(* rudimentary recfile parser
     https://www.gnu.org/software/recutils/manual/recutils.html#The-Rec-Format *)

open Angstrom

type field = {
  k: string;
  v: string;
}

type record = field list

let newline = char '\n'
let is_newline = function '\n' -> true | _ -> false

let field_name =
  let first_char =
    satisfy (function 'a' .. 'z' | 'A' .. 'Z' | '%' -> true | _ -> false)
  in
  let subsequent_char =
    satisfy (function
      | 'a' .. 'z' | 'A' .. 'Z' | '0' .. '9' | '_' -> true
      | _ -> false)
  in
  lift2
    (fun hd tl -> String.of_seq (List.to_seq (hd :: tl)))
    first_char (many subsequent_char)

(* todo handle '\' escape and '+' on next line *)
let field_value = take_till is_newline <* newline

let field =
  let blank = satisfy (function ' ' | '\t' -> true | _ -> false) in
  let blanks = skip_many1 blank in
  lift3 (fun k () v -> { k; v }) field_name (char ':' *> blanks) field_value

let blank = newline *> return ()
let comment = (char '#' *> take_till is_newline <* newline) *> return ()
let record = many1 field

let records =
  let sep =
    (* at least one blank line *)
    skip_many comment *> blank *> skip_many (comment <|> blank)
  in
  sep_by1 sep record

let recfile : record list t =
  skip_many (comment <|> blank) *> records <* skip_many (comment <|> blank)

let parse s = parse_string ~consume:All recfile s