Instead of offsetting the screenshot by the position of the document
element, just take a screenshot of the entire content rect. After
looking at Acid3 a bit more, this more accurately represents what the
user actually sees on the webpage.
This moves setting the navigator.webdriver flag from after WebContent
creates the WebDriver connection, to its own IPC to be triggered from
WebDriver. This is closer to the spec, but mostly serves as an easy
test to validate the connection.
This just sets up the infrastructure for the WebContent process to house
WebDriver IPCs, and adds an IPC for WebContent to create the WebDriver
connection. The WebDriverConnection class inside WebContent ultimately
will contain most of what is currently in WebDriver::Session (so the
copyright attributions are copied here as well).
The socket created by WebDriver is currently /tmp/browser_webdriver
(formatted with some IDs). This will be moved to the /tmp/webdriver
folder, as WebDriver will create multiple sockets to communicate with
both Browser and WebContent as the IPCs are iteratively moved to
WebContent. That path is unveiled here, though it is unused as of this
commit.
The full-page screenshot was missing horizontal overflow content on most
web pages. Change the dimensions of the captured screenshot to match the
actual content size.
Note that even though there is no response, this IPC has to be
synchronous to allow all scroll events to trigger before returning to
the calling WebDriver process.
This cannot be done on the Browser or WebDriver ends, or via the
existing run_javascript() IPC endpoint, as we cannot transfer JS objects
through the IPC boundary (yet), only serialized JSON, so the individual
WebDriver steps around script execution need to run in the WebContent
process.
To use the `GET /session/{id}/element/{id}/css/{property name}`
WebDriver endpoint, two new IPC calls through the Browser are
implemented:
- get_active_documents_type returns the type of the active document,
which is either "xml" or "html"
- get_computed_value_for_element returns the computed CSS value (as
String) for the given element and CSS property name
This removes a set of complex reference cycles between DOM, layout tree
and browsing context.
It also makes lifetimes much easier to reason about, as the DOM and
layout trees are now free to keep each other alive.
This patch adds `get_document_element()` and `query_selector_all()`
which return Node's IDs.
`get_document_element` returns the ID of the document element
`query_selector_all` returns the IDs of all elements matching the
selector starting at the Node associated with the start_node_id
The basic idea is that a global object cannot just come out of nowhere,
it must be associated to a realm - so get it from there, if needed.
This is to enforce the changes from all the previous commits by not
handing out global objects unless you actually have an initialized
realm (either stored somewhere, or the VM's current realm).
Each of these strings would previously rely on StringView's char const*
constructor overload, which would call __builtin_strlen on the string.
Since we now have operator ""sv, we can replace these with much simpler
versions. This opens the door to being able to remove
StringView(char const*).
No functional changes.
I already fixed the crash from this in #14470, but didn't fully fix
the issue. Currently the browser just avoids sending the
inspect_dom_node() IPC call for non-visible nodes.
The main problem with this is it means the browser keeps displaying
the overlay for the previously selected node. This commit fixes
the crash in the WebContent side, so the IPC call can still be made
and the selection correctly updated.
To achieve this goal:
- The Browser unveils "/tmp/portal/filesystemaccess"
- Pass the page through LoadRequest => ResourceLoader
- ResourceLoader requests a file to the FileSystemAccessServer via IPC
- OutOfProcessWebView handles it and sends a file descriptor back to
the Page.
