Validate HTML/CSS/...

[The-Compiler]

The HTML/CSS/... shipped with qutebrowser should be validated by rendering it (either a minimal version via utils.jinja or by calling the real code) and then be validated.

I can think of various methods to do so:

• Using lxml.etree:

parser = lxml.etree.HTMLParser(recover=False)
lxml.etree.fromstring(html, parser)

• Using PyTidyLib
• Using py_w3c to run the W3C validator (over the network...)
• Using cssutils (for CSS)

[neeasade]

So the files that need to be validated can be located like so?

$ find . | grep -vE "htmlcov|json" | grep -E "\.html|\.css|\.js"

./.tox/py34/lib/python3.4/site-packages/pytest_html/resources/main.js
./.tox/py34/lib/python3.4/site-packages/pytest_html/resources/style.css
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/index.html
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/coverage_html.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/pyfile.html
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/jquery.tablesorter.min.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/jquery.debounce.min.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/jquery.min.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/jquery.hotkeys.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/jquery.isonscreen.js
./.tox/py34/lib/python3.4/site-packages/coverage/htmlfiles/style.css
./.tox/py34/lib/python3.4/site-packages/qutebrowser/javascript/position_caret.js
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/error.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/dirbrowser.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/COPYING.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/base.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/settings.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/log.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/pre.html
./.tox/py34/lib/python3.4/site-packages/qutebrowser/html/version.html
./tests/html/jsconfirm.html
./tests/html/jsprompt.html
./tests/html/jsalert_multiline.html
./tests/unit/javascript/base.html
./tests/unit/javascript/position_caret/simple.html
./tests/unit/javascript/position_caret/scrolled_down_img.html
./tests/unit/javascript/position_caret/scrolled_down.html
./tests/unit/javascript/position_caret/invisible.html
./tests/unit/browser/test_webelem_jsescape.html
./qutebrowser/javascript/position_caret.js
./qutebrowser/html/error.html
./qutebrowser/html/dirbrowser.html
./qutebrowser/html/COPYING.html
./qutebrowser/html/base.html
./qutebrowser/html/settings.html
./qutebrowser/html/log.html
./qutebrowser/html/pre.html
./qutebrowser/html/version.html

[The-Compiler]

It's mainly the stuff in qutebrowser/javascript and qutebrowser/html - though some of it needs to be rendered via jinja to (hopefully) be valid

[neeasade]

Alright, so this would be in a new test file and I could use the jinja_test for reference on rendering to test?

[The-Compiler]

I don't think the jinja tests help much here - you'd have a custom jinja2.Environment for the tests (take a look at unit/javascript/conftest.py maybe, which does something similar) and then render the templates with some dummy values.

How are you planning to do the validation though?

[neeasade]

I was going to look into using py_w3c

[The-Compiler]

This uses the network unfortunately, but I think it's the best solution - maybe you could take a look at the other solutions in my original post as well? It'd be nice to have something to run offline automated, and then the W3C validator e.g. manually

[neeasade]

irc snippet for posterity

    neeasade │ The-Compiler: when you say manually on the w3c    
             │ validator, you mean the w3c validator offline?
The-Compiler │ neeasade: the issue is that I sometimes push a dozen
             │ times per day, so it'd be nice to have something which
             │ runs offline to always run with the automated checks
    neeasade │ The-Compiler: yeah I understand - I didn't consider
             │ that when I first posted                                 
The-Compiler │ and then something using W3C via the network only when
             │ there actually was a .html modified, or by hand from
             │ time to time
    neeasade │ ahhhh okay :)
    neeasade │ I will look more into the other links from OP
    neeasade │ at a glancepy_w3c looked most convienent (aside from
             │ network)
The-Compiler │ I imagine it with a parametrized pytest fixture, which
             │ checks a HTML via various ways, and skips w3c unless
             │ some commandline arg was given
The-Compiler │ I wonder if this is a good issue for you to work on,
             │ though - I have some ideas how to make it work, and
             │ some of the work also is integrating it with the CI
    neeasade │ I will likely be skimming the easy tagged issues before
             │ I do anything with this, I was curious about it

[The-Compiler]

Some links about running the W3C validator on Travis:

https://validator.github.io/validator/
https://github.com/validator/validator/releases
https://validator.w3.org/nu/about.html
https://w3c.github.io/developers/tools/
https://github.com/svenkreiss/html5validator