OLD | NEW |
---|---|
(Empty) | |
1 // Copyright 2016 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-unicode-regexps --harmony-regexp-lookbehind | |
6 | |
7 function execl(expectation, regexp, subject) { | |
8 if (regexp instanceof String) regexp = new RegExp(regexp, "u"); | |
9 assertEquals(expectation, regexp.exec(subject)); | |
10 } | |
11 | |
12 function execs(expectation, regexp_source, subject) { | |
13 execl(expectation, new RegExp(regexp_source, "u"), subject); | |
14 } | |
15 | |
16 // Character ranges. | |
17 execl(["A"], /[A-D]/u, "A"); | |
18 execs(["A"], "[A-D]", "A"); | |
19 execl(["ABCD"], /[A-D]+/u, "ZABCDEF"); | |
20 execs(["ABCD"], "[A-D]+", "ZABCDEF"); | |
21 | |
22 execl(["\u{12345}"], /[\u1234-\u{12345}]/u, "\u{12345}"); | |
23 execs(["\u{12345}"], "[\u1234-\u{12345}]", "\u{12345}"); | |
24 execl(null, /[^\u1234-\u{12345}]/u, "\u{12345}"); | |
25 execs(null, "[^\u1234-\u{12345}]", "\u{12345}"); | |
26 | |
27 execl(["\u{1234}"], /[\u1234-\u{12345}]/u, "\u{1234}"); | |
28 execs(["\u{1234}"], "[\u1234-\u{12345}]", "\u{1234}"); | |
29 execl(null, /[^\u1234-\u{12345}]/u, "\u{1234}"); | |
30 execs(null, "[^\u1234-\u{12345}]", "\u{1234}"); | |
31 | |
32 execl(null, /[\u1234-\u{12345}]/u, "\u{1233}"); | |
33 execs(null, "[\u1234-\u{12345}]", "\u{1233}"); | |
34 execl(["\u{1233}"], /[^\u1234-\u{12345}]/u, "\u{1233}"); | |
35 execs(["\u{1233}"], "[^\u1234-\u{12345}]", "\u{1233}"); | |
36 | |
37 execl(["\u{12346}"], /[^\u1234-\u{12345}]/u, "\u{12346}"); | |
38 execs(["\u{12346}"], "[^\u1234-\u{12345}]", "\u{12346}"); | |
39 execl(null, /[\u1234-\u{12345}]/u, "\u{12346}"); | |
40 execs(null, "[\u1234-\u{12345}]", "\u{12346}"); | |
41 | |
42 execl(["\u{12342}"], /[\u{12340}-\u{12345}]/u, "\u{12342}"); | |
43 execs(["\u{12342}"], "[\u{12340}-\u{12345}]", "\u{12342}"); | |
44 execl(null, /[^\u{12340}-\u{12345}]/u, "\u{12342}"); | |
45 execs(null, "[^\u{12340}-\u{12345}]", "\u{12342}"); | |
mathias
2016/01/21 12:44:36
To test the equivalent regular expression using su
| |
46 | |
47 execl(["\u{ffff}"], /[\u{ff80}-\u{12345}]/u, "\u{ffff}"); | |
48 execs(["\u{ffff}"], "[\u{ff80}-\u{12345}]", "\u{ffff}"); | |
49 execl(null, /[^\u{ff80}-\u{12345}]/u, "\u{ffff}"); | |
50 execs(null, "[^\u{ff80}-\u{12345}]", "\u{ffff}"); | |
mathias
2016/01/21 12:44:36
Same here:
execl(["\u{ffff}"], /[\u{ff80}-\u{1234
| |
51 | |
52 // Lone surrogate | |
53 execl(["\ud800"], /[^\u{ff80}-\u{12345}]/u, "\uff99\u{d800}A"); | |
54 execs(["\udc00"], "[^\u{ff80}-\u{12345}]", "\uff99\u{dc00}A"); | |
55 execl(["\udc01"], /[\u0100-\u{10ffff}]/u, "A\udc01"); | |
56 execl(["\udc03"], /[\udc01-\udc03]/u, "\ud801\udc02\udc03"); | |
57 execl(["\ud801"], /[\ud801-\ud803]/u, "\ud802\udc01\ud801"); | |
58 | |
59 // Paired sorrogate. | |
60 execl(null, /[^\u{ff80}-\u{12345}]/u, "\u{d800}\u{dc00}"); | |
61 execs(null, "[^\u{ff80}-\u{12345}]", "\u{d800}\u{dc00}"); | |
62 execl(["\ud800\udc00"], /[\u{ff80}-\u{12345}]/u, "\u{d800}\u{dc00}"); | |
63 execs(["\ud800\udc00"], "[\u{ff80}-\u{12345}]", "\u{d800}\u{dc00}"); | |
64 execl(["foo\u{10e6d}bar"], /foo\ud803\ude6dbar/u, "foo\u{10e6d}bar"); | |
65 | |
66 // Lone surrogates | |
67 execl(["\ud801\ud801"], /\ud801+/u, "\ud801\udc01\ud801\ud801"); | |
68 execl(["\udc01\udc01"], /\udc01+/u, "\ud801\ud801\udc01\udc01\udc01"); | |
69 | |
70 execl(["\udc02\udc03A"], /\W\WA/u, "\ud801\udc01A\udc02\udc03A"); | |
71 execl(["\ud801\ud802"], /\ud801./u, "\ud801\udc01\ud801\ud802"); | |
72 execl(["\udc02\udc03A"], /[\ud800-\udfff][\ud800-\udfff]A/u, | |
73 "\ud801\udc01A\udc02\udc03A"); | |
74 | |
75 // Character classes | |
76 execl(null, /\w/u, "\ud801\udc01"); | |
77 execl(["\ud801"], /[^\w]/, "\ud801\udc01"); | |
78 execl(["\ud801\udc01"], /[^\w]/u, "\ud801\udc01"); | |
79 execl(["\ud801"], /\W/, "\ud801\udc01"); | |
80 execl(["\ud801\udc01"], /\W/u, "\ud801\udc01"); | |
81 | |
82 execl(["\ud800X"], /.X/u, "\ud800XaX"); | |
83 execl(["aX"], /.(?<!\ud800)X/u, "\ud800XaX"); | |
84 execl(["aX"], /.(?<![\ud800-\ud900])X/u, "\ud800XaX"); | |
85 | |
86 execl(null, /[]/u, "\u1234"); | |
87 execl(["0abc"], /[^]abc/u, "0abc"); | |
88 execl(["\u1234abc"], /[^]abc/u, "\u1234abc"); | |
89 execl(["\u{12345}abc"], /[^]abc/u, "\u{12345}abc"); | |
90 | |
91 // Backward matches of lone surrogates. | |
92 execl(["B", "\ud803A"], /(?<=([\ud800-\ud900]A))B/u, | |
93 "\ud801\udc00AB\udc00AB\ud802\ud803AB"); | |
94 execl(["B", "\udc00A"], /(?<=([\ud800-\u{10300}]A))B/u, | |
95 "\ud801\udc00AB\udc00AB\ud802\ud803AB"); | |
96 execl(["B", "\udc11A"], /(?<=([\udc00-\udd00]A))B/u, | |
97 "\ud801\udc00AB\udc11AB\ud802\ud803AB"); | |
98 execl(["X", "\ud800C"], /(?<=(\ud800\w))X/u, | |
99 "\ud800\udc00AX\udc11BX\ud800\ud800CX"); | |
100 execl(["C", "\ud800\ud800"], /(?<=(\ud800.))\w/u, | |
101 "\ud800\udc00AX\udc11BX\ud800\ud800CX"); | |
102 execl(["X", "\udc01C"], /(?<=(\udc01\w))X/u, | |
103 "\ud800\udc01AX\udc11BX\udc01\udc01CX"); | |
104 execl(["C", "\udc01\udc01"], /(?<=(\udc01.))./u, | |
105 "\ud800\udc01AX\udc11BX\udc01\udc01CX"); | |
106 | |
107 | |
108 var L = "\ud800"; | |
109 var T = "\udc00"; | |
110 var X = "X"; | |
111 | |
112 // Test string contains only match. | |
113 function testw(expect, src, subject) { | |
114 var re = new RegExp("^" + src + "$", "u"); | |
115 assertEquals(expect, re.test(subject)); | |
116 } | |
117 | |
118 // Test string starts with match. | |
119 function tests(expect, src, subject) { | |
120 var re = new RegExp("^" + src, "u"); | |
121 assertEquals(expect, re.test(subject)); | |
122 } | |
123 | |
124 testw(true, X, X); | |
125 testw(true, L, L); | |
126 testw(true, T, T); | |
127 testw(true, L + T, L + T); | |
128 testw(true, T + L, T + L); | |
129 testw(false, T, L + T); | |
130 testw(false, L, L + T); | |
131 testw(true, ".(?<=" + L + ")", L); | |
132 testw(true, ".(?<=" + T + ")", T); | |
133 testw(true, ".(?<=" + L + T + ")", L + T); | |
134 testw(true, ".(?<=" + L + T + ")", L + T); | |
135 tests(true, ".(?<=" + T + ")", T + L); | |
136 tests(false, ".(?<=" + L + ")", L + T); | |
137 tests(false, ".(?<=" + T + ")", L + T); | |
138 tests(true, "..(?<=" + T + ")", T + T + L); | |
139 tests(true, "..(?<=" + T + ")", X + T + L); | |
140 tests(true, "...(?<=" + L + ")", X + T + L); | |
141 tests(false, "...(?<=" + T + ")", X + L + T) | |
142 tests(true, "..(?<=" + L + T + ")", X + L + T) | |
143 tests(true, "..(?<=" + L + T + "(?<=" + L + T + "))", X + L + T); | |
144 tests(false, "..(?<=" + L + "(" + T + "))", X + L + T); | |
145 tests(false, ".*" + L, X + L + T); | |
146 tests(true, ".*" + L, X + L + L + T); | |
147 tests(false, ".*" + L, X + L + T + L + T); | |
148 tests(false, ".*" + T, X + L + T + L + T); | |
149 tests(true, ".*" + T, X + L + T + T + L + T); | |
OLD | NEW |