Line data Source code
1 : /* tst_idna2.c --- Self tests for idna_to_ascii_8z().
2 : * Copyright (C) 2002-2020 Simon Josefsson
3 : *
4 : * This file is part of GNU Libidn.
5 : *
6 : * This program is free software: you can redistribute it and/or modify
7 : * it under the terms of the GNU General Public License as published by
8 : * the Free Software Foundation, either version 3 of the License, or
9 : * (at your option) any later version.
10 : *
11 : * This program is distributed in the hope that it will be useful,
12 : * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 : * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 : * GNU General Public License for more details.
15 : *
16 : * You should have received a copy of the GNU General Public License
17 : * along with this program. If not, see <http://www.gnu.org/licenses/>.
18 : *
19 : */
20 :
21 : #ifdef HAVE_CONFIG_H
22 : # include "config.h"
23 : #endif
24 :
25 : #include <stdio.h>
26 : #include <stdlib.h>
27 : #include <stdarg.h>
28 : #include <string.h>
29 :
30 : #include <idna.h>
31 :
32 : #include "utils.h"
33 :
34 : struct idna
35 : {
36 : const char *in;
37 : const char *out;
38 : };
39 :
40 : static const struct idna idna[] = {
41 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xc3\xad\x64\x6e", "example.xn--dn-mja"
42 : /* 1-1-1 Has an IDN in just the TLD */
43 : },
44 : {"\xc3\xab\x78\x2e\xc3\xad\x64\x6e", "xn--x-ega.xn--dn-mja"
45 : /* 1-1-2 Has an IDN in the TLD and SLD */
46 : },
47 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xc3\xa5\xc3\xbe\xc3\xa7",
48 : "example.xn--5cae2e"
49 : /* 1-2-1 Latin-1 TLD */
50 : },
51 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xc4\x83\x62\xc4\x89",
52 : "example.xn--b-rhat"
53 : /* 1-2-2 Latin Extended A TLD */
54 : },
55 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xc8\xa7\xc6\x80\xc6\x88",
56 : "example.xn--lhaq98b"
57 : /* 1-2-3 Latin Extended B TLD */
58 : },
59 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\xb8\x81\xe1\xb8\x83\xe1\xb8\x89",
60 : "example.xn--2fges"
61 : /* 1-2-4 Latin Extended Additional TLD */
62 : },
63 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe4\xb8\xbf\xe4\xba\xba\xe5\xb0\xb8",
64 : "example.xn--xiqplj17a"
65 : /* 1-3-1 Han TLD */
66 : },
67 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe3\x81\x8b\xe3\x81\x8c\xe3\x81\x8d",
68 : "example.xn--u8jcd"
69 : /* 1-3-2 Hiragana TLD */
70 : },
71 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe3\x82\xab\xe3\x82\xac\xe3\x82\xad",
72 : "example.xn--lckcd"
73 : /* 1-3-3 Katakana TLD */
74 : },
75 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\x84\x80\xe1\x85\xa1\xe1\x86\xa8",
76 : "example.xn--p39a"
77 : /* 1-3-4 Hangul Jamo TLD */
78 : /* Don't resolve as example.xn--ypd8qrh */
79 : },
80 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xea\xb1\xa9\xeb\x93\x86\xec\x80\xba",
81 : "example.xn--o69aq2nl0j"
82 : /* 1-3-5 Hangul TLD */
83 : },
84 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xea\x80\x8a\xea\x80\xa0\xea\x8a\xb8",
85 : "example.xn--6l7arby7j"
86 : /* 1-3-6 Yi TLD */
87 : },
88 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xce\xb1\xce\xb2\xce\xb3",
89 : "example.xn--mxacd"
90 : /* 1-3-7 Greek TLD */
91 : },
92 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\xbc\x82\xe1\xbc\xa6\xe1\xbd\x95",
93 : "example.xn--fng7dpg"
94 : /* 1-3-8 Greek Extended TLD */
95 : },
96 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xd0\xb0\xd0\xb1\xd0\xb2",
97 : "example.xn--80acd"
98 : /* 1-3-9 Cyrillic TLD */
99 : },
100 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xd5\xa1\xd5\xa2\xd5\xa3",
101 : "example.xn--y9acd"
102 : /* 1-3-10 Armeian TLD */
103 : },
104 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\x83\x90\xe1\x83\x91\xe1\x83\x92",
105 : "example.xn--lodcd"
106 : /* 1-3-11 Georgian TLD */
107 : },
108 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe2\x88\xa1\xe2\x86\xba\xe2\x8a\x82",
109 : "example.xn--b7gxomk"
110 : /* 1-4-1 Symbols TLD */
111 : },
112 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xa4\x95\xe0\xa4\x96\xe0\xa4\x97",
113 : "example.xn--11bcd"
114 : /* 1-5-1 Devanagari TLD */
115 : },
116 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xa6\x95\xe0\xa6\x96\xe0\xa6\x97",
117 : "example.xn--p5bcd"
118 : /* 1-5-2 Bengali TLD */
119 : },
120 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xa8\x95\xe0\xa8\x96\xe0\xa8\x97",
121 : "example.xn--d9bcd"
122 : /* 1-5-3 Gurmukhi TLD */
123 : },
124 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xaa\x95\xe0\xaa\x96\xe0\xaa\x97",
125 : "example.xn--0dccd"
126 : /* 1-5-4 Gujarati TLD */
127 : },
128 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xac\x95\xe0\xac\x96\xe0\xac\x97",
129 : "example.xn--ohccd"
130 : /* 1-5-5 Oriya TLD */
131 : },
132 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xae\x95\xe0\xae\x99\xe0\xae\x9a",
133 : "example.xn--clcid"
134 : /* 1-5-6 Tamil TLD */
135 : },
136 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xb0\x95\xe0\xb0\x96\xe0\xb0\x97",
137 : "example.xn--zoccd"
138 : /* 1-5-7 Telugu TLD */
139 : },
140 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xb2\x95\xe0\xb2\x96\xe0\xb2\x97",
141 : "example.xn--nsccd"
142 : /* 1-5-8 Kannada TLD */
143 : },
144 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xb4\x95\xe0\xb4\x96\xe0\xb4\x97",
145 : "example.xn--bwccd"
146 : /* 1-5-9 Malayalam TLD */
147 : },
148 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xb6\x9a\xe0\xb6\x9b\xe0\xb6\x9c",
149 : "example.xn--3zccd"
150 : /* 1-5-10 Sinhala TLD */
151 : },
152 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xb8\x81\xe0\xb8\x82\xe0\xb8\x83",
153 : "example.xn--12ccd"
154 : /* 1-5-11 Thai TLD */
155 : },
156 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xba\x81\xe0\xba\x82\xe0\xba\x84",
157 : "example.xn--p6ccg"
158 : /* 1-5-12 Lao TLD */
159 : },
160 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe0\xbd\x80\xe0\xbd\x81\xe0\xbd\x82",
161 : "example.xn--5cdcd"
162 : /* 1-5-13 Tibetan TLD */
163 : },
164 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\x80\x80\xe1\x80\x81\xe1\x80\x82",
165 : "example.xn--nidcd"
166 : /* 1-5-14 Myanmar TLD */
167 : },
168 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\x9e\x80\xe1\x9e\x81\xe1\x9e\x82",
169 : "example.xn--i2ecd"
170 : /* 1-5-15 Khmer TLD */
171 : },
172 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xe1\xa0\xa0\xe1\xa0\xa1\xe1\xa0\xa2",
173 : "example.xn--26ecd"
174 : /* 1-5-16 Mongolian TLD */
175 : },
176 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xd8\xa7\xd8\xa8\xd8\xa9",
177 : "example.xn--mgbcd"
178 : /* 1-6-1 Arabic TLD */
179 : },
180 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xd7\x90\xd7\x91\xd7\x92",
181 : "example.xn--4dbcd"
182 : /* 1-6-2 Hebrew TLD */
183 : },
184 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xdc\x90\xdc\x91\xdc\x92",
185 : "example.xn--9mbcd"
186 : /* 1-6-3 Syriac TLD */
187 : },
188 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\x61\x62\x63\xe3\x82\xab\xe3\x82\xac\xe3\x82\xad",
189 : "example.xn--abc-mj4bfg"
190 : /* 1-7-1 ASCII and non-Latin TLD */
191 : },
192 : {"\x65\x78\x61\x6d\x70\x6c\x65\x2e\xc3\xa5\xc3\xbe\xc3\xa7\xe3\x82\xab\xe3\x82\xac\xe3\x82\xad",
193 : "example.xn--5cae2e328wfag"
194 : /* 1-7-2 Latin (non-ASCII) and non-Latin TLD */
195 : },
196 : {"\xc3\xad\x21\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
197 : /* 2-3-1-1 Includes ! before Nameprep */
198 : /* Don't resolve as xn--!dn-qma.example */
199 : },
200 : {"\xc3\xad\x24\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
201 : /* 2-3-1-2 Includes $ before Nameprep */
202 : /* Don't resolve as xn--$dn-qma.example */
203 : },
204 : {"\xc3\xad\x2b\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
205 : /* 2-3-1-3 Includes + before Nameprep */
206 : /* Don't resolve as xn--+dn-qma.example */
207 : },
208 : {"\x2d\xc3\xad\x31\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
209 : /* 2-3-2-1 Leading hyphen before Nameprep */
210 : /* Don't resolve as xn---1dn-vpa.example */
211 : },
212 : {"\xc3\xad\x31\x64\x6e\x2d\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
213 : /* 2-3-2-2 Trailing hyphen before Nameprep */
214 : /* Don't resolve as xn--1dn--upa.example */
215 : },
216 : {"\xc3\xad\xef\xbc\x8b\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
217 : /* 2-3-3-1 Gets a + after Nameprep */
218 : /* Don't resolve as xn--dn-mja0331x.example */
219 : },
220 : {"\xc3\xad\xe2\x81\xbc\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
221 : /* 2-3-3-2 Gets a = after Nameprep */
222 : /* Don't resolve as xn--dn-mja0343a.example */
223 : },
224 : {"\xef\xb9\xa3\xc3\xad\x32\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
225 : /* 2-3-4-1 Leading hyphen after Nameprep */
226 : /* Don't resolve as xn--2dn-qma32863a.example */
227 : /* Don't resolve as xn---2dn-vpa.example */
228 : },
229 : {"\xc3\xad\x32\x64\x6e\xef\xbc\x8d\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
230 : /* 2-3-4-2 Trailing hyphen after Nameprep */
231 : /* Don't resolve as xn--2dn-qma79363a.example */
232 : /* Don't resolve as xn--2dn--upa.example */
233 : },
234 : {"\xc2\xb9\x31\x2e\x65\x78\x61\x6d\x70\x6c\x65", "11.example"
235 : /* 2-4-1 All-ASCII check, Latin */
236 : },
237 : {"\xe2\x85\xa5\x76\x69\x2e\x65\x78\x61\x6d\x70\x6c\x65", "vivi.example"
238 : /* 2-4-2 All-ASCII check, symbol */
239 : },
240 : {"\xc3\x9f\x73\x73\x2e\x65\x78\x61\x6d\x70\x6c\x65", "ssss.example"
241 : /* 2-4-3 All-ASCII check, sharp S */
242 : },
243 : {"\x78\x6e\x2d\x2d\xc3\xaf\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
244 : /* 2-5-1 ACE prefix before Nameprep, body */
245 : /* Don't resolve as xn--xn--dn-sja.example */
246 : /* Don't resolve as xn--dn-sja.example */
247 : },
248 : {"\xe2\x85\xb9\x6e\x2d\x2d\xc3\xa4\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
249 : ""
250 : /* 2-5-2 ACE prefix before Nameprep, prefix */
251 : /* Don't resolve as xn--xn--dn-uia.example */
252 : /* Don't resolve as xn--dn-uia.example */
253 : },
254 : {"", ""
255 : /* 2-8-1 Zero-length label after Nameprep */
256 : /* Don't resolve as xn--kba.example */
257 : /* Don't resolve as xn--.example */
258 : },
259 : {"\x33\x30\x30\x32\x2d\x74\x65\x73\x74\xe3\x80\x82\xc3\xad\x64\x6e",
260 : "3002-test.xn--dn-mja"
261 : /* 2-9-1 U+3002 acts as a label separator */
262 : /* Don't resolve as xn--3002-testdn-wcb2087m.example */
263 : },
264 : {"\x66\x66\x30\x65\x2d\x74\x65\x73\x74\xef\xbc\x8e\xc3\xad\x64\x6e",
265 : "ff0e-test.xn--dn-mja"
266 : /* 2-9-2 U+FF0E acts as a label separator */
267 : /* Don't resolve as xn--ff0e-testdn-wcb45865f.example */
268 : },
269 : {"\x66\x66\x36\x31\x2d\x74\x65\x73\x74\xef\xbd\xa1\xc3\xad\x64\x6e",
270 : "ff61-test.xn--dn-mja"
271 : /* 2-9-3 U+FF61 acts as a label separator */
272 : /* Don't resolve as xn--ff61-testdn-wcb33975f.example */
273 : },
274 : {"\x30\x30\x61\x64\x6f\x75\x74\xc2\xad\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
275 : "xn--00adoutdn-m5a.example"
276 : /* 4-1-1-1 00adout<00AD><00ED>dn.example -> 00adout<00ED>dn.example */
277 : /* Don't resolve as xn--00adoutdn-cna81e.example */
278 : },
279 : {"\x32\x30\x30\x64\x6f\x75\x74\xe2\x80\x8d\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
280 : "xn--200doutdn-m5a.example"
281 : /* 4-1-1-2 200dout<200D><00ED>dn.example -> 200dout<00ED>dn.example */
282 : /* Don't resolve as xn--200doutdn-m5a1678f.example */
283 : },
284 : {"\x73\x69\x6d\x70\x6c\x65\x63\x61\x70\x44\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
285 : "xn--simplecapddn-1fb.example"
286 : /* 4-1-2-1 simplecap<0044><00ED>dn.example -> simplecap<0064><00ED>dn.example */
287 : },
288 : {"\x6c\x61\x74\x69\x6e\x74\x6f\x67\x72\x65\x65\x6b\xc2\xb5\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
289 : "xn--latintogreekdn-cmb716i.example"
290 : /* 4-1-2-2 latintogreek<00B5><00ED>dn.example -> latintogreek<03BC><00ED>dn.example */
291 : /* Don't resolve as xn--latintogreekdn-cxa01g.example */
292 : },
293 : {"\x6c\x61\x74\x69\x6e\x65\x78\x74\xc3\x87\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
294 : "xn--latinextdn-v6a6e.example"
295 : /* 4-1-2-3 latinext<00C7><00ED>dn.example -> latinext<00E7><00ED>dn.example */
296 : /* Don't resolve as xn--latinextdn-twa07b.example */
297 : },
298 : {"\x73\x68\x61\x72\x70\x73\xc3\x9f\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
299 : "xn--sharpsssdn-r8a.example"
300 : /* 4-1-2-4 sharps<00DF><00ED>dn.example -> sharpsss<00ED>dn.example */
301 : /* Don't resolve as xn--sharpsdn-vya4l.example */
302 : },
303 : {"\x74\x75\x72\x6b\x69\x73\x68\x69\xc4\xb0\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
304 : "xn--turkishiidn-wcb701e.example"
305 : /* 4-1-2-5 turkishi<0130><00ED>dn.example -> turkishi<0069><0307><00ED>dn.example */
306 : /* Don't resolve as xn--turkishidn-r8a71f.example */
307 : },
308 : {"\x65\x78\x70\x74\x77\x6f\xc5\x89\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
309 : "xn--exptwondn-m5a502c.example"
310 : /* 4-1-2-6 exptwo<0149><00ED>dn.example -> exptwo<02BC><006E><00ED>dn.example */
311 : /* Don't resolve as xn--exptwodn-h2a33g.example */
312 : },
313 : {"\x61\x64\x64\x66\x6f\x6c\x64\xcf\x92\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
314 : "xn--addfolddn-m5a121f.example"
315 : /* 4-1-2-7 addfold<03D2><00ED>dn.example -> addfold<03C5><00ED>dn.example */
316 : /* Don't resolve as xn--addfolddn-m5a462f.example */
317 : },
318 : {"\x65\x78\x70\x74\x68\x72\x65\x65\xe1\xbd\x92\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
319 : "xn--expthreedn-r8a5844g.example"
320 : /* 4-1-2-8 expthree<1F52><00ED>dn.example -> expthree<03C5><0313><0300><00ED>dn.example */
321 : },
322 : {"\x6e\x6f\x6e\x62\x6d\x70\xf0\x90\x90\x80\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
323 : "xn--nonbmpdn-h2a34747d.example"
324 : /* 4-1-2-9 nonbmp<10400><00ED>dn.example -> nonbmp<10428><00ED>dn.example */
325 : /* Don't resolve as xn--nonbmpdn-h2a37046d.example */
326 : },
327 : {"\x6e\x6f\x6e\x62\x6d\x70\x74\x6f\x61\x73\x63\x69\x69\xf0\x9d\x90\x80\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
328 : "xn--nonbmptoasciiadn-msb.example"
329 : /* 4-1-2-10 nonbmptoascii<1D400><00ED>dn.example -> nonbmptoasciia<00ED>dn.example */
330 : /* Don't resolve as xn--nonbmptoasciidn-hpb54112i.example */
331 : },
332 : {"\x72\x65\x67\x63\x6f\x6d\x62\x65\xcc\x81\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
333 : "xn--regcombdn-h4a8b.example"
334 : /* 4-2-1-1 regcomb<0065><0301><00ED>dn.example -> regcomb<00E9><00ED>dn.example */
335 : /* Don't resolve as xn--regcombedn-r8a794d.example */
336 : },
337 : {"\x63\x6f\x6d\x62\x61\x6e\x64\x63\x61\x73\x65\x45\xcc\x81\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
338 : "xn--combandcasedn-lhb4d.example"
339 : /* 4-2-1-2 combandcase<0045><0301><00ED>dn.example -> combandcase<00E9><00ED>dn.example */
340 : /* Don't resolve as xn--combandcaseedn-cmb526f.example */
341 : },
342 : {"\x61\x64\x6a\x63\x6f\x6d\x62\xc2\xba\xcc\x81\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
343 : "xn--adjcombdn-m5a9d.example"
344 : /* 4-2-1-3 adjcomb<00BA><0301><00ED>dn.example -> adjcomb<00F3><00ED>dn.example */
345 : /* Don't resolve as xn--adjcombdn-1qa57cp3r.example */
346 : },
347 : {"\x65\x78\x74\x63\x6f\x6d\x62\x6f\x63\xcc\x81\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
348 : "xn--extcombodn-r8a52a.example"
349 : /* 4-2-1-4 extcombo<0063><0301><00ED>dn.example -> extcombo<0107><00ED>dn.example */
350 : /* Don't resolve as xn--extcombocdn-wcb920e.example */
351 : },
352 : {"\x64\x6f\x75\x62\x6c\x65\x64\x69\x61\x63\x31\x75\xcc\x88\xcc\x81\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
353 : "xn--doublediac1dn-6ib836a.example"
354 : /* 4-2-1-5 doublediac1<0075><0308><0301><00ED>dn.example -> doublediac2<01D8><00ED>dn.example */
355 : /* Don't resolve as xn--doublediac1udn-cmb526fnd.example */
356 : },
357 : {"\x64\x6f\x75\x62\x6c\x65\x64\x69\x61\x63\x32\x75\xcc\x81\xcc\x88\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
358 : "xn--doublediac2dn-6ib8qs73a.example"
359 : /* 4-2-1-6 doublediac2<0075><0301><0308><00ED>dn.example -> doublediac2<01D8><00ED>dn.example */
360 : /* Don't resolve as xn--doublediac2udn-cmb526fod.example */
361 : },
362 : {"\x6e\x65\x77\x6e\x6f\x72\x6d\xf0\xaf\xa1\xb4\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
363 : "xn--newnormdn-m5a7856x.example"
364 : /* 4-2-2-1 newnorm<2F874><00ED>dn.example -> newnorm<5F33><00ED>dn.example should not become <5F53> */
365 : /* Don't resolve as xn--newnormdn-m5a9396x.example */
366 : /* Don't resolve as xn--newnormdn-m5a9968x.example */
367 : },
368 : {"\xe2\x80\x80\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
369 : /* 4-3-1 Spacing */
370 : /* Don't resolve as xn--dn-mja3392a.example */
371 : },
372 : {"\xdb\x9d\xc3\xad\x64\x6e\x2d\x32\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
373 : /* 4-3-2 Control */
374 : /* Don't resolve as xn--dn-2-upa332g.example */
375 : },
376 : {"\xee\x80\x85\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
377 : /* 4-3-3 Private use */
378 : /* Don't resolve as xn--dn-mja1659t.example */
379 : },
380 : {"\xf3\xb0\x80\x85\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
381 : /* 4-3-4 Private use, non-BMP */
382 : /* Don't resolve as xn--dn-mja7922x.example */
383 : },
384 : {"\xef\xb7\x9d\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
385 : /* 4-3-5 Non-character */
386 : /* Don't resolve as xn--dn-mja1210x.example */
387 : },
388 : {"\xf0\x9f\xbf\xbe\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
389 : /* 4-3-6 Non-character, non-BMP */
390 : /* Don't resolve as xn--dn-mja7922x.example */
391 : },
392 : {"\xef\xbf\xbd\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
393 : /* 4-3-7 Surrogate points */
394 : /* Don't resolve as xn--dn-mja7922x.example */
395 : },
396 : {"\xef\xbf\xba\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
397 : /* 4-3-8 Inappropriate for plain */
398 : /* Don't resolve as xn--dn-mja5822x.example */
399 : },
400 : {"\xe2\xbf\xb5\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
401 : /* 4-3-9 Inappropriate for canonical */
402 : /* Don't resolve as xn--dn-mja3729b.example */
403 : },
404 : {"\xe2\x81\xaa\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
405 : /* 4-3-10 Change display simple */
406 : /* Don't resolve as xn--dn-mja7533a.example */
407 : },
408 : {"\xe2\x80\x8f\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
409 : /* 4-3-11 Change display RTL */
410 : /* Don't resolve as xn--dn-mja3992a.example */
411 : },
412 : {"\xf3\xa0\x80\x81\xf3\xa0\x81\x85\xf3\xa0\x81\x8e\x68\x69\x69\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
413 : ""
414 : /* 4-3-12 Language tags */
415 : /* Don't resolve as xn--hiidn-km43aaa.example */
416 : },
417 : {"\xd8\xa8\x6f\xd8\xb8\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
418 : /* 4-4-1 Arabic RandALCat-LCat-RandALCat */
419 : /* Don't resolve as xn--o-0mc3c.example */
420 : },
421 : {"\xd8\xa8\xd8\xb8\x6f\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
422 : /* 4-4-2 Arabic RandALCat-RandALCat-other */
423 : /* Don't resolve as xn--o-0mc2c.example */
424 : },
425 : {"\x6f\xd8\xa8\xd8\xb8\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
426 : /* 4-4-3 Arabic other-RandALCat-RandALCat */
427 : /* Don't resolve as xn--o-1mc2c.example */
428 : },
429 : {"\xd7\x91\x6f\xd7\xa1\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
430 : /* 4-4-4 Hebrew RandALCat-LCat-RandALCat */
431 : /* Don't resolve as xn--o-1hc3c.example */
432 : },
433 : {"\xd7\x91\xd7\xa1\x6f\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
434 : /* 4-4-5 Hebrew RandALCat-RandALCat-other */
435 : /* Don't resolve as xn--o-1hc2c.example */
436 : },
437 : {"\x6f\xd7\x91\xd7\xa1\x2e\x65\x78\x61\x6d\x70\x6c\x65", ""
438 : /* 4-4-6 Hebrew other-RandALCat-RandALCat */
439 : /* Don't resolve as xn--o-2hc2c.example */
440 : },
441 : {"\xc8\xb7\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
442 : "xn--dn-mja33k.example"
443 : /* 5-1-1 Unassigned in BMP; zone editors should reject */
444 : },
445 : {"\xf0\x90\x88\x85\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
446 : "xn--dn-mja7734x.example"
447 : /* 5-1-2 Unassinged outside BMP; zone editors should reject */
448 : /* Don't resolve as xn--dn-mja7922x.example */
449 : },
450 : {"\xc8\xb4\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
451 : "xn--dn-mja12k.example"
452 : /* 5-2-1 Newly assigned in BMP; zone editors should reject */
453 : },
454 : {"\xf0\x90\x80\x85\xc3\xad\x64\x6e\x2e\x65\x78\x61\x6d\x70\x6c\x65",
455 : "xn--dn-mja9232x.example"
456 : /* 5-2-2 Newly assigned outside of BMP; zone editors should reject */
457 : /* Don't resolve as xn--dn-mja7922x.example */
458 : }
459 : };
460 :
461 : void
462 1 : doit (void)
463 : {
464 : unsigned i;
465 : char *out;
466 : int rc;
467 :
468 99 : for (i = 0; i < sizeof (idna) / sizeof (idna[0]); i++)
469 : {
470 98 : if (debug)
471 0 : printf ("IDNA2 entry %u\n", i);
472 :
473 98 : if (debug)
474 : {
475 : uint32_t *p;
476 : size_t len;
477 0 : printf ("in: %s\n", idna[i].in);
478 0 : hexprint (idna[i].in, strlen (idna[i].in));
479 0 : escapeprint (idna[i].in, strlen (idna[i].in));
480 0 : p = stringprep_utf8_to_ucs4 (idna[i].in, -1, &len);
481 0 : ucs4print (p, len);
482 0 : free (p);
483 : }
484 :
485 98 : rc = idna_to_ascii_8z (idna[i].in, &out,
486 : IDNA_ALLOW_UNASSIGNED |
487 : IDNA_USE_STD3_ASCII_RULES);
488 98 : if (rc != IDNA_SUCCESS && strlen (idna[i].out) > 0)
489 : {
490 0 : fail ("IDNA2 entry %u failed: %d\n", i, rc);
491 0 : continue;
492 : }
493 :
494 98 : if (debug && rc == IDNA_SUCCESS)
495 : {
496 0 : printf ("computed out: %s\n", out);
497 0 : printf ("expected out: %s\n", idna[i].out);
498 : }
499 98 : else if (debug)
500 0 : printf ("returned %d\n", rc);
501 :
502 98 : if (rc == IDNA_SUCCESS)
503 : {
504 69 : if (strlen (idna[i].out) != strlen (out) ||
505 69 : strcasecmp (idna[i].out, out) != 0)
506 : {
507 0 : fail ("IDNA2 entry %u failed\n", i);
508 0 : if (debug)
509 0 : printf ("ERROR\n");
510 : }
511 69 : else if (debug)
512 0 : printf ("OK\n");
513 :
514 69 : free (out);
515 69 : out = NULL;
516 : }
517 29 : else if (rc != IDNA_SUCCESS && strlen (idna[i].out) == 0 && debug)
518 0 : printf ("OK (fail)\n");
519 29 : else if (debug)
520 0 : printf ("OK\n");
521 : }
522 1 : }
|