mlox/lib/parser.ml

192 lines
6 KiB
OCaml
Raw Normal View History

let ( let* ) = Result.bind
2024-08-12 16:31:28 +02:00
open Error
open Expr
open Lexer
type parse_result = (expr_node, parser_error list) result
type expr_result = (expr_node, parser_error) result
2024-08-12 16:31:28 +02:00
type state = { tokens : token list; errors_rev : parser_error list }
let is_at_end state = (List.hd !state.tokens).token_type == Eof
let append_error msg pos state =
let e = ParserError.make pos msg in
{ state with errors_rev = e :: state.errors_rev }
let advance state = state := { !state with tokens = List.tl !state.tokens }
let peek state = List.hd !state.tokens
let next state =
assert (not ((List.hd !state.tokens).token_type == Eof));
let token = List.hd !state.tokens in
advance state;
token
let advance_if state tt =
if (peek state).token_type == tt then (
advance state;
true)
else false
let matches state tts =
let f = ( == ) (peek state).token_type in
Array.fold_left (fun acc tt -> acc || f tt) false tts
let cur_pos state = (peek state).pos
let collect_chain (state : state ref) (tts : token_type array)
(higher_prec : state ref -> expr_result) : ((expr_node * token) array, parser_error) result =
let rec collect_chain_rec (acc : (expr_node * token) list) =
if (not (is_at_end state)) && matches state tts then
let token = next state in
let* expr = higher_prec state in
let acc = (expr, token) :: acc in
collect_chain_rec acc
else Ok acc
in
collect_chain_rec [] |> Result.map (fun l -> Array.of_list (List.rev l))
let primary (state : state ref) : expr_result =
let pos = cur_pos state in
match (peek state).token_type with
| Number x ->
advance state;
Ok (make_number pos x)
| String s ->
advance state;
Ok (make_string pos s)
| True ->
advance state;
Ok (make_bool pos true)
| False ->
advance state;
Ok (make_bool pos false)
| Nil ->
advance state;
Ok (make_nil pos)
| tt ->
let msg = Printf.sprintf "Unexpected %s, expected valid expression" (show_token_type tt) in
let pos = (peek state).pos in
Error { msg; pos }
let rec grouping (state : state ref) : expr_result =
if matches state [| LeftParen |] then (
advance state;
let* expr = expression state in
if advance_if state RightParen then Ok expr (* expect a ) here *)
else
let pos = cur_pos state in
let tt = (peek state).token_type in
let msg = Printf.sprintf "Expected RightParen, got %s instead" (show_token_type tt) in
Error { pos; msg })
else primary state
and neg_not (state : state ref) : expr_result =
if matches state [| Bang; Minus |] then
let token = next state in
let pos = token.pos in
let* expr = neg_not state in
let op =
match token.token_type with
| Bang -> Not
| Minus -> Neg
| _ -> assert false (* should only be here if tt is - ! *)
2024-08-12 16:31:28 +02:00
in
let expr = make_unary pos op expr in
Ok expr
else grouping state
and mul_or_div (state : state ref) : expr_result =
let* expr = neg_not state in
let* exprs_tokens = collect_chain state [| Star; Slash |] neg_not in
let f acc (expr, token) =
let pos = token.pos in
let op : binary_op =
match token.token_type with
| Star -> Mul
| Slash -> Div
| _ -> assert false (* should only be here if tt is * / *)
2024-08-12 18:28:21 +02:00
in
make_binary pos op acc expr
in
let expr = Array.fold_left f expr exprs_tokens in
Ok expr
and sum_or_diff (state : state ref) : expr_result =
let* expr = mul_or_div state in
let* exprs_tokens = collect_chain state [| Plus; Minus |] mul_or_div in
let f acc (expr, token) =
let pos = token.pos in
let op : binary_op =
match token.token_type with
| Plus -> Plus
| Minus -> Minus
| _ -> assert false (* should only be here if tt is + - *)
2024-08-12 18:28:21 +02:00
in
make_binary pos op acc expr
in
let expr = Array.fold_left f expr exprs_tokens in
Ok expr
and inequality (state : state ref) : expr_result =
(* TODO: maybe rework to only have Less and Greater as ops; performance? *)
let* expr = sum_or_diff state in
let* exprs_tokens =
collect_chain state [| Greater; GreaterEqual; Less; LessEqual |] sum_or_diff
in
let f acc (expr, token) =
let pos = token.pos in
let (op : binary_op) =
2024-08-12 18:28:21 +02:00
match token.token_type with
| Greater -> Greater
| Less -> Less
| GreaterEqual -> GreaterEqual
| LessEqual -> LessEqual
| _ -> assert false (* should only be here if tt is > < >= <= *)
2024-08-12 18:28:21 +02:00
in
make_binary pos op acc expr
in
let expr = Array.fold_left f expr exprs_tokens in
Ok expr
and equality (state : state ref) : expr_result =
let* expr = inequality state in
let* exprs_tokens = collect_chain state [| EqualEqual; BangEqual |] inequality in
let f acc (expr, token) =
let pos = token.pos in
match token.token_type with
| EqualEqual -> make_binary pos Equal acc expr
| BangEqual ->
let expr = make_binary pos Equal acc expr in
make_unary pos Not expr
| _ -> assert false (* should only be here if tt is == != *)
in
let expr = Array.fold_left f expr exprs_tokens in
Ok expr
2024-08-12 16:31:28 +02:00
and expression (state : state ref) : expr_result = equality state
let rec synchronise (state : state ref) =
match (peek state).token_type with
| Semicolon -> advance state
| Class | Fun | Var | For | If | While | Print | Return | Eof -> ()
| _ ->
advance state;
synchronise state
2024-08-12 16:31:28 +02:00
let parse (tokens : token list) : parse_result =
let state = ref { tokens; errors_rev = [] } in
let result = expression state |> Result.map_error (fun e -> [ e ]) in
assert (Result.is_error result || (peek state).token_type = Eof);
result
(* let expr = State.expression state in
let state =
if not (State.is_at_end state) then
let tt = (State.peek state).token_type in
let msg = Printf.sprintf "Unexpected %s at end" (show_token_type tt) in
State.append_error msg (State.peek state).pos state
else state
in
(* if List.length state.errors_rev != 0 then Ok expr else Error (List.rev state.errors_rev) *)
match state.errors_rev with [] -> Ok expr | es -> Error (List.rev es) *)