Parse string with Tag and create node tree (support nest)

[yamanetaisei]

I want to identify the text between the Tags.
For that I am trying to generate a NodeTree.

I have text like this

"aaa<small>bbb</small>ccc*ddddd*" 

The definition of Node is as this.
Node Tree is represented by an array of Node.

enum Node: Equatable {
    case text(_ text: String)
    case bold(children: [Self]) // between **
    case small(children: [Self]) // between <small></small>
    case italic(children: [Self]) // between <i></i>
}

I defined a Parser like this.

This discussion was used as a reference for how to make it
Parse string with HTML tag and transform to Text

class MyParcer {
    lazy var body =  Many {
        smallTag
        italicTag
        boldAsta
        textNormal
        Rest().compactMap {
                $0.isEmpty
                ? nil
                : Node.text(String($0))
            }
    }

    private lazy var smallTag = Parse {
        "<small>"
        PrefixUpTo("</small>")
            .compactMap { [weak self] in
                return $0.isEmpty
                ? nil
                : Node.small(
                    children: self?.parseChildren(String($0)) ?? [.text(String($0))]
                )
            }
        "</small>"
    }

    private lazy var boldAsta = Parse {
        "*"
        PrefixUpTo("*")
        "*"
    }.compactMap { [weak self] content in
        return content.isEmpty
        ? nil
        : Node.bold(
            children: self?.parseChildren(String(content)) ?? [.text(String(content))]
        )
    }

    private lazy var italicTag = Parse {
        "<i>"
        PrefixUpTo("</i>")
            .compactMap { [weak self] in
                return $0.isEmpty
                ? nil
                : Node.italic(
                    children: self?.parseChildren(String($0)) ?? [.text(String($0))]
                )
            }
        "</i>"
    }

    private func parseChildren(_ content: String) -> [Node] {
        do {
            return try body.parse(content)
        } catch {
            return [.text(content)]
        }
    }

     private lazy var textNormal = OneOf {
        PrefixUpTo("*")
            .compactMap {
              $0.isEmpty
              ? nil
                : Node.text(String($0))
            } 

       PrefixUpTo("<i>")
            .compactMap {
              $0.isEmpty
              ? nil
                : Node.text(String($0))
            }

        PrefixUpTo("<small>")
            .compactMap {
              $0.isEmpty
              ? nil
                : Node.text(String($0))
            }
    }
}

This Parser can nest text with different attributes

// This test is success
func testNest1() throws {
        let p = p.body
        let input = "<small>abc*123*abc</small>"
        let output = [
            Node.small(
                children: [
                    .text("abc"),
                    .italic(children: [.text("123")]),
                    .text("abc")
                ]
            )
        ]

        XCTAssertEqual(
            try p.parse(input),
            output
        )
    }

However, nesting text with the same attributes will not parse correctly

// This test is failure
func testNest2() throws {
        let p = p.body
        let input = "<small>abc<small>123</small>abc</small>" // using same attributes
        let output = [
            MfmNode.small(
                children: [
                    .text("abc"),
                    .small(children: [.text("123")]),
                    .text("abc")
                ]
            )
        ]

        XCTAssertEqual(
            try p.parse(input), // [.small(children: [.text("abc<small>123"), ]), .text("abc</small>")]
            output
        )
    }

The problem is that </small> , which should be a child element, is used as the parent's closeTag

Others such as these are not parsed correctly
<small>abc<small>def</small>ghi<small>jkl</small>mno</small>

Is there anyone who can give me any advice?
Thank you.

[yamanetaisei]

@aarsland @mbrandonw @acosmicflamingo
Sorry to call on you after so many months.

Those who took part in this discussion may have already moved on.

If you don't mind, can you give me some ideas?

[acosmicflamingo]

I unfortunately had a lot of issues designing a parser that supported nested tags like that. Because of my use-case though, it turns out that trying to create that kind of parser was overkill anyway.

[aarsland]

Sorry, I did not do more than what is in that discussion.