feat: serialize AST node back to wikitext string

[linonetwo]

closes #8255

Currently just a demo, I will gradually move code from https://github.com/tiddly-gittly/wikiast/tree/master/packages/wikiast-util-to-wikitext to here.

Try it with tiddler (Well, if official website have share plugin installed, I can put a link here):

AAA test

---

\```js
var match = reEnd.exec(this.parser.source)
\```

end

and run this in console:

$tw.utils.serializeParseTree($tw.wiki.parseTiddler('New Tiddler').tree) 

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Updated (UTC)
tiddlywiki5	✅ Ready (Inspect)	Visit Preview	Jun 12, 2024 5:39pm

[pmario]

The problem here is, that the output of you test does not produce the same rendered HTML output as the original source. See the screenshot.

[linonetwo]

First line is wrapped by a p element, but I haven't find the rule that produce the p element. And there was no p html in the source, so I may need to add some attribute to ast to show this p element is "fake".

Anyway, this is just a demo for API, do you like the API name and their usage? The further task is just adding exports.serailize = xxx to each rule.

[pmario]

First line is wrapped by a p element, but I haven't find the rule that produce the p element. And there was no p html in the source, so I may need to add some attribute to ast to show this p element is "fake".

IMO the current parser has a "systemic" problem with redundant P-tags, that we probably may need to solve first.

Working around this problem with "fake" elements will just increase your code complexity, instead of solving the underlaying problem. It will also create an other dependency, that will make solving the problem, in the core, in backwards compatible way even harder as it is now.

[pmario]

IMO your code will throw an RSOD if options is undefined
So options = options || {} should be used instead. We use this pattern all over the places.

[linonetwo]

I copy $tw.utils.extend({},options) from somewhere nearby, I thought this will work.

[linonetwo]

redundant P-tags

Can't remove then, many of my plugin style sheet is already depending on this structure. And p > text for block is following HTML5 semantic.

Another option is handle it like text node, regard it as a special case.

[pmario]

And p > text for block is following HTML5 semantic.

That's true, but we do have a lot of p > div which should not happen at all.

[linonetwo]

Handle the block rule by adding a rule: 'parseBlock' to WikiParser.prototype.parseBlock.

I've added some test for it.

[linonetwo]

The LLM prompt

update this to add exports.serialize = function(tree, serialize) { that accept node and a serialize that accept array of node or a node and return a string ,Use ES5 only in the serialize , and only output this method. No explain but add comments for internal variable's possible example value and intention.Don't add comment that is already explained by variable name. And add comment on top of the line. No space after comma. And provide a test in another codeblock, use example in the top comment from rule code. Not like serialize code, No space and extra comment in test code

[linonetwo]

All tests passed, opened for review.