Open YakovL opened 4 years ago
Actually JavaScript code in content-script.js
does have access to the page -exactly same as Firefox. So by logical reasoning, it ought to work in the same way in both browsers.
Here's the problem: in Firefox
window.mozillaLoadFile
and others;in Chrome patch-classic-io.js is injected, but it gets a sandboxed context and window
is not the same as window
, so the change that patch-classic-io.js introduces doesn't really affect window.mozillaLoadFile
etc of the page itself. That answer at SO suggests a hack that can help though. I'll probably give it a try soon; this will make the code ugly though because the whole patch-classic-io.js should be put as a JS string inside content-script.js.
What if we add the code in the patch to the content-script itself - as a function? Instead of inject, we can call it as a function, right?
Not sure, could you provide a sample code?
The following seems to work:
change injectExtensionScript
to this:
function injectExtensionScript(text) { const scriptElement = document.createElement("script"); scriptElement.innerHTML = text; document.head.appendChild(scriptElement); scriptElement.remove(); }
call it (instead of injectExtensionScript('patch-classic.io)
) like this:
injectExtensionScript(`
window.mozillaLoadFile = function(path) {
try {
// Just read the file synchronously
var xhReq = new XMLHttpRequest();
xhReq.open("GET", "file:///" + encodeURIComponent(path), false);
xhReq.send(null);
return xhReq.responseText;
} catch(ex) {
return false;
}
};
window.mozillaSaveFile = function(path, content) {
var messageBox = document.getElementById("tiddlyfox-message-box");
if(!messageBox) return false;
// Create the message element and put it into the message box
var message = document.createElement("div");
message.setAttribute("data-tiddlyfox-path", path);
message.setAttribute("data-tiddlyfox-content", content);
messageBox.appendChild(message);
// Create and dispatch the custom event to the extension
var event = document.createEvent("Events");
event.initEvent("tiddlyfox-save-file", true, false);
message.dispatchEvent(event);
return true;
};
// expose the supported I/O events
window.eventBasedIO = {
save: {
name: 'tiddlyfox-save-file'
},
saved: {
name: 'tiddlyfox-have-saved-file'
}
};`);
(sorry for ugly formatting, github markup makes it a bit difficult)
However, while this injecting works in extension edited locally, it doesn't seem to find the Timimi backend, so I've only succeeded with saving when activated both current Timimi extension and my local version. Any idea how can I fix this?
You mean you want to create a local environment for testing purposes?
Yes, that would be helpful (so that I can create PRs with more confidence)
Actually, that worked in Firefox: I loaded Timimi extension from cloned and edited repo on local file system and it interacted with Timimi backend just like the main extension. Not sure why this is not the case in Chrome (well, in Vivaldi, I haven't tested this in Chrome yet)
Yeah. Chrome works a little differently. You need to edit the timimi.json for it to work with chrome in testing environment.
Here is what you have to do. Uninstall timimi that you may have installed from chrome webstore Load the unpacked extension to Chrome. In the extension page, timimi will have an extension id. Copy that. Now go to timimi.json and replace the extension id in that with the value you copied
well, this doesn't seem to be working. To be sure, I've retested in Chrome (instead of Vivaldi), but it's the same:
<user folder>\AppData\Roaming\Timimi-Chrome\Timimi.json
(is it the one?), changed allowed_origins
by substituting the idmozillaSaveFile
changed etc), added some changes, (unchecked saving backups,) called savingA message (more likely browser than native, white one with Timimi icon) appeared saying "Timimi save FAILED Error on contacting Timimi host" and the changes are not saved. I'll probably create a PR in parallel (not sure where the changes should go though: they can be used for Firefox too and it's not nice if the code is duplicated anyway, since further changes should be remembered to be put into both)
Hi Riz, any comments on this? What do you think about merging #48 so that we can use this minimal implementation in the next release and improve it further if needed?
Merged. Now the only thing we have to see is if it passes the chrome store's security check.
Great, thanks! Yeah, the security check is indeed is an interesting question because such code injection may be considered "hacky".
By the way, do you have any plans to de-duplicate code for Firefox and Chrome extensions? Like generate one from another, so that it is enough to introduce new changes only in one place.
The plan is to make the code similar in both eventually.
Hi Riz,
we have to see is if it passes the chrome store's security check
any news on this? What's the usual time they take to review an updated version?
Currently, saving in Chrome doesn't work for TWC. Here's a quick note from #32, I'll expand it at some point (additional research required):