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 3c1cf9d0 2021-01-11 op
17 3c1cf9d0 2021-01-11 op #include <ctype.h>
18 3c1cf9d0 2021-01-11 op #include <string.h>
19 3c1cf9d0 2021-01-11 op
20 3c1cf9d0 2021-01-11 op #include "gmid.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 3c1cf9d0 2021-01-11 op static int
79 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
80 3c1cf9d0 2021-01-11 op {
81 9a672b37 2021-01-28 op if (p->iri[0] != '%')
82 3c1cf9d0 2021-01-11 op return 0;
83 3c1cf9d0 2021-01-11 op
84 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
85 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
86 3c1cf9d0 2021-01-11 op return 0;
87 3c1cf9d0 2021-01-11 op }
88 3c1cf9d0 2021-01-11 op
89 3c1cf9d0 2021-01-11 op sscanf(p->iri+1, "%2hhx", p->iri);
90 3c1cf9d0 2021-01-11 op memmove(p->iri+1, p->iri+3, strlen(p->iri+3)+1);
91 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
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 3c1cf9d0 2021-01-11 op return 1;
97 3c1cf9d0 2021-01-11 op }
98 3c1cf9d0 2021-01-11 op
99 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
100 3c1cf9d0 2021-01-11 op static int
101 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
102 3c1cf9d0 2021-01-11 op {
103 3c1cf9d0 2021-01-11 op p->parsed->schema = p->iri;
104 3c1cf9d0 2021-01-11 op
105 3c1cf9d0 2021-01-11 op if (!isalpha(*p->iri)) {
106 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
107 3c1cf9d0 2021-01-11 op return 0;
108 3c1cf9d0 2021-01-11 op }
109 3c1cf9d0 2021-01-11 op
110 de428fff 2021-01-13 op do {
111 de428fff 2021-01-13 op /* normalize the scheme (i.e. lowercase it)
112 de428fff 2021-01-13 op *
113 de428fff 2021-01-13 op * XXX: since we cannot have good things, tolower
114 881dc835 2021-01-16 op * behaviour depends on the LC_CTYPE locale. The good
115 881dc835 2021-01-16 op * news is that we're sure p->iri points to something
116 881dc835 2021-01-16 op * that's in the ASCII range, so tolower can't
117 881dc835 2021-01-16 op * mis-behave on some systems due to the locale. */
118 de428fff 2021-01-13 op *p->iri = tolower(*p->iri);
119 de428fff 2021-01-13 op p->iri++;
120 de428fff 2021-01-13 op } while (isalnum(*p->iri)
121 3c1cf9d0 2021-01-11 op || *p->iri == '+'
122 3c1cf9d0 2021-01-11 op || *p->iri == '-'
123 de428fff 2021-01-13 op || *p->iri == '.');
124 3c1cf9d0 2021-01-11 op
125 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
126 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
127 3c1cf9d0 2021-01-11 op return 0;
128 3c1cf9d0 2021-01-11 op }
129 3c1cf9d0 2021-01-11 op
130 3c1cf9d0 2021-01-11 op *p->iri = '\0';
131 f7b816dc 2021-01-15 op if (p->iri[1] != '/' || p->iri[2] != '/') {
132 3c1cf9d0 2021-01-11 op p->err = "invalid marker after scheme";
133 3c1cf9d0 2021-01-11 op return 0;
134 3c1cf9d0 2021-01-11 op }
135 3c1cf9d0 2021-01-11 op
136 f7b816dc 2021-01-15 op p->iri += 3;
137 3c1cf9d0 2021-01-11 op return 1;
138 3c1cf9d0 2021-01-11 op }
139 3c1cf9d0 2021-01-11 op
140 3c1cf9d0 2021-01-11 op /* *DIGIT */
141 3c1cf9d0 2021-01-11 op static int
142 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
143 3c1cf9d0 2021-01-11 op {
144 3c1cf9d0 2021-01-11 op uint32_t i = 0;
145 3c1cf9d0 2021-01-11 op
146 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
147 3c1cf9d0 2021-01-11 op
148 3c1cf9d0 2021-01-11 op for (; isdigit(*p->iri); p->iri++) {
149 3c1cf9d0 2021-01-11 op i = i * 10 + *p->iri - '0';
150 3c1cf9d0 2021-01-11 op if (i > UINT16_MAX) {
151 3c1cf9d0 2021-01-11 op p->err = "port number too large";
152 3c1cf9d0 2021-01-11 op return 0;
153 3c1cf9d0 2021-01-11 op }
154 3c1cf9d0 2021-01-11 op }
155 3c1cf9d0 2021-01-11 op
156 3c1cf9d0 2021-01-11 op if (*p->iri != '/' && *p->iri != '\0') {
157 3c1cf9d0 2021-01-11 op p->err = "illegal character in port number";
158 3c1cf9d0 2021-01-11 op return 0;
159 3c1cf9d0 2021-01-11 op }
160 3c1cf9d0 2021-01-11 op
161 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
162 3c1cf9d0 2021-01-11 op
163 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
164 3c1cf9d0 2021-01-11 op *p->iri = '\0';
165 3c1cf9d0 2021-01-11 op p->iri++;
166 3c1cf9d0 2021-01-11 op }
167 3c1cf9d0 2021-01-11 op
168 3c1cf9d0 2021-01-11 op return 1;
169 3c1cf9d0 2021-01-11 op }
170 3c1cf9d0 2021-01-11 op
171 3c1cf9d0 2021-01-11 op /* TODO: add support for ip-literal and ipv4addr ? */
172 3c1cf9d0 2021-01-11 op /* *( unreserved / sub-delims / pct-encoded ) */
173 3c1cf9d0 2021-01-11 op static int
174 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
175 3c1cf9d0 2021-01-11 op {
176 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
177 3c1cf9d0 2021-01-11 op
178 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
179 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
180 117ac52c 2021-01-29 op || parse_pct_encoded(p)
181 117ac52c 2021-01-29 op || valid_multibyte_utf8(p)) {
182 e4d82bec 2021-01-15 op /* normalize the host name. */
183 e4d82bec 2021-01-15 op if (*p->iri < 0x7F)
184 e4d82bec 2021-01-15 op *p->iri = tolower(*p->iri);
185 3c1cf9d0 2021-01-11 op p->iri++;
186 e4d82bec 2021-01-15 op }
187 3c1cf9d0 2021-01-11 op
188 3c1cf9d0 2021-01-11 op if (p->err != NULL)
189 3c1cf9d0 2021-01-11 op return 0;
190 3c1cf9d0 2021-01-11 op
191 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
192 3c1cf9d0 2021-01-11 op *p->iri = '\0';
193 3c1cf9d0 2021-01-11 op p->iri++;
194 3c1cf9d0 2021-01-11 op return parse_port(p);
195 b777bf4b 2021-01-15 op } else
196 b777bf4b 2021-01-15 op p->parsed->port_no = 1965;
197 3c1cf9d0 2021-01-11 op
198 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
199 3c1cf9d0 2021-01-11 op *p->iri = '\0';
200 3c1cf9d0 2021-01-11 op p->iri++;
201 3c1cf9d0 2021-01-11 op return 1;
202 3c1cf9d0 2021-01-11 op }
203 3c1cf9d0 2021-01-11 op
204 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
205 3c1cf9d0 2021-01-11 op return 1;
206 3c1cf9d0 2021-01-11 op
207 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
208 3c1cf9d0 2021-01-11 op return 0;
209 3c1cf9d0 2021-01-11 op }
210 3c1cf9d0 2021-01-11 op
211 3c1cf9d0 2021-01-11 op /* Routine for path_clean. Elide the pointed .. with the preceding
212 3c1cf9d0 2021-01-11 op * element. Return 0 if it's not possible. incr is the length of
213 3c1cf9d0 2021-01-11 op * the increment, 3 for ../ and 2 for .. */
214 3c1cf9d0 2021-01-11 op static int
215 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
216 3c1cf9d0 2021-01-11 op {
217 3c1cf9d0 2021-01-11 op char *j;
218 3c1cf9d0 2021-01-11 op
219 3c1cf9d0 2021-01-11 op if (i == path)
220 3c1cf9d0 2021-01-11 op return 0;
221 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
222 3c1cf9d0 2021-01-11 op /* noop */ ;
223 3c1cf9d0 2021-01-11 op if (*j == '/')
224 3c1cf9d0 2021-01-11 op j++;
225 3c1cf9d0 2021-01-11 op i += incr;
226 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
227 3c1cf9d0 2021-01-11 op return 1;
228 3c1cf9d0 2021-01-11 op }
229 3c1cf9d0 2021-01-11 op
230 3c1cf9d0 2021-01-11 op /*
231 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
232 3c1cf9d0 2021-01-11 op *
233 3c1cf9d0 2021-01-11 op * 1. Replace multiple slashes with a single slash
234 3c1cf9d0 2021-01-11 op * 2. Eliminate each . path name element
235 3c1cf9d0 2021-01-11 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
236 3c1cf9d0 2021-01-11 op * 4. Eliminate trailing .. if possible or error (go would only discard)
237 3c1cf9d0 2021-01-11 op *
238 3c1cf9d0 2021-01-11 op * Unlike path.Clean, this function return the empty string if the
239 3c1cf9d0 2021-01-11 op * original path is equivalent to "/".
240 3c1cf9d0 2021-01-11 op */
241 3c1cf9d0 2021-01-11 op static int
242 3c1cf9d0 2021-01-11 op path_clean(char *path)
243 3c1cf9d0 2021-01-11 op {
244 3c1cf9d0 2021-01-11 op char *i;
245 3c1cf9d0 2021-01-11 op
246 3c1cf9d0 2021-01-11 op /* 1. replace multiple slashes with a single one */
247 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
248 3c1cf9d0 2021-01-11 op if (*i == '/' && *(i+1) == '/') {
249 3c1cf9d0 2021-01-11 op memmove(i, i+1, strlen(i)); /* move also the \0 */
250 3c1cf9d0 2021-01-11 op i--;
251 3c1cf9d0 2021-01-11 op }
252 3c1cf9d0 2021-01-11 op }
253 3c1cf9d0 2021-01-11 op
254 3c1cf9d0 2021-01-11 op /* 2. eliminate each . path name element */
255 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
256 6a9ae707 2021-01-11 op if ((i == path || *i == '/') &&
257 6a9ae707 2021-01-11 op *i != '.' && i[1] == '.' && i[2] == '/') {
258 3c1cf9d0 2021-01-11 op /* move also the \0 */
259 3c1cf9d0 2021-01-11 op memmove(i, i+2, strlen(i)-1);
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 if (!strcmp(path, ".") || !strcmp(path, "/.")) {
264 3c1cf9d0 2021-01-11 op *path = '\0';
265 3c1cf9d0 2021-01-11 op return 1;
266 3c1cf9d0 2021-01-11 op }
267 3c1cf9d0 2021-01-11 op
268 3c1cf9d0 2021-01-11 op /* 3. eliminate each inner .. along with the preceding non-.. */
269 3c1cf9d0 2021-01-11 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
270 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 3))
271 3c1cf9d0 2021-01-11 op return 0;
272 3c1cf9d0 2021-01-11 op
273 3c1cf9d0 2021-01-11 op /* 4. eliminate trailing ..*/
274 3c1cf9d0 2021-01-11 op if ((i = strstr(path, "..")) != NULL)
275 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 2))
276 3c1cf9d0 2021-01-11 op return 0;
277 3c1cf9d0 2021-01-11 op
278 3c1cf9d0 2021-01-11 op return 1;
279 3c1cf9d0 2021-01-11 op }
280 3c1cf9d0 2021-01-11 op
281 3c1cf9d0 2021-01-11 op static int
282 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
283 3c1cf9d0 2021-01-11 op {
284 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
285 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
286 3c1cf9d0 2021-01-11 op return 1;
287 3c1cf9d0 2021-01-11 op
288 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
289 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
290 3c1cf9d0 2021-01-11 op || *p->iri == '/'
291 3c1cf9d0 2021-01-11 op || *p->iri == '?'
292 19e7bd00 2021-02-06 op || *p->iri == ':'
293 19e7bd00 2021-02-06 op || *p->iri == '@'
294 8404ec30 2021-02-05 op || valid_pct_encoded(p)
295 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
296 3c1cf9d0 2021-01-11 op p->iri++;
297 3c1cf9d0 2021-01-11 op
298 3c1cf9d0 2021-01-11 op if (p->err != NULL)
299 3c1cf9d0 2021-01-11 op return 0;
300 3c1cf9d0 2021-01-11 op
301 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
302 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
303 3c1cf9d0 2021-01-11 op return 0;
304 3c1cf9d0 2021-01-11 op }
305 3c1cf9d0 2021-01-11 op
306 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
307 3c1cf9d0 2021-01-11 op *p->iri = '\0';
308 3c1cf9d0 2021-01-11 op p->iri++;
309 3c1cf9d0 2021-01-11 op }
310 3c1cf9d0 2021-01-11 op
311 3c1cf9d0 2021-01-11 op return 1;
312 3c1cf9d0 2021-01-11 op }
313 3c1cf9d0 2021-01-11 op
314 3c1cf9d0 2021-01-11 op /* don't even bother */
315 3c1cf9d0 2021-01-11 op static int
316 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
317 3c1cf9d0 2021-01-11 op {
318 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
319 3c1cf9d0 2021-01-11 op return 1;
320 3c1cf9d0 2021-01-11 op }
321 3c1cf9d0 2021-01-11 op
322 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
323 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
324 3c1cf9d0 2021-01-11 op static int
325 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
326 3c1cf9d0 2021-01-11 op {
327 3c1cf9d0 2021-01-11 op char c;
328 42bbdc79 2021-01-21 op
329 42bbdc79 2021-01-21 op /* trim initial slashes */
330 42bbdc79 2021-01-21 op while (*p->iri == '/')
331 42bbdc79 2021-01-21 op p->iri++;
332 3c1cf9d0 2021-01-11 op
333 3c1cf9d0 2021-01-11 op p->parsed->path = p->iri;
334 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
335 3c1cf9d0 2021-01-11 op p->parsed->query = p->parsed->fragment = p->iri;
336 3c1cf9d0 2021-01-11 op return 1;
337 3c1cf9d0 2021-01-11 op }
338 3c1cf9d0 2021-01-11 op
339 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
340 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
341 3c1cf9d0 2021-01-11 op || *p->iri == '/'
342 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
343 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
344 3c1cf9d0 2021-01-11 op p->iri++;
345 3c1cf9d0 2021-01-11 op
346 3c1cf9d0 2021-01-11 op if (p->err != NULL)
347 3c1cf9d0 2021-01-11 op return 0;
348 3c1cf9d0 2021-01-11 op
349 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
350 3c1cf9d0 2021-01-11 op p->err = "illegal character in path";
351 3c1cf9d0 2021-01-11 op return 0;
352 3c1cf9d0 2021-01-11 op }
353 3c1cf9d0 2021-01-11 op
354 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
355 3c1cf9d0 2021-01-11 op c = *p->iri;
356 3c1cf9d0 2021-01-11 op *p->iri = '\0';
357 3c1cf9d0 2021-01-11 op p->iri++;
358 3c1cf9d0 2021-01-11 op
359 3c1cf9d0 2021-01-11 op if (c == '#') {
360 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
361 3c1cf9d0 2021-01-11 op return 0;
362 3c1cf9d0 2021-01-11 op } else
363 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
364 3c1cf9d0 2021-01-11 op return 0;
365 3c1cf9d0 2021-01-11 op }
366 3c1cf9d0 2021-01-11 op
367 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
368 3c1cf9d0 2021-01-11 op p->err = "illegal path";
369 3c1cf9d0 2021-01-11 op return 0;
370 3c1cf9d0 2021-01-11 op }
371 3c1cf9d0 2021-01-11 op
372 3c1cf9d0 2021-01-11 op return 1;
373 3c1cf9d0 2021-01-11 op }
374 3c1cf9d0 2021-01-11 op
375 3c1cf9d0 2021-01-11 op int
376 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
377 3c1cf9d0 2021-01-11 op {
378 3c1cf9d0 2021-01-11 op char *end;
379 3c1cf9d0 2021-01-11 op struct parser p = {iri, ret, NULL};
380 3c1cf9d0 2021-01-11 op
381 3c1cf9d0 2021-01-11 op bzero(ret, sizeof(*ret));
382 3c1cf9d0 2021-01-11 op
383 3c1cf9d0 2021-01-11 op /* initialize optional stuff to the empty string */
384 3c1cf9d0 2021-01-11 op end = iri + strlen(iri);
385 57d0d0ad 2021-01-31 op p.parsed->host = end;
386 3c1cf9d0 2021-01-11 op p.parsed->port = end;
387 3c1cf9d0 2021-01-11 op p.parsed->path = end;
388 3c1cf9d0 2021-01-11 op p.parsed->query = end;
389 3c1cf9d0 2021-01-11 op p.parsed->fragment = end;
390 3c1cf9d0 2021-01-11 op
391 3c1cf9d0 2021-01-11 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
392 3c1cf9d0 2021-01-11 op *err_ret = p.err;
393 3c1cf9d0 2021-01-11 op return 0;
394 3c1cf9d0 2021-01-11 op }
395 3c1cf9d0 2021-01-11 op
396 3c1cf9d0 2021-01-11 op *err_ret = NULL;
397 3c1cf9d0 2021-01-11 op return 1;
398 3c1cf9d0 2021-01-11 op }
399 3c1cf9d0 2021-01-11 op
400 3c1cf9d0 2021-01-11 op int
401 c4f682f8 2021-01-27 op trim_req_iri(char *iri, const char **err)
402 3c1cf9d0 2021-01-11 op {
403 3c1cf9d0 2021-01-11 op char *i;
404 3c1cf9d0 2021-01-11 op
405 c4f682f8 2021-01-27 op if ((i = strstr(iri, "\r\n")) == NULL) {
406 c4f682f8 2021-01-27 op *err = "missing CRLF";
407 3c1cf9d0 2021-01-11 op return 0;
408 c4f682f8 2021-01-27 op }
409 3c1cf9d0 2021-01-11 op *i = '\0';
410 3c1cf9d0 2021-01-11 op return 1;
411 2fafa2d2 2021-02-01 op }
412 2fafa2d2 2021-02-01 op
413 2fafa2d2 2021-02-01 op
414 2fafa2d2 2021-02-01 op int
415 2fafa2d2 2021-02-01 op serialize_iri(struct iri *i, char *buf, size_t len)
416 2fafa2d2 2021-02-01 op {
417 2fafa2d2 2021-02-01 op size_t l;
418 2fafa2d2 2021-02-01 op
419 2fafa2d2 2021-02-01 op /* in ex.c we receive empty "" strings as NULL */
420 2fafa2d2 2021-02-01 op if (i->schema == NULL || i->host == NULL) {
421 2fafa2d2 2021-02-01 op memset(buf, 0, len);
422 2fafa2d2 2021-02-01 op return 0;
423 2fafa2d2 2021-02-01 op }
424 2fafa2d2 2021-02-01 op
425 2fafa2d2 2021-02-01 op strlcpy(buf, i->schema, len);
426 2fafa2d2 2021-02-01 op strlcat(buf, "://", len);
427 2fafa2d2 2021-02-01 op strlcat(buf, i->host, len);
428 2fafa2d2 2021-02-01 op strlcat(buf, "/", len);
429 2fafa2d2 2021-02-01 op
430 2fafa2d2 2021-02-01 op if (i->path != NULL)
431 2fafa2d2 2021-02-01 op l = strlcat(buf, i->path, len);
432 2fafa2d2 2021-02-01 op
433 2fafa2d2 2021-02-01 op if (i->query != NULL && *i->query != '\0') {
434 2fafa2d2 2021-02-01 op strlcat(buf, "?", len);
435 2fafa2d2 2021-02-01 op l = strlcat(buf, i->query, len);
436 2fafa2d2 2021-02-01 op }
437 2fafa2d2 2021-02-01 op
438 2fafa2d2 2021-02-01 op return l < len;
439 3c1cf9d0 2021-01-11 op }