I'm trying to create a secure Darknet with WebRTC DataChannels in pure HTML, so I'm not interested about to know when an iframe has been fully loaded, but instead I'm interested to capture the iframe elements (inline images and so) using a custom scheme so I can be able from the parent page (the one connected to the Darknet) to do the real request and response with the actual data. With FirefoxOS mozbrowserlocationchange event of the Browser API objects (an extension to iframes) I could be able to capture the user navigation, cancel it, do the real request on the Darknet and later inject the iframe with the real content fetched by the parent page, but how could I be able to do the same with inline images and scripts on this loaded page? Or is this not currently possible and should I ask to them about add this functionality?
Obviously, I don't have any control about the iframes content pages, so they would be created by whatever and in any way, and also the usage of Browser API is just because seems to be the most useful to whan I'm trying to do, ideally it would be perfect if this is possible to achieve with plain iframes... :-)
Update:
A half-solution I have thinking about would be since I could be able to capture the mozbrowserlocationchange event to do the real content request of the HTML page and before fill the iframe with it do the request of their linked images and scripts and set them inline to prevent the iframe from doing more request. This would only lead to somewhat very simple web pages compared to current web standards (no AJAX, no async loading of script tags...) but definitely it would be usable up to some point :-)
Anyway, is there any other better alternative?
That sounds like something, that would be possible (straightforward, even) as soon as Service Controllers (previously known as NavigationControllers) are implemented, but I do not know any way to accomplish this via any currently available method.
No wonder you didn't find info about this - the proposal is called "Service workers" (though, previously this was called Event workers, and even before that, they were called - guess what - navigation controllers). This is a lively spec! ;) Find the working draft on GitHub: https://github.com/slightlyoff/ServiceWorker/ with a lengthy explainer document that should get you going.
Also, there is a document with the current Chrome (blink) implementation plans.