, it is not too complicated indeed, so be my guests, knock
yourselves out ;)

[Sebastian Noack, 2017/09/07 20:39:00]

Well, all we need at the current point, is extracting the image size of an PNG
image. This is a trivial thing to do by parsing the relevant bytes from the PNG
header, and would avoid having us to instruct users to install either PIL or
Pillow, and testing our code with both libraries.

[Sebastian Noack, 2017/09/08 21:07:12]

It is related to the tests, as it simplifies the tests. If we keep using
PIL/Pillow, my initial comment in this thread is still valid. But in order to
make sure everything works with both PIL and Pillow, we wouldn't only have to
add a couple additional lines to tox.ini, but also cause the tests to take twice
as long to run. Not to mention an additional dependency for users of buildtools.

Also note that detecting the size of a PNG image takes the same amount of code,
with PIL/Pillow than it does without:

  width, height = struct.unpack('>ii', files[filename][16:24])

  width, height = Image.open(StringIO(files[filename])).size

Anyway, if you still prefer to use PIL/Pillow, make at least sure that the code
is tested with both libraries.

[Vasily Kuznetsov, 2017/09/11 12:50:06]

Seems like your proposal would work fine and it's simple and compact indeed, so
I don't mind. And eliminating dependency on PIL/Pillow is a good thing!

     pytest
+    pytest-cov
     jinja2
     # flake8-putty is not yet compatible with flake8 3.0
     flake8==2.*
     flake8-putty
     pep8-naming
     hg+https://hg.adblockplus.org/codingtools#egg=flake8-eyeo&subdirectory=flake8-eyeo
 commands =
-    py.test tests
+    pytest tests --capture=sys --cov=buildtools --cov-report=term-missing
     flake8