From f6875ea1f5f1d8f75ec93a55f7b17f7aabaf5e1b Mon Sep 17 00:00:00 2001
From: Sergei Zharinov <zharinov@users.noreply.github.com>
Date: Tue, 6 Jan 2026 09:33:31 -0300
Subject: [PATCH] fix: ban empty anonymous nodes with diagnostic

---
 crates/plotnik-lib/src/diagnostics/message.rs |  3 ++
 .../plotnik-lib/src/parser/grammar/atoms.rs   | 34 ++++++++++++++++++-
 .../parser/tests/recovery/coverage_tests.rs   |  6 ++--
 .../parser/tests/recovery/validation_tests.rs | 32 +++++++++++++++++
 4 files changed, 72 insertions(+), 3 deletions(-)

diff --git a/crates/plotnik-lib/src/diagnostics/message.rs b/crates/plotnik-lib/src/diagnostics/message.rs
index d36a9fd9..9e2b187c 100644
--- a/crates/plotnik-lib/src/diagnostics/message.rs
+++ b/crates/plotnik-lib/src/diagnostics/message.rs
@@ -30,6 +30,7 @@ pub enum DiagnosticKind {
 
     // User wrote something that doesn't belong
     EmptyTree,
+    EmptyAnonymousNode,
     BareIdentifier,
     InvalidSeparator,
     AnchorInAlternation,
@@ -144,6 +145,7 @@ impl DiagnosticKind {
             Self::ExpectedTypeName => Some("e.g., `::MyType` or `::string`"),
             Self::ExpectedFieldName => Some("e.g., `-value`"),
             Self::EmptyTree => Some("use `(_)` to match any named node, or `_` for any node"),
+            Self::EmptyAnonymousNode => Some("use a valid anonymous node or remove it"),
             Self::TreeSitterSequenceSyntax => Some("use `{...}` for sequences"),
             Self::NegationSyntaxDeprecated => Some("use `-field` instead of `!field`"),
             Self::MixedAltBranches => {
@@ -182,6 +184,7 @@ impl DiagnosticKind {
 
             // Invalid syntax
             Self::EmptyTree => "empty `()` is not allowed",
+            Self::EmptyAnonymousNode => "empty anonymous node",
             Self::BareIdentifier => "bare identifier is not valid",
             Self::InvalidSeparator => "unexpected separator",
             Self::AnchorInAlternation => "anchors cannot appear directly in alternations",
diff --git a/crates/plotnik-lib/src/parser/grammar/atoms.rs b/crates/plotnik-lib/src/parser/grammar/atoms.rs
index e0fa046c..daaf23bd 100644
--- a/crates/plotnik-lib/src/parser/grammar/atoms.rs
+++ b/crates/plotnik-lib/src/parser/grammar/atoms.rs
@@ -1,3 +1,6 @@
+use rowan::TextRange;
+
+use crate::diagnostics::DiagnosticKind;
 use crate::parser::Parser;
 use crate::parser::cst::SyntaxKind;
 
@@ -10,9 +13,38 @@ impl Parser<'_, '_> {
 
     /// `"if"` | `'+'`
     pub(crate) fn parse_str(&mut self) {
+        let start = self.current_span().start();
         self.start_node(SyntaxKind::Str);
-        self.bump_string_tokens();
+
+        let open_quote = self.current();
+        self.bump(); // opening quote
+
+        let has_content = self.currently_is(SyntaxKind::StrVal);
+        if has_content {
+            self.bump();
+        }
+
+        let closing = self.current();
+        assert_eq!(
+            closing, open_quote,
+            "parse_str: expected closing {:?} but found {:?} \
+             (lexer should only produce quote tokens from complete strings)",
+            open_quote, closing
+        );
+        let end = self.current_span().end();
+        self.bump(); // closing quote
+
         self.finish_node();
+
+        if !has_content {
+            self.diagnostics
+                .report(
+                    self.source_id,
+                    DiagnosticKind::EmptyAnonymousNode,
+                    TextRange::new(start, end),
+                )
+                .emit();
+        }
     }
 
     /// Consume string tokens (quote + optional content + quote) without creating a node.
diff --git a/crates/plotnik-lib/src/parser/tests/recovery/coverage_tests.rs b/crates/plotnik-lib/src/parser/tests/recovery/coverage_tests.rs
index 60a29153..3b3f4bb8 100644
--- a/crates/plotnik-lib/src/parser/tests/recovery/coverage_tests.rs
+++ b/crates/plotnik-lib/src/parser/tests/recovery/coverage_tests.rs
@@ -200,7 +200,8 @@ fn empty_double_quote_string() {
     Q = (a "")
     "#};
 
-    let res = Query::expect_valid_cst(input);
+    // Empty anonymous nodes are now invalid, but CST structure is still correct
+    let res = Query::expect(input).dump_cst();
 
     insta::assert_snapshot!(res, @r#"
     Root
@@ -223,7 +224,8 @@ fn empty_single_quote_string() {
     Q = (a '')
     "#};
 
-    let res = Query::expect_valid_cst(input);
+    // Empty anonymous nodes are now invalid, but CST structure is still correct
+    let res = Query::expect(input).dump_cst();
 
     insta::assert_snapshot!(res, @r#"
     Root
diff --git a/crates/plotnik-lib/src/parser/tests/recovery/validation_tests.rs b/crates/plotnik-lib/src/parser/tests/recovery/validation_tests.rs
index 6b4a161d..07d63838 100644
--- a/crates/plotnik-lib/src/parser/tests/recovery/validation_tests.rs
+++ b/crates/plotnik-lib/src/parser/tests/recovery/validation_tests.rs
@@ -1407,3 +1407,35 @@ fn negation_syntax_deprecated_warning() {
     help: use `-field` instead of `!field`
     ");
 }
+
+#[test]
+fn empty_anonymous_node_double_quotes() {
+    let input = r#"(node "")"#;
+
+    let res = Query::expect_invalid(input);
+
+    insta::assert_snapshot!(res, @r#"
+    error: empty anonymous node
+      |
+    1 | (node "")
+      |       ^^
+      |
+    help: use a valid anonymous node or remove it
+    "#);
+}
+
+#[test]
+fn empty_anonymous_node_single_quotes() {
+    let input = "(node '')";
+
+    let res = Query::expect_invalid(input);
+
+    insta::assert_snapshot!(res, @r"
+    error: empty anonymous node
+      |
+    1 | (node '')
+      |       ^^
+      |
+    help: use a valid anonymous node or remove it
+    ");
+}