Parse blocks to generate the excerpt

[pento]

Now that #11141 has landed, we can use the same logic when generating an excerpt, which allows us to deprecate get_dynamic_blocks_regex().

By using a list of allowed blocks, we can avoid the other issue that strip_dynamic_blocks() is intended to address, when a dynamic block can call get_the_excerpt(), causing an infinite loop.

Out of interest, 198c82f is an alternate method, which just tweaks strip_dynamic_blocks() to use the parser, but I think this is the better method.

[dmsnell]

This hits top-level blocks but in some cases I think it might be problematic (such as the column block).

This would be a good use-case for the filter I've toyed with in #10108. If we applied a pre-render filter to the blocks then we could whitelist or blacklist down the tree. Then we wouldn't have odd behaviors with top-level vs. nested blocks.

[pento]

Thanks for the review @dmsnell, that's a good point about nested blocks.

I've actually added render_block filter to Core, which would do a similar job.

https://github.com/WordPress/wordpress-develop/blob/114a1eeb3f96ab33b5432b3296fb2ffb3c5e1bdc/src/wp-includes/blocks.php#L194-L202

[catehstn]

No rush to get this merged today, pushing to 4.6.

[catehstn]

@pento is this still relevant or is it replaced by the change in core?

[pento]

Core uses the same code as in this PR, so both Gutenberg and core are currently wrong, in their own special way. 🙂

I think we can just implement something like #10108, and modify this change to use that new filter. This can be done in core, we don't need to bother in Gutenberg.

(Let's leave this PR open until there's a corresponding core ticket.)

[gziolo]

I opened a corresponding ticket in core, which means this one can be closed as mentioned above. Let's continue on trac 👍

[Soean]

The new ticket: https://core.trac.wordpress.org/ticket/46133

[gziolo]

Thanks @Soean, I planned to share the link but somehow I missed it :)