| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
| |
Due to sandboxing we can no longer pass pac file.
Pass it as data url instead by reading the pac file and
encoding into base64.
Fix failing proxypac url on windows.
Fixes: QTBUG-81557
Change-Id: I3dc3da4fbd3cce4e903c75022b8e9fe5faf71604
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Mojo data pipes are non-blocking, meaning we have to wait until there's room in
a buffer before we can transfer data from the QIODevice to the pipe.
Use mojo::SimpleWatcher to monitor the pipe for readiness and use the two-phase
BeginWriteData/EndWriteData API to let the QIODevice write directly into the
pipe's internal buffer, avoiding a copy.
Fixes: QTBUG-82244
Change-Id: I65e69ce72d0e99bc047c57b5a22531c0891c553a
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|\
| |
| |
| |
| |
| |
| |
| |
| | |
Blacklisted one test requiring an update of qt5.git.
Conflicts:
.qmake.conf
Change-Id: I75e55a1c5f8840cde55ddb60d632287b2affadeb
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Setting an empty string instead of "()" fixes that case where the file
input doesn't have "accept" attribute.
Task-number: QTBUG-82109
Change-Id: I8a72f819fa6d8bbab4e5f1067b38ad75ff11e118
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Ammends 445235bc01. Preserve some previous assumptions on order of
change events for url and title. Fixes flaky failures in previously
written tests like WebEngineViewSource::test_viewSource.
Fixes: QTBUG-81855
Change-Id: I487d27d594d5a0d74d39b7b58e815e5c75a73fb3
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
These features all failed to work with our old compositing
implementation, but works with viz compositor, and due to being default
in Chrome, might work more reliable being on, than off.
Change-Id: Ifab1913fd4bc20f295dc7c2ccffa510610b23796
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
Update the headers as asked for in FollowRedirect.
Change-Id: I86d241d52abe8bd9d082b7ad49e921a955dc1403
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
|\ \ |
|
| |\|
| | |
| | |
| | | |
Change-Id: I349a4ecbbd9d3d121ca6564db77e417872246554
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
We changed RootCompositorFrameSinks to be destroyed asynchronously (in
HostFrameSinkManager::InvalidateFrameSinkId) which means that one can still
exist during shutdown in GpuThreadControllerQt::destroyGpuProcess. This results
in a deadlock in single threaded GPU mode: in destroyGpuProcess we wait for the
viz thread to exit, but the FrameSinkManagerImpl on the viz thread will try to
destroy the RootCompositorFrameSink, which waits for work to be done on the
GPU=UI thread, which is waiting for the viz thread to exit.
Fix by destroying all RootCompositorFrameSinks before destroyGpuProcess.
Change-Id: I4cf135f29b90ae0bf78525d5747567dc10a775e6
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
macOS Accessibility queries the window for the focused accessibility
element. The window forwards the query to the widget with active focus.
This widget is the RWHVQtDelegateWidget if a web element is focused
in QWebEngineView. Therefore, a QAccessibleWidget interface has been
implemented for the RWHVQtDelegateWidget to forward the request to the
QWebEngineView.
The focused accessibility element expected to be returned by the
QAccessibleInterface::focusChild() method. In case of the macOS accessibility
backend, it is called by the accessibilityFocusedUIElement() NSAccessibility
API function. It expects the focused web accessibility element otherwise
VoiceOver won't focus properly.
The focused web accessiblity element is looked up by the new
BrowserAccessibilityQt::focusChild() method.
RenderWidgetHostviewQtDelegateWidget::focusChild() and
QWebengineViewAccessible::focusChild() methods have been also implemented
to forward it.
This patch depends on a focusChild() fix in qtbase:
a132e02540 Fix QAccessibleWidget::focusChild() to return focused descendant
Microsoft Narrator also uses focusChild() to query the current focused
element when it starts but it is still functional without this fix.
Task-number: QTBUG-78284
Task-number: QTBUG-81539
Change-Id: I3c4861e58622ccbb5046c60c4efcc19842400a88
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
|/ /
| |
| |
| |
| |
| |
| |
| | |
Removes much of the specific code for non-networksupport, classes that
won't work or even compile with 78/79 based.
Change-Id: I150794a55164998fa3d0031757718081fa7d7d9a
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This can useful for e.g. implementing something like the "Task manager" in
Chromium or otherwise interacting with the render process (e.g. to kill it for
some reason while debugging).
[ChangeLog] Add a renderProcessPid() getter to (Q)WebEnginePage which allows
getting the process ID of the underlying render process.
Change-Id: Id5d59be9b6bd46ffc3a6aa480cb5ff7bd3b8aa31
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Task-number: QTBUG-79890
Task-number: QTBUG-81556
Task-number: QTBUG-81557
Task-number: QTBUG-81614
Change-Id: I4c81ca1b1fb9c9595c1f670706948fcbe65bcd33
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | | |
Initial merge of QtPdf into QtWebEngine.
Fixes: QTBUG-69519
Change-Id: I48dc25a59f2c161bb231bd0fa60392eb70fe4e7d
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This change adds two new modules to qtwebengine repository.
New modules do not depend on webengine module, however webengine
chromium source code and Chromium "gn" configuration is
required to build QtPdf.
Adding two unrelated modules to webengine might look crazy:
however sharing gn build configuration and Chromium code base
with necessary qt adaptations simplifies code maintenance and
minimises required code checkouts. Back porting of security
patches for Chromium also affects Pdfium. Moreover, Pdfium is no
longer a separate project, but integrated into Chromium:
therefore moving it out of Chromium source tree would require
extra effort.
Rename webengine-core feature to build-qtwebengine-core, this
makes consistent feature naming with build-qtpdf
At the moment two new modules have integrated build, with possible
shortcuts:
qmake -- --no-build-qtwebengine-core
qmake -- --no-build-qtpdf
Webengine build is disabled by default now.
Change-Id: Iac3d9927d51f3ac316db0148d275eda843dcc19b
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This is essential when Chromium tries to load media files in multiple
jobs
over custom protocols, like qrc. Allow subsequent jobs to continue
reading
media files from specified positions to avoid media glitches and errors.
Implemented with network service.
Change-Id: I8df963ce94015ff65316dd6e0b6f2fcc77c20427
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\ \ \
| | |/
| |/|
| | | |
Change-Id: I1230ec3a854c0798fa4f7960642e988b917f431e
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Fixes: QTBUG-81521
Change-Id: I8ca82224cd834b667471d1e96a44430164d3669e
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| |\ \ |
|
| | |\ \
| | | | |
| | | | |
| | | | | |
Change-Id: Ic2b5f2a3e6f5af56d92652e57c05a631481201b3
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Change-Id: Ifb5f7c866c26746f989ab4e8c49c48bcbdb50077
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This is essential when Chromium tries to load media files in multiple jobs
over custom protocols, like qrc. Allow subsequent jobs to continue reading
media files from specified positions to avoid media glitches and errors.
Task-number: QTBUG-80234
Change-Id: I9a7e98c0cb08b2399b7928ecf026c0deb90a1bcb
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Windows IME does not support hidden text therefore IME input is disabled
on password fields. The shortcuts are supposed to be overridden in input
fields. Checking the keyboard focus on an input field is done by
verifying if the IME is enabled. This won't work with password fields on
platforms where hidden text is not supported, so also check if the
Qt::ImhHiddenText IME hint is set.
Fixes: QTBUG-81206
Change-Id: I81870beb556a9dda67295496dad8b672fbc5eba2
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Task-number: QTBUG-78284
Change-Id: Ie3bf247752308fb104ab0f244736bd3a8d070762
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Selective application of clang-format.
Change-Id: Ied39e0bcf842219b789354bd14ff8f8f18e8fb60
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
|\ \ \ \
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | | |
Conflicts:
.qmake.conf
Change-Id: I2f035e56cc23bdb600f93d601f0c1163ba32a1d5
|
| |\| |
| | | |
| | | |
| | | | |
Change-Id: Ibc671c7a5ac4b070f3406c41598d071fd978e420
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
TouchCancel events have an empty touchPoints() list, which first trips
when accessing touchPoints[0], and later on crashes Chromium if we pass
the empty list to m_touchSelectionController.
Rework handleTouchEvent() to route TouchCancel events like other touch
events, and make sure we pass a non-empty touchpoints list to Chromium.
Task-number: QTBUG-80893
Change-Id: Ie8396a1191f72b5bbb2b047f131794b37cfded48
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
| |/ /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Due security changes to prevent url spoofing, our implementation
is getting extra invalidate url requests. Unfortunately, this breaks our
url handling, which now gets lots of new back and fort url changed signals and
make several unit test failures. After tedious investigation of Chromium
omnibox handing and trying out different approaches, it seems that
only sensible solution is to follow Chromium logic and make
NavigationStateChanged to update 'ui' in asynchronous matter.
This change tries not break any tests and simplify url handling.
The only side effect of this change is that WebEnginePage::setContent will
get extra 'url' signal of initial 'urlData' and later 'baseUrl' change
is emitted.
Fix one of qml tests which did not expect to have url on LoadStartedStatus.
Task-number: QTBUG-63388
Task-number: QTBUG-48995
Change-Id: Id347f4325c036e16bfae7bf2f694905e0f21f8d7
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Was bool but it's really a string.
Change-Id: I79eace4892e8048212ce3296b28e342d72c43667
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | | |
Task-number: QTBUG-74602
Change-Id: I8ba358e03472e58c40363fede56e87dfd665ae83
Reviewed-by: Tony Sarajärvi <tony.sarajarvi@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Enable sandboxing on Windows. Enable heterogeneous sandbox symbol
resolution and statically link the sandboxing code into the helper
process. This means we have two copies of the sandboxing code,
one statically linked in the executable and one in the shared
library. Since they are not exported they don't conflict, but
we need to take to initialize the right version in the helper process
binary, and pass its sandbox interface to shared library using it.
For sandbox debug output, we also need to initialize the second copy
of the commandlineparser and logging system.
Fixes: QTBUG-51170
Change-Id: I8f503c8d6b40674465f32772ef906817dad2b449
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Follow Chromium's aura, android, windows, and fuchsia implementations of
GetNativeImageNamed by forwarding the call to GetImageNamed.
Fixes DCHECK when enabling TouchEmulator in devtools.
Change-Id: If5fdc30293bc18843c85d932f75f7cb2c70e5815
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
FlingScheduler, in viz mode, is not able to progress the animation unless it has
access to our ui::Compositor in RHWVQt for observing OnAnimationStep.
Add a GetCompositor virtual method to RHWV and override in RHWVQt.
Fixes: QTBUG-80089
Change-Id: I4f52a83d29f5ae4c7c9e8cc4908384d1c7e81a82
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Extension resources are now loaded through ExtensionsBrowserClientQt's
LoadResourceFromResourceBundle, and permission to load extra schemes
is now handled by installing subresource URL loaders.
Change-Id: Id0445088607c9be019fbfc134db4e60b94e54479
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This should also work with network-service
Change-Id: I1d38112b76f9a656cc50ffbb826db28b41cf2f85
Fixes: QTBUG-80462
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Selective application of clang-format in src/core/renderer_host
Change-Id: I6fc78374be914dca15b15ab8b8f4b7bed83f21af
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
We now get LaunchURL calls for the RPH schemes and have to manually filter them.
Change-Id: Ia3f3e9cde392f95d2247899640f4551ae0b9f1a5
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | | |
Change-Id: If51e4ef40172ade46d776a384b449b22afbe73c4
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
QVariant seems to be missing from the indirect includes now.
Change-Id: Ic945b99bbe34726b28d1e272988d8cb0d4c2ea76
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|\| |
| | |
| | |
| | |
| | | |
Fixes: QTBUG-80555
Change-Id: I4c7f69f697c09526f5f927948e6a6bf1c43fac17
|
| |\ \
| | | |
| | | |
| | | | |
Change-Id: I4f73d4b11bee795185d4eaae718d4cfdb3112100
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Fixes regression with Amazon Prime and Netflix.
Change-Id: Ic96b4fa94fd06bfb9f7d75ca36e7eeddcda81218
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Chromium's DownloadManager doesn't create its download items before
path for saving page is confirmed. So assert inside updateDownloadPath
was not correct. Moreover, the name is confusing because it's not really
updating anything. Remove it and use ProfileAdapterClient::DownloadInfo
timestamp to determine updated filename after directory change.
Ammends recent new api for changing download directory 0884fab3b1.
Fixes: QTBUG-80372
Change-Id: If9efb52979deb3cf21fc4e12989173c85e04e090
Reviewed-by: Jüri Valdmann <juri.valdmann@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If the new test became flaky it might happen because the first text
search finished before the second findText() call. This is very
unlikely, but in this case the test should be modified to not to check
if the first find failed. The point is to check we get the correct
amount of signals and the second search doesn't assert.
If the callbacks will be removed in Qt6, it should be re-considered to
remove the "unfinished find" workaround and trigger the first successful
findTextFinished() signal even if it happens in the middle of another
search.
Fixes: QTBUG-80086
Change-Id: I9c1ce20fc43fd81e8af784385a00ac2e7f7603b7
Reviewed-by: Kirill Burtsev <kirill.burtsev@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
With the adaptations for Chromium 76, RenderWidgetHostViewQt was changed to
become a RenderProcessHostObserver with the renderProcessTerminated signal being
emitted from the override of RenderProcessHostObserver::RenderProcessExited.
The problem with this can be seen by setting a breakpoint on the
RenderProcessGone override in RenderWidgetHostViewQt. We then get the trace:
QtWebEngineCore::RenderWidgetHostViewQt::RenderProcessGone
content::RenderWidgetHostImpl::RendererExited()
content::RenderViewHostImpl::RenderProcessExited
content::RenderProcessHostImpl::ProcessDied
ProcessDied iterates over all the observers and calls RenderProcessExited. Both
the RenderViewHostImpl and our RWHVQt are observers, but the RVHImpl comes
first. The RVHImpl then calls RendererExited, which calls our RenderProcessGone,
which does a 'delete this'. Now our RenderProcessExited override can never be
called because we have already deleted our observer.
Fix by moving the RenderProcessGone code to WebContentsDelegateQt and getting
the exit code from WebContents::GetCrashedErrorCode. Also add test.
Task-number: QTBUG-80085
Change-Id: I434744286df97a37b64722d7c15a1d4ee11c8af6
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Fixes: QTBUG-78280
Change-Id: Ic202314bb7935741791fa8c747e255b10dc7dc61
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Reviewed-by: Florian Bruhin <qt-project.org@the-compiler.org>
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Re-enables using proprietary codecs used by many popular platforms
for streaming video.
Fixes: QTBUG-80352
Change-Id: Id1730576a98cfc4b23b523e61c1d8168414e3a48
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| | |\ \ |
|