Adding support for Record Java extension and accent

semgrep-core/src/api/AST_generic_to_v1.ml

@@ -923,8 +923,10 @@ and
                          cimplements = v_cimplements;
                          cbody = v_cbody;
                          cmixins = v_cmixins;
+                         cparams = v_cparams;
                        } =
   let v_cbody = map_bracket (map_of_list map_field) v_cbody in
+  let v_cparams = map_parameters v_cparams in
   let v_cmixins = map_of_list map_type_ v_cmixins in
   let v_cimplements = map_of_list map_type_ v_cimplements in
   let v_cextends = map_of_list map_type_ v_cextends in
@@ -935,6 +937,7 @@ and
     cimplements = v_cimplements;
     cbody = v_cbody;
     cmixins = v_cmixins;
+    cparams = v_cparams;
   }
 
 and map_class_kind = function
@@ -943,6 +946,7 @@ and map_class_kind = function
   | Trait -> `Trait
   | Object -> `Object
   | AtInterface -> `AtInterface
+  | RecordClass -> `RecordClass
 
 and map_directive =
   function

semgrep-core/src/api/AST_generic_v1.atd

@@ -934,6 +934,8 @@ type class_definition = {
   (* class_kind in type_ is usually a Trait *)
   cmixins:      type_ list; (* PHP 'uses' *)
 
+  (* for Java Record or Scala Classes; we could transpile them into fields *)
+  cparams: parameters;
   (* newscope: note: this can be an empty fake bracket when used in Partial.*)
   cbody: field list bracket;
 }
@@ -944,6 +946,8 @@ type class_kind = [
   | Trait
   (* Kotlin, Scala *)
   | Object
+  (* Java 'record', Scala 'case class' *)
+  | RecordClass
   (* java: *)
   | AtInterface (* @interface, a.k.a annotation type declaration *)
   ]

semgrep-core/src/core/ast/AST_generic.ml

@@ -1500,6 +1500,8 @@ and class_definition = {
   (* class_kind in type_ is usually a Trait *)
   cmixins:      type_ list; (* PHP 'uses' *)
 
+  (* for Java Record or Scala Classes; we could transpile them into fields *)
+  cparams: parameters;
   (* newscope:
    * note: this can be an empty fake bracket when used in Partial.
    * TODO? use an option here?
@@ -1515,6 +1517,8 @@ and class_kind =
   | Trait
   (* Kotlin, Scala *)
   | Object
+  (* Java 'record', Scala 'case class' *)
+  | RecordClass
   (* java: *)
   | AtInterface (* @interface, a.k.a annotation type declaration *)
 (*e: type [[AST_generic.class_kind]] *)

semgrep-core/src/core/ast/Map_AST.ml

@@ -853,18 +853,21 @@ let (mk_visitor: visitor_in -> visitor_out) = fun vin ->
       cimplements = v_cimplements;
       cbody = v_cbody;
       cmixins = v_cmixins;
+      cparams;
     } =
     let v_cbody = map_bracket (map_of_list map_field) v_cbody in
     let v_cmixins = map_of_list map_type_ v_cmixins in
     let v_cimplements = map_of_list map_type_ v_cimplements in
     let v_cextends = map_of_list map_type_ v_cextends in
     let v_ckind = map_class_kind v_ckind in
+    let cparams = map_parameters cparams in
     {
       ckind = v_ckind;
       cextends = v_cextends;
       cimplements = v_cimplements;
       cbody = v_cbody;
       cmixins = v_cmixins;
+      cparams;
     }
 
   and map_class_kind (x,t) =

semgrep-core/src/core/ast/Meta_AST.ml

@@ -1139,11 +1139,15 @@ and
     cimplements = v_cimplements;
     cbody = v_cbody;
     cmixins = v_cmixins;
+    cparams;
   } =
   let bnds = [] in
   let arg = vof_bracket (OCaml.vof_list vof_field) v_cbody in
   let bnd = ("cbody", arg) in
   let bnds = bnd :: bnds in
+  let arg = vof_parameters cparams in
+  let bnd = ("cparams", arg) in
+  let bnds = bnd :: bnds in
   let arg = OCaml.vof_list vof_type_ v_cmixins in
   let bnd = ("cmixins", arg) in
   let bnds = bnd :: bnds in
@@ -1163,6 +1167,7 @@ and vof_class_kind_bis =
   | Trait -> OCaml.VSum ("Trait", [])
   | AtInterface -> OCaml.VSum ("AtInterface", [])
   | Object -> OCaml.VSum ("Object", [])
+  | RecordClass -> OCaml.VSum ("RecordClass", [])
 and vof_ident_and_id_info (v1, v2) =
   let v1 = vof_ident v1 in
   let v2 = vof_id_info v2 in

semgrep-core/src/core/ast/Visitor_AST.ml

@@ -880,11 +880,13 @@ let (mk_visitor: visitor_in -> visitor_out) = fun vin ->
       cimplements = v_cimplements;
       cmixins = v_mixins;
       cbody = v_cbody;
+      cparams;
     } =
       let arg = v_class_kind v_ckind in
       let arg = v_list v_type_ v_cextends in
       let arg = v_list v_type_ v_cimplements in
       let arg = v_list v_type_ v_mixins in
+      v_parameters cparams;
       let arg = v_bracket (v_list v_field) v_cbody in
       ()
     in

semgrep-core/src/matching/Generic_vs_generic.ml

@@ -2553,19 +2553,20 @@ and m_list__m_type_any_order (xsa: A.type_ list) (xsb: A.type_ list) =
 (*s: function [[Generic_vs_generic.m_class_definition]] *)
 and m_class_definition a b =
   match a, b with
-    { A. ckind = a1; cextends = a2; cimplements = a3; cbody = a4;
-      cmixins = a5;
+    { A. ckind = a1; cextends = a2; cimplements = a3; cmixins = a5;
+      cbody = a4; cparams = a6;
     },
-    { B. ckind = b1; cextends = b2; cimplements = b3; cbody = b4;
-      cmixins = b5;
+    { B. ckind = b1; cextends = b2; cimplements = b3; cmixins = b5;
+      cbody = b4; cparams = b6;
     } ->
       m_class_kind a1 b1 >>= (fun () ->
         (* TODO: use also m_list_in_any_order? regressions for python? *)
         (m_list__m_type_) a2 b2 >>= (fun () ->
           (m_list__m_type_any_order) a3 b3 >>= (fun () ->
             (m_list__m_type_any_order) a5 b5 >>= (fun () ->
-              m_bracket (m_fields) a4 b4
-            ))))
+              m_parameters a6 b6 >>= (fun () ->
+                m_bracket (m_fields) a4 b4
+              )))))
 (*e: function [[Generic_vs_generic.m_class_definition]] *)
 
 (*s: function [[Generic_vs_generic.m_class_kind]] *)
@@ -2577,9 +2578,10 @@ and m_class_kind_bis a b =
   | A.Trait, B.Trait
   | A.AtInterface, B.AtInterface
   | A.Object, B.Object
+  | A.RecordClass, B.RecordClass
     -> return ()
   | A.Class, _ | A.Interface, _ | A.Trait, _
-  | A.AtInterface, _ | A.Object, _
+  | A.AtInterface, _ | A.Object, _ | A.RecordClass, _
     -> fail ()
 (*e: function [[Generic_vs_generic.m_class_kind]] *)
 

semgrep-core/src/parsing/Parse_target.ml

@@ -207,9 +207,8 @@ let just_parse_with_lang lang file =
          * an invoke because of a segfault/memory-leak), but when both parsers
          * fail, it's better to give the tree-sitter parsing error now.
         *)
-        Pfff (throw_tokens Parse_java.parse);
-        (* TODO: move before, but need fix tree-sitter-java first *)
         TreeSitter Parse_java_tree_sitter.parse;
+        Pfff (throw_tokens Parse_java.parse);
       ]
         Java_to_generic.program
   | Lang.Go ->

semgrep-core/src/parsing/pfff/Python_to_generic.ml

@@ -537,7 +537,7 @@ and stmt_aux x =
       in
       let ent = G.basic_entity v1 v4 in
       let def = { G.ckind = (G.Class, v0); cextends = v2;
-                  cimplements = []; cmixins = [];
+                  cimplements = []; cmixins = []; cparams = [];
                   cbody = fb (v3 |> List.map(fun x ->G.FieldStmt x);)
                 } in
       [G.DefStmt (ent, G.ClassDef def) |> G.s]

semgrep-core/src/parsing/pfff/java_to_generic.ml

@@ -229,12 +229,13 @@ and expr e =
       (match v3 with
        | None -> G.Call (G.IdSpecial (G.New, v0), (lp,(G.ArgType v1)::v2,rp))
        | Some decls ->
-           let anonclass = G.AnonClass { G.
-                                         ckind = (G.Class, v0);
-                                         cextends = [v1];
-                                         cimplements = []; cmixins = [];
-                                         cbody = decls |> bracket (List.map (fun x -> G.FieldStmt x))
-                                       }
+           let anonclass =
+             G.AnonClass { G.
+                           ckind = (G.Class, v0);
+                           cextends = [v1];
+                           cimplements = []; cmixins = []; cparams = [];
+                           cbody = decls |> bracket (List.map (fun x -> G.FieldStmt x))
+                         }
            in
            G.Call (G.IdSpecial (G.New, v0), (lp,(G.Arg anonclass)::v2,rp))
       )
@@ -315,7 +316,7 @@ and expr e =
       let v2 = typ v2 in
       G.TypedMetavar (v1, Parse_info.fake_info " ", v2)
   | Lambda (v1, t, v2) ->
-      let v1 = params v1 in
+      let v1 = parameters v1 in
       let v2 = stmt v2 in
       G.Lambda { G.fparams = v1; frettype = None; fbody = v2;
                  fkind = (G.Arrow, t); }
@@ -455,7 +456,7 @@ and init =
   | ArrayInit v1 -> let v1 = bracket (list init) v1 in
       G.Container (G.Array, v1)
 
-and params v = List.map parameter_binding v
+and parameters v = List.map parameter_binding v
 and parameter_binding = function
   | ParamClassic v | ParamReceiver v ->
       let (ent, t) = var v in
@@ -474,7 +475,7 @@ and
     m_body = m_body
   } =
   let ent, rett = var m_var in
-  let v2 = params m_formals in
+  let v2 = parameters m_formals in
   let v3 = list typ m_throws in
   let v4 = stmt m_body in
   let throws = v3 |> List.map (fun t ->
@@ -516,14 +517,16 @@ and class_decl {
   cl_mods = cl_mods;
   cl_extends = cl_extends;
   cl_impls = cl_impls;
-  cl_body = cl_body
+  cl_body = cl_body;
+  cl_formals;
 } =
   let v1 = ident cl_name in
   let v2 = class_kind cl_kind in
   let v3 = list type_parameter cl_tparams in
   let v4 = modifiers cl_mods in
   let v5 = option typ cl_extends in
   let v6 = list ref_type cl_impls in
+  let cparams = parameters cl_formals in
   let v7 = class_body cl_body in
   let fields = v7 |> bracket (List.map (fun x -> G.FieldStmt x)) in
   let ent = { (G.basic_entity v1 v4) with
@@ -533,6 +536,7 @@ and class_decl {
                cextends = Common.opt_to_list v5;
                cimplements = v6;
                cmixins = [];
+               cparams;
                cbody = fields;
              } in
   ent, cdef
@@ -544,6 +548,7 @@ and class_kind (x, t) =
    | ClassRegular ->  G.Class
    | Interface -> G.Interface
    | AtInterface -> G.AtInterface
+   | Record -> G.RecordClass
   ), t
 
 and decl decl =

semgrep-core/src/parsing/pfff/js_to_generic.ml

@@ -549,7 +549,7 @@ and class_ { c_extends; c_implements; c_body; c_kind; c_attrs;  } =
   let attrs = list attribute c_attrs in
   let cimplements = list type_ c_implements in
   { G.ckind = H.conv_class_kind c_kind; cextends; cimplements;
-    cmixins = []; cbody = v2;}, attrs
+    cmixins = []; cparams = []; cbody = v2;}, attrs
 
 
 and field_classic {fld_name=v1; fld_attrs = v2; fld_type = vt; fld_body = v3} =

semgrep-core/src/parsing/pfff/php_to_generic.ml

@@ -544,6 +544,7 @@ and class_def {
               cextends = extends |> Common.opt_to_list;
               cimplements = implements;
               cmixins = uses;
+              cparams = [];
               cbody = t1,
                       fields |> List.map (fun def ->
                         G.FieldStmt (G.DefStmt def |> G.s)

semgrep-core/src/parsing/pfff/ruby_to_generic.ml

@@ -523,7 +523,7 @@ and definition def =
            in
            let def = { G.ckind = (G.Class, t); cextends = extends;
                        (* TODO: this is done by special include/require builtins *)
-                       cimplements = []; cmixins = [];
+                       cimplements = []; cmixins = []; cparams = [];
                        cbody = fb ([G.FieldStmt body]);
                      } in
            G.DefStmt (ent, G.ClassDef def) |> G.s

semgrep-core/src/parsing/tree_sitter/Parse_csharp_tree_sitter.ml

@@ -1079,7 +1079,7 @@ and expression (env : env) (x : CST.expression) : AST.expr =
        let v5 = token env v5 (* "}" *) in
        AnonClass {
          ckind = (Class, v1);
-         cextends = []; cimplements = []; cmixins = [];
+         cextends = []; cimplements = []; cmixins = []; cparams = [];
          cbody = (v2, v3, v5);
        }
    | `Array_crea_exp (v1, v2, v3) ->
@@ -2496,8 +2496,7 @@ and class_interface_struct (env : env) class_kind (v1, v2, v3, v4, v5, v6, v7, v
   AST.DefStmt (ent, AST.ClassDef {
     ckind = (class_kind, v3);
     cextends = v6;
-    cimplements = [];
-    cmixins = [];
+    cimplements = []; cmixins = []; cparams = [];
     cbody = (open_bra, fields, close_bra);
   }) |> AST.s
 

semgrep-core/src/parsing/tree_sitter/Parse_java_tree_sitter.ml

@@ -1283,8 +1283,22 @@ and enum_body (env : env) ((v1, v2, v3, v4, v5) : CST.enum_body) =
   let _v5 = token env v5 (* "}" *) in
   v2, v4
 
+and record_declaration (env : env) ((v1, v2, v3, v4, v5) : CST.record_declaration) : class_decl =
+  let v1 =
+    (match v1 with
+     | Some x -> modifiers env x
+     | None -> [])
+  in
+  let v2 = token env v2 (* "record" *) in
+  let v3 = identifier env v3 (* pattern [a-zA-Z_]\w* *) in
+  let v4 = formal_parameters env v4 in
+  let v5 = class_body env v5 in
+  { cl_name = v3; cl_kind = (Record, v2); cl_tparams = [];
+    cl_mods = v1; cl_extends = None; cl_impls = []; cl_formals = v4;
+    cl_body = v5 }
 
 and class_body_decl env = function
+  | `Record_decl x -> [Class (record_declaration env x)]
   | `Field_decl x -> field_declaration env x
   | `Meth_decl x -> [Method (method_declaration env x)]
   | `Class_decl x -> [Class (class_declaration env x)]
@@ -1349,7 +1363,7 @@ and class_declaration (env : env) ((v1, v2, v3, v4, v5, v6, v7) : CST.class_decl
   in
   let v7 = class_body env v7 in
   { cl_name = v3; cl_kind = (ClassRegular, v2); cl_tparams = v4;
-    cl_mods = v1; cl_extends = v5; cl_impls = v6;
+    cl_mods = v1; cl_extends = v5; cl_impls = v6; cl_formals = [];
     cl_body = v7 }
 
 
@@ -1550,7 +1564,7 @@ and annotation_type_declaration (env : env) ((v1, v2, v3, v4) : CST.annotation_t
   let v3 = identifier env v3 (* pattern [a-zA-Z_]\w* *) in
   let v4 = annotation_type_body env v4 in
   { cl_mods = v1; cl_name = v3; cl_body = v4; cl_kind = (AtInterface, v2);
-    cl_tparams = []; cl_extends = None; cl_impls = [];
+    cl_tparams = []; cl_extends = None; cl_impls = []; cl_formals = [];
   }
 
 
@@ -1622,7 +1636,7 @@ and interface_declaration (env : env) ((v1, v2, v3, v4, v5, v6) : CST.interface_
   in
   let v6 = interface_body env v6 in
   { cl_name = v3; cl_kind = (Interface, v2); cl_tparams = v4;
-    cl_mods = v1; cl_extends = None; cl_impls = v5;
+    cl_mods = v1; cl_extends = None; cl_impls = v5; cl_formals = [];
     cl_body = v6 }