Cannot get the start point in the node at path [1] because it has no start text node.

Do you want to request a feature or report a bug?

Report a bug

What’s the current behavior?

Bug GIF:

https://codesandbox.io/s/slate-reproductions-cm3r7?file=/index.js

Slate: 0.57.1 Browser: Chrome OS: Mac

What’s the expected behavior?

The expected behavior is that the string can be parsed to Slate object without throwing an error.

---

I found that the bug is in this function

https://github.com/ianstormtaylor/slate/blob/f6bfe034d707693488c38da77537fd36cb8856cf/packages/slate-hyperscript/src/creators.ts#L27-L81

because the p tag has no text, the children parameter is an empty array, therefore the for loop is never performed and returns the variable node as an empty array.

[cameracker]

Hi @lavaldi , the issue is that the paragraph blocks the deserialize function generates are allowed to have no children. This is an invalid state for slate; all blocks need to have at least one text node, even if it is empty. The deserializer function should be updated to correct that issue.

As for images, you should be able to pass a void flag to the image block to indicate that it is not intended to have text children. See the image example for how some examples of void blocks.

[TheSpyder]

Perhaps a validation should be added that if children is an empty array that it returns [{ text: '' }] 🤔

If #4208 is merged, this will happen when Editor.normalize is used (for example Editor.normalize(editor, { force: true }) can be used to normalize an entire document).