From 586b60542973c4b32f3bde49656042fba6941804 Mon Sep 17 00:00:00 2001
From: Mohamed Boudra <boudra.moha@gmail.com>
Date: Sun, 8 Nov 2026 14:22:56 +0100
Subject: [PATCH] fix(highlight): handle Unicode heredoc delimiters and
 repeated template markers

Heredoc delimiters only accepted ASCII, so a Unicode delimiter was recorded
as empty and the rest of the file rendered as a string. `$$${` consumed the
second marker before confirming `{` followed, which turned an escape into an
interpolation.
---
 .../src/__tests__/highlighter.test.ts         | 16 ++++++
 packages/highlight/src/hcl/tokens.ts          | 53 +++++++------------
 2 files changed, 34 insertions(+), 35 deletions(-)

diff --git a/packages/highlight/src/__tests__/highlighter.test.ts b/packages/highlight/src/__tests__/highlighter.test.ts
index 7c5e710504..2a277d0391 100644
--- a/packages/highlight/src/__tests__/highlighter.test.ts
+++ b/packages/highlight/src/__tests__/highlighter.test.ts
@@ -401,6 +401,22 @@ describe("Terraform", () => {
     expect(styleOf(result[3], "3")).toBe("number");
   });
 
+  it("ends a heredoc with a Unicode delimiter", () => {
+    const result = highlightCode("x = <<ÉOF\nfoo\nÉOF\ny = 1", "main.tf");
+
+    expect(styleOf(result[3], "y")).toBe("definition");
+    expect(styleOf(result[3], "1")).toBe("number");
+  });
+
+  it.each([
+    ['x = "$$${var.name}"', "$"],
+    ['x = "%%%{ if true }"', "%"],
+  ])("keeps escaped template text in %s as a string", (code) => {
+    const line = highlightCode(code, "main.tf")[0];
+
+    expect(line[line.length - 1]).toEqual({ text: code.slice(4), style: "string" });
+  });
+
   it("parses a real configuration without error nodes", () => {
     const code = [
       "terraform {",
diff --git a/packages/highlight/src/hcl/tokens.ts b/packages/highlight/src/hcl/tokens.ts
index 3d4ca15685..d88d45e7fb 100644
--- a/packages/highlight/src/hcl/tokens.ts
+++ b/packages/highlight/src/hcl/tokens.ts
@@ -164,6 +164,21 @@ function scanHeredocBody(input: InputStream, expected: string): void {
   }
 }
 
+// `$${` and `%%{` are escapes. The doubled marker counts only when `{` follows it, so in `$$${`
+// the first `$` is literal text and the rest is the escape.
+function scanTemplateMarker(input: InputStream, marker: string, startToken: number): void {
+  input.advance();
+  if (String.fromCharCode(input.next) === "{") {
+    input.advance();
+    return input.acceptToken(startToken);
+  }
+  if (String.fromCharCode(input.next) === marker && String.fromCharCode(input.peek(1)) === "{") {
+    input.advance();
+    input.advance();
+  }
+  input.acceptToken(templateLiteralChunk);
+}
+
 export const scanQuotedTemplateStart = new ExternalTokenizer(
   (input, stack) => {
     if (
@@ -225,23 +240,7 @@ export const scanTemplateInterpolationStart = new ExternalTokenizer(
       !(stack.context && stack.context.type === ContextType.TEMPLATE_INTERPOLATION) &&
       String.fromCharCode(input.next) === "$"
     ) {
-      input.advance();
-      if (String.fromCharCode(input.next) === "{") {
-        input.advance();
-        return input.acceptToken(TemplateInterpolationStart);
-      }
-
-      // escape sequence
-      if (String.fromCharCode(input.next) === "$") {
-        input.advance();
-        if (String.fromCharCode(input.next) === "{") {
-          // $${
-          input.advance();
-          return input.acceptToken(templateLiteralChunk);
-        }
-      }
-
-      return input.acceptToken(templateLiteralChunk);
+      return scanTemplateMarker(input, "$", TemplateInterpolationStart);
     }
   },
   { contextual: true, fallback: true },
@@ -268,23 +267,7 @@ export const scanTemplateDirectiveStart = new ExternalTokenizer(
       !(stack.context && stack.context.type === ContextType.TEMPLATE_DIRECTIVE) &&
       String.fromCharCode(input.next) === "%"
     ) {
-      input.advance();
-      if (String.fromCharCode(input.next) === "{") {
-        input.advance();
-        return input.acceptToken(TemplateDirectiveStart);
-      }
-
-      // escape sequence
-      if (String.fromCharCode(input.next) === "%") {
-        input.advance();
-        if (String.fromCharCode(input.next) === "{") {
-          // $${
-          input.advance();
-          return input.acceptToken(templateLiteralChunk);
-        }
-      }
-
-      return input.acceptToken(templateLiteralChunk);
+      return scanTemplateMarker(input, "%", TemplateDirectiveStart);
     }
   },
   { contextual: true, fallback: true },
@@ -312,7 +295,7 @@ export const scanHeredocIdentifier = new ExternalTokenizer(
       input.next !== -1
     ) {
       let next = String.fromCharCode(input.next);
-      while (/[0-9a-zA-Z]/.test(next) || next === "_" || next === "-") {
+      while (/[\p{L}\p{N}_-]/u.test(next)) {
         input.advance();
         next = String.fromCharCode(input.next);
       }