2024-08-14 14:15:01 +02:00
|
|
|
let ( let* ) = Result.bind
|
|
|
|
|
|
2024-08-12 16:31:28 +02:00
|
|
|
open Error
|
|
|
|
|
open Expr
|
|
|
|
|
open Lexer
|
2024-08-25 02:12:51 +02:00
|
|
|
open Stmt
|
2024-08-12 16:31:28 +02:00
|
|
|
|
2024-08-25 02:12:51 +02:00
|
|
|
type parse_result = (stmt_node list, parser_error list) result
|
|
|
|
|
type stmt_result = (stmt_node, parser_error) result
|
2024-08-17 13:32:59 +02:00
|
|
|
type expr_result = (expr_node, parser_error) result
|
2024-08-25 03:07:18 +02:00
|
|
|
type state = { tokens : token list }
|
2024-08-12 16:31:28 +02:00
|
|
|
|
2024-08-14 14:15:01 +02:00
|
|
|
let is_at_end state = (List.hd !state.tokens).token_type == Eof
|
2024-08-25 03:07:18 +02:00
|
|
|
let advance state = state := { tokens = List.tl !state.tokens }
|
2024-08-14 14:15:01 +02:00
|
|
|
|
|
|
|
|
let next state =
|
|
|
|
|
assert (not ((List.hd !state.tokens).token_type == Eof));
|
|
|
|
|
let token = List.hd !state.tokens in
|
|
|
|
|
advance state;
|
|
|
|
|
token
|
|
|
|
|
|
2024-08-25 03:07:18 +02:00
|
|
|
let peek state = List.hd !state.tokens
|
|
|
|
|
let peek_tt (state : state ref) : token_type = (peek state).token_type
|
|
|
|
|
let cur_pos state = (peek state).pos
|
|
|
|
|
|
2024-08-14 14:15:01 +02:00
|
|
|
let advance_if state tt =
|
2024-08-25 02:12:51 +02:00
|
|
|
if peek_tt state == tt then (
|
2024-08-14 14:15:01 +02:00
|
|
|
advance state;
|
|
|
|
|
true)
|
|
|
|
|
else false
|
|
|
|
|
|
2024-08-25 02:12:51 +02:00
|
|
|
let consume state tt =
|
|
|
|
|
if advance_if state tt then Ok ()
|
|
|
|
|
else
|
|
|
|
|
Error
|
|
|
|
|
(ParserError.make (cur_pos state)
|
|
|
|
|
(Printf.sprintf "Expected %s, but got %s" (show_token_type tt)
|
|
|
|
|
(show_token_type (peek_tt state))))
|
|
|
|
|
|
2024-08-14 14:15:01 +02:00
|
|
|
let matches state tts =
|
2024-08-25 02:12:51 +02:00
|
|
|
let f = ( == ) (peek_tt state) in
|
2024-08-14 14:15:01 +02:00
|
|
|
Array.fold_left (fun acc tt -> acc || f tt) false tts
|
|
|
|
|
|
|
|
|
|
let collect_chain (state : state ref) (tts : token_type array)
|
2024-08-17 13:32:59 +02:00
|
|
|
(higher_prec : state ref -> expr_result) : ((expr_node * token) array, parser_error) result =
|
|
|
|
|
let rec collect_chain_rec (acc : (expr_node * token) list) =
|
2024-08-14 14:15:01 +02:00
|
|
|
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 =
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = cur_pos state in
|
2024-08-25 02:12:51 +02:00
|
|
|
match peek_tt state with
|
2024-08-17 13:32:59 +02:00
|
|
|
| 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)
|
2024-08-14 14:15:01 +02:00
|
|
|
| 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
|
2024-08-17 13:32:59 +02:00
|
|
|
if advance_if state RightParen then Ok expr (* expect a ) here *)
|
2024-08-14 12:36:36 +02:00
|
|
|
else
|
2024-08-14 14:15:01 +02:00
|
|
|
let pos = cur_pos state in
|
2024-08-25 02:12:51 +02:00
|
|
|
let tt = peek_tt state in
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = token.pos in
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-17 13:32:59 +02:00
|
|
|
let expr = make_unary pos op expr in
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-25 02:12:51 +02:00
|
|
|
let f acc (expr, (token : token)) =
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = token.pos in
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-17 13:32:59 +02:00
|
|
|
make_binary pos op acc expr
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-25 02:12:51 +02:00
|
|
|
let f acc (expr, (token : token)) =
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = token.pos in
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-17 13:32:59 +02:00
|
|
|
make_binary pos op acc expr
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-25 02:12:51 +02:00
|
|
|
let f acc (expr, (token : token)) =
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = token.pos in
|
2024-08-14 14:15:01 +02:00
|
|
|
let (op : binary_op) =
|
2024-08-12 18:28:21 +02:00
|
|
|
match token.token_type with
|
2024-08-14 14:15:01 +02:00
|
|
|
| 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
|
2024-08-17 13:32:59 +02:00
|
|
|
make_binary pos op acc expr
|
2024-08-14 14:15:01 +02:00
|
|
|
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
|
2024-08-25 02:12:51 +02:00
|
|
|
let f acc (expr, (token : token)) =
|
2024-08-17 13:32:59 +02:00
|
|
|
let pos = token.pos in
|
2024-08-14 14:15:01 +02:00
|
|
|
match token.token_type with
|
2024-08-17 13:32:59 +02:00
|
|
|
| EqualEqual -> make_binary pos Equal acc expr
|
2024-08-14 14:15:01 +02:00
|
|
|
| BangEqual ->
|
2024-08-17 13:32:59 +02:00
|
|
|
let expr = make_binary pos Equal acc expr in
|
|
|
|
|
make_unary pos Not expr
|
2024-08-14 14:15:01 +02:00
|
|
|
| _ -> 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
|
|
|
|
2024-08-14 14:15:01 +02:00
|
|
|
and expression (state : state ref) : expr_result = equality state
|
|
|
|
|
|
2024-08-25 02:12:51 +02:00
|
|
|
let statement (state : state ref) : stmt_result =
|
|
|
|
|
let pos = cur_pos state in
|
|
|
|
|
match peek_tt state with
|
|
|
|
|
| Print ->
|
|
|
|
|
advance state;
|
|
|
|
|
let* expr = expression state in
|
|
|
|
|
let* _ = consume state Semicolon in
|
|
|
|
|
let stmt = make_print pos expr in
|
|
|
|
|
Ok stmt
|
|
|
|
|
| tt ->
|
2024-08-25 03:07:18 +02:00
|
|
|
advance state;
|
2024-08-25 02:12:51 +02:00
|
|
|
let msg =
|
|
|
|
|
Printf.sprintf "Statement stating with %s not yet implemented" (show_token_type tt)
|
|
|
|
|
in
|
|
|
|
|
Error (ParserError.make pos msg)
|
|
|
|
|
|
2024-08-14 14:15:01 +02:00
|
|
|
let rec synchronise (state : state ref) =
|
2024-08-25 02:12:51 +02:00
|
|
|
match peek_tt state with
|
2024-08-14 14:15:01 +02:00
|
|
|
| Semicolon -> advance state
|
|
|
|
|
| Class | Fun | Var | For | If | While | Print | Return | Eof -> ()
|
|
|
|
|
| _ ->
|
|
|
|
|
advance state;
|
|
|
|
|
synchronise state
|
2024-08-12 16:31:28 +02:00
|
|
|
|
2024-08-25 03:07:18 +02:00
|
|
|
let rec parse_impl (state : state ref) : parse_result =
|
2024-08-25 02:12:51 +02:00
|
|
|
let result = statement state in
|
|
|
|
|
match result with
|
|
|
|
|
| Ok stmt when peek_tt state == Eof -> Ok [ stmt ]
|
|
|
|
|
| Ok stmt ->
|
2024-08-25 03:07:18 +02:00
|
|
|
print_endline (show_stmt stmt.stmt);
|
|
|
|
|
let* stmts = parse_impl state in
|
2024-08-25 02:12:51 +02:00
|
|
|
Ok (stmt :: stmts)
|
|
|
|
|
| Error e -> (
|
|
|
|
|
synchronise state;
|
|
|
|
|
if peek_tt state == Eof then Error [ e ]
|
|
|
|
|
else
|
2024-08-25 03:07:18 +02:00
|
|
|
let tail_result = parse_impl state in
|
2024-08-25 02:12:51 +02:00
|
|
|
match tail_result with Ok _ -> Error [ e ] | Error es -> Error (e :: es))
|
2024-08-25 03:07:18 +02:00
|
|
|
|
|
|
|
|
let parse (tokens : token list) : parse_result =
|
|
|
|
|
(* filter out all the comment tokens *)
|
|
|
|
|
let tokens =
|
|
|
|
|
List.filter (fun tok -> match tok.token_type with Comment _ -> false | _ -> true) tokens
|
|
|
|
|
in
|
|
|
|
|
let state = ref { tokens } in
|
|
|
|
|
parse_impl state
|