captureVisibleTab in new tab overriden by "chrome_url_overrides"

[Robbi]

I was wondering if chrome.tabs. captureVisibleTab works in a  new tab overriden by "chrome_url_overrides".

It'd seem not.

Is there a workaround or some web api that I could use as an alternative?

I would like to create a rudimentary magnifying glass that is activated by clicking on an image (not browser action).

Do I have any chance?

Thanks

[wOxxOm]

If it doesn't work with "permissions": ["<all_urls>"] then it's a bug, which you can report on https://crbug.com. Meanwhile you can try opening a new window that will invoke the API, or add an iframe with a chrome-extension:// URL and call the API there.

[Robbi]

Hi wOxxOm,

I tried all the workarounds you suggested me, but unfortunately none of them works.

It seems the method fails not only on new tabs but also on all extension pages (chrome://extensions_id/*).
With other normal web pages it works instead.

The error I get when I invoke it when a extension page is selected on target windos is "The 'activeTab' permission is not in effect because this extension has not been in invoked." Note that I add "<all_urls>" permission in manifest file.

At this point, I suspect this method was designed to work only in content scripts.

[wOxxOm]

Content scripts can't use chrome.tabs API so no, it's definitely a bug. So far the only workaround is to use browser_action, which you don't want though.

[Robbi]

You're right, content scripts can't use chrome.tabs API.

I wanted to refer to all those extensions in the store that perform the function of magnifier.

I have installed a couple of these.
These are activated on the browserAction and work on all normal web pages except pages whose URL starts with chrome: //.
However, I see no reason to ban this method to extension pages.
It would be advisable to grant use to our own pages without giving <all_urls> permission which in these use cases would also be excessive.
Do you think it is appropriate to open a bug or ask for a new feature?

[wOxxOm]

Since it worked before https://crrev.com/548891 landed in Chrome 67, it's technically a bug.

[hrg...@gmail.com]

According to the tabs.captureVisibleTab documentation, "sensitive sites can only be captured with the activeTab permission".

Sensitive sites are "chrome:-scheme pages, other extensions' pages, and data: URLs".

Since the URL of the new tab page is chrome://newtab (whether it's overridden or not), the limitation you are facing is documented.

[wOxxOm]

> the limitation you are facing is documented 

No, the extension has access to its own tab so it definitely should be capable of capturing it even without activeTab.

[wOxxOm]

To elaborate, I guess the bug in Chrome is caused by the fact that the newtab has a chrome://newtab URL internally but the person who implemented https://crrev.com/548891 forgot to add a check for an overridden newtab's effective origin which is the extension's own origin. P.S. as mentioned earlier, a workaround is possible via browser_action.

[hrg...@gmail.com]

That's not clear in the documentation. An overridden chrome:// URL is a special case and the documentation doesn't specify any special behavior for that case. So the current API behavior matches the documentation.

I agree that this should be corrected, but the documentation must be clear about it first.

[Robbi]

I've been banging my head since this morning.

It would seem that <all_urls> allows you to capture ONLY the ACTIVE tab without necessarily having to click on the browseAction.

I tried via tab.update to divert the focus to another tab and then to take the photo and it works ONLY if the page schema is NOT chrome://

With the activeTab permission, on the other hand, it also works on pages with a chrome:// schema as long as the "activeTab" grant is given to the tab by clicking on the browserAction.

After the tab has acquired that grant, captureVisibleTab can be used through a simple button.

Behavior, in my opinion, very strange.

I develop a small extension for testing purpouse.

Here is the link: LINK

Clicking the browserAction the first time, a page extension opens with same buttons

whereas the following clicks (on browserAction) activate the tabs.captureVisibleTab on current tab.

In manifest file there is already <all_urls> permission.

It can  obviously comment this permission and add "activeTab" to perform all combination tests

[wOxxOm]

It's a bug, simple as that, because the API must work with the activeTab permission even when invoked from an overridden chrome://newtab page as stated in the documentation. I would have reported it myself right away but dozens of the reports I made aren't fixed yet so I'm pessimistic about it.

As for a workaround you can use the extension's icon itself to initiate capturing instead of using a button inside your newtab page. I guess you would need to declare a background script and use chrome.browserAction.onClicked inside.