markdown-parsing/Sources/main.swift

import Foundation
@preconcurrency import Parsing

let yamlString = """
---
author: Michael Housh
categories:
  - HVAC
  - General
  - Programming
copy: true
draft: false
date: 2023-10-21
lastmod: 2023-10-21
image: banner.png
featuredImage: banner.png
series:
  - General
tags:
  - HVAC
  - General
  - Programming

---
"""

_ = """
title: "You Should Learn Markdown"
slug: "you-should-learn-markdown"
"""

enum YamlItem {
  case boolean(Bool)
  case date(year: Int, month: Int, day: Int)
  case string(String)
  case listItem(String)

  private static let dateParser = Parse(input: Substring.UTF8View.self) {
    Digits(4)
    "-".utf8
    Digits(2)
    "-".utf8
    Digits(2)
  }.map { Self.date(year: $0.0, month: $0.1, day: $0.2) }

  private static let booleanParser = Parse(input: Substring.UTF8View.self) {
    Bool.parser()
  }.map { Self.boolean($0) }

  private static let quotedString = Parse(input: Substring.UTF8View.self) {
    "\"".utf8
    PrefixUpTo("\"".utf8)
    "\"".utf8
  }

  private static let stringParser = Parse(input: Substring.UTF8View.self) {
    OneOf {
      quotedString
      Rest()
    }
  }.map { Self.string(String(Substring($0))) }

  private static let listItemParser = Parse(input: Substring.UTF8View.self) {
    Skip { Whitespace() }
    "- ".utf8
    Rest()
  }.map { Self.listItem(String(Substring($0))) }

  static let parser = OneOf {
    dateParser
    booleanParser
    listItemParser
    stringParser
  }
}

let yamlBlockParser = Parse(input: Substring.UTF8View.self) {
  "---\n".utf8
  Many {
    PrefixUpTo(":".utf8)
    ":".utf8
    Optionally { YamlItem.parser }
    // PrefixUpTo("\n".utf8)
    // From(.substring) { YamlLine.parser }
  } separator: {
    "\n".utf8
  } terminator: {
    // "\n".utf8
    "---".utf8
  }
}

let stringLine = #"author: "Michael Housh""#
// let parsedLine = try yamlLineParser.parse(stringLine[...].utf8)
// print(parsedLine)

let stringItem = "foo"
let quotedItem = "\"foo\""
let dateItem = "2025-01-29"
let boolItem = "true"

try print(YamlItem.parser.parse(stringItem[...]))
try print(YamlItem.parser.parse(quotedItem[...]))
try print(YamlItem.parser.parse(dateItem[...]))
try print(YamlItem.parser.parse(boolItem[...]))

let kvLine = "author: Michael Housh"
var listLine = #"""
  - HVAC
  - General
  - Programming

"""#[...]

struct YamlList: Parser {

  var body: some Parser<Substring.UTF8View, [String]> {
    Many(into: [String]()) { array, string in
      array.append(string)
    } element: {
      Whitespace()
      "-".utf8
      Whitespace()
      PrefixUpTo("\n".utf8).map(.string)
      "\n".utf8
    }
  }
}

try print(YamlList().parse(listLine))

var yamlLines = """
author: Michael Housh
test:
  - HVAC
  - General
  - Programming

"""[...]

struct SimpleYamlValue: Parser {
  enum Output: Equatable {
    case array([String])
    case bool(Bool)
    case date(year: Int, month: Int, day: Int)
    case string(String)
  }

  var body: some Parser<Substring.UTF8View, Output> {
    OneOf {
      // YamlList().map { Output.array($0) }
      DateParser()
      Bool.parser().map(.case(Output.bool))
      PrefixUpTo("\n".utf8).map(.string).map(.case(Output.string))
      Rest().map(.string).map(.case(Output.string))
    }
  }

  struct DateParser: Parser {
    var body: some Parser<Substring.UTF8View, Output> {
      Parse {
        Digits(4)
        "-".utf8
        Digits(2)
        "-".utf8
        Digits(2)
      }.map(.case(Output.date))
    }
  }
}

// This works, but doesn't handle lists well.
struct SimpleYamlParser: Parser {
  var body: some Parser<Substring.UTF8View, [String: SimpleYamlValue.Output]> {
    Many(into: [String: SimpleYamlValue.Output]()) { (dict: inout [String: SimpleYamlValue.Output], pair: (String, SimpleYamlValue.Output)) in
      let (key, value) = pair
      dict[key] = value
    } element: {
      PrefixUpTo(":".utf8).map(.string)
      ":".utf8
      Whitespace()
      SimpleYamlValue()
    } separator: {
      "\n".utf8
    } terminator: {
      End()
    }
  }
}

let simpleYamlString = #"""
author: Michael Housh
copy: true
draft: false
date: 2023-10-21
lastmod: 2023-10-21
image: banner.png
featuredImage: banner.png
"""#

let parsed = try SimpleYamlParser().parse(simpleYamlString[...].utf8)
// print(parsed)

for (key, value) in parsed {
  print("\(key): \(value)")
}

// try print(SimpleYamlParser().parse(simpleYamlString[...].utf8))