I've done my research and struggled with this for a while, but I need your help.
I'm building a Chrome DevTools extension. It should should pass the currently selected element from the 'Elements' panel as a reference to a JS object defined in a content script.
It is important that I pass the reference to the selected element, or some other way of identifying the element from the content script.
I understand the workflow with 'isolated worlds' in Chrome DevTools. I also understand messaging between extension pages, background page and content scripts. This only happens with JSON primitives, hence no JS scope passing.
How can I pass the element selected in devtools Elements panel to the content script that lives in the inspected page?
Edit
Here's what I know so far:
Getting a reference to the selected element:
chrome.devtools.inspectedWindow.eval("(" + function(){ console.log($0) }.toString() + ")()")
That function expression will run in the context of the inspected page, not in the context of the devtools extension and not in the context of the 'isolated world' of the content script. I don't believe it is possible to pass in a reference to a different context using closures.
The reference to the selected DOM element
$0
can't be returned because it can't be serialized to JSON due to circular references.The
chrome.devtools
namespace isn't available outside the devtools extension page. The$0
reference can't be used outside the evaluated expression inchrome.devtools.inspectedWindow
Workaround
As a workaround, I chose to use the shared DOM to mark the selected element with a data attribute and use that to re-select it in the context of the content script. Messaging is used to pass the data attribute marker around.
Here's a simplified version of the code:
In the devtools extension page:
// setup a communication port
port = chrome.runtime.connect({name: "devtools"});
chrome.devtools.panels.elements.onSelectionChanged.addListener(function(){
// expression to run in the context of the inspected page
var expression = "(" + mark.toString() + ")()"
// evaluate the expression and handle the result
chrome.devtools.inspectedWindow.eval(expression, dispatchToContentScript)
});
function mark(){
// mark the currently selected element
$0.setAttribute('data-selected')
// send the marker to the callback
return { marker: 'data-selected' }
}
function dispatchToContentScript(data){
// dispatch data to the content script which is listening to the same port.
port.postMessage(data)
}
In the content script:
var port = chrome.runtime.connect({name: "devtools"});
port.onMessage.addListener(function(data) {
// re-select the element in the context of the content script
var el = document.querySelector('['+ data.marker +']')
})
It's not a clean solution but I can use it for my needs.
Is there a simpler way to achieve the same result - identify from a content script the element selected in the devtools 'Elements' panel?