regexp-lookbehind.html (12110B)
1 <!DOCTYPE html> 2 <meta charset="utf-8"> 3 <!-- 4 Copyright (C) 2017 the V8 project authors. All rights reserved. 5 This code is governed by the BSD license found in the LICENSE file. 6 --> 7 <title>JavaScript RegExp lookbehind assertions: alternations</title> 8 <script src="/resources/testharness.js"></script> 9 <script src="/resources/testharnessreport.js"></script> 10 11 <script> 12 test(t => { 13 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/alternations.js 14 assert_array_equals("xabcd".match(/.*(?<=(..|...|....))(.*)/), ["xabcd", "cd", ""], "#1"); 15 assert_array_equals("xabcd".match(/.*(?<=(xx|...|....))(.*)/), ["xabcd", "bcd", ""], "#2"); 16 assert_array_equals("xxabcd".match(/.*(?<=(xx|...))(.*)/), ["xxabcd", "bcd", ""], "#3"); 17 assert_array_equals("xxabcd".match(/.*(?<=(xx|xxx))(.*)/), ["xxabcd", "xx", "abcd"], "#4"); 18 }, "Alternations are tried left to right, with no backtracking into a lookbehind"); 19 20 test(t => { 21 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/back-references.js 22 assert_array_equals("abb".match(/(.)(?<=(\1\1))/), ["b", "b", "bb"], "#1"); 23 assert_array_equals("abB".match(/(.)(?<=(\1\1))/i), ["B", "B", "bB"], "#2"); 24 assert_array_equals("aabAaBa".match(/((\w)\w)(?<=\1\2\1)/i), ["aB", "aB", "a"], "#3"); 25 assert_array_equals("aabAaBa".match(/(\w(\w))(?<=\1\2\1)/i), ["Ba", "Ba", "a"], "#4"); 26 assert_array_equals("abaBbAa".match(/(?=(\w))(?<=(\1))./i), ["b", "b", "B"], "#5"); 27 assert_array_equals(" 'foo' ".match(/(?<=(.))(\w+)(?=\1)/), ["foo", "'", "foo"], "#6"); 28 assert_array_equals(" \"foo\" ".match(/(?<=(.))(\w+)(?=\1)/), ["foo", "\"", "foo"], "#7"); 29 assert_array_equals("abbb".match(/(.)(?<=\1\1\1)/), ["b", "b"], "#8"); 30 assert_array_equals("fababab".match(/(..)(?<=\1\1\1)/), ["ab", "ab"], "#9"); 31 assert_equals(" .foo\" ".match(/(?<=(.))(\w+)(?=\1)/), null, "#10"); 32 assert_equals("ab".match(/(.)(?<=\1\1\1)/), null, "#11"); 33 assert_equals("abb".match(/(.)(?<=\1\1\1)/), null, "#12"); 34 assert_equals("ab".match(/(..)(?<=\1\1\1)/), null, "#13"); 35 assert_equals("abb".match(/(..)(?<=\1\1\1)/), null, "#14"); 36 assert_equals("aabb".match(/(..)(?<=\1\1\1)/), null, "#15"); 37 assert_equals("abab".match(/(..)(?<=\1\1\1)/), null, "#16"); 38 assert_equals("fabxbab".match(/(..)(?<=\1\1\1)/), null, "#17"); 39 assert_equals("faxabab".match(/(..)(?<=\1\1\1)/), null, "#18"); 40 }, "Back-references"); 41 42 test(t => { 43 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/back-references-to-captures.js 44 assert_array_equals("abcCd".match(/(?<=\1(\w))d/i), ["d", "C"], "#1"); 45 assert_array_equals("abxxd".match(/(?<=\1([abx]))d/), ["d", "x"], "#2"); 46 assert_array_equals("ababc".match(/(?<=\1(\w+))c/), ["c", "ab"], "#3"); 47 assert_array_equals("ababbc".match(/(?<=\1(\w+))c/), ["c", "b"], "#4"); 48 assert_equals("ababdc".match(/(?<=\1(\w+))c/), null, "#5"); 49 assert_array_equals("ababc".match(/(?<=(\w+)\1)c/), ["c", "abab"], "#6"); 50 }, "Back-references to captures inside the lookbehind"); 51 52 test(t => { 53 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/captures.js 54 const str = "abcdef"; 55 assert_array_equals(str.match(/(?<=(c))def/), ["def", "c"], "#1"); 56 assert_array_equals(str.match(/(?<=(\w{2}))def/), ["def", "bc"], "#2"); 57 assert_array_equals(str.match(/(?<=(\w(\w)))def/), ["def", "bc", "c"], "#3"); 58 assert_array_equals(str.match(/(?<=(\w){3})def/), ["def", "a"], "#4"); 59 assert_array_equals(str.match(/(?<=(bc)|(cd))./), ["d", "bc", undefined], "#5"); 60 assert_array_equals(str.match(/(?<=([ab]{1,2})\D|(abc))\w/), ["c", "a", undefined], "#6"); 61 assert_array_equals(str.match(/\D(?<=([ab]+))(\w)/), ["ab", "a", "b"], "#7"); 62 assert_array_equals(str.match(/(?<=b|c)\w/g), ["c", "d"], "#8"); 63 assert_array_equals(str.match(/(?<=[b-e])\w{2}/g), ["cd", "ef"], "#9"); 64 }, "Capturing matches"); 65 66 test(t => { 67 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/captures-negative.js 68 assert_array_equals("abcdef".match(/(?<!(^|[ab]))\w{2}/), ["de", undefined]); 69 }, "Captures inside negative lookbehind"); 70 71 test(t => { 72 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/do-not-backtrack.js 73 // The lookbehind captures "abc" so that \1 does not match. We do not backtrack 74 // to capture only "bc" in the lookbehind. 75 assert_equals("abcdbc".match(/(?<=([abc]+)).\1/), null); 76 }, "Do not backtrack into a lookbehind"); 77 78 test(t => { 79 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/greedy-loop.js 80 assert_array_equals("abbbbbbc".match(/(?<=(b+))c/), ["c", "bbbbbb"], "#1"); 81 assert_array_equals("ab1234c".match(/(?<=(b\d+))c/), ["c", "b1234"], "#2"); 82 assert_array_equals("ab12b23b34c".match(/(?<=((?:b\d{2})+))c/), ["c", "b12b23b34"], "#3"); 83 }, "Greedy loop"); 84 85 test(t => { 86 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/misc.js 87 assert_equals("abcdef".match(/(?<=$abc)def/), null, "#1"); 88 assert_equals("fno".match(/^f.o(?<=foo)$/), null, "#2"); 89 assert_equals("foo".match(/^foo(?<!foo)$/), null, "#3"); 90 assert_equals("foo".match(/^f.o(?<!foo)$/), null, "#4"); 91 assert_array_equals("foo".match(/^foo(?<=foo)$/), ["foo"], "#5"); 92 assert_array_equals("foo".match(/^f.o(?<=foo)$/), ["foo"], "#6"); 93 assert_array_equals("fno".match(/^f.o(?<!foo)$/), ["fno"], "#7"); 94 assert_array_equals("foooo".match(/^foooo(?<=fo+)$/), ["foooo"], "#8"); 95 assert_array_equals("foooo".match(/^foooo(?<=fo*)$/), ["foooo"], "#9"); 96 assert_array_equals(/(abc\1)/.exec("abc"), ["abc", "abc"], "#10"); 97 assert_array_equals(/(abc\1)/.exec("abc\u1234"), ["abc", "abc"], "#11"); 98 assert_array_equals(/(abc\1)/i.exec("abc"), ["abc", "abc"], "#12"); 99 assert_array_equals(/(abc\1)/i.exec("abc\u1234"), ["abc", "abc"], "#13"); 100 }, "Miscellaneous"); 101 102 test(t => { 103 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/mutual-recursive.js 104 assert_array_equals(/(?<=a(.\2)b(\1)).{4}/.exec("aabcacbc"), ["cacb", "a", ""], "#1"); 105 assert_array_equals(/(?<=a(\2)b(..\1))b/.exec("aacbacb"), ["b", "ac", "ac"], "#2"); 106 assert_array_equals(/(?<=(?:\1b)(aa))./.exec("aabaax"), ["x", "aa"], "#3"); 107 assert_array_equals(/(?<=(?:\1|b)(aa))./.exec("aaaax"), ["x", "aa"], "#4"); 108 }, "Mutual recursive capture/back references"); 109 110 test(t => { 111 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/negative.js 112 assert_array_equals("abcdef".match(/(?<!abc)\w\w\w/), ["abc"], "#1"); 113 assert_array_equals("abcdef".match(/(?<!a.c)\w\w\w/), ["abc"], "#2"); 114 assert_array_equals("abcdef".match(/(?<!a\wc)\w\w\w/), ["abc"], "#3"); 115 assert_array_equals("abcdef".match(/(?<!a[a-z])\w\w\w/), ["abc"], "#4"); 116 assert_array_equals("abcdef".match(/(?<!a[a-z]{2})\w\w\w/), ["abc"], "#5"); 117 assert_equals("abcdef".match(/(?<!abc)def/), null, "#6"); 118 assert_equals("abcdef".match(/(?<!a.c)def/), null, "#7"); 119 assert_equals("abcdef".match(/(?<!a\wc)def/), null, "#8"); 120 assert_equals("abcdef".match(/(?<!a[a-z][a-z])def/), null, "#9"); 121 assert_equals("abcdef".match(/(?<!a[a-z]{2})def/), null, "#10"); 122 assert_equals("abcdef".match(/(?<!a{1}b{1})cde/), null, "#11"); 123 assert_equals("abcdef".match(/(?<!a{1}[a-z]{2})def/), null, "#12"); 124 }, "Negative lookbehinds"); 125 126 test(t => { 127 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/nested-lookaround.js 128 assert_array_equals("abcdef".match(/(?<=ab(?=c)\wd)\w\w/), ["ef"], "#1"); 129 assert_array_equals("abcdef".match(/(?<=a(?=([^a]{2})d)\w{3})\w\w/), ["ef", "bc"], "#2"); 130 assert_array_equals("abcdef".match(/(?<=a(?=([bc]{2}(?<!a{2}))d)\w{3})\w\w/), ["ef", "bc"], "#3"); 131 assert_array_equals("faaao".match(/^faaao?(?<=^f[oa]+(?=o))/), ["faaa"], "#4"); 132 assert_equals("abcdef".match(/(?<=a(?=([bc]{2}(?<!a*))d)\w{3})\w\w/), null, "#5"); 133 }, "Nested lookaround"); 134 135 test(t => { 136 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/simple-fixed-length.js 137 assert_equals("b".match(/^.(?<=a)/), null, "#1"); 138 assert_equals("boo".match(/^f\w\w(?<=\woo)/), null, "#2"); 139 assert_equals("fao".match(/^f\w\w(?<=\woo)/), null, "#3"); 140 assert_equals("foa".match(/^f\w\w(?<=\woo)/), null, "#4"); 141 assert_array_equals("a".match(/^.(?<=a)/), ["a"], "#5"); 142 assert_array_equals("foo1".match(/^f..(?<=.oo)/), ["foo"], "#6"); 143 assert_array_equals("foo2".match(/^f\w\w(?<=\woo)/), ["foo"], "#7"); 144 assert_array_equals("abcdef".match(/(?<=abc)\w\w\w/), ["def"], "#8"); 145 assert_array_equals("abcdef".match(/(?<=a.c)\w\w\w/), ["def"], "#9"); 146 assert_array_equals("abcdef".match(/(?<=a\wc)\w\w\w/), ["def"], "#10"); 147 assert_array_equals("abcdef".match(/(?<=a[a-z])\w\w\w/), ["cde"], "#11"); 148 assert_array_equals("abcdef".match(/(?<=a[a-z][a-z])\w\w\w/), ["def"], "#12"); 149 assert_array_equals("abcdef".match(/(?<=a[a-z]{2})\w\w\w/), ["def"], "#13"); 150 assert_array_equals("abcdef".match(/(?<=a{1})\w\w\w/), ["bcd"], "#14"); 151 assert_array_equals("abcdef".match(/(?<=a{1}b{1})\w\w\w/), ["cde"], "#15"); 152 assert_array_equals("abcdef".match(/(?<=a{1}[a-z]{2})\w\w\w/), ["def"], "#16"); 153 }, "Simple fixed-length matches"); 154 155 test(t => { 156 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/sliced-strings.js 157 const oob_subject = "abcdefghijklmnabcdefghijklmn".slice(14); 158 assert_equals(oob_subject.match(/(?=(abcdefghijklmn))(?<=\1)a/i), null, ""); 159 assert_equals(oob_subject.match(/(?=(abcdefghijklmn))(?<=\1)a/), null, ""); 160 assert_equals("abcdefgabcdefg".slice(1).match(/(?=(abcdefg))(?<=\1)/), null, ""); 161 }, "Sliced strings"); 162 163 test(t => { 164 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/start-of-line.js 165 assert_equals("abcdef".match(/(?<=^[^a-c]{3})def/), null, "#1"); 166 assert_equals("foooo".match(/"^foooo(?<=^o+)$/), null, "#2"); 167 assert_equals("foooo".match(/"^foooo(?<=^o*)$/), null, "#3"); 168 assert_array_equals("abcdef".match(/(?<=^abc)def/), ["def"], "#4"); 169 assert_array_equals("abcdef".match(/(?<=^[a-c]{3})def/), ["def"], "#5"); 170 assert_array_equals("xyz\nabcdef".match(/(?<=^[a-c]{3})def/m), ["def"], "#6"); 171 assert_array_equals("ab\ncd\nefg".match(/(?<=^)\w+/gm), ["ab", "cd", "efg"], "#7"); 172 assert_array_equals("ab\ncd\nefg".match(/\w+(?<=$)/gm), ["ab", "cd", "efg"], "#8"); 173 assert_array_equals("ab\ncd\nefg".match(/(?<=^)\w+(?<=$)/gm), ["ab", "cd", "efg"], "#9"); 174 assert_array_equals("foo".match(/^foo(?<=^fo+)$/), ["foo"], "#10"); 175 assert_array_equals("foooo".match(/^foooo(?<=^fo*)/), ["foooo"], "#11"); 176 assert_array_equals("foo".match(/^(f)oo(?<=^\1o+)$/), ["foo", "f"], "#12"); 177 assert_array_equals("foo".match(/^(f)oo(?<=^\1o+)$/i), ["foo", "f"], "#13"); 178 assert_array_equals("foo\u1234".match(/^(f)oo(?<=^\1o+).$/i), ["foo\u1234", "f"], "#14"); 179 assert_array_equals("abcdefdef".match(/(?<=^\w+)def/), ["def"], "#15"); 180 assert_array_equals("abcdefdef".match(/(?<=^\w+)def/g), ["def", "def"], "#16"); 181 }, "Start of line matches"); 182 183 test(t => { 184 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/sticky.js 185 const re1 = /(?<=^(\w+))def/g; 186 assert_array_equals(re1.exec("abcdefdef"), ["def", "abc"], "#1"); 187 assert_array_equals(re1.exec("abcdefdef"), ["def", "abcdef"], "#2"); 188 const re2 = /\Bdef/g; 189 assert_array_equals(re2.exec("abcdefdef"), ["def"], "#3"); 190 assert_array_equals(re2.exec("abcdefdef"), ["def"], "#4"); 191 }, "Sticky matches"); 192 193 test(t => { 194 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/variable-length.js 195 assert_array_equals("abcdef".match(/(?<=[a|b|c]*)[^a|b|c]{3}/), ["def"], "#1"); 196 assert_array_equals("abcdef".match(/(?<=\w*)[^a|b|c]{3}/), ["def"], "#2"); 197 }, "Variable-length matches"); 198 199 test(t => { 200 // https://github.com/tc39/test262/blob/main/test/built-ins/RegExp/lookBehind/word-boundary.js 201 assert_array_equals("abc def".match(/(?<=\b)[d-f]{3}/), ["def"], "#1"); 202 assert_array_equals("ab cdef".match(/(?<=\B)\w{3}/), ["def"], "#2"); 203 assert_array_equals("ab cdef".match(/(?<=\B)(?<=c(?<=\w))\w{3}/), ["def"], "#3"); 204 assert_equals("abcdef".match(/(?<=\b)[d-f]{3}/), null, "#4"); 205 }, "Word boundary matches"); 206 </script>