module Lexer =

    open FParsec.Primitives
    open FParsec.CharParsers

    let escapeCharacters<'a> : Parser<string, 'a> =
        ["+"; "-"; "&&"; "||"; "!"; "("; ")"; "{"; "}"; "["; "]"; "^"; "\""; "~"; "*"; "?"; ":"; "\\"]  |> Seq.map pstring |> Seq.reduce (<|>)

    let wordOperator<'a> : Parser<Token, 'a> =
        pipe3 spaces1 (pstring "TO" <|> pstring "OR" <|> pstring "AND" <|> pstring "NOT") spaces1  (fun _ s _ -> Operator s)

    let operator<'a> : Parser<Token, 'a> =
        (escapeCharacters |>> Operator) <|> attempt wordOperator <|> (spaces1 |>> fun c -> Operator "OR")

    let term<'a> : Parser<Token, 'a> =
        many1Chars (notFollowedBy operator >>. anyChar) |>> Term

    let phrase<'a> : Parser<Token, 'a> =
        between (pchar '\"') (pchar '\"') (manyStrings ((notFollowedBy operator >>. anyString 1) <|> (skipString "\\" >>. escapeCharacters))) |>> Phrase

    let exec<'a> : Parser<list<Token>, 'a> =
        manyTill (term <|> phrase <|> operator) eof

    let tokenize input =
        let r = runParserOnString exec "" "" input
        match r with
        | Success (a, b, c) ->
            a
        | Failure (a, b, c) ->
            failwith a