1 files changed, 59 insertions, 9 deletions

diff --git a/integration/bunjs-only-snippets/escapeHTML.test.js b/integration/bunjs-only-snippets/escapeHTML.test.js
index 13ff138c9..6c709bf76 100644
--- a/integration/bunjs-only-snippets/escapeHTML.test.js
+++ b/integration/bunjs-only-snippets/escapeHTML.test.js
@@ -2,7 +2,19 @@ import { describe, it, expect } from "bun:test";
 import { gcTick } from "./gc";
 
 describe("escapeHTML", () => {
+  // The matrix of cases we need to test for:
+  // 1. Works with short strings
+  // 2. Works with long strings
+  // 3. Works with latin1 strings
+  // 4. Works with utf16 strings
+  // 5. Works when the text to escape is somewhere in the middle
+  // 6. Works when the text to escape is in the beginning
+  // 7. Works when the text to escape is in the end
+  // 8. Returns the same string when there's no need to escape
   it("works", () => {
+    expect(escapeHTML("absolutely nothing to do here")).toBe(
+      "absolutely nothing to do here"
+    );
     expect(escapeHTML("<script>alert(1)</script>")).toBe(
       "&lt;script&gt;alert(1)&lt;/script&gt;"
     );
@@ -18,16 +30,10 @@ describe("escapeHTML", () => {
     expect(escapeHTML("\v")).toBe("\v");
     expect(escapeHTML("\b")).toBe("\b");
     expect(escapeHTML("\u00A0")).toBe("\u00A0");
+    expect(escapeHTML("<script>ab")).toBe("&lt;script&gt;ab");
+    expect(escapeHTML("<script>")).toBe("&lt;script&gt;");
+    expect(escapeHTML("<script><script>")).toBe("&lt;script&gt;&lt;script&gt;");
 
-    // The matrix of cases we need to test for:
-    // 1. Works with short strings
-    // 2. Works with long strings
-    // 3. Works with latin1 strings
-    // 4. Works with utf16 strings
-    // 5. Works when the text to escape is somewhere in the middle
-    // 6. Works when the text to escape is in the beginning
-    // 7. Works when the text to escape is in the end
-    // 8. Returns the same string when there's no need to escape
     expect(escapeHTML("lalala" + "<script>alert(1)</script>" + "lalala")).toBe(
       "lalala&lt;script&gt;alert(1)&lt;/script&gt;lalala"
     );
@@ -39,6 +45,13 @@ describe("escapeHTML", () => {
       "lalala" + "&lt;script&gt;alert(1)&lt;/script&gt;"
     );
 
+    expect(escapeHTML("What does 😊 mean?")).toBe("What does 😊 mean?");
+    const output = escapeHTML("<What does 😊");
+    expect(output).toBe("&lt;What does 😊");
+    expect(escapeHTML("<div>What does 😊 mean in text?")).toBe(
+      "&lt;div&gt;What does 😊 mean in text?"
+    );
+
     expect(
       escapeHTML(
         ("lalala" + "<script>alert(1)</script>" + "lalala").repeat(900)
@@ -50,5 +63,42 @@ describe("escapeHTML", () => {
     expect(
       escapeHTML(("lalala" + "<script>alert(1)</script>").repeat(900))
     ).toBe(("lalala" + "&lt;script&gt;alert(1)&lt;/script&gt;").repeat(900));
+
+    // the positions of the unicode codepoint are important
+    // our simd code for U16 is at 8 bytes, so we need to especially check the boundaries
+    expect(
+      escapeHTML("😊lalala" + "<script>alert(1)</script>" + "lalala")
+    ).toBe("😊lalala&lt;script&gt;alert(1)&lt;/script&gt;lalala");
+    expect(escapeHTML("<script>😊alert(1)</script>" + "lalala")).toBe(
+      "&lt;script&gt;😊alert(1)&lt;/script&gt;lalala"
+    );
+    expect(escapeHTML("<script>alert(1)😊</script>" + "lalala")).toBe(
+      "&lt;script&gt;alert(1)😊&lt;/script&gt;lalala"
+    );
+    expect(escapeHTML("<script>alert(1)</script>" + "😊lalala")).toBe(
+      "&lt;script&gt;alert(1)&lt;/script&gt;😊lalala"
+    );
+    expect(escapeHTML("<script>alert(1)</script>" + "lal😊ala")).toBe(
+      "&lt;script&gt;alert(1)&lt;/script&gt;lal😊ala"
+    );
+    expect(
+      escapeHTML("<script>alert(1)</script>" + "lal😊ala".repeat(10))
+    ).toBe("&lt;script&gt;alert(1)&lt;/script&gt;" + "lal😊ala".repeat(10));
+
+    for (let i = 1; i < 10; i++)
+      expect(escapeHTML("<script>alert(1)</script>" + "la😊".repeat(i))).toBe(
+        "&lt;script&gt;alert(1)&lt;/script&gt;" + "la😊".repeat(i)
+      );
+
+    expect(escapeHTML("la😊" + "<script>alert(1)</script>")).toBe(
+      "la😊" + "&lt;script&gt;alert(1)&lt;/script&gt;"
+    );
+    expect(
+      escapeHTML(("lalala" + "<script>alert(1)</script>😊").repeat(1))
+    ).toBe(("lalala" + "&lt;script&gt;alert(1)&lt;/script&gt;😊").repeat(1));
+
+    expect(escapeHTML("😊".repeat(100))).toBe("😊".repeat(100));
+    expect(escapeHTML("😊<".repeat(100))).toBe("😊&lt;".repeat(100));
+    expect(escapeHTML("<😊>".repeat(100))).toBe("&lt;😊&gt;".repeat(100));
   });
 });