Add support for parsing Markdown inside HTML

[adam-fowler]

See CommonMark example 108 https://spec.commonmark.org/0.18/#example-108.

Added GroupFragment protocol that can hold multiple fragments. HTML fragment comforms to this as it needs to hold its HTML plus any Markdown fragments included inside the HTML block.

HTML parsing will set state possibleMarkdown whenever it comes across two newlines in a row. At this point it checks for markdown characters and tries to parse them.

Checkout the tests I added for examples of markdown inside HTML.

I specifically added this so I could create a div with a class attribute containing images as follows. Thus allowing me to use css to style the images.

<div class="myimageclass">

![My image](image.jpg)

</div>

[john-mueller]

Hi @adam-fowler, thanks for putting this together! I'm hoping @JohnSundell has time to look at this soon, but I found a couple of things that need to be fixed for this to be merged.

First, could you rebase this on master, so that it will merge nicely with other pending PRs?

Second, there is a string index out-of-bounds error I found when running this against the CommonMark test suite. I added a guard statement in the review comments. I also added a couple nitpicky formatting details based on previous comments by John.

With those two changes, this PR causes CommonMark tests 122, 125, and 157 to pass, with no newly failing tests. Being able to wrap elements in styling <div> tags will be great!

[john-mueller]

Suggested change

	guard !reader.didReachEnd else { break }

The following input (example 125 in the current CommonMark spec) causes an out-of-bounds error:

<div>
*foo*

*bar*

This guard statement fixes the issue, and all other current spec tests run without crashing.

[adam-fowler]

I had to put the guard statement slightly further up to get it to work. ie above the double newline check

[john-mueller]

Oops, that's where it was supposed to be.

[john-mueller]

Suggested change

Picky, but I think John prefers not to have this indentation.

[john-mueller]

Suggested change

	func testUnclosedHTMLWithDoubleNewline() {
	let html = MarkdownParser().html(from: """
	<div>
	*foo*
	*bar*
	""")
	XCTAssertEqual(html, "<div>\n*foo*<p><em>bar</em></p>")
	}

Here's an extra test that catches the out-of-bounds error.

[john-mueller]

Suggested change

	("testHTMLWithDoubleNewline", testHTMLWithDoubleNewline),
	("testHTMLWithDoubleNewline", testHTMLWithDoubleNewline),
	("testUnclosedHTMLWithDoubleNewline", testUnclosedHTMLWithDoubleNewline),

Add extra text case.

[adam-fowler]

You don't seem to have included the test. I'll add it

[john-mueller]

Looks good to me!

[alexito4]

I think this would help too for cases where you want to have some code hidden:

<details><summary>See the entire config</summary>
-> start code inside details ```
code
-> end code inside details ```
</details>