Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 1 | // Copyright 2015 the V8 project authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | // Flags: --harmony-regexp-lookbehind |
| 6 | |
| 7 | // Simple fixed-length matches. |
| 8 | assertEquals(["a"], "a".match(/^.(?<=a)/)); |
| 9 | assertNull("b".match(/^.(?<=a)/)); |
| 10 | assertEquals(["foo"], "foo1".match(/^f..(?<=.oo)/)); |
| 11 | assertEquals(["foo"], "foo2".match(/^f\w\w(?<=\woo)/)); |
| 12 | assertNull("boo".match(/^f\w\w(?<=\woo)/)); |
| 13 | assertNull("fao".match(/^f\w\w(?<=\woo)/)); |
| 14 | assertNull("foa".match(/^f\w\w(?<=\woo)/)); |
| 15 | assertEquals(["def"], "abcdef".match(/(?<=abc)\w\w\w/)); |
| 16 | assertEquals(["def"], "abcdef".match(/(?<=a.c)\w\w\w/)); |
| 17 | assertEquals(["def"], "abcdef".match(/(?<=a\wc)\w\w\w/)); |
| 18 | assertEquals(["cde"], "abcdef".match(/(?<=a[a-z])\w\w\w/)); |
| 19 | assertEquals(["def"], "abcdef".match(/(?<=a[a-z][a-z])\w\w\w/)); |
| 20 | assertEquals(["def"], "abcdef".match(/(?<=a[a-z]{2})\w\w\w/)); |
| 21 | assertEquals(["bcd"], "abcdef".match(/(?<=a{1})\w\w\w/)); |
| 22 | assertEquals(["cde"], "abcdef".match(/(?<=a{1}b{1})\w\w\w/)); |
| 23 | assertEquals(["def"], "abcdef".match(/(?<=a{1}[a-z]{2})\w\w\w/)); |
| 24 | |
| 25 | // Variable-length matches. |
| 26 | assertEquals(["def"], "abcdef".match(/(?<=[a|b|c]*)[^a|b|c]{3}/)); |
| 27 | assertEquals(["def"], "abcdef".match(/(?<=\w*)[^a|b|c]{3}/)); |
| 28 | |
| 29 | // Start of line matches. |
| 30 | assertEquals(["def"], "abcdef".match(/(?<=^abc)def/)); |
| 31 | assertEquals(["def"], "abcdef".match(/(?<=^[a-c]{3})def/)); |
| 32 | assertEquals(["def"], "xyz\nabcdef".match(/(?<=^[a-c]{3})def/m)); |
| 33 | assertEquals(["ab", "cd", "efg"], "ab\ncd\nefg".match(/(?<=^)\w+/gm)); |
| 34 | assertEquals(["ab", "cd", "efg"], "ab\ncd\nefg".match(/\w+(?<=$)/gm)); |
| 35 | assertEquals(["ab", "cd", "efg"], "ab\ncd\nefg".match(/(?<=^)\w+(?<=$)/gm)); |
| 36 | assertNull("abcdef".match(/(?<=^[^a-c]{3})def/)); |
| 37 | assertNull("foooo".match(/"^foooo(?<=^o+)$/)); |
| 38 | assertNull("foooo".match(/"^foooo(?<=^o*)$/)); |
| 39 | assertEquals(["foo"], "foo".match(/^foo(?<=^fo+)$/)); |
| 40 | assertEquals(["foooo"], "foooo".match(/^foooo(?<=^fo*)/)); |
| 41 | assertEquals(["foo", "f"], "foo".match(/^(f)oo(?<=^\1o+)$/)); |
| 42 | assertEquals(["foo", "f"], "foo".match(/^(f)oo(?<=^\1o+)$/i)); |
| 43 | assertEquals(["foo\u1234", "f"], "foo\u1234".match(/^(f)oo(?<=^\1o+).$/i)); |
| 44 | assertEquals(["def"], "abcdefdef".match(/(?<=^\w+)def/)); |
| 45 | assertEquals(["def", "def"], "abcdefdef".match(/(?<=^\w+)def/g)); |
| 46 | |
| 47 | // Word boundary matches. |
| 48 | assertEquals(["def"], "abc def".match(/(?<=\b)[d-f]{3}/)); |
| 49 | assertEquals(["def"], "ab cdef".match(/(?<=\B)\w{3}/)); |
| 50 | assertEquals(["def"], "ab cdef".match(/(?<=\B)(?<=c(?<=\w))\w{3}/)); |
| 51 | assertNull("abcdef".match(/(?<=\b)[d-f]{3}/)); |
| 52 | |
| 53 | // Negative lookbehind. |
| 54 | assertEquals(["abc"], "abcdef".match(/(?<!abc)\w\w\w/)); |
| 55 | assertEquals(["abc"], "abcdef".match(/(?<!a.c)\w\w\w/)); |
| 56 | assertEquals(["abc"], "abcdef".match(/(?<!a\wc)\w\w\w/)); |
| 57 | assertEquals(["abc"], "abcdef".match(/(?<!a[a-z])\w\w\w/)); |
| 58 | assertEquals(["abc"], "abcdef".match(/(?<!a[a-z]{2})\w\w\w/)); |
| 59 | assertNull("abcdef".match(/(?<!abc)def/)); |
| 60 | assertNull("abcdef".match(/(?<!a.c)def/)); |
| 61 | assertNull("abcdef".match(/(?<!a\wc)def/)); |
| 62 | assertNull("abcdef".match(/(?<!a[a-z][a-z])def/)); |
| 63 | assertNull("abcdef".match(/(?<!a[a-z]{2})def/)); |
| 64 | assertNull("abcdef".match(/(?<!a{1}b{1})cde/)); |
| 65 | assertNull("abcdef".match(/(?<!a{1}[a-z]{2})def/)); |
| 66 | |
| 67 | // Capturing matches. |
| 68 | assertEquals(["def", "c"], "abcdef".match(/(?<=(c))def/)); |
| 69 | assertEquals(["def", "bc"], "abcdef".match(/(?<=(\w{2}))def/)); |
| 70 | assertEquals(["def", "bc", "c"], "abcdef".match(/(?<=(\w(\w)))def/)); |
| 71 | assertEquals(["def", "a"], "abcdef".match(/(?<=(\w){3})def/)); |
| 72 | assertEquals(["d", "bc", undefined], "abcdef".match(/(?<=(bc)|(cd))./)); |
| 73 | assertEquals(["c", "a", undefined], |
| 74 | "abcdef".match(/(?<=([ab]{1,2})\D|(abc))\w/)); |
| 75 | assertEquals(["ab", "a", "b"], "abcdef".match(/\D(?<=([ab]+))(\w)/)); |
| 76 | assertEquals(["c", "d"], "abcdef".match(/(?<=b|c)\w/g)); |
| 77 | assertEquals(["cd", "ef"], "abcdef".match(/(?<=[b-e])\w{2}/g)); |
| 78 | |
| 79 | // Captures inside negative lookbehind. (They never capture.) |
| 80 | assertEquals(["de", undefined], "abcdef".match(/(?<!(^|[ab]))\w{2}/)); |
| 81 | |
| 82 | // Nested lookaround. |
| 83 | assertEquals(["ef"], "abcdef".match(/(?<=ab(?=c)\wd)\w\w/)); |
| 84 | assertEquals(["ef", "bc"], "abcdef".match(/(?<=a(?=([^a]{2})d)\w{3})\w\w/)); |
| 85 | assertEquals(["ef", "bc"], |
| 86 | "abcdef".match(/(?<=a(?=([bc]{2}(?<!a{2}))d)\w{3})\w\w/)); |
| 87 | assertNull("abcdef".match(/(?<=a(?=([bc]{2}(?<!a*))d)\w{3})\w\w/)); |
| 88 | assertEquals(["faaa"], "faaao".match(/^faaao?(?<=^f[oa]+(?=o))/)); |
| 89 | |
| 90 | // Back references. |
| 91 | assertEquals(["b", "b", "bb"], "abb".match(/(.)(?<=(\1\1))/)); |
| 92 | assertEquals(["B", "B", "bB"], "abB".match(/(.)(?<=(\1\1))/i)); |
| 93 | assertEquals(["aB", "aB", "a"], "aabAaBa".match(/((\w)\w)(?<=\1\2\1)/i)); |
| 94 | assertEquals(["Ba", "Ba", "a"], "aabAaBa".match(/(\w(\w))(?<=\1\2\1)/i)); |
| 95 | assertEquals(["b", "b", "B"], "abaBbAa".match(/(?=(\w))(?<=(\1))./i)); |
| 96 | assertEquals(["foo", "'", "foo"], " 'foo' ".match(/(?<=(.))(\w+)(?=\1)/)); |
| 97 | assertEquals(["foo", "\"", "foo"], " \"foo\" ".match(/(?<=(.))(\w+)(?=\1)/)); |
| 98 | assertNull(" .foo\" ".match(/(?<=(.))(\w+)(?=\1)/)); |
| 99 | assertNull("ab".match(/(.)(?<=\1\1\1)/)); |
| 100 | assertNull("abb".match(/(.)(?<=\1\1\1)/)); |
| 101 | assertEquals(["b", "b"], "abbb".match(/(.)(?<=\1\1\1)/)); |
| 102 | assertNull("ab".match(/(..)(?<=\1\1\1)/)); |
| 103 | assertNull("abb".match(/(..)(?<=\1\1\1)/)); |
| 104 | assertNull("aabb".match(/(..)(?<=\1\1\1)/)); |
| 105 | assertNull("abab".match(/(..)(?<=\1\1\1)/)); |
| 106 | assertNull("fabxbab".match(/(..)(?<=\1\1\1)/)); |
| 107 | assertNull("faxabab".match(/(..)(?<=\1\1\1)/)); |
| 108 | assertEquals(["ab", "ab"], "fababab".match(/(..)(?<=\1\1\1)/)); |
| 109 | |
| 110 | // Back references to captures inside the lookbehind. |
| 111 | assertEquals(["d", "C"], "abcCd".match(/(?<=\1(\w))d/i)); |
| 112 | assertEquals(["d", "x"], "abxxd".match(/(?<=\1([abx]))d/)); |
| 113 | assertEquals(["c", "ab"], "ababc".match(/(?<=\1(\w+))c/)); |
| 114 | assertEquals(["c", "b"], "ababbc".match(/(?<=\1(\w+))c/)); |
| 115 | assertNull("ababdc".match(/(?<=\1(\w+))c/)); |
| 116 | assertEquals(["c", "abab"], "ababc".match(/(?<=(\w+)\1)c/)); |
| 117 | |
| 118 | // Alternations are tried left to right, |
| 119 | // and we do not backtrack into a lookbehind. |
| 120 | assertEquals(["xabcd", "cd", ""], "xabcd".match(/.*(?<=(..|...|....))(.*)/)); |
| 121 | assertEquals(["xabcd", "bcd", ""], "xabcd".match(/.*(?<=(xx|...|....))(.*)/)); |
| 122 | assertEquals(["xxabcd", "bcd", ""], "xxabcd".match(/.*(?<=(xx|...))(.*)/)); |
| 123 | assertEquals(["xxabcd", "xx", "abcd"], "xxabcd".match(/.*(?<=(xx|xxx))(.*)/)); |
| 124 | |
| 125 | // We do not backtrack into a lookbehind. |
| 126 | // The lookbehind captures "abc" so that \1 does not match. We do not backtrack |
| 127 | // to capture only "bc" in the lookbehind. |
| 128 | assertNull("abcdbc".match(/(?<=([abc]+)).\1/)); |
| 129 | |
| 130 | // Greedy loop. |
| 131 | assertEquals(["c", "bbbbbb"], "abbbbbbc".match(/(?<=(b+))c/)); |
| 132 | assertEquals(["c", "b1234"], "ab1234c".match(/(?<=(b\d+))c/)); |
| 133 | assertEquals(["c", "b12b23b34"], "ab12b23b34c".match(/(?<=((?:b\d{2})+))c/)); |
| 134 | |
| 135 | // Sticky |
| 136 | var re1 = /(?<=^(\w+))def/g; |
| 137 | assertEquals(["def", "abc"], re1.exec("abcdefdef")); |
| 138 | assertEquals(["def", "abcdef"], re1.exec("abcdefdef")); |
| 139 | var re2 = /\Bdef/g; |
| 140 | assertEquals(["def"], re2.exec("abcdefdef")); |
| 141 | assertEquals(["def"], re2.exec("abcdefdef")); |
| 142 | |
| 143 | // Misc |
| 144 | assertNull("abcdef".match(/(?<=$abc)def/)); |
| 145 | assertEquals(["foo"], "foo".match(/^foo(?<=foo)$/)); |
| 146 | assertEquals(["foo"], "foo".match(/^f.o(?<=foo)$/)); |
| 147 | assertNull("fno".match(/^f.o(?<=foo)$/)); |
| 148 | assertNull("foo".match(/^foo(?<!foo)$/)); |
| 149 | assertNull("foo".match(/^f.o(?<!foo)$/)); |
| 150 | assertEquals(["fno"], "fno".match(/^f.o(?<!foo)$/)); |
| 151 | assertEquals(["foooo"], "foooo".match(/^foooo(?<=fo+)$/)); |
| 152 | assertEquals(["foooo"], "foooo".match(/^foooo(?<=fo*)$/)); |
| 153 | assertEquals(["abc", "abc"], /(abc\1)/.exec("abc")); |
| 154 | assertEquals(["abc", "abc"], /(abc\1)/.exec("abc\u1234")); |
| 155 | assertEquals(["abc", "abc"], /(abc\1)/i.exec("abc")); |
| 156 | assertEquals(["abc", "abc"], /(abc\1)/i.exec("abc\u1234")); |
| 157 | var oob_subject = "abcdefghijklmnabcdefghijklmn".substr(14); |
| 158 | assertNull(oob_subject.match(/(?=(abcdefghijklmn))(?<=\1)a/i)); |
| 159 | assertNull(oob_subject.match(/(?=(abcdefghijklmn))(?<=\1)a/)); |
| 160 | |
| 161 | // Mutual recursive capture/back references |
| 162 | assertEquals(["cacb", "a", ""], /(?<=a(.\2)b(\1)).{4}/.exec("aabcacbc")); |
| 163 | assertEquals(["b", "ac", "ac"], /(?<=a(\2)b(..\1))b/.exec("aacbacb")); |
| 164 | assertEquals(["x", "aa"], /(?<=(?:\1b)(aa))./.exec("aabaax")); |
| 165 | assertEquals(["x", "aa"], /(?<=(?:\1|b)(aa))./.exec("aaaax")); |