Get cell by ID?

[krassowski]

Problem

Currently the API supports getting the cell by index. Index is not a stable identifier in a notebook.

Proposed Solution

It would be perfect to be able to fetch a cell by ID, without the overhead of converting the entire notebook to Python objects

Additional context

jupyter_ydoc/jupyter_ydoc/ynotebook.py

Lines 92 to 115 in 9223523

	def get_cell(self, index: int) -> Dict[str, Any]:
	"""
	Returns a cell.
	:param index: The index of the cell.
	:type index: int
	:return: A cell.
	:rtype: Dict[str, Any]
	"""
	meta = self._ymeta.to_py()
	cell = self._ycells[index].to_py()
	cell.pop("execution_state", None)
	cast_all(cell, float, int) # cells coming from Yjs have e.g. execution_count as float
	if "id" in cell and meta["nbformat"] == 4 and meta["nbformat_minor"] <= 4:
	# strip cell IDs if we have notebook format 4.0-4.4
	del cell["id"]
	if (
	"attachments" in cell
	and cell["cell_type"] in ("raw", "markdown")
	and not cell["attachments"]
	):
	del cell["attachments"]
	return cell

[krassowski]

Since self._ycells is a list this may require holding a hashmap/dict from cell id to cell index. This may be hard, but ultimately may be worth it.

[davidbrochart]

Would something like that be enough?

def get_cell_by_id(self, id):
    for ycell in self.ycells:
        _id = ycell.get("id")
        if _id == id:
            return ycell

[krassowski]

This is what I am doing right now, this request is more about having a way to fetch it in O(1) rather than O(n).