The `cursor` property accepts a list of possible cursors, which behave
as a fallback: We use whichever cursor is the first available one. This
is a little complicated because initially, any remote images have not
loaded, so we need to use the fallback standard cursor, and then switch
to another when it loads.
So, ComputedValues stores a Vector of cursors, and then in EventHandler
we scan down that list until we find a cursor that's ready for use.
The spec defines cursors as being `<url>`, but allows for `<image>`
instead. That includes functions like `linear-gradient()`.
This commit implements image cursors in the Qt UI, but not AppKit.
This supports evaluating the script and replying with the result. We
currently serialize JS objects to a string, but we will need to support
dynamic interaction with the objects over IPC. This does not yet support
sending console messages to DevTools.
Our existing WebContentConsoleClient is very specific to our home-grown
Inspector. It renders console output to an HTML string. For DevTools, we
will not want this behavior; we will want to send representations of raw
JS values.
This patch makes WebContentConsoleClient a base class to handle console
input from the user, either from the Inspector or from DevTools. It then
moves the HTML rendering needed for the Inspector to a new class,
InspectorConsoleClient. And we add a DevToolsConsoleClient (currently
just stubbed) to handle needs specific to DevTools.
We choose at runtime which console client to install, based on the
--devtools command line flag.
Our own Inspector differs from most other DevTools implementations with
regard to highlighting DOM nodes as you hover elements in the inspected
DOM tree. In other implementations, as you change the hovered node, the
browser will render a box model overlay onto the page for that node. We
currently don't do this; we wait until you click the node, at which
point we both paint the overlay and inspect the node's properties.
This patch does not change that behavior, but separates the IPCs and
internal tracking of inspected nodes to support the standard DevTools
behavior. So the DOM document now stores an inspected node and a
highlighted node. The former is used for features such as "$0" in the
JavaScript console, and the latter is used for the box model overlay.
Our Inspector continues to set these to the same node.
This reduces the number of `.cpp` files that need to be recompiled when
one of the below header files changes as follows:
CSS/ComputedProperties.h: 1113 -> 49
CSS/ComputedValues.h: 1120 -> 209
This is required to store Content Security Policies, as their
Directives are implemented as subclasses with overridden virtual
functions. Thus, they cannot be stored as generic Directive classes, as
it'll lose the ability to call overridden functions when they are
copied.
This implementation can be better improved in the future by ripping
out a lot of the manual logic in LibWebSocket and rely on libcurl to
parse our message payloads. But for now, this uses the 'raw mode' of
curl websockets in connect-only mode to allow for somewhat seamless
integration into our event loop.
Explicitly link final targets with OpenSSL to ensure that the vcpkg
version is loaded instead of the system one.
Before this change we would inherit `libcrypto.so` and `libssl.so` from
other dependencies, like Qt, that do not have their RPATH rewritten.
This would cause the loader to prefer the system libraries over the
vcpkg ones causing all sorts of version mismatch issues.
The effectiveness of this change can be verified with
`readelf -d ./bin/Ladybird` showing `libcrypto.so` and `libssl.so` as
direct dependencies, before they would not appear. Additionally, `ldd`
will show `libcrypto.so` and `libssl.so` pointing to the vcpkg builds.
Our existing coalescing mechanism for input events didn't prevent
multiple mousemove/mousewheel events from being processed between paint
cycles. Since handling these events can trigger style & layout updates
solely for hit-testing purposes, we might end up doing work that won't
be observable by a user and could be avoided by shceduling input events
processing to happen right before painting the next frame.
We can't simply walk the element tree and pass in `pseudo_element` each
time. Instead, we want to look at:
1. Element's pseudo-element
2. Element
3. Element's parents
We don't yet support a proxy configuration, but we can still validate
the capability received from the WebDriver client. We should also fail
to create a WebDriver session if a proxy configuration is present.
The WebDriver spec now separately tracks an active HTTP session list,
which will contains all non-BiDi WebDriver sessions by default. There
may only be one active HTTP session at a time.
See: https://github.com/w3c/webdriver/commit/63a397f
Session management is a bit awkward right now in that the list of active
sessions is managed by Client, resulting in operations like closing a
session being split between several functions in Client and Session.
This patch moves all session management to the Session class. Closing a
session is now entirely in Session::close().
This will make managing a separate HTTP session list a bit simpler.
Executing scripts via WebDriver has a bit of awkwardness around dealing
with user dialogs that open during script execution. When this happens,
we must return control back to the client immediately with a null
response, while allowing the script to continue executing. When the
script completes, we must then ignore its result.
We've previously handled this by tracking a boolean for the ongoing
script execution, set to true when the script begins and false when it
ends (either via normal script completion or the above dialog handling).
However, this failed to handle the following scenario, running two
scripts in a row:
execute_script("alert('hi'); return 1;")
execute_script("return 2;")
The first script would execute and open a dialog, and thus return a null
response to the client while the script continued and the dialog remains
open. The second script would "handle any user prompts", which closes
the dialog. This would end the execution of the first script. But since
we're now executing a script again, the boolean flag is true, and we'd
return the result of the first script back to the client. The client
would then think this is the result of the second script.
So we now track script execution with a simple ID. If a script completes
whose execution ID is not the ID of the currently executing script, we
drop the result.
Lots of editorial spec bugs here, but these changes largely affect how
the unhandledPromptBehavior capability is handled. We also now set an
additional capability for the default User Agent string.
WebDriver script authors may now provide either:
* A user prompt handler configuration to be used for all prompt types.
* A set of per-prompt-type user prompt handlers.
This also paves the way for interaction with the beforeunload prompt,
though we do not yet support that feature in LibWeb.
See: https://github.com/w3c/webdriver/commit/43903d0
Before, if something went wrong with DNS lookup and there were unrelated
records (i.e. not A or AAAA) then we would still attempt to build a
resolve list. This resulted in curl errors related to the option itself
and displayed as "unknown network error" to the user.
Before this change, `m_needs_repaint` was reset in
`Document::record_display_list()` only when the cached display list was
absent. This meant that if the last triggered repaint used the cached
display list, we would keep repainting indefinitely until the display
list was invalidated (We schedule a task that checks if repainting is
required 60/s).
This change also moves `m_needs_repaint` from Document to
TraversableNavigable as we only ever need to repaint a document that
belongs to traversable.
