Copy `config.js` to a single `config.js-old` file whenever a
`config.template` is present, instead of using `Date()` as a suffix.
Creating files with a timestamp suffix means that whenever Magic Mirror
is restarted or recreated a new file is written into the config
directory.
Benefits:
* Single backup file will avoid excessive writing of files
* Saves space and usage on SD cards.
* Makes the folder cleaner and easier to overview, compared to ending up
with something like this as time goes on:
While waiting for the easterbunny I cleaned up some bad coding practice
:-)
Very open for comments especially regarding the places I commented
myself...
---------
Co-authored-by: veeck <michael@veeck.de>
- order (external first)
- remove superfluous file extensions
- new line after imports
- deconstruct (only one time (in `check_config.js`))
- fix path (only one time (in `global-setup.js`))
We have used it inconsistently till now. Template literals are more
modern and easier to maintain in my opinion.
Because that's a large amount of changes, here's a way to reproduce it:
I added the rule `"prefer-template": "error"` to the `.eslintrc.json`
and did an autofix. Since this caused a new problem in line 409 of
`newsfeed.js`, I reversed it in that line and also removed the rule from
the eslint config file.
The rule is described here:
https://eslint.org/docs/latest/rules/prefer-template
Note: I've played around with some other linter rules as well, and some
seem to point to some specific, non-cosmetic, issues. But before I dive
even deeper and then introduce even bigger and hardly understandable
changes at once, I thought I'd start with this simple cosmetic rule.
Percentage should be always rounded so that we dont get something like
"47.0000000001 %"
Some small typo and naming fixes also while I am here
---------
Co-authored-by: veeck <michael@veeck.de>
Async node_helper dont have to finish immediately in loadModules. So the
start callback in the app.js with the config isnt called for some time.
But the electron ready event can already be fired in the meantime.
This lead to the electron app starting but without a config (which is
provded by the node_helper callback) therefor crashing.
This PR fixes#2487 by moving the callback call out of the loadModules
block, therefor the config is provided in time.
If any new async node_helper doesnt like this, we will see it :-)
Co-authored-by: veeck <michael@veeck.de>
Many users like me do have the problem that they want to embed other
sites to their mirror by "iframe".
As some developers set the "x-frame-options" and
"content-security-policy" for security reasons these sites can not be
embedded.
Electron provides the "webview" element additionally to "iframe" which
allows to embed these sites although. The main difference is that a new
process is started which handles the "webview" element.
BUT: As the "webview" process needs to be started and is isolated
"webview" is slower and the elements can not be accessed from the
embedding website.
As an alternative i implemented a small callback function in electron.js
which removes the response headers that forbid the embedding.
The removing can be controlled with the new config options:
* ignoreXOriginHeader
* ignoreContentSecurityPolicy
Adds support for sending and receiving HTTP-headers when using the
CORS-method.
This change is required for the Yr weather-provider introduced in
https://github.com/MichMich/MagicMirror/pull/2948.
To make it easier to add unit tests I moved the server-functions into a
separate file.
When trying to debug why the tests broke for
https://github.com/MichMich/MagicMirror/pull/2946 I found that the tests
does not wait for the app to start and close. So if the startup isn't
blocking that would fail.
So I added a callback for `close()` too and converted them to promises
for the `startApplication()` and `stopApplication()` and updated all the
e2e tests to await both. Will try to refactor all these callbacks to
promises in a later PR.
small update to the fetchData method to use the fetch helper instead of
the old XCMLHttpRequest.
Also fixes some typos :-)
Co-authored-by: veeck <michael@veeck.de>
