whatpm/t/URIChecker.t

#!/usr/bin/perl
use strict;

use lib qw[/home/wakaba/work/manakai/lib];
## ISSUE: Message::URI::URIReference module.

use Test;
BEGIN { plan tests => 81 }

my $Cases = [
  {
    data => q<http://localhost/%40/>,
    errors => [],
  },
  {
    data => q<>,
    errors => [],
  },
  {
    data => q<abcdef>,
    errors => [],
  },
  {
    data => q<%7Fabc:efg>,
    errors => ['m::syntax error:iriref3987'],
  },
  {
    data => q<http://test/[53::0]>,
    errors => ['m::syntax error:iriref3987'],
  },
  {
    data => q<100%!>,
    errors => ['m::syntax error:iriref3987'],
  },
  {
    data => q<%a2>,
    errors => ['w:0.2:URL:lowercase hexadecimal digit'],
  },
  {
    data => q<%a2?>,
    errors => ['w:0.2:URL:lowercase hexadecimal digit'],
  },
  {
    data => q<?%a2>,
    errors => ['w:1.3:URL:lowercase hexadecimal digit'],
  },
  {
    data => q<%a2%1b>,
    errors => ['w:0.2:URL:lowercase hexadecimal digit',
               'w:3.5:URL:lowercase hexadecimal digit'],
  },
  {
    data => q<http://%5b/>,
    errors => ['w:7.9:URL:lowercase hexadecimal digit',
               'w::URL:non-DNS host'],
  },
  {
    data => q<http://%5b/%25a/bv/%cc>,
    errors => ['w:7.9:URL:lowercase hexadecimal digit',
               'w:19.21:URL:lowercase hexadecimal digit',
               'w::URL:non-DNS host'],
  },
  {
    data => q<%41>,
    errors => ['w:0.2:URL:percent-encoded unreserved'],
  },
  {
    data => q<%41%7E>,
    errors => ['w:0.2:URL:percent-encoded unreserved',
               'w:3.5:URL:percent-encoded unreserved'],
  },
  {
    data => q</%41>,
    errors => ['w:1.3:URL:percent-encoded unreserved'],
  },
  {
    data => q<http://%5a/>,
    errors => ['w:7.9:URL:lowercase hexadecimal digit',
               'w:7.9:URL:percent-encoded unreserved'],
  },
  {
    data => q<./%2E%2E>,
    errors => ['w:2.4:URL:percent-encoded unreserved',
               'w:5.7:URL:percent-encoded unreserved'],
  },
  {
    data => q<http://www.example.com/%7Euser/>,
    errors => ['w:23.25:URL:percent-encoded unreserved'],
  },
  {
    data => q<HTTP://example/>,
    errors => ['w::URL:uppercase scheme name'],
  },
  {
    data => q<Http://example/>,
    errors => ['w::URL:uppercase scheme name'],
  },
  {
    data => q<datA:,>,
    errors => ['w::URL:uppercase scheme name'],
  },
  {
    data => q<dat%41:,>,
    errors => ['m::syntax error:iriref3987',
               'w::URL:uppercase scheme name',
               'w:3.5:URL:percent-encoded unreserved'],
  },
  {
    data => q<g%5A:,>,
    errors => ['m::syntax error:iriref3987',
               'w::URL:uppercase scheme name',
               'w:1.3:URL:percent-encoded unreserved'],
  },
  {
    data => q<g%7A:,>,
    errors => ['m::syntax error:iriref3987',
               'w:1.3:URL:percent-encoded unreserved'],
  },
  {
    data => q<http://www.test:2222/>,
    errors => [],
  },
  {
    data => q<http://www.example:0/>,
    errors => [],
  },
  {
    data => q<http://www@example:/>,
    errors => ['w::URL:empty port'],
  },
  {
    data => q<http://www.test:/>,
    errors => ['w::URL:empty port'],
  },
  {
    data => q<http://user:password@example/>,
    errors => ['w::URL:password'],
  },
  {
    data => q<http://EXAMPLE/>,
    errors => ['w::URL:uppercase host'],
  },
  {
    data => q<http://USER@example/>,
    errors => [],
  },
  {
    data => q<http://[v0.aaa]/>,
    errors => ['w:1.2:URL:address format:v0'],
  },
  {
    data => q<http://user@[v0.aaa]/>,
    errors => ['w:1.2:URL:address format:v0'],
  },
  {
    data => q<http://user@[V0A.aaa]/>,
    errors => ['w:1.3:URL:address format:V0A',
               'w::URL:uppercase host'],
  },
  {
    data => q<http://127.0.0.1/>,
    errors => [],
  },
  {
    data => q<http://123456789012345678901234567890123456789012345678901234567890123.test/>,
    errors => [],
  },
  {
    data => q<http://1234567890123456789012345678901234567890123456789012345678901234.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123/>,
    errors => [],
  },
  {
    data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123./>,
    errors => ['w:256.256:URL:long host'],
  },
  {
    data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.456789/>,
    errors => ['w:256.262:URL:long host'],
  },
  {
    data => q<http://a_b.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => q<http://%61.test/>,
    errors => ['w:7.9:URL:percent-encoded unreserved'],
  },
  {
    data => q<http://a.test/>,
    errors => [],
  },
  {
    data => q<http://1.test/>,
    errors => [],
  },
  {
    data => q<http://a-1.test/>,
    errors => [],
  },
  {
    data => q<http://1-a.test/>,
    errors => [],
  },
  {
    data => q<http://a-.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => q<http://-a.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => q<http://a.b.test/>,
    errors => [],
  },
  {
    data => q<http://a.bc.test/>,
    errors => [],
  },
  {
    data => q<http://a.b-c.test/>,
    errors => [],
  },
  {
    data => qq<http://\x{123}\x{456}.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => qq<http://\x{4E00}%80.test/>,
    errors => ['w::URL:non-DNS host',
               'm::URL:non UTF-8 host'],
  },
  {
    data => q<http://a.%E3%81%82%E3%81%84.test/>,
    errors => ['w::URL:non-DNS host'],
  },
  {
    data => q<example://a/b/c/%7Bfoo%7D>,
    errors => [],
  },
  {
    data => q<eXAMPLE://a/./b/../b/%63/%7bfoo%7d>,
    errors => ['w::URL:uppercase scheme name',
               'w:21.23:URL:percent-encoded unreserved',
               'w:25.27:URL:lowercase hexadecimal digit',
               'w:31.33:URL:lowercase hexadecimal digit',
               'w::URL:dot-segment'],
  },
  {
    data => q<example://a/.htaccess>,
    errors => [],
  },
  {
    data => q<example://a/.>,
    errors => ['w::URL:dot-segment'],
  },
  {
    data => q<http://example.com>,
    errors => ['w::URL:empty path'],
  },
  {
    data => q<http://example.com/>,
    errors => [],
  },
  {
    data => q<http://example.com:/>,
    errors => ['w::URL:empty port'],
  },
  {
    data => q<http://example.com:80/>,
    errors => ['w::URL:default port'],
  },
  {
    data => q<hTTP://example.com:80/>,
    errors => ['w::URL:uppercase scheme name',
               'w::URL:default port'],
  },
  {
    data => q<%68ttp://example.com:80/>,
    errors => ['m::syntax error:iriref3987',
               'w:0.2:URL:percent-encoded unreserved',
               'w::URL:default port'],
  },
  {
    data => q<file://user@/>,
    errors => ['w::URL:empty host'],
  },
  {
    data => q<http://example.com/?>,
    errors => [],
  },
  {
    data => q<mailto:Joe@Example.COM>,
    errors => [],
  },
  {
    data => q<mailto:Joe@example.com>,
    errors => [],
  },
  {
    data => q<http://example.com/data>,
    errors => [],
  },
  {
    data => q<ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm>,
    errors => [],
  },
  {
    data => qq<http://r\xE9sum\xE9.example.org>,
    errors => ['w::URL:non-DNS host',
               'w::URL:empty path'],
  },
  {
    data => qq<http://validator.w3.org/check?uri=http%3A%2F%2Fr\xE9;sum\xE9.example.com>,
    errors => [],
  },
  {
    data => q<http://validator.w3.org/check?uri=http%3A%2F%2Fr%C3%A9sum%C3%A9.example.com>,
    errors => [],
  },
  {
    data => qq<http://example.com/\x{10300}\x{10301}\x{10302}>,
    errors => [],
  },
  {
    data => q<http://example.com/%F0%90%8C%80%F0%90%8C%81%F0%90%8C%82>,
    errors => [],
  },
  {
    data => q<http://www.example.org/r%E9sum%E9.html>,
    errors => [],
  },
  {
    data => q<http://xn--99zt52a.example.org/%e2%80%ae>,
    errors => ['w:31.33:URL:lowercase hexadecimal digit',
               'w:37.39:URL:lowercase hexadecimal digit'],
  },
  {
    data => qq<example://a/b/c/%7Bfoo%7D/ros\xE9>,
    errors => [],
  },
  {
    data => qq<eXAMPLE://a/./b/../b/%63/%7bfoo%7d/ros%C3%A9>,
    errors => ['w::URL:uppercase scheme name',
               'w:21.23:URL:percent-encoded unreserved',
               'w:25.27:URL:lowercase hexadecimal digit',
               'w:31.33:URL:lowercase hexadecimal digit',
               'w::URL:dot-segment'],
  },
  {
    data => qq<http://www.example.org/..abc/>,
    errors => [],
  },
  {
    data => qq<http://www.example.org/../abc/>,
    errors => ['w::URL:dot-segment'],
  },
  {
    data => qq<http://www.example.org/r\xE9sum\xE9.html>,
    errors => [],
  },
  {
    data => qq<http://www.example.org/re\x{301}sume\x{301}.html>,
    errors => [], ## TODO: not in NFC
  },
  {
    data => q<http://www.example.org/r%E9sum%E9.xml#r%C3%A9sum%C3%A9>,
    errors => [],
  },
];

require Whatpm::URIChecker;

for my $test (@$Cases) {
  @{$test->{errors}} = sort {$a cmp $b} @{$test->{errors}};
  my @errors;
  Whatpm::URIChecker->check_iri_reference ($test->{data}, sub {
    my %opt = @_;
    push @errors, $opt{level} . ':' .
        (defined $opt{pos_start} ? $opt{pos_start} . '.' . $opt{pos_end} : '') . ':' . 
        $opt{type} .
        (defined $opt{text} ? ':' . $opt{text} : '');
  });
  @errors = sort {$a cmp $b} @errors;

  ok join ("\n", @errors), join ("\n", @{$test->{errors}}), $test->{data};
}
1	#!/usr/bin/perl
2	use strict;
3
4	use lib qw[/home/wakaba/work/manakai/lib];
5	## ISSUE: Message::URI::URIReference module.
6
7	use Test;
8	BEGIN { plan tests => 81 }
9
10	my $Cases = [
11	{
12	data => q<http://localhost/%40/>,
13	errors => [],
14	},
15	{
16	data => q<>,
17	errors => [],
18	},
19	{
20	data => q<abcdef>,
21	errors => [],
22	},
23	{
24	data => q<%7Fabc:efg>,
25	errors => ['m::syntax error:iriref3987'],
26	},
27	{
28	data => q<http://test/[53::0]>,
29	errors => ['m::syntax error:iriref3987'],
30	},
31	{
32	data => q<100%!>,
33	errors => ['m::syntax error:iriref3987'],
34	},
35	{
36	data => q<%a2>,
37	errors => ['w:0.2:URL:lowercase hexadecimal digit'],
38	},
39	{
40	data => q<%a2?>,
41	errors => ['w:0.2:URL:lowercase hexadecimal digit'],
42	},
43	{
44	data => q<?%a2>,
45	errors => ['w:1.3:URL:lowercase hexadecimal digit'],
46	},
47	{
48	data => q<%a2%1b>,
49	errors => ['w:0.2:URL:lowercase hexadecimal digit',
50	'w:3.5:URL:lowercase hexadecimal digit'],
51	},
52	{
53	data => q<http://%5b/>,
54	errors => ['w:7.9:URL:lowercase hexadecimal digit',
55	'w::URL:non-DNS host'],
56	},
57	{
58	data => q<http://%5b/%25a/bv/%cc>,
59	errors => ['w:7.9:URL:lowercase hexadecimal digit',
60	'w:19.21:URL:lowercase hexadecimal digit',
61	'w::URL:non-DNS host'],
62	},
63	{
64	data => q<%41>,
65	errors => ['w:0.2:URL:percent-encoded unreserved'],
66	},
67	{
68	data => q<%41%7E>,
69	errors => ['w:0.2:URL:percent-encoded unreserved',
70	'w:3.5:URL:percent-encoded unreserved'],
71	},
72	{
73	data => q</%41>,
74	errors => ['w:1.3:URL:percent-encoded unreserved'],
75	},
76	{
77	data => q<http://%5a/>,
78	errors => ['w:7.9:URL:lowercase hexadecimal digit',
79	'w:7.9:URL:percent-encoded unreserved'],
80	},
81	{
82	data => q<./%2E%2E>,
83	errors => ['w:2.4:URL:percent-encoded unreserved',
84	'w:5.7:URL:percent-encoded unreserved'],
85	},
86	{
87	data => q<http://www.example.com/%7Euser/>,
88	errors => ['w:23.25:URL:percent-encoded unreserved'],
89	},
90	{
91	data => q<HTTP://example/>,
92	errors => ['w::URL:uppercase scheme name'],
93	},
94	{
95	data => q<Http://example/>,
96	errors => ['w::URL:uppercase scheme name'],
97	},
98	{
99	data => q<datA:,>,
100	errors => ['w::URL:uppercase scheme name'],
101	},
102	{
103	data => q<dat%41:,>,
104	errors => ['m::syntax error:iriref3987',
105	'w::URL:uppercase scheme name',
106	'w:3.5:URL:percent-encoded unreserved'],
107	},
108	{
109	data => q<g%5A:,>,
110	errors => ['m::syntax error:iriref3987',
111	'w::URL:uppercase scheme name',
112	'w:1.3:URL:percent-encoded unreserved'],
113	},
114	{
115	data => q<g%7A:,>,
116	errors => ['m::syntax error:iriref3987',
117	'w:1.3:URL:percent-encoded unreserved'],
118	},
119	{
120	data => q<http://www.test:2222/>,
121	errors => [],
122	},
123	{
124	data => q<http://www.example:0/>,
125	errors => [],
126	},
127	{
128	data => q<http://www@example:/>,
129	errors => ['w::URL:empty port'],
130	},
131	{
132	data => q<http://www.test:/>,
133	errors => ['w::URL:empty port'],
134	},
135	{
136	data => q<http://user:password@example/>,
137	errors => ['w::URL:password'],
138	},
139	{
140	data => q<http://EXAMPLE/>,
141	errors => ['w::URL:uppercase host'],
142	},
143	{
144	data => q<http://USER@example/>,
145	errors => [],
146	},
147	{
148	data => q<http://[v0.aaa]/>,
149	errors => ['w:1.2:URL:address format:v0'],
150	},
151	{
152	data => q<http://user@[v0.aaa]/>,
153	errors => ['w:1.2:URL:address format:v0'],
154	},
155	{
156	data => q<http://user@[V0A.aaa]/>,
157	errors => ['w:1.3:URL:address format:V0A',
158	'w::URL:uppercase host'],
159	},
160	{
161	data => q<http://127.0.0.1/>,
162	errors => [],
163	},
164	{
165	data => q<http://123456789012345678901234567890123456789012345678901234567890123.test/>,
166	errors => [],
167	},
168	{
169	data => q<http://1234567890123456789012345678901234567890123456789012345678901234.test/>,
170	errors => ['w::URL:non-DNS host'],
171	},
172	{
173	data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123/>,
174	errors => [],
175	},
176	{
177	data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123./>,
178	errors => ['w:256.256:URL:long host'],
179	},
180	{
181	data => q<http://123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.123456789012345678901234567890123456789012345678901234567890123.456789/>,
182	errors => ['w:256.262:URL:long host'],
183	},
184	{
185	data => q<http://a_b.test/>,
186	errors => ['w::URL:non-DNS host'],
187	},
188	{
189	data => q<http://%61.test/>,
190	errors => ['w:7.9:URL:percent-encoded unreserved'],
191	},
192	{
193	data => q<http://a.test/>,
194	errors => [],
195	},
196	{
197	data => q<http://1.test/>,
198	errors => [],
199	},
200	{
201	data => q<http://a-1.test/>,
202	errors => [],
203	},
204	{
205	data => q<http://1-a.test/>,
206	errors => [],
207	},
208	{
209	data => q<http://a-.test/>,
210	errors => ['w::URL:non-DNS host'],
211	},
212	{
213	data => q<http://-a.test/>,
214	errors => ['w::URL:non-DNS host'],
215	},
216	{
217	data => q<http://a.b.test/>,
218	errors => [],
219	},
220	{
221	data => q<http://a.bc.test/>,
222	errors => [],
223	},
224	{
225	data => q<http://a.b-c.test/>,
226	errors => [],
227	},
228	{
229	data => qq<http://\x{123}\x{456}.test/>,
230	errors => ['w::URL:non-DNS host'],
231	},
232	{
233	data => qq<http://\x{4E00}%80.test/>,
234	errors => ['w::URL:non-DNS host',
235	'm::URL:non UTF-8 host'],
236	},
237	{
238	data => q<http://a.%E3%81%82%E3%81%84.test/>,
239	errors => ['w::URL:non-DNS host'],
240	},
241	{
242	data => q<example://a/b/c/%7Bfoo%7D>,
243	errors => [],
244	},
245	{
246	data => q<eXAMPLE://a/./b/../b/%63/%7bfoo%7d>,
247	errors => ['w::URL:uppercase scheme name',
248	'w:21.23:URL:percent-encoded unreserved',
249	'w:25.27:URL:lowercase hexadecimal digit',
250	'w:31.33:URL:lowercase hexadecimal digit',
251	'w::URL:dot-segment'],
252	},
253	{
254	data => q<example://a/.htaccess>,
255	errors => [],
256	},
257	{
258	data => q<example://a/.>,
259	errors => ['w::URL:dot-segment'],
260	},
261	{
262	data => q<http://example.com>,
263	errors => ['w::URL:empty path'],
264	},
265	{
266	data => q<http://example.com/>,
267	errors => [],
268	},
269	{
270	data => q<http://example.com:/>,
271	errors => ['w::URL:empty port'],
272	},
273	{
274	data => q<http://example.com:80/>,
275	errors => ['w::URL:default port'],
276	},
277	{
278	data => q<hTTP://example.com:80/>,
279	errors => ['w::URL:uppercase scheme name',
280	'w::URL:default port'],
281	},
282	{
283	data => q<%68ttp://example.com:80/>,
284	errors => ['m::syntax error:iriref3987',
285	'w:0.2:URL:percent-encoded unreserved',
286	'w::URL:default port'],
287	},
288	{
289	data => q<file://user@/>,
290	errors => ['w::URL:empty host'],
291	},
292	{
293	data => q<http://example.com/?>,
294	errors => [],
295	},
296	{
297	data => q<mailto:Joe@Example.COM>,
298	errors => [],
299	},
300	{
301	data => q<mailto:Joe@example.com>,
302	errors => [],
303	},
304	{
305	data => q<http://example.com/data>,
306	errors => [],
307	},
308	{
309	data => q<ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm>,
310	errors => [],
311	},
312	{
313	data => qq<http://r\xE9sum\xE9.example.org>,
314	errors => ['w::URL:non-DNS host',
315	'w::URL:empty path'],
316	},
317	{
318	data => qq<http://validator.w3.org/check?uri=http%3A%2F%2Fr\xE9;sum\xE9.example.com>,
319	errors => [],
320	},
321	{
322	data => q<http://validator.w3.org/check?uri=http%3A%2F%2Fr%C3%A9sum%C3%A9.example.com>,
323	errors => [],
324	},
325	{
326	data => qq<http://example.com/\x{10300}\x{10301}\x{10302}>,
327	errors => [],
328	},
329	{
330	data => q<http://example.com/%F0%90%8C%80%F0%90%8C%81%F0%90%8C%82>,
331	errors => [],
332	},
333	{
334	data => q<http://www.example.org/r%E9sum%E9.html>,
335	errors => [],
336	},
337	{
338	data => q<http://xn--99zt52a.example.org/%e2%80%ae>,
339	errors => ['w:31.33:URL:lowercase hexadecimal digit',
340	'w:37.39:URL:lowercase hexadecimal digit'],
341	},
342	{
343	data => qq<example://a/b/c/%7Bfoo%7D/ros\xE9>,
344	errors => [],
345	},
346	{
347	data => qq<eXAMPLE://a/./b/../b/%63/%7bfoo%7d/ros%C3%A9>,
348	errors => ['w::URL:uppercase scheme name',
349	'w:21.23:URL:percent-encoded unreserved',
350	'w:25.27:URL:lowercase hexadecimal digit',
351	'w:31.33:URL:lowercase hexadecimal digit',
352	'w::URL:dot-segment'],
353	},
354	{
355	data => qq<http://www.example.org/..abc/>,
356	errors => [],
357	},
358	{
359	data => qq<http://www.example.org/../abc/>,
360	errors => ['w::URL:dot-segment'],
361	},
362	{
363	data => qq<http://www.example.org/r\xE9sum\xE9.html>,
364	errors => [],
365	},
366	{
367	data => qq<http://www.example.org/re\x{301}sume\x{301}.html>,
368	errors => [], ## TODO: not in NFC
369	},
370	{
371	data => q<http://www.example.org/r%E9sum%E9.xml#r%C3%A9sum%C3%A9>,
372	errors => [],
373	},
374	];
375
376	require Whatpm::URIChecker;
377
378	for my $test (@$Cases) {
379	@{$test->{errors}} = sort {$a cmp $b} @{$test->{errors}};
380	my @errors;
381	Whatpm::URIChecker->check_iri_reference ($test->{data}, sub {
382	my %opt = @_;
383	push @errors, $opt{level} . ':' .
384	(defined $opt{pos_start} ? $opt{pos_start} . '.' . $opt{pos_end} : '') . ':' .
385	$opt{type} .
386	(defined $opt{text} ? ':' . $opt{text} : '');
387	});
388	@errors = sort {$a cmp $b} @errors;
389
390	ok join ("\n", @errors), join ("\n", @{$test->{errors}}), $test->{data};
391	}