1// Copyright 2016 The Go Authors. All rights reserved. 2// Use of this source code is governed by a BSD-style 3// license that can be found in the LICENSE file. 4 5// +build go1.10 6 7package idna 8 9import "testing" 10 11// TestLabelErrors tests strings returned in case of error. All results should 12// be identical to the reference implementation and can be verified at 13// https://unicode.org/cldr/utility/idna.jsp. The reference implementation, 14// however, seems to not display Bidi and ContextJ errors. 15// 16// In some cases the behavior of browsers is added as a comment. In all cases, 17// whenever a resolve search returns an error here, Chrome will treat the input 18// string as a search string (including those for Bidi and Context J errors), 19// unless noted otherwise. 20func TestLabelErrors(t *testing.T) { 21 encode := func(s string) string { s, _ = encode(acePrefix, s); return s } 22 type kind struct { 23 name string 24 f func(string) (string, error) 25 } 26 punyA := kind{"PunycodeA", punycode.ToASCII} 27 resolve := kind{"ResolveA", Lookup.ToASCII} 28 display := kind{"ToUnicode", Display.ToUnicode} 29 p := New(VerifyDNSLength(true), MapForLookup(), BidiRule()) 30 lengthU := kind{"CheckLengthU", p.ToUnicode} 31 lengthA := kind{"CheckLengthA", p.ToASCII} 32 p = New(MapForLookup(), StrictDomainName(false)) 33 std3 := kind{"STD3", p.ToASCII} 34 p = New(MapForLookup(), CheckHyphens(false)) 35 hyphens := kind{"CheckHyphens", p.ToASCII} 36 37 testCases := []struct { 38 kind 39 input string 40 want string 41 wantErr string 42 }{ 43 {lengthU, "", "", "A4"}, // From UTS 46 conformance test. 44 {lengthA, "", "", "A4"}, 45 46 {lengthU, "xn--", "", "A4"}, 47 {lengthU, "foo.xn--", "foo.", "A4"}, // TODO: is dropping xn-- correct? 48 {lengthU, "xn--.foo", ".foo", "A4"}, 49 {lengthU, "foo.xn--.bar", "foo..bar", "A4"}, 50 51 {display, "xn--", "", ""}, 52 {display, "foo.xn--", "foo.", ""}, // TODO: is dropping xn-- correct? 53 {display, "xn--.foo", ".foo", ""}, 54 {display, "foo.xn--.bar", "foo..bar", ""}, 55 56 {lengthA, "a..b", "a..b", "A4"}, 57 {punyA, ".b", ".b", ""}, 58 // For backwards compatibility, the Punycode profile does not map runes. 59 {punyA, "\u3002b", "xn--b-83t", ""}, 60 {punyA, "..b", "..b", ""}, 61 62 {lengthA, ".b", ".b", "A4"}, 63 {lengthA, "\u3002b", ".b", "A4"}, 64 {lengthA, "..b", "..b", "A4"}, 65 {lengthA, "b..", "b..", ""}, 66 67 // Sharpened Bidi rules for Unicode 10.0.0. Apply for ALL labels in ANY 68 // of the labels is RTL. 69 {lengthA, "\ufe05\u3002\u3002\U0002603e\u1ce0", "..xn--t6f5138v", "A4"}, 70 {lengthA, "FAX\u2a77\U0001d186\u3002\U0001e942\U000e0181\u180c", "", "B6"}, 71 72 {resolve, "a..b", "a..b", ""}, 73 // Note that leading dots are not stripped. This is to be consistent 74 // with the Punycode profile as well as the conformance test. 75 {resolve, ".b", ".b", ""}, 76 {resolve, "\u3002b", ".b", ""}, 77 {resolve, "..b", "..b", ""}, 78 {resolve, "b..", "b..", ""}, 79 {resolve, "\xed", "", "P1"}, 80 81 // Raw punycode 82 {punyA, "", "", ""}, 83 {punyA, "*.foo.com", "*.foo.com", ""}, 84 {punyA, "Foo.com", "Foo.com", ""}, 85 86 // STD3 rules 87 {display, "*.foo.com", "*.foo.com", "P1"}, 88 {std3, "*.foo.com", "*.foo.com", ""}, 89 90 // Hyphens 91 {display, "r3---sn-apo3qvuoxuxbt-j5pe.googlevideo.com", "r3---sn-apo3qvuoxuxbt-j5pe.googlevideo.com", "V2"}, 92 {hyphens, "r3---sn-apo3qvuoxuxbt-j5pe.googlevideo.com", "r3---sn-apo3qvuoxuxbt-j5pe.googlevideo.com", ""}, 93 {display, "-label-.com", "-label-.com", "V3"}, 94 {hyphens, "-label-.com", "-label-.com", ""}, 95 96 // Don't map U+2490 (DIGIT NINE FULL STOP). This is the behavior of 97 // Chrome, Safari, and IE. Firefox will first map ⒐ to 9. and return 98 // lab9.be. 99 {resolve, "lab⒐be", "xn--labbe-zh9b", "P1"}, // encode("lab⒐be") 100 {display, "lab⒐be", "lab⒐be", "P1"}, 101 102 {resolve, "plan⒐faß.de", "xn--planfass-c31e.de", "P1"}, // encode("plan⒐fass") + ".de" 103 {display, "Plan⒐faß.de", "plan⒐faß.de", "P1"}, 104 105 // Chrome 54.0 recognizes the error and treats this input verbatim as a 106 // search string. 107 // Safari 10.0 (non-conform spec) decomposes "⒈" and computes the 108 // punycode on the result using transitional mapping. 109 // Firefox 49.0.1 goes haywire on this string and prints a bunch of what 110 // seems to be nested punycode encodings. 111 {resolve, "日本⒈co.ßßß.de", "xn--co-wuw5954azlb.ssssss.de", "P1"}, 112 {display, "日本⒈co.ßßß.de", "日本⒈co.ßßß.de", "P1"}, 113 114 {resolve, "a\u200Cb", "ab", ""}, 115 {display, "a\u200Cb", "a\u200Cb", "C"}, 116 117 {resolve, encode("a\u200Cb"), encode("a\u200Cb"), "C"}, 118 {display, "a\u200Cb", "a\u200Cb", "C"}, 119 120 {resolve, "grﻋﺮﺑﻲ.de", "xn--gr-gtd9a1b0g.de", "B"}, 121 { 122 // Notice how the string gets transformed, even with an error. 123 // Chrome will use the original string if it finds an error, so not 124 // the transformed one. 125 display, 126 "gr\ufecb\ufeae\ufe91\ufef2.de", 127 "gr\u0639\u0631\u0628\u064a.de", 128 "B", 129 }, 130 131 {resolve, "\u0671.\u03c3\u07dc", "xn--qib.xn--4xa21s", "B"}, // ٱ.σߜ 132 {display, "\u0671.\u03c3\u07dc", "\u0671.\u03c3\u07dc", "B"}, 133 134 // normalize input 135 {resolve, "a\u0323\u0322", "xn--jta191l", ""}, // ạ̢ 136 {display, "a\u0323\u0322", "\u1ea1\u0322", ""}, 137 138 // Non-normalized strings are not normalized when they originate from 139 // punycode. Despite the error, Chrome, Safari and Firefox will attempt 140 // to look up the input punycode. 141 {resolve, encode("a\u0323\u0322") + ".com", "xn--a-tdbc.com", "V1"}, 142 {display, encode("a\u0323\u0322") + ".com", "a\u0323\u0322.com", "V1"}, 143 } 144 145 for _, tc := range testCases { 146 doTest(t, tc.f, tc.name, tc.input, tc.want, tc.wantErr) 147 } 148} 149