We now create a new instance of atom::api::DesktopCapturer for every
request instead of weirdly re-using the same instance and queuing
requests. This means there is now a 1:1 relationship between request
and DesktopCapturer so there isn't a race condition between the observer
for one request calling back before the observer of another. This is an
issue ever since the backing APIs moved to worker threads.
This also does a few things to ensure memory management
* Only ever listen to one event per-request, after that we wipe the emit
function to ignore all future events
* Ensures we clean up the window_capturer_, screen_capturer_ and
captured_sources_ in native land once the request is over.
This _in theory_ fixes a flake we've been seeing on CI where we try to
resolve the promise for a request that no longerr exists.
* spec: clean up after a failed window count assertion
Previously when this assertion failed all tests that ran after the
failed assertion also failed. This ensure that the assertion fails for
the test that actually caused the issue but cleans up the left-over
windows so that future tests do not fail.
* fix: maintain a ref count for objects sent over remote
Previously there was a race condition where a GC could occur in the
renderer process between the main process sending a meta.id and the
renderer pulling the proxy out its weakmap to stop it being GC'ed.
This fixes that race condition by maintaining a "sent" ref count in the
object registry and a "received" ref count in the object cache on the
renderer side. The deref request now sends the number of refs the
renderer thinks it owns, if the number does not match the value in the
object registry it is assumed that there is an IPC message containing a
new reference in flight and this race condition was hit.
The browser side ref count is then reduced and we wait for the new deref
message. This guaruntees that an object will only be removed from the
registry if every reference we sent has been guarunteed to be unreffed.
By default the Chromedriver will send remote-debugging-port=0 to let the
browser choose a free port to listen on. The chosen port is written to
a known file in the user data dir that is passed to the app through the
CLI.
This PR does two things.
1. Correctly passes the USER_DATA_DIR to the remote debugging server so
it knows where to write the file
2. Adds support for --user-data-dir as we did not support that CLI
argument and Chromedriver relies on being able to tell the "browser"
where to write this file.
Fixes#17354
* refactor: use mojo for electron internal IPC
* add sender_id, drop MessageSync
* remove usages of AtomFrameMsg_Message
* iwyu
* first draft of renderer->browser direction
* refactor to reuse a single ipc interface
* implement TakeHeapSnapshot through mojo
* the rest of the owl^WtakeHeapSnapshot mojofication
* remove no-op overrides in AtomRendererClient
* delete renderer-side ElectronApiServiceImpl when its pipe is destroyed
* looks like we don't need to overlay the renderer manifest after all
* don't try to send 2 replies to a sync rpc
* undo changes to manifests.cc
* unify sandboxed + unsandboxed ipc events
* lint
* register ElectronBrowser mojo service on devtools WebContents
* fix takeHeapSnapshopt failure paths
* {electron_api => atom}::mojom
* add send_to_all to ElectronRenderer::Message
* keep interface alive until callback is called
* review comments
* use GetContext from RendererClientBase
* robustify a test that uses window.open
* MessageSync posts a task to put sync messages in the same queue as async ones
* add v8::MicrotasksScope and node::CallbackScope
* iwyu
* use weakptr to api::WebContents instead of Unretained
* make MessageSync an asynchronous message & use non-associated interface
* iwyu + comments
* remove unused WeakPtrFactory
* inline OnRendererMessage[Sync]
* cleanups & comments
* use helper methods instead of inline lambdas
* remove unneeded async in test
* add mojo to manifests deps
* add gn check for //electron/manifests and mojo
* don't register renderer side service until preload has been run
* update gn check targets list
* move interface registration back to RenderFrameCreated
* fix: match net module headers & http.IncomingMessage headers
* update net doc for cleanliness
* address feedback from review
* Update spec/api-net-spec.js
Co-Authored-By: codebytere <codebytere@github.com>
* add special cookie case
* feat: add support for content scripts 'all_frames' option
* merged content script tests
'all_frames' test now runs on all variants of sandbox/contentIsolation configurations :D
`powerMonitor.querySystemIdleState` and `powerMonitor.querySystemIdleTime` had async backing APIs in chromium (https://chromium-review.googlesource.com/c/chromium/src/+/1379183). However, that has changed in ch73. So, this PR deprecates the old async APIs and adds new sync APIs.
* chore: refactor browser IPC into typescript
* chore: refactor app.ts into Typescript
* Refactors app.dock into cpp
* Removes app.launcher which has not existed for 3 years
* Removes 2 deprecated APIs (that have been deprecated for more than one
major)
* Refactors deprecate.ts as well
* feat: promisify webContents.savePage()
* no need to make compatible w callbacks, we're breaking it
* fix promise resolve type
* address feedback from review
* fix promise return text
* update smoke test
* feat: promisify the Cookie API
* chore: update specs to test promisified cookies
* chore: add deprecate wrapper for cookie callback API
* docs: update docs to cookie promise changes
* chore: remove redundant namespace use
* docs: improve cookie example
* docs: restore docs for cookie callback API
* chore: restore cookie callback tests
* fix: syntax of cookie promise return types
* feat: add support for node / preloads in subframes
This feature has delibrately been built / implemented in such a way
that it has minimum impact on existing apps / code-paths.
Without enabling the new "nodeSupportInSubFrames" option basically none of this
new code will be hit.
The things that I believe need extra scrutiny are:
* Introduction of `event.reply` for IPC events and usage of `event.reply` instead of `event.sender.send()`
* Usage of `node::FreeEnvironment(env)` when the new option is enabled in order to avoid memory leaks. I have tested this quite a bit and haven't managed to cause a crash but it is still feature flagged behind the "nodeSupportInSubFrames" flag to avoid potential impact.
Closes#10569Closes#10401Closes#11868Closes#12505Closes#14035
* feat: add support preloads in subframes for sandboxed renderers
* spec: add tests for new nodeSupportInSubFrames option
* spec: fix specs for .reply and ._replyInternal for internal messages
* chore: revert change to use flag instead of environment set size
* chore: clean up subframe impl
* chore: apply suggestions from code review
Co-Authored-By: MarshallOfSound <samuel.r.attard@gmail.com>
* chore: clean up reply usage
* chore: fix TS docs generation
* chore: cleanup after rebase
* chore: rename wrap to add in event fns
Unify the behavior between default app and packaged apps:
- create default application menu unless the app has one
- default window-all-closed handling unless the app handles the event
* fix: use Chromium's determined new site instance as candidate when navigating.
When navigating to a new address, consider using Chromium's determined site instance
for the new page as it should belong to an existing browsing instance when the
navigation was triggered by window.open().
fixes 8100.
* Revert "fix: use Chromium's determined new site instance as candidate when navigating."
This reverts commit eb95f935654a2c4d4457821297670836c10fdfd5.
* fix: delegate site instance creation back to content when sandboxed.
* fix: ensure site isolation is on
* test: adapt ut for cross-site navigation
* fix: register pending processes during a navigation.
* refactor: dont call loadURL for a window constructed from an existing webContents.
* test: add sandboxed affinity UT's.
* fix: check affinity before deciding if to force a new site instance.
* chore: adapt subsequent patch.
* refactor: constify logically const methods.
* fix: do not reuse site instances when navigation redirects cross-site.
* test: ensure localStorage accessible after x-site redirect.
* test: adapt localStorage acess denied UT for site isolation.
* fix: do not send render-view-deleted for speculative frames.
* chore: amend tests after rebase.
* test: add ut for webContents' render-view-deleted emission
* fix: introduce current-render-view-deleted for current RVH's deletions.
Revert render-view-deleted to being emitted with any RVH's deletion.
current-render-view-deleted is emitted only when the RVH being deleted
is the current one.
* refactor: style and comments fixed.
* feat: add registerAccelerator flag to allow menu items to skip registration
* docs: add docs for registerAccelerator
* docs: re-add accidentally removed line
Extend the existing win.setBounds functionality by allowing developers to partially update bounds without being forced to pass in all four bounds values. No existing functionality is altere
* refactor: remove -new-contents-created event
Chromium expects us to take ownership of WebContents in AddNewContents,
we should not create V8 wrapper in WebContentsCreated, otherwise we
would have WebContents being managed by 2 unique_ptr at the same time.
* refactor: make CreateAndTake take unique_ptr
* Fix loading of devtools extensions on startup
The persisted DevTools Extensions were not being loaded correctly at startup. The `addDevToolsExtension` function was not defined when it was being called. An error was being thrown and ignored, so the whole thing would fail silently. I moved the code to load the extensions to the end of the event handler, so now it works.
* fixup: remove trailing spaces to unblock CI
* fixup: add logging when the Electron Enable Logging env var is set
* Fix linter error on undefined srcDirectory
* fixup: catch exception when loading extension
* Revert "fixup: catch exception when loading extension"
This reverts commit 42c2cf95bcaab8abfc5fbecbe4365d3adfe36d5b.
* remove printing related things from chromium_src
* chore: add printing build flag and patch
* fix: include PrintingService on other platforms too
* fix: printing_handler is only needed on Windows
* fix: format BUILD.gn properly
* fix: rename printing build flag to avoid conflict with chromium
* fix: place previously missed printing calls behind build flag
* fix: accidentally renamed flag in patch file
* fix: don't include all printing strings
* fix: allow ShowItemInFolder and OpenItem to block, fixing a DCHECK crash
* fix: make things compile, some changes got lost while rebasing
* fix: remove rogue line from BUILD.gn
* chore: update patch description
* style: lint fix
* chore: use chromium printing buildflag, move node related stuff out of patch
* revert: remove ScopedAllowBlockingForTesting call
* fix: fix my rebase blooper
* fix: re-add header lost during rebase, update patch
* fix: add <map> include, tweak the patch a bit
* revert: remove rogue diff from patch
* fix: clean up after rebase
* [ci skip] refactor: create request context from network context
* [ci skip] refactor: subscribe to mojo cookiemanager for cookie changes
* [ci skip] refactor: manage the lifetime of custom URLRequestJobFactory
* refactor: use OOP mojo proxy resolver
* revert: add support for kIgnoreCertificateErrorsSPKIList
* build: provide service manifest overlays for content services
* chore: gn format
* fix: log-net-log switch not working as expected
* spec: verify proxy settings are respected from pac script with session.setProxy
* chore: use chrome constants where possible
* fix: initialize request context for global cert fetcher
* refactor: fix destruction of request context getters
* spec: use custom session for proxy tests
* fix: queue up additional stop callbacks while net log is being stopped
* fix: Add CHECK for cookie manager retrieval
* chore: add helper to retrieve logging state for net log module
* fix: ui::ResourceBundle::GetRawDataResourceForScale => GetRawDataResource
* style: comment unused parameters
* build: move //components/certificate_transparency deps from //brightray
* chore: update gritsettings_resource_ids patch
* chore: update api for chromium 68
* fix: net log instance is now a property of session
* perf: use an internal module resolver instead of relative requires
* perf: memoize the results of getting exported Electron properties
* perf: make internal module changes consistent across sandboxed / bundled files
* remove obsolete tests
* remove unused deprecate API
* make a warnOnce helper for the deprecate methods
* misc. copyediting, e.g. variable names, whitespace
* test that any deprecation warns once at most
* use strict
* fix: use render client id to track deleted render process hosts
Instead of relying on OS process id, which may not be unique
when a process is reused, we rely on the renderer client id
passed by the content layer when starting the renderer process
which is guaranteed to be unique for the lifetime of the app.
* fix: store context id as int64_t
Ensuring that it doesn't wrap easily with a large number
of context creation on some malformed web pages.
* feat: deprecate default value of nodeIntegration
* Use DeprecationStatus::Stable as the default instead of shadowing
* change wording of deprecations
* chore: also deprecate kWebviewTag and kContextIsolation
* chore: do as we preach, lets be secure-by-default in the default app
RenderProcessHost switch can happen between ipc calls when
speculative process are invvolved, which will lead to deletion
of entries on current context. Use OS process handles to
uniquely associate a destruction handler for a render process.
* fix: use OOIF for webview tag
* fix: do not call GetNativeView for webview
* fix: OOIPF webview's WebContents is managed by embedder frame
* fix: guest view can not be focused
* fix: clear zoom controller when guest is destroyed
* fix: implement the webview resize event
The webview is no longer a browser plugin with the resize event, use
ResizeObserver instead.
* test: disable failed tests due to OOPIF webview
* fix: embedder can be destroyed earlier than guest
This happens when embedder is manually destroyed.
* fix: don't double attach
* fix: recreate iframe when webview is reattached
* fix: resize event may happen very early
* test: some tests are working after OOPIF webview
* chore: remove unused browser plugin webview code
* fix: get embedder via closure
When the "destroyed" event is emitted, the entry in guestInstances would be
cleared.
* chore: rename browserPluginNode to internalElement
* test: make the visibilityState test more robust
* chore: guestinstance can not work with OOPIF webview
* fix: element could be detached before got response from browser
After after using `processId-contextCounter` as contextId, it may happen
that contexts in different WebContents sharing the same renderer process
get the same contextId. Using webContentsId as part of key in
ObjectsRegistry can fix this.
Replaces 'did-finish-load' with 'did-stop-loading' which semantically
maps to the events inside Chromium. Before I think we were relying
on a natural 99% winnable race condition.
Fixes#13504
After the page does navigations, garbage collection can still happen in
the old context. This commit changes to store references to remote objects
by _pages_, instead of by _WebContents_.
* Add promise helper and change whenReady to be native impl
* remove commented code
* add GetInner helper to dedupe promise code
* add Promise.reject helper to be consistent with JS
* fix linting
* update promise impl per feedback
* remove param name from unused isolate
* Use non-depreceated resolvers for promises
* Add thread dchecks for promise helper, intiialize promise pointer to nullptr
* Introduce `net.{start|stop}Logging()`
- Slight regression right now as Electron won't automatically start logging net-logs at launch, will soon be fixed
- To implement callback for async controls
* Add `net.isLogging` & optional callback param for `net.stopLogging()`
* Fix small regression on --log-net-log
--log-net-log should work again
* Error on empty file path
* Only start with valid file path
* Remove unused var
* Allow setting log file path before URLRequestContextGetter starts logging
* Add net log tests
* Remove redundant checks
* Use brightray::NetLog
* Clean up code
* Should automatically stop listening
* 🎨 Attempt to fix styles
* Only run non-null callback
* Dump file to tmpdir
* Simplify net log spec
Spawned Electron process on Linux CI can fail to launch
* Separate netLog module
* Remove net logging test from net spec
* Add tests for netLog
* Fix header guard
* Clean up code
* Add netLog.currentlyLoggingPath
* Callback with filepath
* Add test for case when only .stopLogging() is called
* Add docs
* Reintroduce error on invalid arg
* Update copyright
* Update error message
* Juggle file path string types
* Make it possible to disable a module for a renderer
* Put DesktopCapturer API under a build flag
The name is "enable_desktop_capturer".
Enabled by default.
* Don't use JSON to send the result of `ipcRenderer.sendSync`.
- Change the return type of AtomViewHostMsg_Message_Sync from `base::string16`
to `base::ListValue`
- Adjust lib/browser/api/web-contents.js and /lib/renderer/api/ipc-renderer.js
to wrap/unwrap return values to/from array, instead of
serializing/deserializing JSON.
This change can greatly improve `ipcRenderer.sendSync` calls where the return
value contains Buffer instances, because those are converted to Array before
being serialized to JSON(which has no efficient way of representing byte
arrays).
A simple benchmark where remote.require('fs') was used to read a 16mb file got
at least 5x faster, not to mention it used a lot less memory. This difference
tends increases with larger buffers.
* Don't base64 encode Buffers
* Don't allocate V8ValueConverter on the heap
* Replace hidden global.sandbox with NodeBindings::IsInitialized()
* Refactoring: check NodeBindings::IsInitialized() in V8ValueConverter
* Refactor problematic test to make it more reliable
* Add tests for NaN and Infinity
Use a single synchronous IPC call to retrieve data required by early
sandbox scripts. This has two purposes:
- Optimize preload script initialization by:
- Using one synchronous IPC call to retrieve preload script,
webContentsId (more on that later), process.{platform,execPath,env}
- Lazy loading as many modules as possible.
- Fix#12316 for sandbox. @MarshallOfSound addressed the issue in
#12342, but it was still present in sandbox mode. By loading
webContentsId very early and skipping remote module at early
startup, we fix it for sandbox.
- introduce a currentPlatformSupportsAppIndicator() function determining
if app indicators are supported here.
- handle undefined process.env.XDG_CURRENT_DESKTOP
- added some comments to ensure that the intents are clear
Thanks MarshallOfSound
