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 3c1cf9d0 2021-01-11 op }
 47 3c1cf9d0 2021-01-11 op 
 48 3c1cf9d0 2021-01-11 op static int
 49 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
 50 3c1cf9d0 2021-01-11 op {
 51 3c1cf9d0 2021-01-11 op 	if (*p->iri != '%')
 52 3c1cf9d0 2021-01-11 op 		return 0;
 53 3c1cf9d0 2021-01-11 op 
 54 3c1cf9d0 2021-01-11 op 	if (!isxdigit(*(p->iri+1)) || !isxdigit(*(p->iri+2))) {
 55 3c1cf9d0 2021-01-11 op 		p->err = "illegal percent-encoding";
 56 3c1cf9d0 2021-01-11 op 		return 0;
 57 3c1cf9d0 2021-01-11 op 	}
 58 3c1cf9d0 2021-01-11 op 
 59 3c1cf9d0 2021-01-11 op 	sscanf(p->iri+1, "%2hhx", p->iri);
 60 3c1cf9d0 2021-01-11 op 	memmove(p->iri+1, p->iri+3, strlen(p->iri+3)+1);
 61 3c1cf9d0 2021-01-11 op 	if (*p->iri == '\0') {
 62 3c1cf9d0 2021-01-11 op 		p->err = "illegal percent-encoding";
 63 3c1cf9d0 2021-01-11 op 		return 0;
 64 3c1cf9d0 2021-01-11 op 	}
 65 3c1cf9d0 2021-01-11 op 
 66 3c1cf9d0 2021-01-11 op 	return 1;
 67 3c1cf9d0 2021-01-11 op }
 68 3c1cf9d0 2021-01-11 op 
 69 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
 70 3c1cf9d0 2021-01-11 op static int
 71 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
 72 3c1cf9d0 2021-01-11 op {
 73 3c1cf9d0 2021-01-11 op 	p->parsed->schema = p->iri;
 74 3c1cf9d0 2021-01-11 op 
 75 3c1cf9d0 2021-01-11 op 	if (!isalpha(*p->iri)) {
 76 3c1cf9d0 2021-01-11 op 		p->err = "illegal character in scheme";
 77 3c1cf9d0 2021-01-11 op 		return 0;
 78 3c1cf9d0 2021-01-11 op 	}
 79 3c1cf9d0 2021-01-11 op 
 80 de428fff 2021-01-13 op 	do {
 81 de428fff 2021-01-13 op 		/* normalize the scheme (i.e. lowercase it)
 82 de428fff 2021-01-13 op 		 *
 83 de428fff 2021-01-13 op 		 * XXX: since we cannot have good things, tolower
 84 de428fff 2021-01-13 op 		 * depends on the LC_CTYPE locale.  The good things is
 85 de428fff 2021-01-13 op 		 * that we're sure p->iri points to something in the
 86 de428fff 2021-01-13 op 		 * ASCII range, so it shouldn't do weird stuff. */
 87 de428fff 2021-01-13 op 		*p->iri = tolower(*p->iri);
 88 de428fff 2021-01-13 op 		p->iri++;
 89 de428fff 2021-01-13 op 	} while (isalnum(*p->iri)
 90 3c1cf9d0 2021-01-11 op 	    || *p->iri == '+'
 91 3c1cf9d0 2021-01-11 op 	    || *p->iri == '-'
 92 de428fff 2021-01-13 op 	    || *p->iri == '.');
 93 3c1cf9d0 2021-01-11 op 
 94 3c1cf9d0 2021-01-11 op 	if (*p->iri != ':') {
 95 3c1cf9d0 2021-01-11 op 		p->err = "illegal character in scheme";
 96 3c1cf9d0 2021-01-11 op 		return 0;
 97 3c1cf9d0 2021-01-11 op 	}
 98 3c1cf9d0 2021-01-11 op 
 99 3c1cf9d0 2021-01-11 op 	*p->iri = '\0';
100 3c1cf9d0 2021-01-11 op 	if (*(++p->iri) != '/' || *(++p->iri) != '/') {
101 3c1cf9d0 2021-01-11 op 		p->err = "invalid marker after scheme";
102 3c1cf9d0 2021-01-11 op 		return 0;
103 3c1cf9d0 2021-01-11 op 	}
104 3c1cf9d0 2021-01-11 op 
105 3c1cf9d0 2021-01-11 op 	p->iri++;
106 3c1cf9d0 2021-01-11 op 	return 1;
107 3c1cf9d0 2021-01-11 op }
108 3c1cf9d0 2021-01-11 op 
109 3c1cf9d0 2021-01-11 op /* *DIGIT */
110 3c1cf9d0 2021-01-11 op static int
111 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
112 3c1cf9d0 2021-01-11 op {
113 3c1cf9d0 2021-01-11 op 	uint32_t i = 0;
114 3c1cf9d0 2021-01-11 op 
115 3c1cf9d0 2021-01-11 op 	p->parsed->port = p->iri;
116 3c1cf9d0 2021-01-11 op 
117 3c1cf9d0 2021-01-11 op 	for (; isdigit(*p->iri); p->iri++) {
118 3c1cf9d0 2021-01-11 op 		i = i * 10 + *p->iri - '0';
119 3c1cf9d0 2021-01-11 op 		if (i > UINT16_MAX) {
120 3c1cf9d0 2021-01-11 op 			p->err = "port number too large";
121 3c1cf9d0 2021-01-11 op 			return 0;
122 3c1cf9d0 2021-01-11 op 		}
123 3c1cf9d0 2021-01-11 op 	}
124 3c1cf9d0 2021-01-11 op 
125 3c1cf9d0 2021-01-11 op 	if (*p->iri != '/' && *p->iri != '\0') {
126 3c1cf9d0 2021-01-11 op 		p->err = "illegal character in port number";
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->parsed->port_no = i;
131 3c1cf9d0 2021-01-11 op 
132 3c1cf9d0 2021-01-11 op 	if (*p->iri != '\0') {
133 3c1cf9d0 2021-01-11 op 		*p->iri = '\0';
134 3c1cf9d0 2021-01-11 op 		p->iri++;
135 3c1cf9d0 2021-01-11 op 	}
136 3c1cf9d0 2021-01-11 op 
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 /* TODO: add support for ip-literal and ipv4addr ? */
141 3c1cf9d0 2021-01-11 op /* *( unreserved / sub-delims / pct-encoded ) */
142 3c1cf9d0 2021-01-11 op static int
143 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
144 3c1cf9d0 2021-01-11 op {
145 3c1cf9d0 2021-01-11 op 	p->parsed->host = p->iri;
146 3c1cf9d0 2021-01-11 op 
147 3c1cf9d0 2021-01-11 op 	while (unreserved(*p->iri)
148 3c1cf9d0 2021-01-11 op 	    || sub_delimiters(*p->iri)
149 e4d82bec 2021-01-15 op 	    || parse_pct_encoded(p)) {
150 e4d82bec 2021-01-15 op 		/* normalize the host name. */
151 e4d82bec 2021-01-15 op 		if (*p->iri < 0x7F)
152 e4d82bec 2021-01-15 op 			*p->iri = tolower(*p->iri);
153 3c1cf9d0 2021-01-11 op 		p->iri++;
154 e4d82bec 2021-01-15 op 	}
155 3c1cf9d0 2021-01-11 op 
156 3c1cf9d0 2021-01-11 op 	if (p->err != NULL)
157 3c1cf9d0 2021-01-11 op 		return 0;
158 3c1cf9d0 2021-01-11 op 
159 3c1cf9d0 2021-01-11 op 	if (*p->iri == ':') {
160 3c1cf9d0 2021-01-11 op 		*p->iri = '\0';
161 3c1cf9d0 2021-01-11 op 		p->iri++;
162 3c1cf9d0 2021-01-11 op 		return parse_port(p);
163 3c1cf9d0 2021-01-11 op 	}
164 3c1cf9d0 2021-01-11 op 
165 3c1cf9d0 2021-01-11 op 	if (*p->iri == '/') {
166 3c1cf9d0 2021-01-11 op 		*p->iri = '\0';
167 3c1cf9d0 2021-01-11 op 		p->iri++;
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 	if (*p->iri == '\0')
172 3c1cf9d0 2021-01-11 op 		return 1;
173 3c1cf9d0 2021-01-11 op 
174 3c1cf9d0 2021-01-11 op 	p->err = "illegal character in authority section";
175 3c1cf9d0 2021-01-11 op 	return 0;
176 3c1cf9d0 2021-01-11 op }
177 3c1cf9d0 2021-01-11 op 
178 3c1cf9d0 2021-01-11 op /* Routine for path_clean.  Elide the pointed .. with the preceding
179 3c1cf9d0 2021-01-11 op  * element.  Return 0 if it's not possible.  incr is the length of
180 3c1cf9d0 2021-01-11 op  * the increment, 3 for ../ and 2 for .. */
181 3c1cf9d0 2021-01-11 op static int
182 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
183 3c1cf9d0 2021-01-11 op {
184 3c1cf9d0 2021-01-11 op 	char *j;
185 3c1cf9d0 2021-01-11 op 
186 3c1cf9d0 2021-01-11 op 	if (i == path)
187 3c1cf9d0 2021-01-11 op 		return 0;
188 3c1cf9d0 2021-01-11 op 	for (j = i-2; j != path && *j != '/'; j--)
189 3c1cf9d0 2021-01-11 op                 /* noop */ ;
190 3c1cf9d0 2021-01-11 op 	if (*j == '/')
191 3c1cf9d0 2021-01-11 op 		j++;
192 3c1cf9d0 2021-01-11 op 	i += incr;
193 3c1cf9d0 2021-01-11 op 	memmove(j, i, strlen(i)+1);
194 3c1cf9d0 2021-01-11 op 	return 1;
195 3c1cf9d0 2021-01-11 op }
196 3c1cf9d0 2021-01-11 op 
197 3c1cf9d0 2021-01-11 op /*
198 3c1cf9d0 2021-01-11 op  * Use an algorithm similar to the one implemented in go' path.Clean:
199 3c1cf9d0 2021-01-11 op  *
200 3c1cf9d0 2021-01-11 op  * 1. Replace multiple slashes with a single slash
201 3c1cf9d0 2021-01-11 op  * 2. Eliminate each . path name element
202 3c1cf9d0 2021-01-11 op  * 3. Eliminate each inner .. along with the non-.. element that precedes it
203 3c1cf9d0 2021-01-11 op  * 4. Eliminate trailing .. if possible or error (go would only discard)
204 3c1cf9d0 2021-01-11 op  *
205 3c1cf9d0 2021-01-11 op  * Unlike path.Clean, this function return the empty string if the
206 3c1cf9d0 2021-01-11 op  * original path is equivalent to "/".
207 3c1cf9d0 2021-01-11 op  */
208 3c1cf9d0 2021-01-11 op static int
209 3c1cf9d0 2021-01-11 op path_clean(char *path)
210 3c1cf9d0 2021-01-11 op {
211 3c1cf9d0 2021-01-11 op 	char *i;
212 3c1cf9d0 2021-01-11 op 
213 3c1cf9d0 2021-01-11 op 	/* 1. replace multiple slashes with a single one */
214 3c1cf9d0 2021-01-11 op 	for (i = path; *i; ++i) {
215 3c1cf9d0 2021-01-11 op 		if (*i == '/' && *(i+1) == '/') {
216 3c1cf9d0 2021-01-11 op 			memmove(i, i+1, strlen(i)); /* move also the \0 */
217 3c1cf9d0 2021-01-11 op 			i--;
218 3c1cf9d0 2021-01-11 op 		}
219 3c1cf9d0 2021-01-11 op 	}
220 3c1cf9d0 2021-01-11 op 
221 3c1cf9d0 2021-01-11 op 	/* 2. eliminate each . path name element */
222 3c1cf9d0 2021-01-11 op 	for (i = path; *i; ++i) {
223 6a9ae707 2021-01-11 op 		if ((i == path || *i == '/') &&
224 6a9ae707 2021-01-11 op 		    *i != '.' && i[1] == '.' && i[2] == '/') {
225 3c1cf9d0 2021-01-11 op 			/* move also the \0 */
226 3c1cf9d0 2021-01-11 op 			memmove(i, i+2, strlen(i)-1);
227 3c1cf9d0 2021-01-11 op 			i--;
228 3c1cf9d0 2021-01-11 op 		}
229 3c1cf9d0 2021-01-11 op 	}
230 3c1cf9d0 2021-01-11 op 	if (!strcmp(path, ".") || !strcmp(path, "/.")) {
231 3c1cf9d0 2021-01-11 op 		*path = '\0';
232 3c1cf9d0 2021-01-11 op 		return 1;
233 3c1cf9d0 2021-01-11 op 	}
234 3c1cf9d0 2021-01-11 op 
235 3c1cf9d0 2021-01-11 op 	/* 3. eliminate each inner .. along with the preceding non-.. */
236 3c1cf9d0 2021-01-11 op 	for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
237 3c1cf9d0 2021-01-11 op 		if (!path_elide_dotdot(path, i, 3))
238 3c1cf9d0 2021-01-11 op 			return 0;
239 3c1cf9d0 2021-01-11 op 
240 3c1cf9d0 2021-01-11 op 	/* 4. eliminate trailing ..*/
241 3c1cf9d0 2021-01-11 op 	if ((i = strstr(path, "..")) != NULL)
242 3c1cf9d0 2021-01-11 op 		if (!path_elide_dotdot(path, i, 2))
243 3c1cf9d0 2021-01-11 op 			return 0;
244 3c1cf9d0 2021-01-11 op 
245 3c1cf9d0 2021-01-11 op 	return 1;
246 3c1cf9d0 2021-01-11 op }
247 3c1cf9d0 2021-01-11 op 
248 3c1cf9d0 2021-01-11 op static int
249 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
250 3c1cf9d0 2021-01-11 op {
251 3c1cf9d0 2021-01-11 op 	p->parsed->query = p->iri;
252 3c1cf9d0 2021-01-11 op 	if (*p->iri == '\0')
253 3c1cf9d0 2021-01-11 op 		return 1;
254 3c1cf9d0 2021-01-11 op 
255 3c1cf9d0 2021-01-11 op 	while (unreserved(*p->iri)
256 3c1cf9d0 2021-01-11 op 	    || sub_delimiters(*p->iri)
257 3c1cf9d0 2021-01-11 op 	    || *p->iri == '/'
258 3c1cf9d0 2021-01-11 op 	    || *p->iri == '?'
259 3c1cf9d0 2021-01-11 op 	    || parse_pct_encoded(p)
260 3c1cf9d0 2021-01-11 op 	    || valid_multibyte_utf8(p))
261 3c1cf9d0 2021-01-11 op 		p->iri++;
262 3c1cf9d0 2021-01-11 op 
263 3c1cf9d0 2021-01-11 op 	if (p->err != NULL)
264 3c1cf9d0 2021-01-11 op 		return 0;
265 3c1cf9d0 2021-01-11 op 
266 3c1cf9d0 2021-01-11 op 	if (*p->iri != '\0' && *p->iri != '#') {
267 3c1cf9d0 2021-01-11 op 		p->err = "illegal character in query";
268 3c1cf9d0 2021-01-11 op 		return 0;
269 3c1cf9d0 2021-01-11 op 	}
270 3c1cf9d0 2021-01-11 op 
271 3c1cf9d0 2021-01-11 op 	if (*p->iri != '\0') {
272 3c1cf9d0 2021-01-11 op 		*p->iri = '\0';
273 3c1cf9d0 2021-01-11 op 		p->iri++;
274 3c1cf9d0 2021-01-11 op 	}
275 3c1cf9d0 2021-01-11 op 
276 3c1cf9d0 2021-01-11 op 	return 1;
277 3c1cf9d0 2021-01-11 op }
278 3c1cf9d0 2021-01-11 op 
279 3c1cf9d0 2021-01-11 op /* don't even bother */
280 3c1cf9d0 2021-01-11 op static int
281 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
282 3c1cf9d0 2021-01-11 op {
283 3c1cf9d0 2021-01-11 op 	p->parsed->fragment = p->iri;
284 3c1cf9d0 2021-01-11 op 	return 1;
285 3c1cf9d0 2021-01-11 op }
286 3c1cf9d0 2021-01-11 op 
287 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
288 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
289 3c1cf9d0 2021-01-11 op static int
290 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
291 3c1cf9d0 2021-01-11 op {
292 3c1cf9d0 2021-01-11 op 	char c;
293 3c1cf9d0 2021-01-11 op 
294 3c1cf9d0 2021-01-11 op 	p->parsed->path = p->iri;
295 3c1cf9d0 2021-01-11 op 	if (*p->iri == '\0') {
296 3c1cf9d0 2021-01-11 op 		p->parsed->query = p->parsed->fragment = p->iri;
297 3c1cf9d0 2021-01-11 op 		return 1;
298 3c1cf9d0 2021-01-11 op 	}
299 3c1cf9d0 2021-01-11 op 
300 3c1cf9d0 2021-01-11 op 	while (unreserved(*p->iri)
301 3c1cf9d0 2021-01-11 op 	    || sub_delimiters(*p->iri)
302 3c1cf9d0 2021-01-11 op 	    || *p->iri == '/'
303 3c1cf9d0 2021-01-11 op 	    || parse_pct_encoded(p)
304 3c1cf9d0 2021-01-11 op 	    || valid_multibyte_utf8(p))
305 3c1cf9d0 2021-01-11 op 		p->iri++;
306 3c1cf9d0 2021-01-11 op 
307 3c1cf9d0 2021-01-11 op 	if (p->err != NULL)
308 3c1cf9d0 2021-01-11 op 		return 0;
309 3c1cf9d0 2021-01-11 op 
310 3c1cf9d0 2021-01-11 op 	if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
311 3c1cf9d0 2021-01-11 op 		p->err = "illegal character in path";
312 3c1cf9d0 2021-01-11 op 		return 0;
313 3c1cf9d0 2021-01-11 op 	}
314 3c1cf9d0 2021-01-11 op 
315 3c1cf9d0 2021-01-11 op 	if (*p->iri != '\0') {
316 3c1cf9d0 2021-01-11 op 		c = *p->iri;
317 3c1cf9d0 2021-01-11 op 		*p->iri = '\0';
318 3c1cf9d0 2021-01-11 op 		p->iri++;
319 3c1cf9d0 2021-01-11 op 
320 3c1cf9d0 2021-01-11 op 		if (c == '#') {
321 3c1cf9d0 2021-01-11 op 			if (!parse_fragment(p))
322 3c1cf9d0 2021-01-11 op 				return 0;
323 3c1cf9d0 2021-01-11 op 		} else
324 3c1cf9d0 2021-01-11 op 			if (!parse_query(p) || !parse_fragment(p))
325 3c1cf9d0 2021-01-11 op 				return 0;
326 3c1cf9d0 2021-01-11 op 	}
327 3c1cf9d0 2021-01-11 op 
328 3c1cf9d0 2021-01-11 op 	if (!path_clean(p->parsed->path)) {
329 3c1cf9d0 2021-01-11 op 		p->err = "illegal path";
330 3c1cf9d0 2021-01-11 op 		return 0;
331 3c1cf9d0 2021-01-11 op 	}
332 3c1cf9d0 2021-01-11 op 
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 int
337 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
338 3c1cf9d0 2021-01-11 op {
339 3c1cf9d0 2021-01-11 op 	char *end;
340 3c1cf9d0 2021-01-11 op 	struct parser p = {iri, ret, NULL};
341 3c1cf9d0 2021-01-11 op 
342 3c1cf9d0 2021-01-11 op 	bzero(ret, sizeof(*ret));
343 3c1cf9d0 2021-01-11 op 
344 3c1cf9d0 2021-01-11 op 	/* initialize optional stuff to the empty string */
345 3c1cf9d0 2021-01-11 op 	end = iri + strlen(iri);
346 3c1cf9d0 2021-01-11 op 	p.parsed->port = end;
347 3c1cf9d0 2021-01-11 op 	p.parsed->path = end;
348 3c1cf9d0 2021-01-11 op 	p.parsed->query = end;
349 3c1cf9d0 2021-01-11 op 	p.parsed->fragment = end;
350 3c1cf9d0 2021-01-11 op 
351 3c1cf9d0 2021-01-11 op 	if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
352 3c1cf9d0 2021-01-11 op 		*err_ret = p.err;
353 3c1cf9d0 2021-01-11 op 		return 0;
354 3c1cf9d0 2021-01-11 op 	}
355 3c1cf9d0 2021-01-11 op 
356 3c1cf9d0 2021-01-11 op 	*err_ret = NULL;
357 3c1cf9d0 2021-01-11 op 	return 1;
358 3c1cf9d0 2021-01-11 op }
359 3c1cf9d0 2021-01-11 op 
360 3c1cf9d0 2021-01-11 op int
361 3c1cf9d0 2021-01-11 op trim_req_iri(char *iri)
362 3c1cf9d0 2021-01-11 op {
363 3c1cf9d0 2021-01-11 op 	char *i;
364 3c1cf9d0 2021-01-11 op 
365 3c1cf9d0 2021-01-11 op 	if ((i = strstr(iri, "\r\n")) == NULL)
366 3c1cf9d0 2021-01-11 op 		return 0;
367 3c1cf9d0 2021-01-11 op 	*i = '\0';
368 3c1cf9d0 2021-01-11 op 	return 1;
369 3c1cf9d0 2021-01-11 op }