从Chrome扩展程序内调用DevTools的inspect（）函数

Question

I want to create an extension that can programmatically inspect an element as if the user right-clicked on an element and clicked "Inspect." I want that to happen, preferably, from a devtools page set via devtools_page in the manifest. I want the exact behavior - Elements panel opens, the element is highlighted on the left and its styles on the right sidebar.

How would I achieve that?

What I've tried

As a starting point, I want the extension to attach a debugger and inspect the document body once I click on the extension icon on top. Here's my devtools_page :

chrome.browserAction.onClicked.addListener(function (tab) {
    chrome.debugger.attach({tabId: tab.id}, "1.0", function () {
        chrome.debugger.sendCommand({tabId: tab.id}, 'Runtime.evaluate', {
            expression: 'console.log(document.body); inspect(document.body);'
        }, function () {
            console.log('Result:', arguments);
        }); 
    });
});

If I now look at the DevTools window of the page I'm inspecting, I can correctly see the log of the body and click on it to open the element inside the inspector. However, if I head to the extension's devtools page and inspect that , I can see the "Result:" log there and it reads:

Result:
Arguments(1)
  0:
    exceptionDetails: {columnNumber: 28, exception: {…}, exceptionId: 2, lineNumber: 0, scriptId: "88", …}
    result:
      className: "ReferenceError"
      description: "ReferenceError: inspect is not defined↵    at <anonymous>:1:29"
      objectId: "{"injectedScriptId":1,"id":1}"
      subtype: "error"
      type: "object"
      __proto__: Object
    __proto__: Object

So the evaluated code doesn't have access to the inspect() function, even though it has access to the element I want to inspect. Apparently it executes in the context of the page and not the context of its DevTools.

How can I programmatically set the inspected element in DevTools from my extension?

Answer 1

As @Pa Wa pointed me to, chrome.devtools.inspectedWindow did the job via its eval() method:

chrome.devtools.inspectedWindow.eval('inspect(document.body)');

It doesn't even require a debugger attached.