| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
|
|
|
| |
* fix: solve recipe function calls that were broken cause of js=>ts conversion
* fix: use an interface instead to keep type-safety
* fix: remove faulty test
|
| |
|
| |
|
| |
|
|
|
|
|
| |
* turn actions into typescript
* correct tsconfig
* added TypedStore
|
| |
|
| |
|
|
|
|
| |
This reverts commit 87bf6224518fd39ba7485cde691bad5941216258.
|
|
|
| |
Fix one of the asks in #239
|
|
|
|
|
|
| |
This option can be used to allow the user to choose how to handle links
(whether to open in the default browser or within the Ferdium service itself)
Co-authored-by: André Oliveira <oliveira.andrerodrigues95@gmail.com>
|
|
|
|
| |
Added TODO comments, notes, debug logs.
|
|
|
| |
This reverts commit 8b1bfc356c3182c481394251f3ceeb902f2baec3.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In https://github.com/ferdium/ferdium-app/pull/23 we removed usages of
the debug package due to an electron bug.
This patch aims to restore some debug functionality by introducing a
shim. The shim detect whether if it is being introduced in a preload
script where the electron but would be triggered, and falls back to a
simple replacement for debug. However, in the main and renderer
processes, where a preload script is not being used, we still get full
debug functionality.
In this way, a module can be used both in a preload script and outside
of it, while still preserving debug functionality whenever possible.
Signed-off-by: Kristóf Marussy <kristof@marussy.com>
|
| |
|
| |
|
|
|
|
| |
(#2228)
|
|
|
|
| |
https://github.com/getferdi/ferdi/issues/1280
WhatsApp-like services can set active dialog title to the app title eg. Ferdi - WhatsApp - Contact Name
|
|
|
| |
Co-authored-by: Markus Hatvan <markus_hatvan@aon.at>
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
| |
implements getferdi/recipes#721 (eg: office365-owa)
|
| |
|
| |
|
|
|
|
| |
Added some error checking
|
| |
|
| |
|
|
|
|
| |
services (#1820)
|
|
|
|
| |
(since we are always premium/free)
|
| |
|
|
|
|
|
|
| |
Users can now search for the services not only based on the name, but also the alias (partial match,
case-insensitive), when adding new services.
fixes #1614, #1615, #1291
|
| |
|
|
|
|
| |
(Addendum to getferdi/recipes#605)
|
|
|
|
|
|
| |
'require' (#1752)
* refactor references to 'userData' and 'appData' directories to move hardcoding into single location
* convert to es6 for lower memory usage as per https://codesource.io/the-difference-between-import-and-require-in-javascript/
|
|
|
|
|
|
|
|
|
|
| |
- update .eslintrc to work for .js and .ts
- update devDependencies
- lint properly both root /src and nested /packages
- update webhint recommended setting for tsconfig.json to shrink output
- Manage all eslint rules from the repo root
- escape single quotes in scripts to please windows build
Co-authored-by: Vijay A <avijayr@protonmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Enable service contextIsolation
* Enable contextIsolation on the service webviews
* Expose a new API window.ferdi in the service main world to allow
calling back into the service isolated world
* Expose a new IPC message inject-js-unsafe from the service isolated
world to execute Javascript in the service main world (i.e., run code
without context isolation). While the name contains the "unsafe"
suffix to show the lack of context isolation, this should mostly be
safe, as no nodejs APIs are available in the injected code.
* Refactor the Notifications shim into a part in the isolated world that
handles displaying and modifying notifications, and a shim in the main
world for the Notifications class. The two communicate via the
window.ferdi endpoint and a Promise object can be used to detect
notification clicks.
* Refactor the screen sharing shim into a part in the isolated world
that enumerated shareable screens and windows and a shim in the main
world that displays the media selector and completes the media
selection promise.
* Expose the injectJSUnsafe API to recipes to inject javascript code
into the main world without context isolation.
* Expose setBadge to the main world
The window.ferdi.setBadge API can be used to update the service badge
from injected unsafe Javascript
* Safer script injection into the service main world
Make sure that we don't try to serialize stray objects back from the
main world to the isolated world by always surrounding the script to be
executed by an anonymous function.
* Always read recipe assets as utf8
* Remove window.log from recipes
We didn't use it anywhere and its behavior was confusing in production
mode.
* Inject multiple unsafe scripts at the same time
* Find in page without remote module
Remove the @electron/remote dependency from the find in page (Ctrl+F)
functionality. The remote webContents is replaced with Electron IPC.
Synchronous IPC messages are handled in the main Electron process,
because the renderer process cannot reply to IPC messages synchronously.
* Update to latest contextIsolation recipes
* Fixing issue with missing 'fs' functions.
Co-authored-by: Vijay A <avijayr@protonmail.com>
|
|
|
|
| |
hibernated service (#1680)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Use hibernation strategy from settings instead of hardcoded 5 minutes
* Fix conditions with isHibernationEnabled, previously disableHibernation
* Make service hibernation obey global setting
Also refactors hibernation to move some hibernation enablement logic
into the Service model
* Remove global hibernation enable switch
Implements option 4 from
https://github.com/getferdi/ferdi/pull/1486#issuecomment-860290992
according to
https://github.com/getferdi/ferdi/pull/1486#issuecomment-876558694
* Implements #865 : Add 'hibernate service' and 'wake up service' in the sidebar context menu.
* Removed 'hibernationEnabled' check on main settings screen
Since this is an (imo) incongruous behavior for the first time user.
They will see a message, but with no ability to choose the hibernation strategy.
* Autogenerated files from conflict fixes
Co-authored-by: Kristóf Marussy <kristof@marussy.com>
Co-authored-by: Vijay A <avijayr@protonmail.com>
|
|
|
|
|
|
|
|
| |
- serviceLimit
- planSelection
- trialStatusBar
and other Franz features that were for different tiers of subscription.
|
|
|
|
| |
manually building it. (#1592)
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
* Upgrade various dependencies to latest, remove unnecessary electron-hunspell
- upgrade eslint and friends to latest
- remove deprecated 'node-sass' in favor of 'sass'
- disable new rules from 'eslint-config-airbnb' that are conflicting with current code style
- add workspace config for 'vscode' that silences 'experimentalDecorator' warning and forces 'prettier' to single quote
* Run yarn lint to autofix with new ruleset and worked down lint issues to zero
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Introduced a global and a service-specific 'user-agent' value that can be set by the user from the preference section.
The global override is in Advanced, while the service override was moved to the bottom of the service editor form (and now occupies the full width of the form).
Show 'restart reqd' prompt for user-agent setting.
The order of the user-agent is the following:
1. Any request header overrides in the recipe will still override the user agent string.
2. If the user has set a user agent override for the service, use that.
3. If the recipe has a predefined user agent, use that.
4. If the user has set a global user agent override, use that.
5. Otherwise, use the user agent string corresponding to the underlying Electron version.
If the current webpage is the Google login form, we still remove the Chrome version number from the user agent string.
The value that would be used if no custom override is set (recipe-provided value or global override for the service, current Electron instance for the global override) is shown as a placeholder in the input box on the corresponding setting screen.
Co-authored-by: Kristóf Marussy <kristof@marussy.com>
|