michael@0: // for-of works on strings and String objects.
michael@0: 
michael@0: load(libdir + "string.js");
michael@0: 
michael@0: function test(s, expectedCodePoints) {
michael@0:     var copy = '';
michael@0:     var codepoints = 0;
michael@0:     var singleHighSurrogate = false;
michael@0:     for (var v of s) {
michael@0:         assertEq(typeof v, 'string');
michael@0:         assertEq(v.length, isSurrogatePair(v) ? 2 : 1);
michael@0:         assertEq(false, singleHighSurrogate && isLowSurrogate(v));
michael@0:         copy += v;
michael@0:         codepoints += 1;
michael@0:         singleHighSurrogate = !isSurrogatePair(v) && isHighSurrogate(v);
michael@0:     }
michael@0:     assertEq(copy, String(s));
michael@0:     assertEq(codepoints, expectedCodePoints);
michael@0: }
michael@0: 
michael@0: test('', 0);
michael@0: test('abc', 3);
michael@0: test('a \0 \ufffe \ufeff', 7);
michael@0: 
michael@0: // Non-BMP characters are generally passed to JS in UTF-16, as surrogate pairs.
michael@0: // ES6 requires that such pairs be treated as a single code point in for-of.
michael@0: test('\ud808\udf45', 1);
michael@0: 
michael@0: // Also test invalid surrogate pairs:
michael@0: // (1) High surrogate not followed by low surrogate
michael@0: test('\ud808', 1);
michael@0: test('\ud808\u0000', 2);
michael@0: // (2) Low surrogate not preceded by high surrogate
michael@0: test('\udf45', 1);
michael@0: test('\u0000\udf45', 2);
michael@0: // (3) Low surrogate followed by high surrogate
michael@0: test('\udf45\ud808', 2);
michael@0: 
michael@0: test(new String(''), 0);
michael@0: test(new String('abc'), 3);
michael@0: test(new String('a \0 \ufffe \ufeff'), 7);
michael@0: test(new String('\ud808\udf45'), 1);
michael@0: test(new String('\ud808'), 1);
michael@0: test(new String('\ud808\u0000'), 2);
michael@0: test(new String('\udf45'), 1);
michael@0: test(new String('\u0000\udf45'), 2);
michael@0: test(new String('\udf45\ud808'), 2);