LuaType.js not correctly parsing some complex values

[probablytukars]

I was using LuaType.js for a personal project, and in my testing I found that there were some parsing errors, luckily I was able to make a fix, but note that some of the variables are changed.

const print = console.log

const isPunc = (char) => !!char.match(/[\{\}<>\-\|&<>()\[\]]/)
const isAtom = (char) => !isPunc(char)

function _tokenize(code, isGroup) {
    print("reading", code)

    let position = 0
    const tokens = []

    const next = () => code[position++]
    const peek = () => code[position]

    const read = (condition) => {
        let buffer = ""
        while (peek() && condition(peek())) {buffer += next()}
        return buffer
    }

    const readBalanced = (left, right) => {
        let buffer = ""
        let depth = 0
        while (peek()) {
            if (peek() === left) {
                depth++
            } else if (peek() === right) {
                if (depth === 0) {
                    break
                } else {
                    depth--
                }
            }
            buffer += next()
        }
        return buffer
    }

    while (position < code.length) {
        if (peek() === "(") {
            position++
            tokens.push({
                type: "tuple",
                within: _tokenize(readBalanced("(", ")"), true)
            })
            next()
            continue
        }

        if (peek() === "[") {
            next()
            tokens.push({
                type: "indexer",
                within: _tokenize(readBalanced("[", "]")),
            })
            next()
            continue
        }

        if (peek() === "{") {
            next()
            tokens.push({
                type: "table",
                within: _tokenize(readBalanced("{", "}"), true),
            })
            next()
            continue
        }

        if (isGroup && peek() === ",") {
            next()
            tokens.push({
              type: "separator",
            })
            continue
          }

        if (isPunc(peek())) {
            const punc = next()

            if (punc === "-" && peek() === ">") {
                tokens.push({
                    type: "arrow",
                })
                next()
                continue
            }

            if (punc === "|") {
                tokens.push({ type: "union" })
                continue
            }

            if (punc === "&") {
                tokens.push({ type: "intersection" })
                continue
            }

            tokens.push({
                type: "punc",
                punc,
            })
            continue
        }

        const atom = read((char) =>
            isGroup ? char !== "," && isAtom(char) : isAtom(char)
        )

        if (atom) {
            if (atom.endsWith(":")) {
                tokens.push({ type: "identifier", identifier: atom.slice(0, -1) })
            } else {
                tokens.push({
                    type: "luaType",
                    luaType: atom,
                })
            }
            continue
        }
    }

    return tokens
}

function getTokens(code) {
    const noWhitespace = code.replace(/\s/g, '')
    return _tokenize(noWhitespace)
}

var advanced_test = "(callback:<T>(internal:(k:{string&number})->T)->string)->{string|number}"

print(JSON.stringify(getTokens(advanced_test), null, 4))

I believe the only parts I changed are adding support for intersection type, and expanding the isPunc function to correctly catch all punctuation, otherwise it can fail on sub-types, of course in my test I removed whitespace before parsing, but adding these modifications it should still work with whitespace.

[evaera]

This isn't the correct way to contribute to an open source project. Pull Requests are for submitting changes, not issues. And it's a good idea to include the cases that are failing in your issue to justify why the change needs to be made, rather than just saying that it does. I can't do anything with this code you've given me.

[probablytukars]

Sorry, never created a pull request before, but I created one now and the dif is only 7 characters and includes the use case where it broke too, hope this is ok!

[YetAnotherClown]

@probablytukars I noticed that you deleted your Pull Request. If you would be interested in opening it again, I'll look into it and merge it if approved.