HTML entities can crash Markdown components ("document is not defined")

[shaunanoordin]

Uncategorised Bug?

Package: app-project, possibly others Affects: ⚠️ anything that uses Markdown, e.g. Project About Team page, About Project, Announcements banner, etc

The problem: if a Markdown component tries to render an escaped HTML entity such as & (ampersand), it crashes the whole page.

• The error only occurs if you access the FEM page directly (via URL or via reload). It does not affect the page if you navigate to it (e.g via project links)
• The error only triggers if the escaped HTML entity is bracketed by both & and ;
  • Markdown text such as & and &amp render as & on the webpage correctly.
  • & on the other hand will crash.
  • Note: I don't think we're supposed to use escaped entities like & in Markdown to begin with.
• This issue only affects FEM pages (confirmed on app-project), and not PFE pages.
• This only affects server-side rendering?
• Error message is document is not defined. Rest of the page is blank.
• This is a more general version of #3673

Dev Notes

Some notes from our investigation:

Scenario 1: literal symbols in the Markdown text Play "Dungeons & Dragons" ©, it's > Cthulhu ♥

• This renders fine
• NOTE: trying to add the < character in our Project Builder causes some things to go bonkers. Typing < will transform it into < and convert all > into > , & into & . Copy-pasting, especially when creating a HTML tag, causes even more WTF-ery. This is a separate project builder discussion.

Scenario 2: proper HTML entities Play "Dungeons & Dragons" ©, it's > Cthulhu ♥

• This causes a crash :x:

Scenario 3?: sorta-proper HTML entities Play &quotDungeons &amp Dragons&quot &copy, it's &gt Cthulhu &hearts

• This renders everything except the ♥ , which is just rendered as literally &hearts
• Trivia: looking at the WHATWG webpage, it turns out &amp &quot &copy &gt are all legit, so certain HTML entities apparently don't need the ; . I learn something new every day.

So anyway, it looks like HTML entities only crash our page if they're bracketed by both & and ; , but ALL HTML entities can crash us (not just the & )

Status

Unknown priority.

EDIT: possibly solved by #3678

[eatyourgreens]

Was this closed by #3678? Entities still raise a 'document is not defined' error, but it doesn't crash the page any more.

[shaunanoordin]

I say let's close this issue for now, as the important thing is that the pages aren't crashing. 👌

(Context for anyone who didn't read #3678: instead of crashing the page, HTML entities in Markdown now render as <p>document is not defined</p> on the page, and only if you have JS disabled.)

The only ways I can think of that the new "document is not defined" error could remotely be a problem are if (1) people notice "document is not defined" as the project's preview in their, erm, Google search results? (super unlikely, since our meta-description tags & etc now render fine)) or (2) we have a new logging system that automatically picks up the flood of console.errors in the dev log.

[eatyourgreens]

@shaunanoordin People will see 'document is not defined' when the page loads in their browser. https://www.zooniverse.org/projects/mariaedgeworthletters/maria-edgeworth-letters/about/team

https://user-images.githubusercontent.com/59547/191921740-f59175d5-62a0-4367-b07d-ed27d26d2838.mov