Blame


1 3c1cf9d0 2021-01-11 op /*
2 3c1cf9d0 2021-01-11 op * Copyright (c) 2020 Omar Polo <op@omarpolo.com>
3 3c1cf9d0 2021-01-11 op *
4 3c1cf9d0 2021-01-11 op * Permission to use, copy, modify, and distribute this software for any
5 3c1cf9d0 2021-01-11 op * purpose with or without fee is hereby granted, provided that the above
6 3c1cf9d0 2021-01-11 op * copyright notice and this permission notice appear in all copies.
7 3c1cf9d0 2021-01-11 op *
8 3c1cf9d0 2021-01-11 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 3c1cf9d0 2021-01-11 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 3c1cf9d0 2021-01-11 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 3c1cf9d0 2021-01-11 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 3c1cf9d0 2021-01-11 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 3c1cf9d0 2021-01-11 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 3c1cf9d0 2021-01-11 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 3c1cf9d0 2021-01-11 op */
16 52418c8d 2021-02-12 op
17 52418c8d 2021-02-12 op #include "gmid.h"
18 3c1cf9d0 2021-01-11 op
19 3c1cf9d0 2021-01-11 op #include <ctype.h>
20 3c1cf9d0 2021-01-11 op #include <string.h>
21 3c1cf9d0 2021-01-11 op
22 3c1cf9d0 2021-01-11 op static inline int
23 3c1cf9d0 2021-01-11 op unreserved(int p)
24 3c1cf9d0 2021-01-11 op {
25 3c1cf9d0 2021-01-11 op return isalnum(p)
26 3c1cf9d0 2021-01-11 op || p == '-'
27 3c1cf9d0 2021-01-11 op || p == '.'
28 3c1cf9d0 2021-01-11 op || p == '_'
29 3c1cf9d0 2021-01-11 op || p == '~';
30 3c1cf9d0 2021-01-11 op }
31 3c1cf9d0 2021-01-11 op
32 3c1cf9d0 2021-01-11 op static inline int
33 3c1cf9d0 2021-01-11 op sub_delimiters(int p)
34 3c1cf9d0 2021-01-11 op {
35 3c1cf9d0 2021-01-11 op return p == '!'
36 3c1cf9d0 2021-01-11 op || p == '$'
37 3c1cf9d0 2021-01-11 op || p == '&'
38 3c1cf9d0 2021-01-11 op || p == '\''
39 3c1cf9d0 2021-01-11 op || p == '('
40 3c1cf9d0 2021-01-11 op || p == ')'
41 3c1cf9d0 2021-01-11 op || p == '*'
42 3c1cf9d0 2021-01-11 op || p == '+'
43 3c1cf9d0 2021-01-11 op || p == ','
44 3c1cf9d0 2021-01-11 op || p == ';'
45 3c1cf9d0 2021-01-11 op || p == '=';
46 8404ec30 2021-02-05 op }
47 8404ec30 2021-02-05 op
48 8404ec30 2021-02-05 op static int
49 8404ec30 2021-02-05 op valid_pct_enc_string(char *s)
50 8404ec30 2021-02-05 op {
51 8404ec30 2021-02-05 op if (*s != '%')
52 8404ec30 2021-02-05 op return 1;
53 8404ec30 2021-02-05 op
54 8404ec30 2021-02-05 op if (!isxdigit(s[1]) || !isxdigit(s[2]))
55 8404ec30 2021-02-05 op return 0;
56 8404ec30 2021-02-05 op
57 8404ec30 2021-02-05 op if (s[1] == '0' && s[2] == '0')
58 8404ec30 2021-02-05 op return 0;
59 8404ec30 2021-02-05 op
60 8404ec30 2021-02-05 op return 1;
61 8404ec30 2021-02-05 op }
62 8404ec30 2021-02-05 op
63 8404ec30 2021-02-05 op static int
64 8404ec30 2021-02-05 op valid_pct_encoded(struct parser *p)
65 8404ec30 2021-02-05 op {
66 8404ec30 2021-02-05 op if (p->iri[0] != '%')
67 8404ec30 2021-02-05 op return 0;
68 8404ec30 2021-02-05 op
69 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
70 8404ec30 2021-02-05 op p->err = "illegal percent-encoding";
71 8404ec30 2021-02-05 op return 0;
72 8404ec30 2021-02-05 op }
73 8404ec30 2021-02-05 op
74 8404ec30 2021-02-05 op p->iri += 2;
75 8404ec30 2021-02-05 op return 1;
76 3c1cf9d0 2021-01-11 op }
77 3c1cf9d0 2021-01-11 op
78 9f006a21 2021-02-07 op static void
79 9f006a21 2021-02-07 op pct_decode(char *s)
80 9f006a21 2021-02-07 op {
81 9f006a21 2021-02-07 op sscanf(s+1, "%2hhx", s);
82 9f006a21 2021-02-07 op memmove(s+1, s+3, strlen(s+3)+1);
83 9f006a21 2021-02-07 op }
84 9f006a21 2021-02-07 op
85 3c1cf9d0 2021-01-11 op static int
86 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
87 3c1cf9d0 2021-01-11 op {
88 9a672b37 2021-01-28 op if (p->iri[0] != '%')
89 3c1cf9d0 2021-01-11 op return 0;
90 3c1cf9d0 2021-01-11 op
91 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
92 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
93 3c1cf9d0 2021-01-11 op return 0;
94 3c1cf9d0 2021-01-11 op }
95 3c1cf9d0 2021-01-11 op
96 9f006a21 2021-02-07 op pct_decode(p->iri);
97 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
98 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
99 3c1cf9d0 2021-01-11 op return 0;
100 3c1cf9d0 2021-01-11 op }
101 3c1cf9d0 2021-01-11 op
102 3c1cf9d0 2021-01-11 op return 1;
103 3c1cf9d0 2021-01-11 op }
104 3c1cf9d0 2021-01-11 op
105 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
106 3c1cf9d0 2021-01-11 op static int
107 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
108 3c1cf9d0 2021-01-11 op {
109 3c1cf9d0 2021-01-11 op p->parsed->schema = p->iri;
110 3c1cf9d0 2021-01-11 op
111 3c1cf9d0 2021-01-11 op if (!isalpha(*p->iri)) {
112 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
113 3c1cf9d0 2021-01-11 op return 0;
114 3c1cf9d0 2021-01-11 op }
115 3c1cf9d0 2021-01-11 op
116 de428fff 2021-01-13 op do {
117 a8a1f439 2021-07-07 op /*
118 a8a1f439 2021-07-07 op * normalize the scheme (i.e. lowercase it)
119 de428fff 2021-01-13 op *
120 de428fff 2021-01-13 op * XXX: since we cannot have good things, tolower
121 881dc835 2021-01-16 op * behaviour depends on the LC_CTYPE locale. The good
122 881dc835 2021-01-16 op * news is that we're sure p->iri points to something
123 881dc835 2021-01-16 op * that's in the ASCII range, so tolower can't
124 a8a1f439 2021-07-07 op * mis-behave on some systems due to the locale.
125 a8a1f439 2021-07-07 op */
126 de428fff 2021-01-13 op *p->iri = tolower(*p->iri);
127 de428fff 2021-01-13 op p->iri++;
128 de428fff 2021-01-13 op } while (isalnum(*p->iri)
129 3c1cf9d0 2021-01-11 op || *p->iri == '+'
130 3c1cf9d0 2021-01-11 op || *p->iri == '-'
131 de428fff 2021-01-13 op || *p->iri == '.');
132 3c1cf9d0 2021-01-11 op
133 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
134 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
135 3c1cf9d0 2021-01-11 op return 0;
136 3c1cf9d0 2021-01-11 op }
137 3c1cf9d0 2021-01-11 op
138 3c1cf9d0 2021-01-11 op *p->iri = '\0';
139 f7b816dc 2021-01-15 op if (p->iri[1] != '/' || p->iri[2] != '/') {
140 3c1cf9d0 2021-01-11 op p->err = "invalid marker after scheme";
141 3c1cf9d0 2021-01-11 op return 0;
142 3c1cf9d0 2021-01-11 op }
143 3c1cf9d0 2021-01-11 op
144 f7b816dc 2021-01-15 op p->iri += 3;
145 3c1cf9d0 2021-01-11 op return 1;
146 3c1cf9d0 2021-01-11 op }
147 3c1cf9d0 2021-01-11 op
148 3c1cf9d0 2021-01-11 op /* *DIGIT */
149 3c1cf9d0 2021-01-11 op static int
150 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
151 3c1cf9d0 2021-01-11 op {
152 3c1cf9d0 2021-01-11 op uint32_t i = 0;
153 3c1cf9d0 2021-01-11 op
154 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
155 3c1cf9d0 2021-01-11 op
156 3c1cf9d0 2021-01-11 op for (; isdigit(*p->iri); p->iri++) {
157 3c1cf9d0 2021-01-11 op i = i * 10 + *p->iri - '0';
158 3c1cf9d0 2021-01-11 op if (i > UINT16_MAX) {
159 3c1cf9d0 2021-01-11 op p->err = "port number too large";
160 3c1cf9d0 2021-01-11 op return 0;
161 3c1cf9d0 2021-01-11 op }
162 3c1cf9d0 2021-01-11 op }
163 3c1cf9d0 2021-01-11 op
164 3c1cf9d0 2021-01-11 op if (*p->iri != '/' && *p->iri != '\0') {
165 3c1cf9d0 2021-01-11 op p->err = "illegal character in port number";
166 3c1cf9d0 2021-01-11 op return 0;
167 3c1cf9d0 2021-01-11 op }
168 3c1cf9d0 2021-01-11 op
169 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
170 3c1cf9d0 2021-01-11 op
171 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
172 3c1cf9d0 2021-01-11 op *p->iri = '\0';
173 3c1cf9d0 2021-01-11 op p->iri++;
174 3c1cf9d0 2021-01-11 op }
175 3c1cf9d0 2021-01-11 op
176 3c1cf9d0 2021-01-11 op return 1;
177 3c1cf9d0 2021-01-11 op }
178 3c1cf9d0 2021-01-11 op
179 3c1cf9d0 2021-01-11 op /* TODO: add support for ip-literal and ipv4addr ? */
180 3c1cf9d0 2021-01-11 op /* *( unreserved / sub-delims / pct-encoded ) */
181 3c1cf9d0 2021-01-11 op static int
182 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
183 3c1cf9d0 2021-01-11 op {
184 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
185 3c1cf9d0 2021-01-11 op
186 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
187 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
188 117ac52c 2021-01-29 op || parse_pct_encoded(p)
189 117ac52c 2021-01-29 op || valid_multibyte_utf8(p)) {
190 e4d82bec 2021-01-15 op /* normalize the host name. */
191 e4d82bec 2021-01-15 op if (*p->iri < 0x7F)
192 e4d82bec 2021-01-15 op *p->iri = tolower(*p->iri);
193 3c1cf9d0 2021-01-11 op p->iri++;
194 e4d82bec 2021-01-15 op }
195 3c1cf9d0 2021-01-11 op
196 3c1cf9d0 2021-01-11 op if (p->err != NULL)
197 3c1cf9d0 2021-01-11 op return 0;
198 3c1cf9d0 2021-01-11 op
199 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
200 3c1cf9d0 2021-01-11 op *p->iri = '\0';
201 3c1cf9d0 2021-01-11 op p->iri++;
202 3c1cf9d0 2021-01-11 op return parse_port(p);
203 b777bf4b 2021-01-15 op } else
204 b777bf4b 2021-01-15 op p->parsed->port_no = 1965;
205 3c1cf9d0 2021-01-11 op
206 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
207 3c1cf9d0 2021-01-11 op *p->iri = '\0';
208 3c1cf9d0 2021-01-11 op p->iri++;
209 3c1cf9d0 2021-01-11 op return 1;
210 3c1cf9d0 2021-01-11 op }
211 3c1cf9d0 2021-01-11 op
212 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
213 3c1cf9d0 2021-01-11 op return 1;
214 3c1cf9d0 2021-01-11 op
215 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
216 3c1cf9d0 2021-01-11 op return 0;
217 3c1cf9d0 2021-01-11 op }
218 3c1cf9d0 2021-01-11 op
219 a8a1f439 2021-07-07 op /*
220 a8a1f439 2021-07-07 op * Routine for path_clean. Elide the pointed .. with the preceding
221 3c1cf9d0 2021-01-11 op * element. Return 0 if it's not possible. incr is the length of
222 a8a1f439 2021-07-07 op * the increment, 3 for ../ and 2 for ..
223 a8a1f439 2021-07-07 op */
224 3c1cf9d0 2021-01-11 op static int
225 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
226 3c1cf9d0 2021-01-11 op {
227 3c1cf9d0 2021-01-11 op char *j;
228 3c1cf9d0 2021-01-11 op
229 3c1cf9d0 2021-01-11 op if (i == path)
230 3c1cf9d0 2021-01-11 op return 0;
231 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
232 4842c72d 2021-10-18 op /* noop */ ;
233 3c1cf9d0 2021-01-11 op if (*j == '/')
234 3c1cf9d0 2021-01-11 op j++;
235 3c1cf9d0 2021-01-11 op i += incr;
236 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
237 3c1cf9d0 2021-01-11 op return 1;
238 3c1cf9d0 2021-01-11 op }
239 3c1cf9d0 2021-01-11 op
240 3c1cf9d0 2021-01-11 op /*
241 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
242 3c1cf9d0 2021-01-11 op *
243 3c1cf9d0 2021-01-11 op * 1. Replace multiple slashes with a single slash
244 3c1cf9d0 2021-01-11 op * 2. Eliminate each . path name element
245 3c1cf9d0 2021-01-11 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
246 3c1cf9d0 2021-01-11 op * 4. Eliminate trailing .. if possible or error (go would only discard)
247 3c1cf9d0 2021-01-11 op *
248 3c1cf9d0 2021-01-11 op * Unlike path.Clean, this function return the empty string if the
249 3c1cf9d0 2021-01-11 op * original path is equivalent to "/".
250 3c1cf9d0 2021-01-11 op */
251 3c1cf9d0 2021-01-11 op static int
252 3c1cf9d0 2021-01-11 op path_clean(char *path)
253 3c1cf9d0 2021-01-11 op {
254 3c1cf9d0 2021-01-11 op char *i;
255 3c1cf9d0 2021-01-11 op
256 3c1cf9d0 2021-01-11 op /* 1. replace multiple slashes with a single one */
257 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
258 3c1cf9d0 2021-01-11 op if (*i == '/' && *(i+1) == '/') {
259 3c1cf9d0 2021-01-11 op memmove(i, i+1, strlen(i)); /* move also the \0 */
260 3c1cf9d0 2021-01-11 op i--;
261 3c1cf9d0 2021-01-11 op }
262 3c1cf9d0 2021-01-11 op }
263 3c1cf9d0 2021-01-11 op
264 3c1cf9d0 2021-01-11 op /* 2. eliminate each . path name element */
265 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
266 6a9ae707 2021-01-11 op if ((i == path || *i == '/') &&
267 6a9ae707 2021-01-11 op *i != '.' && i[1] == '.' && i[2] == '/') {
268 3c1cf9d0 2021-01-11 op /* move also the \0 */
269 3c1cf9d0 2021-01-11 op memmove(i, i+2, strlen(i)-1);
270 3c1cf9d0 2021-01-11 op i--;
271 3c1cf9d0 2021-01-11 op }
272 3c1cf9d0 2021-01-11 op }
273 3c1cf9d0 2021-01-11 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
274 3c1cf9d0 2021-01-11 op *path = '\0';
275 3c1cf9d0 2021-01-11 op return 1;
276 3c1cf9d0 2021-01-11 op }
277 3c1cf9d0 2021-01-11 op
278 3c1cf9d0 2021-01-11 op /* 3. eliminate each inner .. along with the preceding non-.. */
279 9d092b60 2021-04-12 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, "..")) {
280 9d092b60 2021-04-12 op /* break if we've found a trailing .. */
281 9d092b60 2021-04-12 op if (i[2] == '\0')
282 9d092b60 2021-04-12 op break;
283 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 3))
284 3c1cf9d0 2021-01-11 op return 0;
285 9d092b60 2021-04-12 op }
286 3c1cf9d0 2021-01-11 op
287 3c1cf9d0 2021-01-11 op /* 4. eliminate trailing ..*/
288 3c1cf9d0 2021-01-11 op if ((i = strstr(path, "..")) != NULL)
289 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 2))
290 3c1cf9d0 2021-01-11 op return 0;
291 3c1cf9d0 2021-01-11 op
292 3c1cf9d0 2021-01-11 op return 1;
293 3c1cf9d0 2021-01-11 op }
294 3c1cf9d0 2021-01-11 op
295 3c1cf9d0 2021-01-11 op static int
296 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
297 3c1cf9d0 2021-01-11 op {
298 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
299 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
300 3c1cf9d0 2021-01-11 op return 1;
301 3c1cf9d0 2021-01-11 op
302 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
303 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
304 3c1cf9d0 2021-01-11 op || *p->iri == '/'
305 3c1cf9d0 2021-01-11 op || *p->iri == '?'
306 19e7bd00 2021-02-06 op || *p->iri == ':'
307 19e7bd00 2021-02-06 op || *p->iri == '@'
308 8404ec30 2021-02-05 op || valid_pct_encoded(p)
309 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
310 3c1cf9d0 2021-01-11 op p->iri++;
311 3c1cf9d0 2021-01-11 op
312 3c1cf9d0 2021-01-11 op if (p->err != NULL)
313 3c1cf9d0 2021-01-11 op return 0;
314 3c1cf9d0 2021-01-11 op
315 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
316 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
317 3c1cf9d0 2021-01-11 op return 0;
318 3c1cf9d0 2021-01-11 op }
319 3c1cf9d0 2021-01-11 op
320 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
321 3c1cf9d0 2021-01-11 op *p->iri = '\0';
322 3c1cf9d0 2021-01-11 op p->iri++;
323 3c1cf9d0 2021-01-11 op }
324 3c1cf9d0 2021-01-11 op
325 3c1cf9d0 2021-01-11 op return 1;
326 3c1cf9d0 2021-01-11 op }
327 3c1cf9d0 2021-01-11 op
328 3c1cf9d0 2021-01-11 op /* don't even bother */
329 3c1cf9d0 2021-01-11 op static int
330 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
331 3c1cf9d0 2021-01-11 op {
332 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
333 3c1cf9d0 2021-01-11 op return 1;
334 3c1cf9d0 2021-01-11 op }
335 3c1cf9d0 2021-01-11 op
336 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
337 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
338 3c1cf9d0 2021-01-11 op static int
339 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
340 3c1cf9d0 2021-01-11 op {
341 3c1cf9d0 2021-01-11 op char c;
342 42bbdc79 2021-01-21 op
343 42bbdc79 2021-01-21 op /* trim initial slashes */
344 42bbdc79 2021-01-21 op while (*p->iri == '/')
345 42bbdc79 2021-01-21 op p->iri++;
346 3c1cf9d0 2021-01-11 op
347 3c1cf9d0 2021-01-11 op p->parsed->path = p->iri;
348 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
349 3c1cf9d0 2021-01-11 op p->parsed->query = p->parsed->fragment = p->iri;
350 3c1cf9d0 2021-01-11 op return 1;
351 3c1cf9d0 2021-01-11 op }
352 3c1cf9d0 2021-01-11 op
353 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
354 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
355 3c1cf9d0 2021-01-11 op || *p->iri == '/'
356 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
357 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
358 3c1cf9d0 2021-01-11 op p->iri++;
359 3c1cf9d0 2021-01-11 op
360 3c1cf9d0 2021-01-11 op if (p->err != NULL)
361 3c1cf9d0 2021-01-11 op return 0;
362 3c1cf9d0 2021-01-11 op
363 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
364 3c1cf9d0 2021-01-11 op p->err = "illegal character in path";
365 3c1cf9d0 2021-01-11 op return 0;
366 3c1cf9d0 2021-01-11 op }
367 3c1cf9d0 2021-01-11 op
368 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
369 3c1cf9d0 2021-01-11 op c = *p->iri;
370 3c1cf9d0 2021-01-11 op *p->iri = '\0';
371 3c1cf9d0 2021-01-11 op p->iri++;
372 3c1cf9d0 2021-01-11 op
373 3c1cf9d0 2021-01-11 op if (c == '#') {
374 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
375 3c1cf9d0 2021-01-11 op return 0;
376 3c1cf9d0 2021-01-11 op } else
377 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
378 3c1cf9d0 2021-01-11 op return 0;
379 3c1cf9d0 2021-01-11 op }
380 3c1cf9d0 2021-01-11 op
381 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
382 3c1cf9d0 2021-01-11 op p->err = "illegal path";
383 3c1cf9d0 2021-01-11 op return 0;
384 3c1cf9d0 2021-01-11 op }
385 3c1cf9d0 2021-01-11 op
386 3c1cf9d0 2021-01-11 op return 1;
387 3c1cf9d0 2021-01-11 op }
388 3c1cf9d0 2021-01-11 op
389 3c1cf9d0 2021-01-11 op int
390 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
391 3c1cf9d0 2021-01-11 op {
392 3c1cf9d0 2021-01-11 op char *end;
393 e15fc957 2021-09-24 op struct parser p = {
394 e15fc957 2021-09-24 op .iri = iri,
395 e15fc957 2021-09-24 op .parsed = ret,
396 e15fc957 2021-09-24 op .err = NULL,
397 e15fc957 2021-09-24 op };
398 3c1cf9d0 2021-01-11 op
399 df0c2926 2021-09-24 op memset(ret, 0, sizeof(*ret));
400 3c1cf9d0 2021-01-11 op
401 3c1cf9d0 2021-01-11 op /* initialize optional stuff to the empty string */
402 3c1cf9d0 2021-01-11 op end = iri + strlen(iri);
403 57d0d0ad 2021-01-31 op p.parsed->host = end;
404 3c1cf9d0 2021-01-11 op p.parsed->port = end;
405 3c1cf9d0 2021-01-11 op p.parsed->path = end;
406 3c1cf9d0 2021-01-11 op p.parsed->query = end;
407 3c1cf9d0 2021-01-11 op p.parsed->fragment = end;
408 3c1cf9d0 2021-01-11 op
409 3c1cf9d0 2021-01-11 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
410 3c1cf9d0 2021-01-11 op *err_ret = p.err;
411 3c1cf9d0 2021-01-11 op return 0;
412 3c1cf9d0 2021-01-11 op }
413 3c1cf9d0 2021-01-11 op
414 3c1cf9d0 2021-01-11 op *err_ret = NULL;
415 3c1cf9d0 2021-01-11 op return 1;
416 3c1cf9d0 2021-01-11 op }
417 3c1cf9d0 2021-01-11 op
418 3c1cf9d0 2021-01-11 op int
419 2fafa2d2 2021-02-01 op serialize_iri(struct iri *i, char *buf, size_t len)
420 2fafa2d2 2021-02-01 op {
421 80fbf1e9 2021-06-16 op size_t l = 0;
422 2fafa2d2 2021-02-01 op
423 2fafa2d2 2021-02-01 op /* in ex.c we receive empty "" strings as NULL */
424 2fafa2d2 2021-02-01 op if (i->schema == NULL || i->host == NULL) {
425 2fafa2d2 2021-02-01 op memset(buf, 0, len);
426 2fafa2d2 2021-02-01 op return 0;
427 2fafa2d2 2021-02-01 op }
428 2fafa2d2 2021-02-01 op
429 2fafa2d2 2021-02-01 op strlcpy(buf, i->schema, len);
430 2fafa2d2 2021-02-01 op strlcat(buf, "://", len);
431 4842c72d 2021-10-18 op strlcat(buf, i->host, len);
432 2fafa2d2 2021-02-01 op strlcat(buf, "/", len);
433 2fafa2d2 2021-02-01 op
434 2fafa2d2 2021-02-01 op if (i->path != NULL)
435 2fafa2d2 2021-02-01 op l = strlcat(buf, i->path, len);
436 2fafa2d2 2021-02-01 op
437 2fafa2d2 2021-02-01 op if (i->query != NULL && *i->query != '\0') {
438 2fafa2d2 2021-02-01 op strlcat(buf, "?", len);
439 2fafa2d2 2021-02-01 op l = strlcat(buf, i->query, len);
440 2fafa2d2 2021-02-01 op }
441 2fafa2d2 2021-02-01 op
442 2fafa2d2 2021-02-01 op return l < len;
443 3c1cf9d0 2021-01-11 op }
444 9f006a21 2021-02-07 op
445 9f006a21 2021-02-07 op char *
446 9f006a21 2021-02-07 op pct_decode_str(char *s)
447 9f006a21 2021-02-07 op {
448 9f006a21 2021-02-07 op char *t;
449 9f006a21 2021-02-07 op
450 9f006a21 2021-02-07 op for (t = s; *t; ++t) {
451 9f006a21 2021-02-07 op if (*t == '%' && valid_pct_enc_string(t))
452 9f006a21 2021-02-07 op pct_decode(t);
453 9f006a21 2021-02-07 op }
454 9f006a21 2021-02-07 op
455 9f006a21 2021-02-07 op return s;
456 9f006a21 2021-02-07 op }