fix: support AnnotatableBlock in the new content runtime

[farhan]

Closes: openedx/openedx-platform#38809

What

Add a parse_xml_new_runtime override to AnnotatableBlock so the block can be loaded by the new openedx_content-based content runtime (Content Libraries).

Why

While extracting the Annotatable block from openedx-platform, we parted ways with RawMixin, which contains the parse_xml_new_runtime method. That method is still required by the extracted block to be compatible with the new content runtime, and it was missed during extraction.

Without it, opening an annotatable block in a Content Library crashes with:

NotImplementedError: XML Serialization is only supported with OpenedXContentRuntime

The new runtime loads blocks via parse_xml_new_runtime. Annotatable only inherited the LegacyXmlMixin fallback, which delegates to the base XBlock.parse_xml. That base parser treats every child XML node as a child block and calls runtime.add_node_as_child(), which the new runtime does not support (it raises the error above). Since annotatable OLX always contains nested markup (<instructions>, <p>, <annotation>), loading always failed.

Fix

Copy the parse_xml_new_runtime method from RawMixin into AnnotatableBlock. It captures the full inner XML into the data field and strips the child nodes before delegating to the base parser.

Case study: Problem block (same pattern, already done)

This is exactly how the extracted Problem block was handled:

• The built-in ProblemBlock (in openedx-platform) extends RawMixin, inheriting parse_xml_new_runtime.
• The extracted ProblemBlock (here) does not extend RawMixin, and instead explicitly copies parse_xml_new_runtime into its own class.

The same was done for the html and video blocks. Annotatable was the one raw-data block where this copy was omitted during extraction; this PR brings it in line.

Manual testing

1. Create an Annotatable Block in the Course
2. Copy the Block from the Course and paste into Library as Component. Or Import the course components into the Library.
3. Before this change: 500 error (NotImplementedError in studio logs). After: the component renders.

Screenshot

Annotatable Block successfully loaded into the Content Library

[bradenmacdonald]

Usually it's better to just implement parse_xml, unless you really need the parsing to work differently in the new runtime vs the old, in which case you can implement both parse_xml_new_runtime and parse_xml with different behavior. But having one simple parse_xml function is better for consistency.

We only introduced parse_xml_new_runtime because some XBlocks like video were doing crazy things in their parse_xml method, and we couldn't change that to be sane without breaking existing stuff.

[farhan]

But having one simple parse_xml function is better for consistency.

We only introduced parse_xml_new_runtime because some XBlocks like video were doing crazy things in their parse_xml method

@bradenmacdonald Thanks for the context — that's really helpful background on why parse_xml_new_runtime exists in the first place.

For the extraction work, our focus was largely a lift-and-shift of the existing logic/methods from edx-platform, so this PR follows that same principle and copies the method over to restore parity.

Current status in xblocks-core: Video, Problem, HTML, and Annotatable now all carry their own parse_xml_new_runtime definitions.

Good point that a single parse_xml is the better long-term shape, with *_new_runtime really only needed for blocks like Video. Since Problem, HTML, and Annotatable may not need the separate method, we can open a follow-up ticket to study consolidating them into one parse_xml — I think keeping this PR scoped to fix sounds reasonable

cc: @kdmccormick

[irfanuddinahmad]

Potential crash when importing a bare <annotatable/> node.

definition_from_xml returns {'data': ''} when the node has no children and no attributes (the len == 0 early-exit branch). The assignment here then unconditionally overwrites whatever super().parse_xml set — including the non-empty field default — with an empty string. Any subsequent call to student_view hits etree.fromstring('') in _render_content and raises XMLSyntaxError: Document is empty.

A bare <annotatable/> is an unusual OLX shape, but the guard is cheap:

if data_field_value:
    block.data = data_field_value

This matches how RawMixin handles the equivalent case in edx-platform.