Blame


1 33d32d1f 2020-12-25 op /*
2 33d32d1f 2020-12-25 op * Copyright (c) 2020 Omar Polo <op@omarpolo.com>
3 33d32d1f 2020-12-25 op *
4 33d32d1f 2020-12-25 op * Permission to use, copy, modify, and distribute this software for any
5 33d32d1f 2020-12-25 op * purpose with or without fee is hereby granted, provided that the above
6 33d32d1f 2020-12-25 op * copyright notice and this permission notice appear in all copies.
7 33d32d1f 2020-12-25 op *
8 33d32d1f 2020-12-25 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 33d32d1f 2020-12-25 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 33d32d1f 2020-12-25 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 33d32d1f 2020-12-25 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 33d32d1f 2020-12-25 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 33d32d1f 2020-12-25 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 33d32d1f 2020-12-25 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 33d32d1f 2020-12-25 op */
16 33d32d1f 2020-12-25 op
17 33d32d1f 2020-12-25 op #include <err.h>
18 33d32d1f 2020-12-25 op #include <stdio.h>
19 33d32d1f 2020-12-25 op #include <string.h>
20 33d32d1f 2020-12-25 op
21 33d32d1f 2020-12-25 op #include "gmid.h"
22 33d32d1f 2020-12-25 op
23 33d32d1f 2020-12-25 op #define TEST(uri, fail, exp, descr) \
24 33d32d1f 2020-12-25 op if (!run_test(uri, fail, exp)) { \
25 33d32d1f 2020-12-25 op fprintf(stderr, "%s:%d: error: %s\n", \
26 33d32d1f 2020-12-25 op __FILE__, __LINE__, descr); \
27 33d32d1f 2020-12-25 op exit(1); \
28 33d32d1f 2020-12-25 op }
29 33d32d1f 2020-12-25 op
30 33d32d1f 2020-12-25 op #define URI(schema, host, port, path, query, frag) \
31 33d32d1f 2020-12-25 op ((struct uri){schema, host, port, 0, path, query, frag})
32 33d32d1f 2020-12-25 op
33 33d32d1f 2020-12-25 op #define DIFF(wanted, got, field) \
34 33d32d1f 2020-12-25 op if (wanted->field == NULL || got->field == NULL || \
35 33d32d1f 2020-12-25 op strcmp(wanted->field, got->field)) { \
36 33d32d1f 2020-12-25 op fprintf(stderr, #field ":\n\tgot: %s\n\twanted: %s\n", \
37 33d32d1f 2020-12-25 op got->field, wanted->field); \
38 33d32d1f 2020-12-25 op return 0; \
39 33d32d1f 2020-12-25 op }
40 33d32d1f 2020-12-25 op
41 33d32d1f 2020-12-25 op #define PASS 0
42 33d32d1f 2020-12-25 op #define FAIL 1
43 33d32d1f 2020-12-25 op
44 33d32d1f 2020-12-25 op int
45 33d32d1f 2020-12-25 op diff_uri(struct uri *p, struct uri *exp)
46 33d32d1f 2020-12-25 op {
47 33d32d1f 2020-12-25 op DIFF(p, exp, schema);
48 33d32d1f 2020-12-25 op DIFF(p, exp, host);
49 33d32d1f 2020-12-25 op DIFF(p, exp, port);
50 33d32d1f 2020-12-25 op DIFF(p, exp, path);
51 33d32d1f 2020-12-25 op DIFF(p, exp, query);
52 33d32d1f 2020-12-25 op DIFF(p, exp, fragment);
53 33d32d1f 2020-12-25 op return 1;
54 33d32d1f 2020-12-25 op }
55 33d32d1f 2020-12-25 op
56 33d32d1f 2020-12-25 op int
57 33d32d1f 2020-12-25 op run_test(const char *uri, int should_fail, struct uri expected)
58 33d32d1f 2020-12-25 op {
59 33d32d1f 2020-12-25 op int failed, ok = 1;
60 33d32d1f 2020-12-25 op char *uri_copy;
61 33d32d1f 2020-12-25 op struct uri parsed;
62 33d32d1f 2020-12-25 op const char *error;
63 33d32d1f 2020-12-25 op
64 33d32d1f 2020-12-25 op if ((uri_copy = strdup(uri)) == NULL)
65 33d32d1f 2020-12-25 op err(1, "strdup");
66 33d32d1f 2020-12-25 op
67 33d32d1f 2020-12-25 op fprintf(stderr, "=> %s\n", uri);
68 33d32d1f 2020-12-25 op failed = !parse_uri(uri_copy, &parsed, &error);
69 33d32d1f 2020-12-25 op
70 33d32d1f 2020-12-25 op if (failed && should_fail)
71 33d32d1f 2020-12-25 op goto done;
72 33d32d1f 2020-12-25 op
73 33d32d1f 2020-12-25 op if (error != NULL)
74 33d32d1f 2020-12-25 op fprintf(stderr, "> %s\n", error);
75 33d32d1f 2020-12-25 op
76 33d32d1f 2020-12-25 op ok = !failed && !should_fail;
77 33d32d1f 2020-12-25 op if (ok)
78 33d32d1f 2020-12-25 op ok = diff_uri(&expected, &parsed);
79 33d32d1f 2020-12-25 op
80 33d32d1f 2020-12-25 op done:
81 33d32d1f 2020-12-25 op free(uri_copy);
82 33d32d1f 2020-12-25 op return ok;
83 33d32d1f 2020-12-25 op }
84 33d32d1f 2020-12-25 op
85 33d32d1f 2020-12-25 op int
86 33d32d1f 2020-12-25 op main(void)
87 33d32d1f 2020-12-25 op {
88 33d32d1f 2020-12-25 op struct uri empty = {"", "", "", PASS, "", "", ""};
89 33d32d1f 2020-12-25 op
90 33d32d1f 2020-12-25 op TEST("http://omarpolo.com",
91 33d32d1f 2020-12-25 op PASS,
92 33d32d1f 2020-12-25 op URI("http", "omarpolo.com", "", "", "", ""),
93 33d32d1f 2020-12-25 op "can parse uri with empty path");
94 33d32d1f 2020-12-25 op
95 33d32d1f 2020-12-25 op /* schema */
96 33d32d1f 2020-12-25 op TEST("omarpolo.com", FAIL, empty, "FAIL when the schema is missing");
97 33d32d1f 2020-12-25 op TEST("gemini:/omarpolo.com", FAIL, empty, "FAIL with invalid marker");
98 33d32d1f 2020-12-25 op TEST("gemini//omarpolo.com", FAIL, empty, "FAIL with invalid marker");
99 33d32d1f 2020-12-25 op TEST("h!!p://omarpolo.com", FAIL, empty, "FAIL with invalid schema");
100 33d32d1f 2020-12-25 op
101 33d32d1f 2020-12-25 op /* authority */
102 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com",
103 33d32d1f 2020-12-25 op PASS,
104 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "", "", ""),
105 33d32d1f 2020-12-25 op "can parse authority with empty path");
106 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/",
107 33d32d1f 2020-12-25 op PASS,
108 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "", "", ""),
109 33d32d1f 2020-12-25 op "can parse authority with empty path (alt)")
110 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com:1965",
111 33d32d1f 2020-12-25 op PASS,
112 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "1965", "", "", ""),
113 33d32d1f 2020-12-25 op "can parse with port and empty path");
114 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com:1965/",
115 33d32d1f 2020-12-25 op PASS,
116 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "1965", "", "", ""),
117 33d32d1f 2020-12-25 op "can parse with port and empty path")
118 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com:196s",
119 33d32d1f 2020-12-25 op FAIL,
120 33d32d1f 2020-12-25 op empty,
121 33d32d1f 2020-12-25 op "FAIL with invalid port number");
122 33d32d1f 2020-12-25 op
123 33d32d1f 2020-12-25 op /* path */
124 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/bar/baz",
125 33d32d1f 2020-12-25 op PASS,
126 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "foo/bar/baz", "", ""),
127 33d32d1f 2020-12-25 op "parse simple paths");
128 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo//bar///baz",
129 33d32d1f 2020-12-25 op PASS,
130 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "foo/bar/baz", "", ""),
131 33d32d1f 2020-12-25 op "parse paths with multiple slashes");
132 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/./bar/./././baz",
133 33d32d1f 2020-12-25 op PASS,
134 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "foo/bar/baz", "", ""),
135 33d32d1f 2020-12-25 op "parse paths with . elements");
136 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/bar/../bar/baz",
137 33d32d1f 2020-12-25 op PASS,
138 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "foo/bar/baz", "", ""),
139 33d32d1f 2020-12-25 op "parse paths with .. elements");
140 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/../foo/bar/../bar/baz/../baz",
141 33d32d1f 2020-12-25 op PASS,
142 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "foo/bar/baz", "", ""),
143 33d32d1f 2020-12-25 op "parse paths with multiple .. elements");
144 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/..",
145 33d32d1f 2020-12-25 op PASS,
146 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "", "", ""),
147 33d32d1f 2020-12-25 op "parse paths with a trailing ..");
148 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/../",
149 33d32d1f 2020-12-25 op PASS,
150 33d32d1f 2020-12-25 op URI("gemini", "omarpolo.com", "", "", "", ""),
151 33d32d1f 2020-12-25 op "parse paths with a trailing ..");
152 33d32d1f 2020-12-25 op TEST("gemini://omarpolo.com/foo/../..",
153 33d32d1f 2020-12-25 op FAIL,
154 33d32d1f 2020-12-25 op empty,
155 33d32d1f 2020-12-25 op "reject paths that would escape the root");
156 df6ca41d 2020-12-25 op TEST("gemini://omarpolo.com/foo/../foo/../././/bar/baz/.././.././/",
157 df6ca41d 2020-12-25 op PASS,
158 df6ca41d 2020-12-25 op URI("gemini", "omarpolo.com", "", "", "", ""),
159 df6ca41d 2020-12-25 op "parse path with lots of cleaning available");
160 33d32d1f 2020-12-25 op
161 33d32d1f 2020-12-25 op /* query */
162 33d32d1f 2020-12-25 op TEST("foo://example.com/foo/?gne",
163 33d32d1f 2020-12-25 op PASS,
164 33d32d1f 2020-12-25 op URI("foo", "example.com", "", "foo/", "gne", ""),
165 33d32d1f 2020-12-25 op "parse query strings");
166 33d32d1f 2020-12-25 op TEST("foo://example.com/foo/?gne&foo",
167 33d32d1f 2020-12-25 op PASS,
168 33d32d1f 2020-12-25 op URI("foo", "example.com", "", "foo/", "gne&foo", ""),
169 33d32d1f 2020-12-25 op "parse query strings");
170 33d32d1f 2020-12-25 op TEST("foo://example.com/foo/?gne%2F",
171 33d32d1f 2020-12-25 op PASS,
172 33d32d1f 2020-12-25 op URI("foo", "example.com", "", "foo/", "gne/", ""),
173 33d32d1f 2020-12-25 op "parse query strings");
174 33d32d1f 2020-12-25 op
175 33d32d1f 2020-12-25 op /* fragment */
176 33d32d1f 2020-12-25 op TEST("foo://bar.co/#foo",
177 33d32d1f 2020-12-25 op PASS,
178 33d32d1f 2020-12-25 op URI("foo", "bar.co", "", "", "", "foo"),
179 33d32d1f 2020-12-25 op "can recognize fragments");
180 33d32d1f 2020-12-25 op
181 33d32d1f 2020-12-25 op /* percent encoding */
182 33d32d1f 2020-12-25 op TEST("foo://bar.com/caf%C3%A8.gmi",
183 33d32d1f 2020-12-25 op PASS,
184 33d32d1f 2020-12-25 op URI("foo", "bar.com", "", "cafè.gmi", "", ""),
185 33d32d1f 2020-12-25 op "can decode");
186 df6ca41d 2020-12-25 op TEST("foo://bar.com/caff%C3%A8%20macchiato.gmi",
187 df6ca41d 2020-12-25 op PASS,
188 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "caffè macchiato.gmi", "", ""),
189 df6ca41d 2020-12-25 op "can decode");
190 df6ca41d 2020-12-25 op TEST("foo://bar.com/caff%C3%A8+macchiato.gmi",
191 df6ca41d 2020-12-25 op PASS,
192 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "caffè+macchiato.gmi", "", ""),
193 df6ca41d 2020-12-25 op "can decode");
194 df6ca41d 2020-12-25 op TEST("foo://bar.com/foo%2F..%2F..",
195 df6ca41d 2020-12-25 op FAIL,
196 df6ca41d 2020-12-25 op empty,
197 df6ca41d 2020-12-25 op "conversion and checking are done in the correct order");
198 df6ca41d 2020-12-25 op TEST("foo://bar.com/foo%00?baz",
199 df6ca41d 2020-12-25 op FAIL,
200 df6ca41d 2020-12-25 op empty,
201 df6ca41d 2020-12-25 op "rejects %00");
202 33d32d1f 2020-12-25 op
203 df6ca41d 2020-12-25 op /* IRI */
204 df6ca41d 2020-12-25 op TEST("foo://bar.com/cafè.gmi",
205 df6ca41d 2020-12-25 op PASS,
206 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "cafè.gmi", "" , ""),
207 df6ca41d 2020-12-25 op "decode IRI (with a 2-byte utf8 seq)");
208 df6ca41d 2020-12-25 op TEST("foo://bar.com/世界.gmi",
209 df6ca41d 2020-12-25 op PASS,
210 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "世界.gmi", "" , ""),
211 df6ca41d 2020-12-25 op "decode IRI");
212 df6ca41d 2020-12-25 op TEST("foo://bar.com/😼.gmi",
213 df6ca41d 2020-12-25 op PASS,
214 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "😼.gmi", "" , ""),
215 df6ca41d 2020-12-25 op "decode IRI (with a 3-byte utf8 seq)");
216 df6ca41d 2020-12-25 op TEST("foo://bar.com/😼/𤭢.gmi",
217 df6ca41d 2020-12-25 op PASS,
218 df6ca41d 2020-12-25 op URI("foo", "bar.com", "", "😼/𤭢.gmi", "" , ""),
219 df6ca41d 2020-12-25 op "decode IRI (with a 3-byte and a 4-byte utf8 seq)");
220 df6ca41d 2020-12-25 op TEST("foo://bar.com/世界/\xC0\x80",
221 df6ca41d 2020-12-25 op FAIL,
222 df6ca41d 2020-12-25 op empty,
223 df6ca41d 2020-12-25 op "reject invalid sequence (overlong NUL)");
224 df6ca41d 2020-12-25 op
225 33d32d1f 2020-12-25 op return 0;
226 33d32d1f 2020-12-25 op }