The namespace data is freed and set to NULL, but there remain references
to the namespace declaration nodes. This (rightfully) confuses libxml2
because its invariants are broken. We also have to remove all remaining
references from the subtree. This fixes the data corruption bug.
Closes GH-12681.
Now it's possible via removeAttribute("xmlns:prefix").
It was not possible to reuse a libxml2 function to reconcile because it
does not align with DOM behaviour.
Closes GH-12542.
This always results in a segfault when trying to instantiate, so this never
worked. At least throw an error instead of segfaulting to prevent developers
from being confused.
Closes GH-12420.
PHP 8.1 introduced a seemingly unintentional BC break in ca94d55a19 by
blocking the (un)serialization of DOM objects.
This was done because the serialization never really worked and just
resulted in an empty object, which upon unserialization just resulted in
an object that you can't use.
Users can however implement their own serialization methods, but the
commit made that impossible as the ACC flag gets passed down to the
child class. An approach was tried in #10307 with a new ACC flag to
selectively allow serialization with subclasses if they implement the
right methods. However, that was found to be too ad hoc.
Instead, let's abuse how the __sleep and __wakeup methods work to throw
the exception instead. If the child class implements the __serialize /
__unserialize method, then the throwing methods won't be called.
Similarly, if the child class implements __sleep and __wakeup, then
they're overridden and it doesn't matter that they throw.
For the user, this PR has the exact same behaviour for (sub)classes that
don't implement the serialization methods: an exception will be thrown.
For code that previously implemented subclasses with these methods, this
approach will make that code work again. This approach should be both BC
preserving and unbreak user's code.
Closes GH-12388.
For the test:
Co-authored-by: wazelin <contact@sergeimikhailov.com>
The original caching implementation had an oversight in combination with
the new lifetime management in DOM for 8.3.
The modification counter is stored on the document object itself, but as
that can get deallocated when all references disappear, stale cache data
can be used. Normally this isn't a problem, unless getElementsByTagName is
called not on the document but on a child node. Fix it by moving caching
data into the ref object, which will outlive all nodes from a document
even if the document object disappears.
Closes GH-12338.
libxml2 prior to 2.9.8 had a different signature for xmlHashScanner.
This signature changed in e03f0a199a
Use an #if to work around the incompatible signature.
Closes GH-12326.
The xmlDOMWrapReconcileNamespaces method we used to fix the namespace
corruption issues in 8.1.21/8.2.8 caused regressions.
Primarily, there is a similar corruption that the xmlReconciliateNs method
used to have in which a namespace is suddenly shifted
(SAML-Toolkits/php-saml#562) and the side-effect of removing redundant
namespaces causes problems when a specific serialization is required.
Closes GH-12308.
As reported in GH-12024, the test fails in the encoding part of the test
file. This is due to a libxml2 bug (that's been fixed in modern
versions, but of course various systems and distros are always behind).
The goal of this part of the test is to check if an encoding declaration
is outputted. So the actual encoding used doesn't matter.
Switch to UTF-8, which seems to always work, to work around the issue.
Because the failure path did not release the string, there was a memory
leak.
As the only valid types for this function are IS_NULL and IS_STRING, we
and IS_NULL is always rejected in practice, solve the issue by not using
a function that increments the refcount in the first place.
Closes GH-12002.
These invalidations only need to happen when elements are added,
removed, or manipulated. Processing instructions are not elements and
their contents are just text.
There are two linked issues:
- Conflicts couldn't be resolved by changing the prefix name.
- Lacking a prefix would shift the namespace as the default namespace,
causing elements to suddenly become part of the namespace instead of
the attributes.
The output could still be improved by removing redundant namespace
declarations, but that's another issue. At least the output is
correct now.
Closes GH-11777.
Because we check the list with dom_sanity_check_node_list_for_insertion()
before dom_is_node_in_list(), then we don't have to check the object
type anymore in dom_is_node_in_list(), because
dom_sanity_check_node_list_for_insertion() will have already done that.
Closes GH-11914.
This method had some useless logic in it. It checked whether the child
node is a child of its parent, which is always true of course.
But I know where this check comes from, if you follow the spec closely
you'll find that the spec used to have explicit child and parent
arguments for the removal algorithm [1].
That's because that algorithm is written in a generic way, where the
parent and child arguments might not come from the same subtree.
However, in this particular case it *is* always the case that the child
is a child of its parent. The checks weren't needed back then for
DOMChildNode::remove(), and are still not needed today.
[1] e.g. https://web.archive.org/web/20180601092634/https://dom.spec.whatwg.org/#concept-node-remove
