Adds a new `mcp` run mode to start an MCP server over stdio.
Implements tools for navigation and JS evaluation, along with
resources for HTML and Markdown page content.
1 - Finalizer callbacks are now give a *Page parameter. Various types no longer
need to maintain a reference to *Page just to finalize
2 - EventManager now handles v8_handoff == false cleanup. This is largely
because of the above change, which would require every:
```
defer if (!event._v8_handoff) event.deinit(false);
```
to be turned into:
```
defer if (!event._v8_handoff) event.deinit(false, page);
```
But the caller might not have a page. Besides this, it makes most uses of Event
simpler. But, in some cases, it could leave a window where the event doesn't
reach the EventManager to be properly managed (though, we have no such cases
as of now).
getImplementation() now returns a cached *DOMImplementation pointer
per Document, matching the getStyleSheets() pattern. This ensures
document.implementation === document.implementation holds true.
Flips dom/nodes/Document-implementation.html (1/2 → 2/2).
- uppercase entire qualified name in tagName (including prefix)
- validate PI data for "?>" and use proper XML Name production with Unicode
- implement willValidate on HTMLInputElement
- throw IndexSizeError DOMException for negative maxLength assignment
flips: Node-nodeName, Document-createProcessingInstruction, button,
maxlength, input-willvalidate (+6 subtests)
Under some conditions, a microtask would be executed for a context that was
already deinit'd, resulting in various use-after-free.
The culprit appears to be WASM compilation being placed in the microtask queue
(by a user-script) and then resolved at some point in the future. We guard the
microtask queue by a context.shutting_down boolean, but v8 doesn't know anything
about this flag. The fact is that, microtasks are tied to an isolate, not a
context.
This commit introduces a number of changes:
1 - It follows 309f254c2c and stores the zig Context inside of an embedder field. This
ensures v8 doesn't consider this when GC'ing, which _could_ extend the
lifetime of the v8::Context beyond what we expect
2 - Most significantly, it introduces per-context microtasks queues. Each
context gets its own queue. This makes cleanup much simpler and reduces the
chance of microtasks outliving the context
3 - pumpMessageLoop is called on context.deinit, this helps to ensure that any
tasks v8 has for our context are processed (e.g. wasm compilation) before
shtudown
4 - The order of context shutdown is important, we notify the isolate of the
context destruction first, then pump the message loop and finally destroy
the context's message loop.
Depends on https://github.com/lightpanda-io/zig-v8-fork/pull/151
- add ChildNode.remove() to DocumentType (flips DocumentType-remove.html)
- return null for MutationRecord.attributeNamespace on non-namespaced
attribute mutations (flips MutationObserver-takeRecords.html)
- stop lowercasing in createElementNS per spec — only createElement
should ASCII-lowercase for HTML namespace (flips
Element/Document-getElementsByTagNameNS.html)
- fix getElementsByTagName to use case-insensitive matching for HTML
namespace elements
Implements the DOM spec algorithms for namespace lookup on all node
types. Stores custom namespace URIs in a page lookup for elements
created via createElementNS with unknown namespaces. Fixes
setAttributeNS to preserve qualified names for xmlns namespace
declarations.
Flips dom/nodes/Node-lookupNamespaceURI.html: 0/75 → 75/75.
after() captured node.nextSibling() once, which went stale when that
sibling was one of the nodes being inserted. Use viableNextSibling() to
find the first following sibling not in the nodes list per the DOM spec.
replaceWith() in CData had the same stale-reference problem and also
removed self before inserting, unlike Element.replaceWith() which keeps
self as the insertion anchor. Adopt the same anchor pattern: insert
before self, then remove self at the end.
Flips ChildNode-after.html from 33/45 to 45/45 and
ChildNode-replaceWith.html from 27/33 to 33/33.
Add all 25 legacy constants (INDEX_SIZE_ERR through DATA_CLONE_ERR)
to DOMException on both constructor and prototype, enabling WPT
assert_throws_dom checks that reference e.code.
1 - Fix an issue where build would persist a value in the call_arena
2 - Remove double allocation (call_arena -> page_arena)
3 - Improve ergonomics of sanitizeValue with a comptime value indicating whether
or not to always dupe the value.
When a WebAPI takes `[]const u8`, we coerce values to strings. But when it
takes a `?[]const u8` how should we handle `null`? Some APIs might want to know
that it was null, others might just want `"null``.
Currently when `null` is passed to `?[]const u8`, we'll get null.
This adds a discriminator type, js.NullableString. When `null` is passed to it
it'll be converted to `"null"`.
This no longer works with frames. Multiple frames could have a scheduled
navigation, so a single arena no longer has a clear lifecycle. Instead an arena
from the pool is used per navigation event, thus the queued_navigation is self-
contained.
This required having libcurl copy the body. Unfortunate. Currently we free the
arena as soon as the navigation begins. This is clean. But it means the body is
immediately freed (thus we need libcurl to copy it). As an alternative, each
page could maintain an optional transfer_arena, which it could free on
httpDone/Error.
page.wait is the only significant difference between the "root" page and a page
for an iframe. I think it's more explicit to move this out of the page and
into the session, which was already the sole entry-point for page.wait.
