Weird `<!--[-->` sytax causing rendering issues on html_render_diff

[Mr0grog]

Version of https://science.nasa.gov/climate-change/adaptation-mitigation/ from December 2024 (and probably much earlier) are causing some weird rendering issues on html_render_diff where we are seeing text like <>>. Screenshot:

(See this live in Scanner)

In the source (archived copy), there are some IE conditional-comment-like syntax that I think is causing the problem:

<!--[--><div class="absolute z-top"><!--[--><!--[-->

That’s not actual conditional comments syntax, though. Maybe an error was causing this output? Or it’s something I haven’t seen before.

Then we get this mess out the other end of the diff:

&lt;<cyfunction 0x76bbfbe6a4d0="" at="" comment="">&gt;<!--<cyfunction Comment at 0x76bbfbe6a4d0-->&gt;&lt;<cyfunction 0x76bbfbe6a4d0="" at="" comment="">&gt;<!--<cyfunction Comment at 0x76bbfbe6a4d0-->&gt;<div><div class="absolute z-top">

Which seems like maybe lxml or html-parser are choking on this in a pretty painful way.

[Mr0grog]

Possibly related, an EDGI analyst noted this problematic diff today, which features these messy comments on the “from” side, and shows the whole page being added on the “to” side: https://monitoring.envirodatagov.org/page/cd80203e-d5cf-4453-8042-fe971141f431/83615878-7465-4d83-ba1d-ebb3f9c626d5..126227d1-4be0-4730-83ad-63179fa376a9

The “from” side stuff is definitely the same as this issue. Not sure if it’s also causing the “to” side problems. On the “to” side, the concrete cause of the out-of-control highlighting is problems re-assembling the markup; the first closing </ins> tag is missing or maybe getting written out as <. Just not sure if the cause of that is these comments or something else.