Add testcase for String.prototype.codePointAt()

tests/api/test-charcodeat.c

@@ -1,5 +1,5 @@
 /*===
-*** test_1 (duk_safe_call)
+*** test_basic (duk_safe_call)
 i=0, n=19, charcode=102
 i=1, n=19, charcode=111
 i=2, n=19, charcode=111
@@ -20,11 +20,17 @@ i=16, n=19, charcode=0
 i=17, n=19, charcode=0
 i=18, n=19, charcode=0
 ==> rc=0, result='undefined'
-*** test_2 (duk_safe_call)
+*** test_invalid_arg (duk_safe_call)
 ==> rc=1, result='TypeError: string required, found 123 (stack index -1)'
+*** test_invalid_utf8 (duk_safe_call)
+index 0: 65
+index 1: 65533
+index 2: 66
+index 3: 0
+==> rc=0, result='undefined'
 ===*/
 
-static duk_ret_t test_1(duk_context *ctx, void *udata) {
+static duk_ret_t test_basic(duk_context *ctx, void *udata) {
 	duk_size_t i, n;
 
 	(void) udata;
@@ -42,7 +48,7 @@ static duk_ret_t test_1(duk_context *ctx, void *udata) {
 	return 0;
 }
 
-static duk_ret_t test_2(duk_context *ctx, void *udata) {
+static duk_ret_t test_invalid_arg(duk_context *ctx, void *udata) {
 	(void) udata;
 
 	/* TypeError for invalid arg type */
@@ -53,7 +59,22 @@ static duk_ret_t test_2(duk_context *ctx, void *udata) {
 	return 0;
 }
 
+static duk_ret_t test_invalid_utf8(duk_context *ctx, void *udata) {
+	char buf[3] = { (char) 0x41, (char) 0xff, (char) 0x42 };
+	size_t i;
+
+	(void) udata;
+
+	duk_push_lstring(ctx, (const char *) buf, sizeof(buf));
+	for (i = 0; i < sizeof(buf) + 1; i++) {  /* overshoot by one on purpose */
+		printf("index %d: %d\n", (int) i, (int) duk_char_code_at(ctx, -1, i));
+	}
+
+	return 0;
+}
+
 void test(duk_context *ctx) {
-	TEST_SAFE_CALL(test_1);
-	TEST_SAFE_CALL(test_2);
+	TEST_SAFE_CALL(test_basic);
+	TEST_SAFE_CALL(test_invalid_arg);
+	TEST_SAFE_CALL(test_invalid_utf8);
 }

tests/ecmascript/test-bi-string-frombuffer.js

@@ -34,19 +34,19 @@ TypeError
 TypeError
 TypeError
 "\u07ad\xbe\xef"
-string "<U+07AD><Error>"
+string "<U+07AD><U+FFFD>"
 |deadbeef|
 "\xff\ue74c"
-string "<Error><U+E74C><U+074C><Error>"
+string "<U+FFFD><U+E74C><U+074C><U+FFFD>"
 |ffeeddcc|
 "abcd"
 string "abcd"
 |61626364|
 "\x00\x01\x02\xfe\xff"
-string "<U+0000><U+0001><U+0002><Error><Error>"
+string "<U+0000><U+0001><U+0002><U+FFFD><U+FFFD>"
 |000102feff|
 "\xff\xff\xff\xffxV4\x12\u03ba\xfe\xca"
-string "<Error><Error><Error><Error>xV4<U+0012><U+03BA><Error><Error><Error>"
+string "<U+FFFD><U+FFFD><U+FFFD><U+FFFD>xV4<U+0012><U+03BA><U+FFFD><U+FFFD><U+FFFD>"
 |ffffffff78563412cefafeca|
 "\x00\x00\x00\x00\x00\x00\x00\x00\x9a\x99\x99\x99\x99\x99\xb9?\x9a\x99\x99\x99\x99\x99\u067f\x00\x00\x00\x00\x00\x00\u0ff333333\u3fc0\x00\x00\x00\x00\x00\U0003f9a6ffff\U001bf000\x00\x00\x00\x00\U00fda659\x99\x99\x99\U01fc0000\x00\x00@\x8f@"
 string "<U+0000><U+0000><U+0000><U+0000><U+0000><U+0000><U+0000><U+0000>?<U+067F>?<U+0000><U+0000><U+0000><U+0000><U+0000><U+0000><U+0FF3>?333333<U+3FC0>?<U+0000><U+0000><U+0000><U+0000><U+0000><U+0000><U+F9A6>?ffffff<U+F000>?<U+0000><U+0000><U+0000><U+0000><U+0000><U+0000><U+A659>?<U+0000>?<U+0000><U+0000><U+0000><U+0000><U+0000>@@"

tests/ecmascript/test-bi-string-proto-codepointat.js

@@ -0,0 +1,90 @@
+/*
+ *  String.prototype.codePointAt()
+ */
+
+/*===
+-3 NaN undefined
+-2 NaN undefined
+-1 NaN undefined
+0 102 102
+1 111 111
+2 111 111
+3 51966 51966
+4 55357 128169
+5 56489 56489
+6 48879 48879
+7 98 98
+8 97 97
+9 114 114
+10 55357 55357
+11 55357 55357
+12 55357 55357
+13 120 120
+14 56489 56489
+15 55357 55357
+16 98 98
+17 97 97
+18 122 122
+19 55357 55357
+20 NaN undefined
+21 NaN undefined
+22 NaN undefined
+[object Object]
+-3 NaN undefined
+-2 NaN undefined
+-1 NaN undefined
+0 91 91
+1 111 111
+2 98 98
+3 106 106
+4 101 101
+5 99 99
+6 116 116
+7 32 32
+8 79 79
+9 98 98
+10 106 106
+11 101 101
+12 99 99
+13 116 116
+14 93 93
+15 NaN undefined
+16 NaN undefined
+17 NaN undefined
+18 NaN undefined
+19 NaN undefined
+===*/
+
+function test() {
+    var x = 'foo' +
+            '\ucafe' +  // simple non-surrogate
+            '\uD83D\uDCA9' +  // valid U+1F4A9 in surrogate pair encoding
+            '\ubeef' +  // simple non-surrogate
+            'bar' +
+            '\ud83d\ud83d' +  // invalid surrogate pair
+            '\ud83dx' +       // -""-
+            '\udca9\ud83d' +  // -""-
+            'baz' +
+            '\ud83d';         // invalid surrogate pair, string ends before pair
+    for (i = -3; i < x.length + 3; i++) {
+        print(i, x.charCodeAt(i), x.codePointAt(i));
+    }
+
+    // .codePointAt() is generic like .charCodeAt(), but they simply String() coerce
+    // their argument first so this is coerced to [object Object] and the char lookup
+    // is then applied.  Codepoints won't be looked up from the object properties.
+    var obj = { 0: 'f', 1: 'o', 2: 'o',
+                3: '\ud83d', 4: '\udca9',
+                5: 'b', 6: 'a', 7: 'r', length: 8 };
+    print(String(obj));
+
+    for (i = -3; i < 20; i++) {
+        print(i, String.prototype.charCodeAt.call(obj, i), String.prototype.codePointAt.call(obj, i));
+    }
+}
+
+try {
+    test();
+} catch (e) {
+    print(e.stack || e);
+}