duktape/ecmascript-testcases/test-builtin-string-proto-replace.js

// FIXME: util

function printCodepoints(x) {
    var i;
    var tmp = [];
    for (i = 0; i < x.length; i++) {
        tmp.push(x.charCodeAt(i));
    }
    print(tmp.join(' '));
}

/*===
basic
fOo
fOO
bar-1123 bar-1234 bar-1345
xfoo
xfoo
xfxoxox
17185 291 50 4660 65244
17185 291 50 4660 65244
17185 291 50 17185 65244
120 9285 120 13654 120 18023
17185 4660 291 65244
17185 4660 291 65244
17185 4660 17185 291 17185 65244 17185
===*/

print('basic');

function basicTest() {
    print('foo'.replace('o', 'O'));

    print('foo'.replace(/o/g, 'O'));

    print('foo-123 foo-234 foo-345'.replace(/foo-(\d+)/g,
        function replacer(matchSub, cap, matchOffset, str) {
            return 'bar-' + String(Number(cap) + 1000);
        })
    );

    // empty search string; matches once at the beginning

    print('foo'.replace('', 'x'));

    // empty non-global regexp match; matches once at the beginning

    print('foo'.replace(/(?:)/, 'x'));

    // empty global regexp match; matches at the beginning (before first
    // char), between every char, and after the last char.  Progress
    // check in the matching process (replace() should perform regexp
    // matching like match() does, according to E5.1) ensures this
    // process terminates and matches only once at each point.

    print('foo'.replace(/(?:)/g, 'x'));

    // non-BMP variants of above tests: these are important because
    // the implementation works with both byte and char offsets

    printCodepoints('\u1234\u0123\u0032\u1234\ufedc'.replace('\u1234', '\u4321'));
    printCodepoints('\u1234\u0123\u0032\u1234\ufedc'.replace(/\u1234/, '\u4321'));
    printCodepoints('\u1234\u0123\u0032\u1234\ufedc'.replace(/\u1234/g, '\u4321'));
    printCodepoints('\u1234\u2345\u1234\u3456\u1234\u4567'.replace(/\u1234(.)/g,
        function replacer(matchSub, cap, matchOffset, str) {
            return 'x' + String.fromCharCode(cap.charCodeAt(0) + 0x100);
        })
    );
    printCodepoints('\u1234\u0123\ufedc'.replace('', '\u4321'));
    printCodepoints('\u1234\u0123\ufedc'.replace(/(?:)/, '\u4321'));
    printCodepoints('\u1234\u0123\ufedc'.replace(/(?:)/g, '\u4321'));
}

try {
    basicTest();
} catch (e) {
    print(e);
}

/*===
dollar replacements
string xooxxr
string $oo$$r
string ffoobbaar
string f[f][fo]bar
string f[obar][bar]bar
string fo$0ar
string foobobar
string foobar
string foob2ar
string foob1ar
string $0abcdefghijklmnopqrstuvwxyzb7b8b9c0c1c2c3c4c5c6c7c8c9d0d1d2d3d4d5d6d7d8d9e0e1e2e3e4e5e6e7e8e9f0f1f2f3f4f5f6f7f8f9g0g1g2g3g4g5g6g7g8g9h0h1h2h3h4h5h6h7h8h9i0i1i2i3i4i5i6i7i8i9j0
string $1-$11,$1-$22
===*/

/* Test all dollar replacement syntax variants */

print('dollar replacements');

function dollarTest() {
    function test(str, searchValue, replaceValue) {
        var t;

        try {
            t = str.replace(searchValue, replaceValue);
            print(typeof t, t);
        } catch (e) {
            print(e.name);
        }
    }

    test('foobar', /([a-f])/g, 'x');

    test('foobar', /([a-f])/g, '$$');

    test('foobar', /([a-f])/g, '$&$&');

    test('foobar', /o/g, '[$`]');

    test('foobar', /o/g, '[$\']');

    test('foobar', /(ob)/g, '$0');   // $0 is not valid, so is interpreted literally as "$0"
    test('foobar', /(ob)/g, '$1$1');
    test('foobar', /(ob)/g, '$01');  // interpreted as "replacement 1"
    test('foobar', /(ob)/g, '$012'); // interpreted as "replacement 01 + '2'"

    // Behavior here is implemented dependent because 11 > NCapturingParens.
    // Rhino and V8 both seem to fall back to interpreting the expression
    // as '$1' followed by a verbatim '1', so we test for that.
    test('foobar', /(ob)/g, '$11');  // interpreted as "replacement 11" not "replacement 1 + '1'"

    // $27 -> $29 are larger than NCapturingParens but $2 is not, so V8/Rhino fall back
    // to interpreting them as $2 followed by a literal digit. Same goes for all matches
    // up to $99.  $100 will be interpreted as $10 followed by a literal '0'.
    test('abcdefghijklmnopqrstuvwxyz',
         /(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)/g,
         '$0$1$2$3$4$5$6$7$8$9$10$11$12$13$14$15$16$17$18$19$20$21$22$23$24$25$26$27$28$29' +
         '$30$31$32$33$34$35$36$37$38$39$40$41$42$43$44$45$46$47$48$49' +
         '$50$51$52$53$54$55$56$57$58$59$60$61$62$63$64$65$66$67$68$69' +
         '$70$71$72$73$74$75$76$77$78$79$80$81$82$83$84$85$86$87$88$89' +
         '$90$91$92$93$94$95$96$97$98$99$100');

    // Specific test form E5.1 Section 15.5.4.11:
    //
    // test("$1,$2", /(\$(\d))/g, "$$1-$1$2");
    //
    // This test cannot be used as is, because "\$" is not a valid RegExp
    // identity escape: identity escape characters cannot come from
    // IdentifierPart; since IdentifierPart includes '$' it must cause a
    // SyntaxError.  On the other hand, a plain '$' has a special meaning,
    // so a dollar sign must be encoded as a hex/unicode escape!

    test("$1,$2", /(\u0024(\d))/g, "$$1-$1$2");
}

try {
    dollarTest();
} catch (e) {
    print(e);
}

/*===
function replacer
repl1 o 1 foobar
string f[repl-1]obar
repl1 o 1 foobar
repl1 o 2 foobar
string f[repl-1][repl-2]bar
repl2 oba o b a 2 foobar
string fo[repl-2]r
repl3 yada-1 ya d 1 0 yada-1 yadda-2 yaddda-3
repl3 yadda-2 ya dd 2 7 yada-1 yadda-2 yaddda-3
repl3 yaddda-3 ya ddd 3 15 yada-1 yadda-2 yaddda-3
string Yadda-2 Yadda-4 Yadda-6
repl_args 3 o <> 1 foobar
repl_args 3 o <> 2 foobar
string f[repl-args-1][repl-args-2]bar
repl_args 6 oba <o:b:a> 2 foobar
string fo[repl-args-2]r
repl4 o o 1 foObar
toString() repl4 retval
repl4 O O 2 foObar
toString() repl4 retval
string f[repl-1 cap=o][repl-2 cap=O]bar
repl2 foo foo foo undefined 0 foobar
repl2 bar bar undefined bar 3 foobar
string [repl-0][repl-3]
===*/

/* Function replacer, regexp captures. */

print('function replacer');

function functionReplacerTest() {
    // zero captures
    function repl1(matchSub, matchOffset, str) {
        print('repl1', matchSub, matchOffset, str);
        return '[repl-' + matchOffset + ']';
    }

    // three captures
    function repl2(matchSub, cap1, cap2, cap3, matchOffset, str) {
        print('repl2', matchSub, cap1, cap2, cap3, matchOffset, str);
        return '[repl-' + matchOffset + ']';
    }

    // three captures
    function repl3(matchSub, cap1, cap2, cap3, matchOffset, str) {
        print('repl3', matchSub, cap1, cap2, cap3, matchOffset, str);
        return 'Yadda-' + Number(cap3) * 2;
    }

    // one capture, returns an object which needs to be string coerced
    function repl4(matchSub, cap, matchOffset, str) {
        print('repl4', matchSub, cap, matchOffset, str);
        return {
            toString: function() { print('toString() repl4 retval');
                                   return '[repl-' + matchOffset + ' cap=' + cap + ']'; },
            valueOf: function() { print('valueOf() repl4 retval');
                                  return 'not called'; }
        };
    }

    // allows any captures, use arguments object to figure out args
    function repl_args() {
        var matchSub, matchOffset, str;
        var caps = [];
        var i, n;

        matchSub = arguments[0];
        matchOffset = arguments[arguments.length - 2];
        str = arguments[arguments.length - 1];
        for (i = 1; i < arguments.length - 2; i++) {
            caps.push(arguments[i]);
        }

        print('repl_args', arguments.length, matchSub, '<' + caps.join(':') + '>', matchOffset, str);

        return '[repl-args-' + matchOffset + ']';
    }

    function test(str, searchValue, replaceValue) {
        var t;

        try {
            t = str.replace(searchValue, replaceValue);
            print(typeof t, t);
        } catch (e) {
            print(e.name);
        }
    }

    test('foobar', 'o', repl1);
    test('foobar', /o/g, repl1);

    test('foobar', /(o)(b)(.)/g, repl2);

    test('yada-1 yadda-2 yaddda-3', /(ya)(d+)a-(\d+)/g, repl3);

    test('foobar', /o/g, repl_args);
    test('foobar', /(o)(b)(.)/g, repl_args);

    // replace function value coercion
    test('foObar', /(o)/gi, repl4);

    // even captures with no match are given to the replacer, as undefined
    test('foobar', /((foo)|(bar))/g, repl2);
}

try {
    functionReplacerTest();
} catch (e) {
    print(e);
}

/*===
regexp details
non-global, match
re: source foo global false ignoreCase false multiline false lastIndex number 0
re: source foo global false ignoreCase false multiline false lastIndex string 123
replace func
re: source foo global false ignoreCase false multiline false lastIndex string 123
bar[repl]barfoo
re: source foo global false ignoreCase false multiline false lastIndex string 123
non-global, no match
re: source goo global false ignoreCase false multiline false lastIndex number 0
re: source goo global false ignoreCase false multiline false lastIndex string 123
barfoobarfoo
re: source goo global false ignoreCase false multiline false lastIndex number 0
global, match
re: source foo global true ignoreCase false multiline false lastIndex number 0
re: source foo global true ignoreCase false multiline false lastIndex string 123
replace func
re: source foo global true ignoreCase false multiline false lastIndex number 6
replace func
re: source foo global true ignoreCase false multiline false lastIndex number 12
replace func
re: source foo global true ignoreCase false multiline false lastIndex number 18
bar[repl]bar[repl]bar[repl]bar
re: source foo global true ignoreCase false multiline false lastIndex number 0
global, no match
re: source goo global true ignoreCase false multiline false lastIndex number 0
re: source goo global true ignoreCase false multiline false lastIndex string 123
barfoobarfoobarfoobar
re: source goo global true ignoreCase false multiline false lastIndex number 0
===*/

/* RegExp details, especially how the RegExp instance properties are
 * changed during non-global or global matching.
 *
 * lastIndex handling:
 *
 *   - If RegExp is not global, E5.1 Section 15.5.4.11 is quite vague
 *     on how lastIndex should behave.  If the normal RegExp exec()
 *     algorithm is used, lastIndex is not updated on a match, but is
 *     written to zero if no match is found (E5.1 Section 15.10.6.2,
 *     step 9.a.i).
 *
 *   - If RegExp is global, E5.1 Section 15.5.4.11 indicates that matching
 *     should proceed similar to what match() does.  Match calls RegExp
 *     exec() which, for global RegExps, updates lastIndex on every call.
 *     On successful matches, lastIndex points to the next character after
 *     the matching part.  On unsuccessful matches, lastIndex is written
 *     to zero (E5.1 Section 15.10.6.2).  When matching starts, lastIndex
 *     is written to zero and any previous value is ignored (and is thus
 *     never coerced).
 *
 * The current implementation will call the replacer while doing the
 * RegExp matching.  This matters for the global case because the replacer
 * will see the intermediate lastIndex values and will even be able to
 * interfere with the matching process.  E5.1 Section 15.5.4.11 can also
 * be read to imply that all matches are first made into some internal
 * value before making any replacer calls.  In this interpretation the
 * replacer would see a lastIndex of zero (since the final non-match will
 * set lastIndex to zero) and although it could tamper with lastIndex, it
 * couldn't affect the replacement process any more.
 *
 * The current test case tests for the former interpretation, i.e. that
 * replacer is called during matching.  Note that at least V8 deviates
 * from this interpretation.
 */

print('regexp details');

function regexpDetailsTest() {
    var re;

    function dumpRe(r) {
        print('re:', 'source', r.source, 'global', r.global, 'ignoreCase', r.ignoreCase,
              'multiline', r.multiline, 'lastIndex', typeof r.lastIndex, r.lastIndex);
    }

    // non-global case with a match: lastIndex is not touched
    print('non-global, match');
    re = /foo/;
    dumpRe(re);
    re.lastIndex = '123';
    dumpRe(re);
    print('barfoobarfoo'.replace(re, function() {
        print('replace func');
        dumpRe(re);
        return '[repl]';
    }));
    dumpRe(re);

    // non-global case with no match: lastIndex is written to zero
    print('non-global, no match');
    re = /goo/;
    dumpRe(re);
    re.lastIndex = '123';
    dumpRe(re);
    print('barfoobarfoo'.replace(re, function() {
        print('replace func');
        dumpRe(re);
        return '[repl]';
    }));
    dumpRe(re);

    // global case with a matches: lastIndex is updated before every
    // replacer call, and written to zero at the end
    print('global, match');
    re = /foo/g;
    dumpRe(re);
    re.lastIndex = '123';
    dumpRe(re);
    print('barfoobarfoobarfoobar'.replace(re, function() {
        print('replace func');
        dumpRe(re);
        return '[repl]';
    }));
    dumpRe(re);

    // global case with no match: lastIndex is written to zero
    print('global, no match');
    re = /goo/g;
    dumpRe(re);
    re.lastIndex = '123';
    dumpRe(re);
    print('barfoobarfoobarfoobar'.replace(re, function() {
        print('replace func');
        dumpRe(re);
        return '[repl]';
    }));
    dumpRe(re);
}

try {
    regexpDetailsTest();
} catch (e) {
    print(e);
}

/*===
coercion
TypeError
TypeError
string tRUE
string faLSE
string 1xy
string qUux
string 1:3
string [object OBJECT]
===*/

/* Coercion of 'this' and arguments. */

print('coercion');

function coercionTest() {
    function test(thisValue, searchValue, replaceValue) {
        var t;

        try {
            t = String.prototype.replace.call(thisValue, searchValue, replaceValue);
            print(typeof t, t);
        } catch (e) {
            print(e.name);
        }
    }

    test(undefined, 'foo', 'bar');
    test(null, 'foo', 'bar');
    test(true, 'rue', 'RUE');
    test(false, 'lse', 'LSE');
    test(123, '23', 'xy');
    test('quux', 'u', 'U');
    test([1,2], ',2', ':3');
    test({ foo:1, bar:2 }, 'Object', 'OBJECT');
}

try {
    coercionTest();
} catch (e) {
    print(e);
}