kdleam

import gleam/int
import gleam/list
import gleam/option.{type Option, None, Some}
import gleam/result
import gleam/string
import kdleam/error.{type KdlError}
import kdleam/scanner.{type Scanner}

@internal
pub fn parse_string(s: Scanner) -> Result(#(String, Scanner), KdlError) {
  use maybe <- result.try(parse_string_if_available(s))
  case maybe {
    #(Some(value), next) -> Ok(#(value, next))
    #(None, next) ->
      case scanner.peek(next) {
        Some(ch) -> Error(scanner.make_error(next, error.UnexpectedChar(ch)))
        None -> Error(scanner.make_error(next, error.UnexpectedEof))
      }
  }
}

@internal
pub fn parse_string_if_available(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  use quoted <- result.try(parse_quoted_string(s))
  case quoted {
    #(Some(_), _) -> Ok(quoted)
    #(None, _) -> {
      use raw <- result.try(parse_raw_string(s))
      case raw {
        #(Some(_), _) -> Ok(raw)
        #(None, _) -> parse_identifier_string(s)
      }
    }
  }
}

@internal
pub fn parse_identifier_string(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s) {
    None -> Ok(#(None, s))
    Some(ch) ->
      case scanner.is_identifier_char(ch) {
        False -> Ok(#(None, s))
        True -> {
          let is_sign = ch == "+" || ch == "-"
          let is_dot = ch == "."
          let is_digit = scanner.is_ascii_digit(ch)
          case is_digit {
            True -> Ok(#(None, s))
            False ->
              case is_sign || is_dot {
                False -> collect_identifier(s)
                True -> collect_signed_or_dotted(s, ch, is_sign, is_dot)
              }
          }
        }
      }
  }
}

fn collect_signed_or_dotted(
  s: Scanner,
  first: String,
  is_sign: Bool,
  is_dot: Bool,
) -> Result(#(Option(String), Scanner), KdlError) {
  let #(_, s1) = scanner.advance(s)
  case is_sign, scanner.peek(s1) {
    True, Some(ch2) ->
      case
        scanner.is_identifier_char(ch2)
        && !scanner.is_ascii_digit(ch2)
        && ch2 != "."
      {
        True -> collect_identifier_continue(s1, [first])
        False ->
          case ch2 == "." {
            True -> {
              let #(_, s2) = scanner.advance(s1)
              collect_identifier_continue(s2, [".", first])
            }
            False -> check_keyword(first, s, s1)
          }
      }
    True, _ -> check_keyword(first, s, s1)
    False, _ ->
      case is_dot {
        True ->
          case scanner.peek(s1) {
            Some(ch2) ->
              case
                scanner.is_identifier_char(ch2) && !scanner.is_ascii_digit(ch2)
              {
                True -> collect_identifier_continue(s1, [first])
                False -> check_keyword(first, s, s1)
              }
            None -> check_keyword(first, s, s1)
          }
        False -> Ok(#(None, s))
      }
  }
}

fn collect_identifier(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  collect_identifier_continue(s, [])
}

fn collect_identifier_continue(
  s: Scanner,
  rev_prefix: List(String),
) -> Result(#(Option(String), Scanner), KdlError) {
  let #(chars, next) = collect_chars(s, rev_prefix)
  let value = chars |> list.reverse |> scanner.join
  case value == "" {
    True -> Ok(#(None, s))
    False -> check_keyword(value, s, next)
  }
}

fn collect_chars(s: Scanner, acc: List(String)) -> #(List(String), Scanner) {
  case scanner.peek(s) {
    Some(ch) ->
      case scanner.is_identifier_char(ch) {
        True -> {
          let #(_, next) = scanner.advance(s)
          collect_chars(next, [ch, ..acc])
        }
        False -> #(acc, s)
      }
    _ -> #(acc, s)
  }
}

fn check_keyword(
  value: String,
  start: Scanner,
  next: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.is_reserved_keyword(value) {
    True -> Error(scanner.make_error(start, error.BareKeyword))
    False -> Ok(#(Some(value), next))
  }
}

@internal
pub fn parse_quoted_string(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s) {
    Some("\"") -> {
      let #(_, s1) = scanner.advance(s)
      case scanner.peek(s1), scanner.peek_next(s1) {
        Some("\""), Some("\"") -> {
          let #(_, s2) = scanner.advance(s1)
          let #(_, s3) = scanner.advance(s2)
          parse_multiline_quoted(s3)
        }
        _, _ -> parse_quoted_body(s1, [])
      }
    }
    _ -> Ok(#(None, s))
  }
}

fn parse_quoted_body(
  s: Scanner,
  acc: List(String),
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s) {
    None -> Error(scanner.make_error(s, error.UnclosedString))
    Some("\"") -> {
      let #(_, next) = scanner.advance(s)
      Ok(#(Some(acc |> list.reverse |> scanner.join), next))
    }
    Some("\\") -> {
      let #(_, s1) = scanner.advance(s)
      use result <- result.try(parse_escape(s1))
      let #(value, s2) = result
      parse_quoted_body(s2, [value, ..acc])
    }
    Some(ch) ->
      case scanner.is_newline(ch) {
        True -> Error(scanner.make_error(s, error.UnclosedString))
        False ->
          case scanner.is_disallowed(ch) {
            True -> Error(scanner.make_error(s, error.DisallowedCodePoint(ch)))
            False -> {
              let #(_, next) = scanner.advance(s)
              parse_quoted_body(next, [ch, ..acc])
            }
          }
      }
  }
}

fn parse_escape(s: Scanner) -> Result(#(String, Scanner), KdlError) {
  case scanner.peek(s) {
    Some("n") -> esc("\n", s)
    Some("r") -> esc("\r", s)
    Some("t") -> esc("\t", s)
    Some("b") -> esc("\u{0008}", s)
    Some("f") -> esc("\u{000C}", s)
    Some("\\") -> esc("\\", s)
    Some("\"") -> esc("\"", s)
    Some("s") -> esc(" ", s)
    Some("u") -> parse_unicode_escape(s)
    Some(ch) ->
      case scanner.is_unicode_space(ch) || scanner.is_newline(ch) {
        True -> {
          Ok(#("", consume_whitespace_escape(s)))
        }
        False -> Error(scanner.make_error(s, error.InvalidEscape))
      }
    None -> Error(scanner.make_error(s, error.UnclosedString))
  }
}

fn esc(value: String, s: Scanner) -> Result(#(String, Scanner), KdlError) {
  let #(_, next) = scanner.advance(s)
  Ok(#(value, next))
}

fn parse_unicode_escape(s: Scanner) -> Result(#(String, Scanner), KdlError) {
  let #(_, s1) = scanner.advance(s)
  use s2 <- result.try(scanner.expect(s1, "{"))
  let #(hex, s3) = collect_until(s2, "}", [])
  use s4 <- result.try(scanner.expect(s3, "}"))
  case valid_unicode_hex(hex) {
    False -> Error(scanner.make_error(s, error.InvalidUnicodeEscape))
    True ->
      case int.base_parse(hex, 16) {
        Ok(i) ->
          case string.utf_codepoint(i) {
            Ok(cp) -> Ok(#(string.from_utf_codepoints([cp]), s4))
            Error(_) -> Error(scanner.make_error(s, error.InvalidUnicodeEscape))
          }
        Error(_) -> Error(scanner.make_error(s, error.InvalidUnicodeEscape))
      }
  }
}

fn valid_unicode_hex(hex: String) -> Bool {
  let chars = string.to_graphemes(hex)
  let len = list.length(chars)
  len > 0 && len <= 6 && list.all(chars, scanner.is_hex_digit)
}

fn collect_until(
  s: Scanner,
  delimiter: String,
  acc: List(String),
) -> #(String, Scanner) {
  case scanner.peek(s) {
    Some(ch) ->
      case ch != delimiter {
        True -> {
          let #(_, next) = scanner.advance(s)
          collect_until(next, delimiter, [ch, ..acc])
        }
        False -> #(acc |> list.reverse |> scanner.join, s)
      }
    _ -> #(acc |> list.reverse |> scanner.join, s)
  }
}

fn parse_multiline_quoted(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  let #(had_newline, s1) = scanner.consume_newline(s)
  case had_newline {
    False -> Error(scanner.make_error(s, error.UnclosedString))
    True -> parse_multiline_quoted_lines(s1, [], "")
  }
}

fn parse_multiline_quoted_lines(
  s: Scanner,
  rev_lines: List(String),
  current_line: String,
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s), scanner.peek_next(s), scanner.at(s.chars, s.pos + 2) {
    Some("\""), Some("\""), Some("\"") -> {
      let #(_, s1) = scanner.advance(s)
      let #(_, s2) = scanner.advance(s1)
      let #(_, s3) = scanner.advance(s2)
      use value <- result.try(process_multiline(
        [current_line, ..rev_lines] |> list.reverse,
        scanner.Quoted,
        s,
      ))
      Ok(#(Some(value), s3))
    }
    Some("\\"), _, _ -> {
      let #(_, s1) = scanner.advance(s)
      case scanner.peek(s1) {
        Some(ch) -> {
          case scanner.is_unicode_space(ch) || scanner.is_newline(ch) {
            True ->
              parse_multiline_quoted_lines(
                consume_whitespace_escape(s1),
                rev_lines,
                current_line,
              )
            False -> {
              let #(_, s2) = scanner.advance(s1)
              parse_multiline_quoted_lines(
                s2,
                rev_lines,
                current_line <> "\\" <> ch,
              )
            }
          }
        }
        None -> Error(scanner.make_error(s1, error.InvalidEscape))
      }
    }
    None, _, _ -> Error(scanner.make_error(s, error.UnclosedString))
    Some(ch), _, _ -> {
      case scanner.is_newline(ch) {
        True -> {
          let #(_, next) = scanner.consume_newline(s)
          parse_multiline_quoted_lines(next, [current_line, ..rev_lines], "")
        }
        False ->
          case scanner.is_disallowed(ch) {
            True -> Error(scanner.make_error(s, error.DisallowedCodePoint(ch)))
            False -> {
              let #(_, next) = scanner.advance(s)
              parse_multiline_quoted_lines(next, rev_lines, current_line <> ch)
            }
          }
      }
    }
  }
}

@internal
pub fn parse_raw_string(
  s: Scanner,
) -> Result(#(Option(String), Scanner), KdlError) {
  let #(hashes, after_hashes) = collect_hashes(s, [])
  case
    scanner.peek(after_hashes),
    scanner.peek_next(after_hashes),
    scanner.at(after_hashes.chars, after_hashes.pos + 2)
  {
    Some("\""), Some("\""), Some("\"") -> {
      let #(_, s1) = scanner.advance(after_hashes)
      let #(_, s2) = scanner.advance(s1)
      let #(_, s3) = scanner.advance(s2)
      parse_multiline_raw_lines(s3, hashes, [], "")
    }
    Some("\""), _, _ -> {
      let #(_, s1) = scanner.advance(after_hashes)
      parse_raw_body(s1, hashes, scanner.SingleLine, [])
    }
    _, _, _ -> Ok(#(None, s))
  }
}

fn parse_multiline_raw_lines(
  s: Scanner,
  hashes: List(String),
  rev_lines: List(String),
  current_line: String,
) -> Result(#(Option(String), Scanner), KdlError) {
  let #(had_newline, s1) = scanner.consume_newline(s)
  case had_newline {
    False -> Error(scanner.make_error(s, error.UnclosedString))
    True -> parse_multiline_raw_body_lines(s1, hashes, rev_lines, current_line)
  }
}

fn parse_multiline_raw_body_lines(
  s: Scanner,
  hashes: List(String),
  rev_lines: List(String),
  current_line: String,
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s) {
    None -> Error(scanner.make_error(s, error.UnclosedString))
    Some("\"") ->
      case raw_closer_at(s, hashes, scanner.Multiline) {
        True -> {
          let next = consume_raw_closer(s, hashes, scanner.Multiline)
          use value <- result.try(process_multiline(
            [current_line, ..rev_lines] |> list.reverse,
            scanner.Raw,
            s,
          ))
          Ok(#(Some(value), next))
        }
        False -> {
          let #(_, next) = scanner.advance(s)
          parse_multiline_raw_body_lines(
            next,
            hashes,
            rev_lines,
            current_line <> "\"",
          )
        }
      }
    Some(ch) ->
      case scanner.is_newline(ch) {
        True -> {
          let #(_, next) = scanner.consume_newline(s)
          parse_multiline_raw_body_lines(
            next,
            hashes,
            [current_line, ..rev_lines],
            "",
          )
        }
        False ->
          case scanner.is_disallowed(ch) {
            True -> Error(scanner.make_error(s, error.DisallowedCodePoint(ch)))
            False -> {
              let #(_, next) = scanner.advance(s)
              parse_multiline_raw_body_lines(
                next,
                hashes,
                rev_lines,
                current_line <> ch,
              )
            }
          }
      }
  }
}

fn collect_hashes(s: Scanner, acc: List(String)) -> #(List(String), Scanner) {
  case scanner.peek(s) {
    Some("#") -> {
      let #(_, next) = scanner.advance(s)
      collect_hashes(next, ["#", ..acc])
    }
    _ -> #(list.reverse(acc), s)
  }
}

fn parse_raw_body(
  s: Scanner,
  hashes: List(String),
  form: scanner.LineForm,
  acc: List(String),
) -> Result(#(Option(String), Scanner), KdlError) {
  case scanner.peek(s) {
    None -> Error(scanner.make_error(s, error.UnclosedString))
    Some("\"") ->
      case raw_closer_at(s, hashes, form) {
        True -> {
          let next = consume_raw_closer(s, hashes, form)
          let value = acc |> list.reverse |> scanner.join
          Ok(#(Some(value), next))
        }
        False -> {
          let #(_, next) = scanner.advance(s)
          parse_raw_body(next, hashes, form, ["\"", ..acc])
        }
      }
    Some(ch) -> {
      case form == scanner.SingleLine && scanner.is_newline(ch) {
        True -> Error(scanner.make_error(s, error.UnclosedString))
        False -> {
          let #(_, next) = scanner.advance(s)
          parse_raw_body(next, hashes, form, [ch, ..acc])
        }
      }
    }
  }
}

fn raw_closer_at(
  s: Scanner,
  hashes: List(String),
  form: scanner.LineForm,
) -> Bool {
  let offset = case form {
    scanner.Multiline -> 3
    scanner.SingleLine -> 1
  }
  case form {
    scanner.Multiline ->
      scanner.peek(s) == Some("\"")
      && scanner.peek_next(s) == Some("\"")
      && scanner.at(s.chars, s.pos + 2) == Some("\"")
      && hashes_match(s, s.pos + offset, hashes)
    scanner.SingleLine -> hashes_match(s, s.pos + offset, hashes)
  }
}

fn hashes_match(s: Scanner, pos: Int, hashes: List(String)) -> Bool {
  case hashes {
    [] -> True
    ["#", ..rest] ->
      scanner.at(s.chars, pos) == Some("#") && hashes_match(s, pos + 1, rest)
    _ -> False
  }
}

fn consume_raw_closer(
  s: Scanner,
  hashes: List(String),
  form: scanner.LineForm,
) -> Scanner {
  let count =
    list.length(hashes)
    + case form {
      scanner.Multiline -> 3
      scanner.SingleLine -> 1
    }
  consume_n(s, count)
}

fn consume_n(s: Scanner, n: Int) -> Scanner {
  case n <= 0 {
    True -> s
    False -> {
      let #(_, next) = scanner.advance(s)
      consume_n(next, n - 1)
    }
  }
}

fn consume_whitespace_escape(s: Scanner) -> Scanner {
  case scanner.peek(s) {
    Some(ch) ->
      case scanner.is_unicode_space(ch) {
        True -> {
          let #(_, next) = scanner.advance(s)
          consume_whitespace_escape(next)
        }
        False ->
          case scanner.is_newline(ch) {
            True -> {
              let #(_, next) = scanner.consume_newline(s)
              consume_whitespace_escape(next)
            }
            False -> s
          }
      }
    _ -> s
  }
}

fn process_multiline(
  lines: List(String),
  kind: scanner.StringKind,
  error_at: Scanner,
) -> Result(String, KdlError) {
  case last_line(lines, []) {
    #(None, _) -> Ok("")
    #(Some(prefix), body_lines) ->
      process_multiline_lines(body_lines, prefix, kind, error_at, True, "")
  }
}

fn last_line(
  lines: List(String),
  rev_body: List(String),
) -> #(Option(String), List(String)) {
  case lines {
    [] -> #(None, list.reverse(rev_body))
    [last] -> #(Some(last), list.reverse(rev_body))
    [line, ..rest] -> last_line(rest, [line, ..rev_body])
  }
}

fn process_multiline_lines(
  lines: List(String),
  prefix: String,
  kind: scanner.StringKind,
  error_at: Scanner,
  first: Bool,
  acc: String,
) -> Result(String, KdlError) {
  case lines {
    [] -> Ok(acc)
    [line, ..rest] -> {
      let next_acc = case first {
        True -> acc
        False -> acc <> "\n"
      }
      case line_is_unicode_space(line) {
        True ->
          process_multiline_lines(rest, prefix, kind, error_at, False, next_acc)
        False ->
          case string.starts_with(line, prefix) {
            False ->
              Error(scanner.make_error(error_at, error.InconsistentIndentation))
            True -> {
              let stripped = case string.split_once(line, prefix) {
                Ok(#("", rest)) -> rest
                _ -> line
              }
              use resolved <- result.try(case kind {
                scanner.Quoted -> resolve_escapes(stripped, error_at)
                scanner.Raw -> Ok(stripped)
              })
              process_multiline_lines(
                rest,
                prefix,
                kind,
                error_at,
                False,
                next_acc <> resolved,
              )
            }
          }
      }
    }
  }
}

fn line_is_unicode_space(line: String) -> Bool {
  line |> string.to_graphemes |> list.all(scanner.is_unicode_space)
}

fn resolve_escapes(
  input: String,
  error_at: Scanner,
) -> Result(String, KdlError) {
  resolve_escapes_loop(scanner.new(input), error_at, "")
}

fn resolve_escapes_loop(
  s: Scanner,
  error_at: Scanner,
  acc: String,
) -> Result(String, KdlError) {
  case scanner.peek(s) {
    None -> Ok(acc)
    Some("\\") -> {
      let #(_, s1) = scanner.advance(s)
      use parsed <- result.try(parse_escape(s1))
      let #(value, s2) = parsed
      resolve_escapes_loop(s2, error_at, acc <> value)
    }
    Some(ch) -> {
      case scanner.is_disallowed(ch) {
        True ->
          Error(scanner.make_error(error_at, error.DisallowedCodePoint(ch)))
        False -> {
          let #(_, next) = scanner.advance(s)
          resolve_escapes_loop(next, error_at, acc <> ch)
        }
      }
    }
  }
}