QUrl::toString reverses the Unicode->ASCII conversion that already
occurred here. The text of m_location_edit is already in the format we
expect, so let's just convert QString->AK::URL directly, instead of
taking the detour QString->QUrl->AK::URL
Some Wayland compositors have support of fractional-scale-v1 protocol.
The protocol allows compositor to announce a preferred fractional scale
on a per-wl_surface basis. Qt forwards these Wayland events to an
application using a usual DevicePixelRatioChange event. However, in
contrast to the other platforms, this DevicePixelRatioChange event is
issued directly on widgets and not screens. Additionally, the exact
fractional scale is stored in QWindow object and not the current screen.
Note that in theory it is possible to obtain per-screen fractional
scaling on Wayland by interpolating data provided by wl_output and
xdg_output events but qtwayland does not do that.
If fractional-scale-v1 is not available, qtwayland will still fire
per-Widget DevicePixelRatioChange events, but, obviously, with the
per-screen possibly larger ceiled scaling.
This whole thing makes handling DPI changes on Wayland really simple.
All we need to do is to intercept DevicePixelRatioChange events firing
on QWindow objects and call the old device_pixel_ratio_changed handler
with the window's devicePixelRatio(). The only caveat here is not forget
to always set QWidget's parent before calling devicePixelRatio() on it.
Previously, on systems where pressing Enter would generate "\r\n", only
the '\r' character was being sent to the event handler. This change
ensures consistent behavior across platforms regardless of their native
line ending characters.
The IPC layer between chromes and LibWeb now understands that multiple
top level traversables can live in each WebContent process.
This largely mechanical change adds a billion page_id/page_index
arguments to make sure that pages that end up opening new WebViews
through mechanisms like window.open() still work properly with those
extra windows.
We currently bundle AK with LibCore on Lagom. This means that to use AK,
all libraries must also depend on LibCore. This will create circular
dependencies when we create LibURL, as LibURL will depend on LibUnicode,
which will depend on LibCore, which will depend on LibURL.
Instead of spawning these processes from the WebContent process, we now
create them in the Browser chrome.
Part 1/N of "all processes are owned by the chrome".
These IPCs are different than other IPCs in that we can't just set up a
callback function to be invoked when WebContent sends us the screenshot
data. There are multiple places that would set that callback, and they
would step on each other's toes.
Instead, the screenshot APIs on ViewImplementation now return a Promise
which callers can interact with to receive the screenshot (or an error).
According to Qt documentation, destruction of a QObject's children may
happen in any order. In our case, the Tab's WebContentView is deleted
before its InspectorWidget. The InspectorWidget performs cleanup on that
WebContentView in its destructor; this causes use-after-free since it
has already been destroyed.
From reading Qt threads, if a particular destruction order is required,
it is okay to enforce that order with manual `delete`s. This patch does
so with the InspectorWidget to ensure it is deleted before the
WebContentView. Qt's object ownership is okay with this - it will remove
the InspectorWidget from the Tab's children, preventing any double
deletion.
With this change, chrome no longer has to ask the WebContent process
to paint the next frame into a specified bitmap. Instead, it allocates
bitmaps and sends them to WebContent, which then lets chrome know when
the painting is done.
This work is a preparation to move the execution of painting commands
into a separate thread. Now, it is much easier to start working on the
next frame while the current one is still rendering. This is because
WebContent does not have to inform chrome that the current frame is
ready before it can request the next frame.
Additionally, as a side bonus, we can now eliminate the
did_invalidate_content_rect and did_change_selection IPC calls. These
were used solely for the purpose of informing chrome that it needed to
request a repaint.
This commit un-deprecates DeprecatedString, and repurposes it as a byte
string.
As the null state has already been removed, there are no other
particularly hairy blockers in repurposing this type as a byte string
(what it _really_ is).
This commit is auto-generated:
$ xs=$(ack -l \bDeprecatedString\b\|deprecated_string AK Userland \
Meta Ports Ladybird Tests Kernel)
$ perl -pie 's/\bDeprecatedString\b/ByteString/g;
s/deprecated_string/byte_string/g' $xs
$ clang-format --style=file -i \
$(git diff --name-only | grep \.cpp\|\.h)
$ gn format $(git ls-files '*.gn' '*.gni')
If the Inspector widget already exists, be sure to inspect the page when
it is re-opened. However, this should be a no-op if the page was already
inspected (as any existing Inspector will be reset if a new page load
began).
Note this is not an issue in the AppKit chrome.
It was a bit short-sighted to combine the tag and attribute names into
one string when the Inspector requests a context menu. We will want both
values for some context menu actions. Send both names, as well as the
attribute value, when requesting the context menu.
Pages like the new tab page, error page, etc. all belong solely to
Ladybird, but are scattered across a couple of subfolders in Base. This
moves them all to Base/res/ladybird.
When a QObject subclass (widgets, etc.) are provided a parent, then
ownership of that object is passed to the parent. Similarly, objects
added to a QLayout are owned by the layout.
Thus, do not store these objects in an OwnPtr.
QString is UTF-16, thus QString::size returns the number of UTF-16 code
units. Thus, we would fail to perform, for example:
ak_string_from_qstring(QString("😀"));
Which is 2 UTF-16 code units, but 4 UTF-8 code units.
There's no need for 2 overloads for String and DeprecatedString, we can
just use a StringView. This also avoids some cases of needlessly
allocating a DeprecatedString from a StringView when calling this
method.
Ladybird on Serenity currently only uses F12, and on other platforms
only uses ctrl+shift+I. Most browsers support both hotkeys, so let's do
the same for consistency.
Note that the AppKit chrome cannot support both shortcuts. macOS does
not allow setting multiple "key equivalent" strings on an action. There
are some questionable hacks we could do to support this eventually, but
for now, just ctrl+shift+I is supported on macOS.
It is currently a bit messy to pass these options along from main() to
where WebContent is actually launched. If a new flag were to be added,
there are a couple dozen files that need to be updated to pass that flag
along. With this change, the flag can just be added to the struct, set
in main(), and handled in launch_web_content_process().
In order for same-origin NavigableContainers (iframe, frame, embed, ...)
and window.open() WindowProxies to have the proper JS access to their
embedder/opener, we need to host multiple top level traversables in the
same WebContent process. As a first step, make WebContent::PageHost hold
a HashMap of PageClient objects, each holding their own Web::Page that
represents a TraversableNavigable's API surface with the UI process.
The `page_did_request_scroll_to` API takes a CSS position, and thus
callers should not scale to device pixels before invoking it. Instead,
align this API with (most) other PageHost APIs which scale to device
pixels before sending the corresponding IPC message.
In the AppKit chrome, convert the provided device pixel position to a
widget position.
This matches the negation of the vertical scroll delta value. This makes
the scroll events behave as follows on macOS:
Natural scrolling enabled:
* Scrolling up on the trackpad scrolls down on the page.
* Scrolling right on the trackpad scrolls left on the page.
Natural scrolling disabled:
* Scrolling up on the trackpad scrolls up on the page.
* Scrolling right on the trackpad scrolls right on the page.
This was used to provided base functionality for model-based chromes for
viewing the DOM and accessibility trees. All chromes now use the WebView
inspector model for those trees, thus this class is unused.
When the `--dump-failed-ref-tests` flag is provided, screenshots of the
actual and reference pages will be placed in
`Build/lagom/ladbybird/test-dumps`. This makes it a lot easier to spot
what's wrong with a failing test. :^)
We call [NSApp activateIgnoringOtherApps:YES] to ensure the application
is brought into view and focused when launched from a terminal. But in
a macOS environment, we're better off using the `open` command to launch
the application (in which case, it does take foreground focus).
We now create a WorkerAgent for the parent context, which is currently
only a Window. Note that Workers can have Workers per the spec.
The WorkerAgent spawns a WebWorker process to hold the actual
script execution of the Worker. This is modeled with the
DedicatedWorkerHost object in the WebWorker process.
A start_dedicated_worker IPC method in the WebWorker IPC creates the
WorkerHost object. Future different worker types may use different IPC
messages to create their WorkerHost instance.
This implementation cannot yet postMessage between the parent and the
child processes.
Co-Authored-By: Andreas Kling <kling@serenityos.org>
This allows Ladybird to be the default browser on macOS, and allows for
opening some file types (.html, .svg, .md, etc.) with Ladybird.
Note this currently only works in the GN build (and thus the Qt chrome).
The CMake build does not set up the Resources directory properly for
macOS to run the Ladybird app without $SERENITY_SOURCE_DIR set.
If PulseAudio is available, the Qt6 audio plugin will never be used. So
let's remove it from the build.
Note that on macOS, the Qt6 audio plugin will be used if the Qt chrome
is enabled. Otherwise, Audio Unit will be used for the AppKit chrome.
This follows the pattern for the other services spawned by WebContent.
The notable quirk about this service is that it's actually spawned by
the ImageCodecPlugin rather than in main.cpp in the non-Android port.
As a result we needed to do some ifdef surgery to get all the pieces
in place. But we can now load images in the Android port again :^).
Instead of relying on AK_OS_LINUX, actually use the more accurate
HAS_ACCELERATED_GRAPHICS define to figure out if we should try to use
the generic LibAccelGfx GPU painter.
This patch brings a service to handle image decompression. With it comes
security enhancement due to the process boundary. Indeed, consequences
of a potential attack is reduced as only the decoder will crash without
perturbing the WebContent process.
It also allows us to display pages containing images that we claim to
support but still make us crash, like for not-finished-yet decoders.
As an example, we can now load https://jpegxl.info/jxl-art.html without
crashing the WebContent process.
The setting for the search engine to use is currently ephemeral. Once we
have a settings dialog, we can implement this setting there, and persist
that setting.
The default behavior of QPushButton is much closer to what we want from
a drop-down menu, as shown in the QPushButton::setMenu documentation:
https://doc.qt.io/qt-6/qpushbutton.html#setMenu
This also results in much less of a "squished" look than before.
The current size is too small to be able to read the new tab URL. Use
the `resize` API rather than setting a fixed-size as well, to allow the
user to resize the dialog themselves.
This lets the user zoom in and out on a web page using the View menu or
keyboard shortcuts. This does not implement zooming with ctrl+scroll.
In the future, it'd be nice to embed the zoom level display inside the
location toolbar. But to do that, we will need to invent our own custom
search field and all of the UI classes (controller, cell, etc.) to draw
the field. So for now, this places the zoom level display to the right
of the location toolbar.
This is in the spirit of commit a4692a6c978a6e66d171e003063449790a6c5879
(and the history behind that commit).
We will need to perform lookups from an integral node ID to the JSON for
that node frequently in the Inspector. We will also need to traverse the
DOM tree from a node, through its ancestors, to the root node. These are
essentially the same maps stored by the Qt Inspector widget.
This commit includes only fetching the DOM tree from the WebContent
process and displaying it in an NSOutlineView. The displayed tree
includes some basic styling (e.g. colors).
Instead of having an annoying loop that constantly reschedules a
Core::EventLoop trigger, have the ALooperEventLoopManager do it itself
in the did_post_event() function.
We cannot simply re-use the Unix implementation directly because that
implementation expects to actually be called all the time in order to
service timers. If you don't call its' pump() method, timers do not get
triggered. So, we do still need the seconary thread for Timers that was
added earlier.
Similar to the RequestServer, bind this from the WebContentService
implementation and have it work the same way. Deduplicate some code
while we're here.
Add a RequestServerService class that uses the LadybirdServiceBase class
added previously. Bind to it from the WebContentService's service_main()
during startup.
Create LadybirdServiceBase to hold the standard "set resource dir" and
"init ipc sockets" service functionality that will be common between the
WebContent, RequestServer, and WebSocket services.
Refactor the handler class slightly to avoid the HandlerLeak lint by
making the class a static class inside the companion object and use a
WeakReference to the service instead of a strong one.
After moving to navigables, we started reusing the code that populates
session history entries with the srcdoc attribute value from iframes
in `Page::load_html()` for loading HTML.
This change addresses a crash in `determine_the_origin` which occurred
because this method expected the URL to be `about:srcdoc` if we also
provided HTML content (previously, it was the URL passed along with the
HTML content into `load_html()`).
The FIXME at the bottom of Timer.nativeRun was on the money, and was
the cause of some leaked timers. Solve the issue by passing the
EventLoopImplementation to the JVM Timer object and retrieving it's
thread_local timer list, and posting the events to the Impl rather than
trying to invoke the receiver's timer event callback directly in
the Timer thread. Because the implementation of
ALooperEventLoopManager::did_post_event() reaches for
EventLoop::current(), we also need to make sure that the timer thread
acutally *has* an EventLoop, even if we don't expect to use it for
anything. And we need to make sure to pump it to clear out any of the
poke() invocations sending 4 bytes to the wake pipe for that thread.
This function used to live in AndroidPlatform.cpp, but was removed
during the transition to the new app. We still need to extract the
assets from the tarball that CMake creates. At least, until we come
up with a generic "Resource" concept in LibCore.
Timers run in their own thread, to take advantage of existing Java
Executor features. By hooking into ALooper, we can spin the main
Activity's UI thread event loop without causing a fuss, or spinning the
CPU by just polling our event loop constantly.
This will let us spawn a new process for an Android Service to handle
all our WebContent needs. The ServiceConnection is manged by each
WebView. The lifecycle of the Service is not quite clear yet, but each
bindService call will get a unique Messenger that can be used to
transfer the WebContent side of the LibIPC socketpair we use in other
ports.
The simple smoke test makes sure that we can boot up an android emulator
with our package in it, and that the WebView is visible on boot.
More tests to come with more features :^)
This folder is gitignored because we copy a generated tar file to it
during the build so that the android build tools will add it to our
APK file. However, the gitkeep was not re-added when the folder moved.
This allows our WebViewImplementationNative class to paint into an
Android Bitmap class from the onDraw() view event. We also set the
viewport geometry from the view since we're setting the Kotlin Bitmap
at the same time.
This ports over the `LADYBIRD_USE_LLD` option from the standalone
Ladybird build and generalizes it to work for mold as well: the
`LAGOM_USE_LINKER` variable can be set to the desired name of the
linker. If it's empty, we default to trying LLD and Mold on ELF
platforms (in this order).
This template app from Android Studio should hopefully be more fun to
work on than the Qt wrapped application we were using before. :^)
It currently builds the native code using gradle rules, and has a stub
WebViewImplementationNative class that will wrap a c++ class of the same
name that inhertis from WebView::ViewImplementation. Spawning helper
processes and creating proper views in Kotlin is next on the list.
These classes are used as-is in all chromes. Move them to LibWebView so
that non-Serenity chromes don't have to awkwardly reach into its headers
and sources.
This adds menu items to open an interactive JavaScript console for a web
page. This more or less mimics the Qt implementation of the console.
Hooks are included to tie the lifetime of the console window with the
tab it belongs to; if the tab is closed, the console window is closed.
This creates WebView::ConsoleClient to handle functionality that will be
common to the JS consoles of all Ladybird chromes. This will let each
chrome focus on just the UI.
Note that this includes the `console.group` functionality that only the
Serenity chrome previously had. This was a FIXME in the Qt chrome, and
it is implemented such that all chromes will receive this functionality
for free.
The LadybirdWebView currently assumed it is viewed with a Tab instance.
This will not be true with the JavaScript console. This patch removes
this assumption by plumbing WebContent callbacks through a new protocol.
The Tab interface then implements this protocol.
For example, the JavaScript console will invoke window.scrollTo(0, INF)
to scroll to the bottom of the console after updating its contents. The
NSScrollView being scrolled here seems to behave oddly if we scroll
beyond its limit (e.g. mouse events stop working). Prevent this by
limiting scrolling to the NSScrollView's document rect.
We were relying on TabController setting the web view's initial viewport
in [TabController windowDidResize], which has been triggered when the
Tab is first created. However, it turns out that only happens due to the
toolbar being added to the Tab window, causing a resize event. When the
web view is embedded in a window without a toolbar, this resize event is
not triggered. Therefore, we must set the viewport ourselves when the
web view is added to a window.
Currently, the only NSWindow type in the AppKit chrome is the Tab. Once
we have other window types (e.g. Inspector), commands which assume they
are used on a Tab will either crash or behave weirdly.
This changes the createNewTab: command to accept the tab from which the
new tab is created, rather than assuming that tab is the key window. So
if some JS on a page calls window.open() while a non-Tab window is key,
the new tab will be opened within the same tab group.
This also changes closeCurrentTab: to work on any key window. Regardless
of whether the key window is a Tab or some other window, pressing cmd+W
should just close that window.
This lets the user choose a color scheme which differs from the active
system theme. Upon changing the color scheme, the scheme is broadcast to
all active tabs, and will be used in new tabs.
Without setting the --resources flag, headless-browser defaults to /res
for all resources it tries to find, including the theme. It will not
find this path on Lagom, so our attempt to load the default theme does
not accomplish anything.
The old names sounded like awkward English to me, as they implied the
WebContent process is asking for information. In reality, WebContent is
*providing* the information.
We can easily add hooks to notify the browsers of these events if any
implementation-specific handling is needed in the future, but for now,
these only repaint the client, which we can do in ViewImplementation.
Storing the backup bitmap is the same across Browser and Ladybird. Just
peform that work in LibWebView, and handle only the implementation-
specific nuances within the browsers.
This also sets the default callback to do what every non-Serenity
browser is doing, rather than copy-pasting this callback into every
implementation. The callback is still available for any platform which
might want to override the default behavior. For example, OOPWV now
overrides this callback to use FileSystemAccessClient.
This adds an alternative Ladybird chrome for macOS using the AppKit
framework. Just about everything needed for normal web browsing has
been implemented. This includes:
* Tabbed, scrollable navigation
* History navigation (back, forward, reload)
* Keyboard / mouse events
* Favicons
* Context menus
* Cookies
* Dialogs (alert, confirm, prompt)
* WebDriver support
This does not include debugging tools like the JavaScript console and
inspector, nor theme support.
The Qt chrome is still used by default. To use the AppKit chrome, set
the ENABLE_QT CMake option to OFF.
By running ctest -C Integration -R WPT, you can run WPT. Adding just
-C Integration will run all other tests *and* the WPT test.
Also add support for ./Meta/serenity.sh test lagom WPT to serenity.sh
The error.html page now uses the resource_directory_url this
variable contains the relative path to /Base/res/ on the host
system as a file:// url. This is needed for future pages to load
resource files like icons. For the error.html page this was not
really needed because it lies over this own URL in FrameLoader.cpp.
Non-Qt chromes will want to use the same Info.plist and bundle info as
the Qt chrome. This patch puts the CMake setup for the bundle into a
function that non-Qt chromes may call in their CMakeLists.txt. The Qt
chrome calls it automatically.
This will allow us to bring the WebContent process into non-Qt macOS
chromes. This branch is only reached when creating an <audio> element,
so while the chrome is heavily under development, we can just avoid
these elements.
We were still missing the resources and the libraries inside the actual
bundle directory. Do it at install time to not make a mess of all the
rules. The gn build lists all the libraries in a massive list, which
is quite a pain. We can over-copy a few libraries like this to make the
install script a bit easier to follow.
We weren't installing a lot of generated sources for the top level Lagom
build or for LibWeb, making it impossible to use LibWeb from a
find_package. ...And also Kernel/API/KeyCode.h, which is included by
no less than 8 different files in Userland/Libraries. We also weren't
installing any Ladybird header files.