Add dict literal syntax (rescript-lang#6774)

* Handle dict literal syntax

Closes rescript-lang#6545

* Update test

* Move isTupleArray to res_parsetree_viewer

* Map locations correctly

* Fix location in parsing comments printing

* Make Dict limit to List grammar-wise

* Add changelog entry

Author: bloodyowl

CHANGELOG.md

@@ -15,6 +15,7 @@
 #### :rocket: New Feature
 
 - Allow coercing polyvariants to variants when we can guarantee that the runtime representation matches. https://github.com/rescript-lang/rescript-compiler/pull/6981
+- Add new dict literal syntax (`dict{"foo": "bar"}`). https://github.com/rescript-lang/rescript-compiler/pull/6774
 
 #### :nail_care: Polish
 

jscomp/syntax/src/res_comments_table.ml

@@ -1345,6 +1345,19 @@ and walk_expression expr t comments =
     walk_list
       [Expression parent_expr; Expression member_expr; Expression target_expr]
       t comments
+  | Pexp_apply
+      ( {
+          pexp_desc =
+            Pexp_ident
+              {
+                txt =
+                  Longident.Ldot
+                    (Longident.Ldot (Lident "Js", "Dict"), "fromArray");
+              };
+        },
+        [(Nolabel, key_values)] )
+    when Res_parsetree_viewer.is_tuple_array key_values ->
+    walk_list [Expression key_values] t comments
   | Pexp_apply (call_expr, arguments) ->
     let before, inside, after = partition_by_loc comments call_expr.pexp_loc in
     let after =

jscomp/syntax/src/res_core.ml

@@ -222,6 +222,7 @@ let get_closing_token = function
   | Lbrace -> Rbrace
   | Lbracket -> Rbracket
   | List -> Rbrace
+  | Dict -> Rbrace
   | LessThan -> GreaterThan
   | _ -> assert false
 
@@ -233,7 +234,7 @@ let rec go_to_closing closing_token state =
   | GreaterThan, GreaterThan ->
     Parser.next state;
     ()
-  | ((Token.Lbracket | Lparen | Lbrace | List | LessThan) as t), _ ->
+  | ((Token.Lbracket | Lparen | Lbrace | List | Dict | LessThan) as t), _ ->
     Parser.next state;
     go_to_closing (get_closing_token t) state;
     go_to_closing closing_token state
@@ -1896,6 +1897,9 @@ and parse_atomic_expr p =
     | List ->
       Parser.next p;
       parse_list_expr ~start_pos p
+    | Dict ->
+      Parser.next p;
+      parse_dict_expr ~start_pos p
     | Module ->
       Parser.next p;
       parse_first_class_module_expr ~start_pos p
@@ -3122,6 +3126,20 @@ and parse_record_expr_row p =
     | _ -> None)
   | _ -> None
 
+and parse_dict_expr_row p =
+  match p.Parser.token with
+  | String s -> (
+    let loc = mk_loc p.start_pos p.end_pos in
+    Parser.next p;
+    let field = Location.mkloc (Longident.Lident s) loc in
+    match p.Parser.token with
+    | Colon ->
+      Parser.next p;
+      let fieldExpr = parse_expr p in
+      Some (field, fieldExpr)
+    | _ -> Some (field, Ast_helper.Exp.ident ~loc:field.loc field))
+  | _ -> None
+
 and parse_record_expr_with_string_keys ~start_pos first_row p =
   let rows =
     first_row
@@ -3903,6 +3921,36 @@ and parse_list_expr ~start_pos p =
             loc))
       [(Asttypes.Nolabel, Ast_helper.Exp.array ~loc list_exprs)]
 
+and parse_dict_expr ~start_pos p =
+  let rows =
+    parse_comma_delimited_region ~grammar:Grammar.DictRows ~closing:Rbrace
+      ~f:parse_dict_expr_row p
+  in
+  let loc = mk_loc start_pos p.end_pos in
+  let to_key_value_pair
+      (record_item : Longident.t Location.loc * Parsetree.expression) =
+    match record_item with
+    | ( {Location.txt = Longident.Lident key; loc = keyLoc},
+        ({pexp_loc = value_loc} as value_expr) ) ->
+      Some
+        (Ast_helper.Exp.tuple
+           ~loc:(mk_loc keyLoc.loc_start value_loc.loc_end)
+           [
+             Ast_helper.Exp.constant ~loc:keyLoc (Pconst_string (key, None));
+             value_expr;
+           ])
+    | _ -> None
+  in
+  let key_value_pairs = List.filter_map to_key_value_pair rows in
+  Parser.expect Rbrace p;
+  Ast_helper.Exp.apply ~loc
+    (Ast_helper.Exp.ident ~loc
+       (Location.mkloc
+          (Longident.Ldot
+             (Longident.Ldot (Longident.Lident "Js", "Dict"), "fromArray"))
+          loc))
+    [(Asttypes.Nolabel, Ast_helper.Exp.array ~loc key_value_pairs)]
+
 and parse_array_exp p =
   let start_pos = p.Parser.start_pos in
   Parser.expect Lbracket p;

jscomp/syntax/src/res_grammar.ml

@@ -59,6 +59,7 @@ type t =
   | Pattern
   | AttributePayload
   | TagNames
+  | DictRows
 
 let to_string = function
   | OpenDescription -> "an open description"
@@ -120,6 +121,7 @@ let to_string = function
   | ExprFor -> "a for expression"
   | AttributePayload -> "an attribute payload"
   | TagNames -> "tag names"
+  | DictRows -> "rows of a dict"
 
 let is_signature_item_start = function
   | Token.At | Let | Typ | External | Exception | Open | Include | Module | AtAt
@@ -136,7 +138,7 @@ let is_atomic_pattern_start = function
 let is_atomic_expr_start = function
   | Token.True | False | Int _ | String _ | Float _ | Codepoint _ | Backtick
   | Uident _ | Lident _ | Hash | Lparen | List | Lbracket | Lbrace | LessThan
-  | Module | Percent | Forwardslash | ForwardslashDot ->
+  | Module | Percent | Forwardslash | ForwardslashDot | Dict ->
     true
   | _ -> false
 
@@ -151,7 +153,7 @@ let is_expr_start = function
   | For | Hash | If | Int _ | Lbrace | Lbracket | LessThan | Lident _ | List
   | Lparen | Minus | MinusDot | Module | Percent | Plus | PlusDot | String _
   | Switch | True | Try | Uident _ | Underscore (* _ => doThings() *)
-  | While | Forwardslash | ForwardslashDot ->
+  | While | Forwardslash | ForwardslashDot | Dict ->
     true
   | _ -> false
 
@@ -219,6 +221,10 @@ let is_mod_expr_start = function
     true
   | _ -> false
 
+let is_dict_row_start = function
+  | Token.String _ -> true
+  | _ -> false
+
 let is_record_row_start = function
   | Token.DotDotDot -> true
   | Token.Uident _ | Lident _ -> true
@@ -260,7 +266,7 @@ let is_block_expr_start = function
   | False | Float _ | For | Forwardslash | ForwardslashDot | Hash | If | Int _
   | Lbrace | Lbracket | LessThan | Let | Lident _ | List | Lparen | Minus
   | MinusDot | Module | Open | Percent | Plus | PlusDot | String _ | Switch
-  | True | Try | Uident _ | Underscore | While ->
+  | True | Try | Uident _ | Underscore | While | Dict ->
     true
   | _ -> false
 
@@ -278,6 +284,7 @@ let is_list_element grammar token =
   | FunctorArgs -> is_functor_arg_start token
   | ModExprList -> is_mod_expr_start token
   | TypeParameters -> is_type_parameter_start token
+  | DictRows -> is_dict_row_start token
   | RecordRows -> is_record_row_start token
   | RecordRowsStringKey -> is_record_row_string_key_start token
   | ArgumentList -> is_argument_start token

jscomp/syntax/src/res_parsetree_viewer.ml

@@ -743,3 +743,13 @@ let is_rewritten_underscore_apply_sugar expr =
   match expr.pexp_desc with
   | Pexp_ident {txt = Longident.Lident "_"} -> true
   | _ -> false
+
+let is_tuple_array (expr : Parsetree.expression) =
+  let is_plain_tuple (expr : Parsetree.expression) =
+    match expr with
+    | {pexp_desc = Pexp_tuple _} -> true
+    | _ -> false
+  in
+  match expr with
+  | {pexp_desc = Pexp_array items} -> List.for_all is_plain_tuple items
+  | _ -> false

jscomp/syntax/src/res_parsetree_viewer.mli

@@ -164,3 +164,5 @@ val has_if_let_attribute : Parsetree.attributes -> bool
 val is_rewritten_underscore_apply_sugar : Parsetree.expression -> bool
 
 val is_fun_newtype : Parsetree.expression -> bool
+
+val is_tuple_array : Parsetree.expression -> bool

jscomp/syntax/src/res_printer.ml

@@ -1406,6 +1406,49 @@ and print_record_declaration ~state (lds : Parsetree.label_declaration list)
          Doc.rbrace;
        ])
 
+and print_literal_dict_expr ~state (e : Parsetree.expression) cmt_tbl =
+  let force_break =
+    e.pexp_loc.loc_start.pos_lnum < e.pexp_loc.loc_end.pos_lnum
+  in
+  let tuple_to_row (e : Parsetree.expression) =
+    match e with
+    | {
+     pexp_desc =
+       Pexp_tuple
+         [
+           {pexp_desc = Pexp_constant (Pconst_string (name, _)); pexp_loc}; value;
+         ];
+    } ->
+      Some ((Location.mkloc (Longident.Lident name) pexp_loc, value), e)
+    | _ -> None
+  in
+  let rows =
+    match e with
+    | {pexp_desc = Pexp_array expressions} ->
+      List.filter_map tuple_to_row expressions
+    | _ -> []
+  in
+  Doc.breakable_group ~force_break
+    (Doc.concat
+       [
+         Doc.indent
+           (Doc.concat
+              [
+                Doc.soft_line;
+                Doc.join
+                  ~sep:(Doc.concat [Doc.text ","; Doc.line])
+                  (List.map
+                     (fun ((row, e) :
+                            (Longident.t Location.loc * Parsetree.expression)
+                            * Parsetree.expression) ->
+                       let doc = print_bs_object_row ~state row cmt_tbl in
+                       print_comments doc cmt_tbl e.pexp_loc)
+                     rows);
+              ]);
+         Doc.trailing_comma;
+         Doc.soft_line;
+       ])
+
 and print_constructor_declarations ~state ~private_flag
     (cds : Parsetree.constructor_declaration list) cmt_tbl =
   let force_break =
@@ -4031,6 +4074,24 @@ and print_pexp_apply ~state expr cmt_tbl =
     | [] -> doc
     | attrs ->
       Doc.group (Doc.concat [print_attributes ~state attrs cmt_tbl; doc]))
+  | Pexp_apply
+      ( {
+          pexp_desc =
+            Pexp_ident
+              {
+                txt =
+                  Longident.Ldot
+                    (Longident.Ldot (Lident "Js", "Dict"), "fromArray");
+              };
+        },
+        [(Nolabel, key_values)] )
+    when Res_parsetree_viewer.is_tuple_array key_values ->
+    Doc.concat
+      [
+        Doc.text "dict{";
+        print_literal_dict_expr ~state key_values cmt_tbl;
+        Doc.rbrace;
+      ]
   | Pexp_apply
       ( {pexp_desc = Pexp_ident {txt = Longident.Ldot (Lident "Array", "get")}},
         [(Nolabel, parent_expr); (Nolabel, member_expr)] )
@@ -4541,7 +4602,7 @@ and print_jsx_name {txt = lident} =
     Doc.join ~sep:Doc.dot segments
 
 and print_arguments_with_callback_in_first_position ~state args cmt_tbl =
-  (* Because the same subtree gets printed twice, we need to copy the cmtTbl.
+  (* Because the same subtree gets printed twice, we need to copy the cmt_tbl.
    * consumed comments need to be marked not-consumed and reprinted…
    * Cheng's different comment algorithm will solve this. *)
   let state = State.next_custom_layout state in
@@ -4624,7 +4685,7 @@ and print_arguments_with_callback_in_first_position ~state args cmt_tbl =
     Doc.custom_layout [Lazy.force fits_on_one_line; Lazy.force break_all_args]
 
 and print_arguments_with_callback_in_last_position ~state args cmt_tbl =
-  (* Because the same subtree gets printed twice, we need to copy the cmtTbl.
+  (* Because the same subtree gets printed twice, we need to copy the cmt_tbl.
    * consumed comments need to be marked not-consumed and reprinted…
    * Cheng's different comment algorithm will solve this. *)
   let state = state |> State.next_custom_layout in
@@ -5822,7 +5883,7 @@ let print_pattern p = print_pattern ~state:(State.init ()) p
 let print_implementation ~width (s : Parsetree.structure) ~comments =
   let cmt_tbl = CommentTable.make () in
   CommentTable.walk_structure s cmt_tbl comments;
-  (* CommentTable.log cmtTbl; *)
+  (* CommentTable.log cmt_tbl; *)
   let doc = print_structure ~state:(State.init ()) s cmt_tbl in
   (* Doc.debug doc; *)
   Doc.to_string ~width doc ^ "\n"

jscomp/syntax/src/res_scanner.ml

@@ -196,11 +196,16 @@ let scan_identifier scanner =
     (String.sub [@doesNotRaise]) scanner.src start_off
       (scanner.offset - start_off)
   in
-  if '{' == scanner.ch && str = "list" then (
+  match (scanner, str) with
+  | {ch = '{'}, "list" ->
     next scanner;
     (* TODO: this isn't great *)
-    Token.lookup_keyword "list{")
-  else Token.lookup_keyword str
+    Token.lookup_keyword "list{"
+  | {ch = '{'}, "dict" ->
+    next scanner;
+    (* TODO: this isn't great *)
+    Token.lookup_keyword "dict{"
+  | _ -> Token.lookup_keyword str
 
 let scan_digits scanner ~base =
   if base <= 10 then

jscomp/syntax/src/res_token.ml

@@ -89,6 +89,7 @@ type t =
   | PercentPercent
   | Comment of Comment.t
   | List
+  | Dict
   | TemplateTail of string * Lexing.position
   | TemplatePart of string * Lexing.position
   | Backtick
@@ -200,6 +201,7 @@ let to_string = function
   | PercentPercent -> "%%"
   | Comment c -> "Comment" ^ Comment.to_string c
   | List -> "list{"
+  | Dict -> "dict{"
   | TemplatePart (text, _) -> text ^ "${"
   | TemplateTail (text, _) -> "TemplateTail(" ^ text ^ ")"
   | Backtick -> "`"
@@ -224,6 +226,7 @@ let keyword_table = function
   | "include" -> Include
   | "let" -> Let
   | "list{" -> List
+  | "dict{" -> Dict
   | "module" -> Module
   | "mutable" -> Mutable
   | "of" -> Of
@@ -242,7 +245,7 @@ let keyword_table = function
 let is_keyword = function
   | Await | And | As | Assert | Constraint | Else | Exception | External | False
   | For | If | In | Include | Land | Let | List | Lor | Module | Mutable | Of
-  | Open | Private | Rec | Switch | True | Try | Typ | When | While ->
+  | Open | Private | Rec | Switch | True | Try | Typ | When | While | Dict ->
     true
   | _ -> false
 

jscomp/syntax/tests/parsing/grammar/expressions/dict.res

@@ -0,0 +1,11 @@
+// empty dict
+let x = dict{}
+
+// one value
+let x = dict{"foo": "bar"}
+
+// two values
+let x = dict{"foo": "bar", "bar": "baz"}
+
+let baz = "foo"
+let x = dict{"foo": "bar", "bar": "baz", "baz": baz}

jscomp/syntax/tests/parsing/grammar/expressions/expected/dict.res.txt

@@ -0,0 +1,7 @@
+let x = Js.Dict.fromArray [||]
+let x = Js.Dict.fromArray [|("foo", {js|bar|js})|]
+let x = Js.Dict.fromArray [|("foo", {js|bar|js});("bar", {js|baz|js})|]
+let baz = {js|foo|js}
+let x =
+  Js.Dict.fromArray
+    [|("foo", {js|bar|js});("bar", {js|baz|js});("baz", baz)|]