Generate & clean html pages with Gulp and Nunchucks

[cbesozzi]

We currently have some consistency issues with empty html tags and orphan links.
Moving the pre-processing from python to Gulp could be useful to clean those up since we would be able to use tools like gulp-cheerio.

[bengolder]

@cbesozzi can you add more details on the problem this solves? You're saying we have some orphan links and empty html tags?
How would these libraries solve this problem?

[cbesozzi]

The final HTML pages have some quirks that are inherited by the word doc parsing, some examples are:

• Empty elements which creates inconsistencies in the layout
• Unused & inconsistent classes
• Links that are in text form but are not clickable
• Markup is hard to customize ad hoc for example to add certain properties

With few cleanup tasks we could address these issues and have clear separation between word doc parsing and manipulation.

I suggested Gulp and few Javascript libraries since I'm familiar with it and I can help but it's less a technology issue than it is a maintainability one.

@bengolder this is a nice-to-have rather than a must-have :)

[bengolder]

Thank for adding the details, this helps. I believe the is a "long tail" of varied and small html quality issues. Ideally, I'd like to pull them out one by one and prioritize based on which issues have the greatest impact on usability. But of course I would welcome any work that improves the quality of the output!

In terms of how to address them, the beautiful soup html parsing library and the jinja2 templating library that we are currently using should be useful in any html cleanup tasks. But if we see tasks that they cannot readily solve, I am open to switching libraries.

I would love to breakout more specific sub issues for HTML cleanup with links to where they can be found, especially if they are causing usability issues (like layout inconsistencies or non-clickable links). I know they exist, but I think the most difficulty lies in identifying the problems. If we have them documented, I should be able to make functions to cleanup most of them.