[java] Semgrep grammar augmentation — pattern parsing fixes

fyi/semgrep-grammars/src/semgrep-java/grammar.js

@@ -41,6 +41,12 @@ module.exports = grammar(base_grammar, {
 
     // this is from adding toplevel_explicit_constructor_invocation
     [$.type_parameter, $._unannotated_type],
+
+    // LANG-467: A bare `$...META` (or `...`) after `{` could be parsed as
+    // either an enum-constants list element or as the body of
+    // `enum_body_declarations`. GLR resolves this; the parses are
+    // operationally equivalent for Semgrep's purposes.
+    [$.enum_body, $.enum_body_declarations],
   ]),
 
   rules: {
@@ -65,9 +71,43 @@ module.exports = grammar(base_grammar, {
       $.toplevel_explicit_constructor_invocation,
     ),
 
+    // LANG-467: Allow `...` and `$...META` in the enum-constants list, and
+    // allow `$...META` (in addition to `...`) inside the post-`;` body region.
+    // Override `enum_body` wholesale since the constants list itself needs
+    // ellipsis support, which can't be reached via a `previous`-style
+    // augmentation.
+    enum_body: $ => seq(
+      '{',
+      commaSep(choice(
+        $.enum_constant,
+        $.semgrep_ellipsis,
+        $.semgrep_named_ellipsis,
+      )),
+      optional(','),
+      optional($.enum_body_declarations),
+      '}',
+    ),
+
     enum_body_declarations: ($, previous) => choice(
       previous,
       $.semgrep_ellipsis,
+      $.semgrep_named_ellipsis,
+    ),
+
+    // LANG-466: allow `...` (and `$...META`) interleaved with arrow-rule
+    // switch entries. The colon-form arm already supports ellipsis via the
+    // `statement` augmentation; the arrow-rule arm needs explicit support.
+    switch_block: $ => seq(
+      '{',
+      choice(
+        repeat($.switch_block_statement_group),
+        repeat(choice(
+          $.switch_rule,
+          $.semgrep_ellipsis,
+          $.semgrep_named_ellipsis,
+        )),
+      ),
+      '}',
     ),
 
     // This is a copy of `explicit_constructor_invocation`,
@@ -111,10 +151,13 @@ module.exports = grammar(base_grammar, {
       '}',
     ),
 
-    // So we can put ellipses within annotation type declarations
+    // So we can put ellipses within annotation type declarations.
+    // LANG-469: also allow `$...META` so `@interface $A { $...ELEMENTS }`
+    // parses cleanly.
     annotation_type_element_declaration: ($, previous) => choice(
       previous,
       $.semgrep_ellipsis,
+      $.semgrep_named_ellipsis,
     ),
 
     semgrep_ellipsis: $ => '...',
@@ -173,9 +216,12 @@ module.exports = grammar(base_grammar, {
       field('field', choice($.identifier, $._reserved_identifier, $.this, '...')),
     ),
 
+    // LANG-469: also allow `$...META` so patterns like
+    // `@$ANN(value = "...", $...PROPS)` parse cleanly.
     element_value_pair: ($, previous) => choice(
       previous,
       $.semgrep_ellipsis,
+      $.semgrep_named_ellipsis,
     ),
 
 

fyi/versions

@@ -19,12 +19,27 @@ Last change in file:
       feat: allow reserved identifiers in `element_value_pair`
 ---
 File: semgrep-grammars/src/semgrep-java/grammar.js
-Git repo name: ocaml-tree-sitter-semgrep
-Latest commit in repo: 78b9c78b6f1e98cd1f1d51d1436dd8b99ac617a7
+Git repo name: agent-a1e125b44664eae6a
+Latest commit in repo: 80708da730e76d4fb9aea75470297ae83b28324b
 Last change in file:
-  commit 78b9c78b6f1e98cd1f1d51d1436dd8b99ac617a7
-  Author: brandonspark <wu.brandonj@gmail.com>
-  Date:   Mon Jun 23 17:14:11 2025 -0700
+  commit 80708da730e76d4fb9aea75470297ae83b28324b
+  Author: brandonspark <brandon@semgrep.com>
+  Date:   Wed Apr 29 17:26:27 2026 -0700
 
-      allow no smemicolon in toplevel typed metavariable decl
+      fix(java): allow Semgrep ellipsis in switch/enum/annotation positions
+
+      Augments the Java grammar so canonical Semgrep patterns parse cleanly:
+
+      - LANG-466: `...` (and `$...META`) between arrow-rule entries in
+        `switch_block`. The colon-form arm already gets ellipsis via the
+        `statement` augmentation; this extends the arrow-rule arm.
+      - LANG-467: `...` and `$...META` inside the enum-constants list of
+        `enum_body`, plus `$...META` in `enum_body_declarations` (`...` was
+        already allowed). Adds a GLR conflict between `enum_body` and
+        `enum_body_declarations` to disambiguate a bare ellipsis after `{`.
+      - LANG-469: `$...META` as an entry in `annotation_argument_list` (via
+        `element_value_pair`) and as an annotation type element declaration in
+        `@interface` bodies (via `annotation_type_element_declaration`).
+
+      Adds 5 corpus tests covering the minimal ticket repros.
 ---

lib/Boilerplate.ml

@@ -423,6 +423,46 @@ and map_annotation_type_element_declaration (env : env) (x : CST.annotation_type
   | `Semg_ellips tok -> R.Case ("Semg_ellips",
       (* "..." *) token env tok
     )
+  | `Semg_named_ellips tok -> R.Case ("Semg_named_ellips",
+      (* pattern \$\.\.\.[A-Z_][A-Z_0-9]* *) token env tok
+    )
+  )
+
+and map_anon_choice_enum_cst_1653aac (env : env) (x : CST.anon_choice_enum_cst_1653aac) =
+  (match x with
+  | `Enum_cst (v1, v2, v3, v4) -> R.Case ("Enum_cst",
+      let v1 =
+        (match v1 with
+        | Some x -> R.Option (Some (
+            map_modifiers env x
+          ))
+        | None -> R.Option None)
+      in
+      let v2 =
+        (* pattern [\p{XID_Start}_$][\p{XID_Continue}\u00A2_$]* *) token env v2
+      in
+      let v3 =
+        (match v3 with
+        | Some x -> R.Option (Some (
+            map_argument_list env x
+          ))
+        | None -> R.Option None)
+      in
+      let v4 =
+        (match v4 with
+        | Some x -> R.Option (Some (
+            map_class_body env x
+          ))
+        | None -> R.Option None)
+      in
+      R.Tuple [v1; v2; v3; v4]
+    )
+  | `Semg_ellips tok -> R.Case ("Semg_ellips",
+      (* "..." *) token env tok
+    )
+  | `Semg_named_ellips tok -> R.Case ("Semg_named_ellips",
+      (* pattern \$\.\.\.[A-Z_][A-Z_0-9]* *) token env tok
+    )
   )
 
 and map_anon_choice_formal_param_3e261ef (env : env) (x : CST.anon_choice_formal_param_3e261ef) =
@@ -1134,6 +1174,9 @@ and map_element_value_pair (env : env) (x : CST.element_value_pair) =
   | `Semg_ellips tok -> R.Case ("Semg_ellips",
       (* "..." *) token env tok
     )
+  | `Semg_named_ellips tok -> R.Case ("Semg_named_ellips",
+      (* pattern \$\.\.\.[A-Z_][A-Z_0-9]* *) token env tok
+    )
   )
 
 and map_enhanced_for_statement (env : env) ((v1, v2, v3, v4, v5, v6, v7, v8, v9) : CST.enhanced_for_statement) =
@@ -1159,11 +1202,11 @@ and map_enum_body (env : env) ((v1, v2, v3, v4, v5) : CST.enum_body) =
   let v2 =
     (match v2 with
     | Some (v1, v2) -> R.Option (Some (
-        let v1 = map_enum_constant env v1 in
+        let v1 = map_anon_choice_enum_cst_1653aac env v1 in
         let v2 =
           R.List (List.map (fun (v1, v2) ->
             let v1 = (* "," *) token env v1 in
-            let v2 = map_enum_constant env v2 in
+            let v2 = map_anon_choice_enum_cst_1653aac env v2 in
             R.Tuple [v1; v2]
           ) v2)
         in
@@ -1200,35 +1243,11 @@ and map_enum_body_declarations (env : env) (x : CST.enum_body_declarations) =
   | `Semg_ellips tok -> R.Case ("Semg_ellips",
       (* "..." *) token env tok
     )
+  | `Semg_named_ellips tok -> R.Case ("Semg_named_ellips",
+      (* pattern \$\.\.\.[A-Z_][A-Z_0-9]* *) token env tok
+    )
   )
 
-and map_enum_constant (env : env) ((v1, v2, v3, v4) : CST.enum_constant) =
-  let v1 =
-    (match v1 with
-    | Some x -> R.Option (Some (
-        map_modifiers env x
-      ))
-    | None -> R.Option None)
-  in
-  let v2 =
-    (* pattern [\p{XID_Start}_$][\p{XID_Continue}\u00A2_$]* *) token env v2
-  in
-  let v3 =
-    (match v3 with
-    | Some x -> R.Option (Some (
-        map_argument_list env x
-      ))
-    | None -> R.Option None)
-  in
-  let v4 =
-    (match v4 with
-    | Some x -> R.Option (Some (
-        map_class_body env x
-      ))
-    | None -> R.Option None)
-  in
-  R.Tuple [v1; v2; v3; v4]
-
 and map_enum_declaration (env : env) ((v1, v2, v3, v4, v5) : CST.enum_declaration) =
   let v1 =
     (match v1 with
@@ -2378,8 +2397,20 @@ and map_switch_block (env : env) ((v1, v2, v3) : CST.switch_block) =
     | `Rep_switch_blk_stmt_group xs -> R.Case ("Rep_switch_blk_stmt_group",
         R.List (List.map (map_switch_block_statement_group env) xs)
       )
-    | `Rep_switch_rule xs -> R.Case ("Rep_switch_rule",
-        R.List (List.map (map_switch_rule env) xs)
+    | `Rep_choice_switch_rule xs -> R.Case ("Rep_choice_switch_rule",
+        R.List (List.map (fun x ->
+          (match x with
+          | `Switch_rule x -> R.Case ("Switch_rule",
+              map_switch_rule env x
+            )
+          | `Semg_ellips tok -> R.Case ("Semg_ellips",
+              (* "..." *) token env tok
+            )
+          | `Semg_named_ellips tok -> R.Case ("Semg_named_ellips",
+              (* pattern \$\.\.\.[A-Z_][A-Z_0-9]* *) token env tok
+            )
+          )
+        ) xs)
       )
     )
   in

lib/CST.ml

@@ -209,6 +209,18 @@ and annotation_type_element_declaration = [
       * Token.t (* ";" *)
     )
   | `Semg_ellips of Token.t (* "..." *)
+  | `Semg_named_ellips of semgrep_named_ellipsis (*tok*)
+]
+
+and anon_choice_enum_cst_1653aac = [
+    `Enum_cst of (
+        modifiers option
+      * identifier (*tok*)
+      * argument_list option
+      * class_body option
+    )
+  | `Semg_ellips of Token.t (* "..." *)
+  | `Semg_named_ellips of semgrep_named_ellipsis (*tok*)
 ]
 
 and anon_choice_formal_param_3e261ef = [
@@ -509,6 +521,7 @@ and element_value_pair = [
         anon_choice_id_662bcdc * Token.t (* "=" *) * element_value
     )
   | `Semg_ellips of Token.t (* "..." *)
+  | `Semg_named_ellips of semgrep_named_ellipsis (*tok*)
 ]
 
 and enhanced_for_statement = (
@@ -526,8 +539,9 @@ and enhanced_for_statement = (
 and enum_body = (
     Token.t (* "{" *)
   * (
-        enum_constant
-      * (Token.t (* "," *) * enum_constant) list (* zero or more *)
+        anon_choice_enum_cst_1653aac
+      * (Token.t (* "," *) * anon_choice_enum_cst_1653aac)
+          list (* zero or more *)
     )
       option
   * Token.t (* "," *) option
@@ -541,15 +555,9 @@ and enum_body_declarations = [
       * class_body_declaration list (* zero or more *)
     )
   | `Semg_ellips of Token.t (* "..." *)
+  | `Semg_named_ellips of semgrep_named_ellipsis (*tok*)
 ]
 
-and enum_constant = (
-    modifiers option
-  * identifier (*tok*)
-  * argument_list option
-  * class_body option
-)
-
 and enum_declaration = (
     modifiers option
   * Token.t (* "enum" *)
@@ -1023,7 +1031,13 @@ and switch_block = (
   * [
         `Rep_switch_blk_stmt_group of
           switch_block_statement_group list (* zero or more *)
-      | `Rep_switch_rule of switch_rule list (* zero or more *)
+      | `Rep_choice_switch_rule of
+          [
+              `Switch_rule of switch_rule
+            | `Semg_ellips of Token.t (* "..." *)
+            | `Semg_named_ellips of semgrep_named_ellipsis (*tok*)
+          ]
+            list (* zero or more *)
     ]
   * Token.t (* "}" *)
 )
@@ -1374,6 +1388,13 @@ type element_value_array_initializer (* inlined *) = (
   * Token.t (* "}" *)
 )
 
+type enum_constant (* inlined *) = (
+    modifiers option
+  * identifier (*tok*)
+  * argument_list option
+  * class_body option
+)
+
 type instanceof_expression (* inlined *) = (
     expression
   * Token.t (* "instanceof" *)