src/defs.js: Fix conversion of non-BMP unicode codepoints

Adds tests for to_luastring

2 files changed, 51 insertions, 1 deletions

diff --git a/src/defs.js b/src/defs.js
index 4d5de3b..f76471a 100644
--- a/src/defs.js
+++ b/src/defs.js
@@ -193,7 +193,7 @@ const to_luastring = function(str, cache) {
         // See http://unicode.org/faq/utf_bom.html#utf16-3
         // For UTF8 byte structure, see http://en.wikipedia.org/wiki/UTF-8#Description and https://www.ietf.org/rfc/rfc2279.txt and https://tools.ietf.org/html/rfc3629
         let u = str.codePointAt(i);
-        if (u >= 0xD800) i++; // If it was a surrogate pair it used up two bytes
+        if (u >= 0x10000) i++; // It was a surrogate pair and hence used up two bytes
         if (u <= 0x7F) {
             outU8Array[outIdx++] = u;
         } else if (u <= 0x7FF) {
diff --git a/tests/defs.js b/tests/defs.js
new file mode 100644
index 0000000..8c86588
--- /dev/null
+++ b/tests/defs.js
@@ -0,0 +1,50 @@
+const test = require('tape');
+
+global.WEB = false;
+const defs = require('../src/defs.js');
+
+test('to_luastring', function (t) {
+    t.deepEqual(
+        defs.to_luastring("foo"),
+        ["f".charCodeAt(0), "o".charCodeAt(0), "o".charCodeAt(0)],
+        "Convert normal ascii string"
+    );
+
+    t.deepEqual(
+        defs.to_luastring("fo\0o"),
+        ["f".charCodeAt(0), "o".charCodeAt(0), 0, "o".charCodeAt(0)],
+        "Convert ascii string containing null byte"
+    );
+
+    t.deepEqual(
+        defs.to_luastring("Café"),
+        [67, 97, 102, 195, 169],
+        "Convert string with BMP unicode chars"
+    );
+
+    t.deepEqual(
+        defs.to_luastring(""),
+        [239, 163, 191],
+        "Convert string with codepoint in PUA (U+E000 to U+F8FF)"
+    );
+
+    t.deepEqual(
+        defs.to_luastring("❤️🍾"),
+        [226, 157, 164, 239, 184, 143, 240, 159, 141, 190],
+        "Convert string with surrogate pair"
+    );
+
+    t.deepEqual(
+        defs.to_luastring("\uD800a"),
+        [237, 160, 128, 97],
+        "Convert string with broken surrogate pair"
+    );
+
+    t.deepEqual(
+        defs.to_luastring("\uD823"),
+        [237, 160, 163],
+        "Convert string with broken surrogate pair at end of string"
+    );
+
+    t.end();
+});