Fix serde, change BlockQuote -> Blockquote

[h7kanna]

• Fixed MDAST: serde Serialization/Deserialization not working #72
• Added a sanity test for the same

NOTE : The included sanity test makes sure that at-least the AST serializes and deserializes. More rigorous tests should be added later. Currently, it works as the construct names 'camelCased' match the Mdast JSON except for Blockquote (changing it will be API change).
As it is still in alpha changing 'BlockQuote' enum to 'Blockquote' will make serde work completely.

NOTE: This PR makes another pending one obsolete #74

[wooorm]

Do you recommend that humans run both commands? One or the other?

[h7kanna]

Can be replaced with 'cargo test --all-features'

[wooorm]

ok, can you do that? One test example, with all-features?

[h7kanna]

done

[wooorm]

Thanks for your PR!

---

(changing it will be API change).
As it is still in alpha changing 'BlockQuote' enum to 'Blockquote' will make serde work completely.

Are you saying that currently (main branch) the type field is not blockquote (the correct value according to mdast)?
And that this branch doesn’t change that either?
I would prefer to fix it: to use blockquote.

---

Are you interested in adding, as you note in your OP, some more tests?

[h7kanna]

Thanks for your PR!

(changing it will be API change).
As it is still in alpha changing 'BlockQuote' enum to 'Blockquote' will make serde work completely.

Are you saying that currently (main branch) the type field is not blockquote (the correct value according to mdast)? And that this branch doesn’t change that either? I would prefer to fix it: to use blockquote.

No, in the main branch it is 'blockquote', and matches the mdast. I had to remove the 'type' on the individual enums and replace with 'rename_all=camelCase' on the 'Node'. It seems like we cannot correctly specify tagging on individual constructs. It is resulting in duplicate tags like shown below. I do not like the solution of depending the convention of the camelCase. But I am assuming it will never change.

{
         "type":"Blockquote",
         "type":"blockquote",

Are you interested in adding, as you note in your OP, some more tests?

Yes I started writing tests. But they are turning really verbose. I am refactoring them.
But I created the PR to ask about the 'Blockquote' change and the assumption of 'camelCase' convention.
Also this PR can be individually merged because currently enabling 'serde' feature is not usable at all because of invalid json output.

[wooorm]

invalid json

I don’t think JSON has a limitation on non-duplicate fields? https://www.json.org/json-en.html.

individually merged

fine!

---

Changing Block Quote is 👍 for me!

[h7kanna]

invalid json

I mean deserialization failing because of above duplicate 'type'.

I don’t think JSON has a limitation on non-duplicate fields? https://www.json.org/json-en.html.

individually merged

fine!

Changing Block Quote is 👍 for me!

Yes, made the change as well.

[wooorm]

I mean deserialization failing because of above duplicate 'type'.

Right, but I am trying to point out that I do not think this is not something JSON defines. I think it is something that some libraries do: in JS, JSON.parse('{"a":1,"a":2}') is {a:2}.

[wooorm]

Looks good. Are you planning to add some more tests in this PR, or separately?

[h7kanna]

Looks good. Are you planning to add some more tests in this PR, or separately?

Definitely adding more test cases. I will do in follow up PR. It is just taking time to write down the fixtures. Most of the content is just spans and points.

[wooorm]

Thanks!

[wooorm]

Looking forward to your other PR!

[wooorm]

@h7kanna Do you perhaps have an idea how to fix this?

markdown-rs/src/mdast.rs

Lines 81 to 82 in 4ad5342

	// To do: this should serialize in serde as `null`.
	None,

It should serialize in JSON as null. Not "none".