[elixir] add $...VAR ellipsis-metavariable support

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

@@ -10,6 +10,11 @@ module.exports = grammar(base_grammar, {
   name: 'elixir',
 
   conflicts: ($, previous) => previous.concat([
+    // '$...VAR' is valid both as an _expression (positional) and inside a
+    // pair (keyword). Either resolution yields ParamEllipsis in the
+    // Generic AST, so the choice is harmless; we just need to declare
+    // the conflict explicitly. See the comment on `semgrep_ellipsis`.
+    [$._expression, $.pair],
   ]),
 
   /*
@@ -48,6 +53,12 @@ module.exports = grammar(base_grammar, {
 
     _semgrep_metavariable: $ => token(/\$[A-Z_][A-Z_0-9]*/),
 
+    // Semgrep ellipsis-metavariable: '$...VAR' matches a variadic
+    // sequence and binds it to a metavariable (e.g. $...ARGS).
+    // Tokenized as a single unit so the leading '$' isn't separated
+    // from '...VAR'.
+    _semgrep_ellipsis_metavariable: $ => token(/\$\.\.\.[A-Z_][A-Z_0-9]*/),
+
     // Ellipsis
     // No need for extensions to _expressions for ellipsis because
     // Elixir already uses "..." as valid identifiers
@@ -62,6 +73,7 @@ module.exports = grammar(base_grammar, {
       return choice(
         previous,
         $.semgrep_ellipsis,
+        $._semgrep_ellipsis_metavariable,
       );
     },
 
@@ -79,6 +91,7 @@ module.exports = grammar(base_grammar, {
     _expression: ($, previous) => choice(
       ...previous.members,
       $.deep_ellipsis,
+      $._semgrep_ellipsis_metavariable,
     ),
 
     // The actual ellipsis rules

fyi/tree-sitter-version

@@ -1 +1 @@
-tree-sitter 0.22.6 (d521f0a0791d94f4442cf9be08322f6aabce20d6)
+tree-sitter 0.24.4

fyi/versions

@@ -19,28 +19,25 @@ Last change in file:
       Format with commas
 ---
 File: semgrep-grammars/src/semgrep-elixir/grammar.js
-Git repo name: ocaml-tree-sitter-semgrep
-Latest commit in repo: b9665cd85a2537da15cdc86f81b83722bc14af67
+Git repo name: agent-a1b1f87efc493b5a7
+Latest commit in repo: 041b61e542d4bcefbb375268dc5aaaf0e98403f5
 Last change in file:
-  commit b52bcb7fcd6caecacf5146d934d45afb5ecd5dfb
-  Author: Amarin (Um) Phaosawasdi <amchiclet@users.noreply.github.com>
-  Date:   Wed Jan 31 17:46:55 2024 +0900
+  commit 041b61e542d4bcefbb375268dc5aaaf0e98403f5
+  Author: brandonspark <brandon@semgrep.com>
+  Date:   Wed Apr 29 17:28:58 2026 -0700
 
-      elixir: support metavariable atoms (#474)
+      fix(elixir): add $...VAR ellipsis-metavariable support
 
-      Elixir has atoms that are pretty much identifiers that start with a colon, e.g. `:foo`.
+      Adds a `_semgrep_ellipsis_metavariable` token (`$\.\.\.[A-Z_][A-Z_0-9]*`)
+      and wires it into `_expression` and `pair` so it parses in every position
+      where users naturally reach for it: call args, def/fn parameters, list and
+      tuple elements, do-block bodies, and keyword-argument positions. Declares
+      the resulting `_expression` vs `pair` conflict explicitly (either resolution
+      becomes ParamEllipsis in the Generic AST). Adds corpus tests covering all
+      six positions from LANG-493.
 
-      Metavars can match expressions, but not atoms
-      * `$ATOM` matches `:foo`
-      * `:$ATOM` doesn't parse
+      LANG-501 (`?`/`!` suffix on metavariables) is deferred — root cause is in
+      the upstream tree-sitter-elixir scanner.cc, not in grammar.js.
 
-      This PR allows metavars as atoms so `:$ATOM` matches `:foo`.
-
-      Also fixed some indentation in some previous code.
-
-      Test plan: added test in https://github.com/semgrep/semgrep-proprietary/pull/1274
-
-      ### Security
-
-      - [x] Change has no security implications (otherwise, ping the security team)
+      Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 ---

lib/Boilerplate.ml

@@ -79,8 +79,8 @@ let map_float_ (env : env) (tok : CST.float_) =
 let map_imm_tok_pat_0db2d54 (env : env) (tok : CST.imm_tok_pat_0db2d54) =
   (* pattern [a-z] *) token env tok
 
-let map_semgrep_metavariable (env : env) (tok : CST.semgrep_metavariable) =
-  (* semgrep_metavariable *) token env tok
+let map_semgrep_ellipsis_metavariable (env : env) (tok : CST.semgrep_ellipsis_metavariable) =
+  (* semgrep_ellipsis_metavariable *) token env tok
 
 let map_quoted_content_single (env : env) (tok : CST.quoted_content_single) =
   (* quoted_content_single *) token env tok
@@ -131,6 +131,9 @@ let map_boolean (env : env) (x : CST.boolean) =
 let map_quoted_content_double (env : env) (tok : CST.quoted_content_double) =
   (* quoted_content_double *) token env tok
 
+let map_semgrep_metavariable (env : env) (tok : CST.semgrep_metavariable) =
+  (* semgrep_metavariable *) token env tok
+
 let map_quoted_content_i_bar (env : env) (tok : CST.quoted_content_i_bar) =
   (* quoted_content_i_bar *) token env tok
 
@@ -249,23 +252,6 @@ let map_quoted_curly (env : env) ((v1, v2, v3, v4) : CST.quoted_curly) =
   let v4 = (* "}" *) token env v4 in
   R.Tuple [v1; v2; v3; v4]
 
-let map_identifier (env : env) (x : CST.identifier) =
-  (match x with
-  | `Choice_pat_cf9c6c3 x -> R.Case ("Choice_pat_cf9c6c3",
-      (match x with
-      | `Pat_cf9c6c3 x -> R.Case ("Pat_cf9c6c3",
-          map_pat_cf9c6c3 env x
-        )
-      | `DOTDOTDOT tok -> R.Case ("DOTDOTDOT",
-          (* "..." *) token env tok
-        )
-      )
-    )
-  | `Semg_meta tok -> R.Case ("Semg_meta",
-      (* semgrep_metavariable *) token env tok
-    )
-  )
-
 let map_quoted_single (env : env) ((v1, v2, v3, v4) : CST.quoted_single) =
   let v1 = (* "'" *) token env v1 in
   let v2 =
@@ -409,6 +395,23 @@ let map_quoted_double (env : env) ((v1, v2, v3, v4) : CST.quoted_double) =
   let v4 = (* "\"" *) token env v4 in
   R.Tuple [v1; v2; v3; v4]
 
+let map_identifier (env : env) (x : CST.identifier) =
+  (match x with
+  | `Choice_pat_cf9c6c3 x -> R.Case ("Choice_pat_cf9c6c3",
+      (match x with
+      | `Pat_cf9c6c3 x -> R.Case ("Pat_cf9c6c3",
+          map_pat_cf9c6c3 env x
+        )
+      | `DOTDOTDOT tok -> R.Case ("DOTDOTDOT",
+          (* "..." *) token env tok
+        )
+      )
+    )
+  | `Semg_meta tok -> R.Case ("Semg_meta",
+      (* semgrep_metavariable *) token env tok
+    )
+  )
+
 let map_quoted_heredoc_single (env : env) ((v1, v2, v3, v4) : CST.quoted_heredoc_single) =
   let v1 = (* "'''" *) token env v1 in
   let v2 =
@@ -1563,6 +1566,9 @@ and map_expression (env : env) (x : CST.expression) =
       let v3 = (* "...>" *) token env v3 in
       R.Tuple [v1; v2; v3]
     )
+  | `Semg_ellips_meta tok -> R.Case ("Semg_ellips_meta",
+      (* semgrep_ellipsis_metavariable *) token env tok
+    )
   )
 
 and map_items_with_trailing_separator (env : env) (v1 : CST.items_with_trailing_separator) =
@@ -1673,6 +1679,9 @@ and map_pair (env : env) (x : CST.pair) =
   | `Semg_ellips tok -> R.Case ("Semg_ellips",
       (* "..." *) token env tok
     )
+  | `Semg_ellips_meta tok -> R.Case ("Semg_ellips_meta",
+      (* semgrep_ellipsis_metavariable *) token env tok
+    )
   )
 
 and map_quoted_i_angle (env : env) ((v1, v2, v3, v4) : CST.quoted_i_angle) =

lib/CST.ml

@@ -49,7 +49,7 @@ type float_ = Token.t
 
 type imm_tok_pat_0db2d54 = Token.t (* pattern [a-z] *)
 
-type semgrep_metavariable = Token.t
+type semgrep_ellipsis_metavariable = Token.t
 
 type quoted_content_single = Token.t
 
@@ -82,6 +82,8 @@ type boolean = [
 
 type quoted_content_double = Token.t
 
+type semgrep_metavariable = Token.t
+
 type quoted_content_i_bar = Token.t
 
 type quoted_content_i_curly = Token.t
@@ -129,14 +131,6 @@ type quoted_curly = (
   * Token.t (* "}" *)
 )
 
-type identifier = [
-    `Choice_pat_cf9c6c3 of [
-        `Pat_cf9c6c3 of pat_cf9c6c3
-      | `DOTDOTDOT of Token.t (* "..." *)
-    ]
-  | `Semg_meta of semgrep_metavariable (*tok*)
-]
-
 type quoted_single = (
     Token.t (* "'" *)
   * quoted_content_single (*tok*) option
@@ -185,6 +179,14 @@ type quoted_double = (
   * Token.t (* "\"" *)
 )
 
+type identifier = [
+    `Choice_pat_cf9c6c3 of [
+        `Pat_cf9c6c3 of pat_cf9c6c3
+      | `DOTDOTDOT of Token.t (* "..." *)
+    ]
+  | `Semg_meta of semgrep_metavariable (*tok*)
+]
+
 type quoted_heredoc_single = (
     Token.t (* "'''" *)
   * quoted_content_heredoc_single (*tok*) option
@@ -616,6 +618,7 @@ and expression = [
   | `Deep_ellips of (
         Token.t (* "<..." *) * expression * Token.t (* "...>" *)
     )
+  | `Semg_ellips_meta of semgrep_ellipsis_metavariable (*tok*)
 ]
 
 and items_with_trailing_separator = [
@@ -657,6 +660,7 @@ and local_call_with_parentheses = (
 and pair = [
     `Kw_exp of (keyword * expression)
   | `Semg_ellips of Token.t (* "..." *)
+  | `Semg_ellips_meta of semgrep_ellipsis_metavariable (*tok*)
 ]
 
 and quoted_i_angle = (

lib/Parse.ml

@@ -44,7 +44,7 @@ let children_regexps : (string * Run.exp option) list = [
       Token (Literal "false");
     |];
   );
-  "semgrep_metavariable", None;
+  "semgrep_ellipsis_metavariable", None;
   "imm_tok_pat_8f9e87e", None;
   "imm_tok_pat_b250293", None;
   "quoted_atom_start", None;
@@ -60,6 +60,7 @@ let children_regexps : (string * Run.exp option) list = [
   "quoted_content_double", None;
   "imm_tok_lpar", None;
   "imm_tok_pat_5eb9c21", None;
+  "semgrep_metavariable", None;
   "quoted_content_slash", None;
   "keyword_", None;
   "quoted_content_i_bar", None;
@@ -87,13 +88,6 @@ let children_regexps : (string * Run.exp option) list = [
   "quoted_content_i_heredoc_double", None;
   "atom_", None;
   "newline_before_binary_operator", None;
-  "metavariable_atom",
-  Some (
-    Seq [
-      Token (Literal ":");
-      Token (Name "semgrep_metavariable");
-    ];
-  );
   "operator_identifier",
   Some (
     Alt [|
@@ -151,6 +145,13 @@ let children_regexps : (string * Run.exp option) list = [
       Token (Literal "->");
     |];
   );
+  "metavariable_atom",
+  Some (
+    Seq [
+      Token (Literal ":");
+      Token (Name "semgrep_metavariable");
+    ];
+  );
   "terminator",
   Some (
     Alt [|
@@ -992,6 +993,7 @@ let children_regexps : (string * Run.exp option) list = [
       Token (Name "access_call");
       Token (Name "anonymous_function");
       Token (Name "deep_ellipsis");
+      Token (Name "semgrep_ellipsis_metavariable");
     |];
   );
   "interpolation",
@@ -1142,6 +1144,7 @@ let children_regexps : (string * Run.exp option) list = [
         Token (Name "expression");
       ];
       Token (Name "semgrep_ellipsis");
+      Token (Name "semgrep_ellipsis_metavariable");
     |];
   );
   "quoted_atom",
@@ -1732,7 +1735,7 @@ let trans_boolean ((kind, body) : mt) : CST.boolean =
       )
   | Leaf _ -> assert false
 
-let trans_semgrep_metavariable ((kind, body) : mt) : CST.semgrep_metavariable =
+let trans_semgrep_ellipsis_metavariable ((kind, body) : mt) : CST.semgrep_ellipsis_metavariable =
   match body with
   | Leaf v -> v
   | Children _ -> assert false
@@ -1812,6 +1815,11 @@ let trans_imm_tok_pat_5eb9c21 ((kind, body) : mt) : CST.imm_tok_pat_5eb9c21 =
   | Leaf v -> v
   | Children _ -> assert false
 
+let trans_semgrep_metavariable ((kind, body) : mt) : CST.semgrep_metavariable =
+  match body with
+  | Leaf v -> v
+  | Children _ -> assert false
+
 let trans_quoted_content_slash ((kind, body) : mt) : CST.quoted_content_slash =
   match body with
   | Leaf v -> v
@@ -1947,19 +1955,6 @@ let trans_newline_before_binary_operator ((kind, body) : mt) : CST.newline_befor
   | Leaf v -> v
   | Children _ -> assert false
 
-let trans_metavariable_atom ((kind, body) : mt) : CST.metavariable_atom =
-  match body with
-  | Children v ->
-      (match v with
-      | Seq [v0; v1] ->
-          (
-            Run.trans_token (Run.matcher_token v0),
-            trans_semgrep_metavariable (Run.matcher_token v1)
-          )
-      | _ -> assert false
-      )
-  | Leaf _ -> assert false
-
 let trans_operator_identifier ((kind, body) : mt) : CST.operator_identifier =
   match body with
   | Children v ->
@@ -2174,6 +2169,19 @@ let trans_operator_identifier ((kind, body) : mt) : CST.operator_identifier =
       )
   | Leaf _ -> assert false
 
+let trans_metavariable_atom ((kind, body) : mt) : CST.metavariable_atom =
+  match body with
+  | Children v ->
+      (match v with
+      | Seq [v0; v1] ->
+          (
+            Run.trans_token (Run.matcher_token v0),
+            trans_semgrep_metavariable (Run.matcher_token v1)
+          )
+      | _ -> assert false
+      )
+  | Leaf _ -> assert false
+
 let trans_terminator ((kind, body) : mt) : CST.terminator =
   match body with
   | Children v ->
@@ -4109,6 +4117,10 @@ and trans_expression ((kind, body) : mt) : CST.expression =
           `Deep_ellips (
             trans_deep_ellipsis (Run.matcher_token v)
           )
+      | Alt (24, v) ->
+          `Semg_ellips_meta (
+            trans_semgrep_ellipsis_metavariable (Run.matcher_token v)
+          )
       | _ -> assert false
       )
   | Leaf _ -> assert false
@@ -4411,6 +4423,10 @@ and trans_pair ((kind, body) : mt) : CST.pair =
           `Semg_ellips (
             trans_semgrep_ellipsis (Run.matcher_token v)
           )
+      | Alt (2, v) ->
+          `Semg_ellips_meta (
+            trans_semgrep_ellipsis_metavariable (Run.matcher_token v)
+          )
       | _ -> assert false
       )
   | Leaf _ -> assert false