ArchiveTeam/wpull

Support for server-side image maps

JustAnotherArchivist opened this issue · 0 comments

I found out today about an obscure feature in HTML: server-side image maps. This is an image with an ismap attribute:

<a href="/foo"><img src="bar.png ismap /></a>

Upon clicking on the image, the coordinates are appended to the URL as a query string of the form ?x,y, e.g. /foo?17,42 when clicking 17 pixels from the left and 42 pixels from the top. Note that this is true even if there is already a query string; if the href is /foo?bar, then the request would be /foo?bar?17,42 (see <a> definition in HTML5.2).

Real-world example using this feature via an ArchiveBot job: http://www.esva.net/cgi-bin/pix?Scrapbooks/Scrapbooks_from_2010/Rally_for_Sanity/rrsf013.jpg (the bar on the left of the picture)

Some thoughts on this:

  • This obviously has the potential to generate a ridiculous amount of URLs, but it could in some cases be necessary to correctly recurse through a site.
  • I'm pretty sure this is extremely rare. Unlike <area>, I don't think I've ever seen ismap before.
  • This would be tricky to implement because you have to know the image size before you can even generate the URLs.
  • If implemented, it should probably be disabled by default and enabled through an option due to the potential URL explosion.

Overall, I'm leaning strongly towards HELL NO!, but I found this interesting enough of a case to mention here. If anyone has a real-world example where this would actually be useful (unlike the one above), please do tell.