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 <ctype.h>
18 33d32d1f 2020-12-25 op #include <string.h>
19 33d32d1f 2020-12-25 op
20 33d32d1f 2020-12-25 op #include "gmid.h"
21 33d32d1f 2020-12-25 op
22 33d32d1f 2020-12-25 op /*
23 33d32d1f 2020-12-25 op * Notes from RFC3986
24 33d32d1f 2020-12-25 op *
25 33d32d1f 2020-12-25 op * => gemini://tanso.net/rfc/rfc3986.txt
26 33d32d1f 2020-12-25 op *
27 33d32d1f 2020-12-25 op *
28 33d32d1f 2020-12-25 op * ABNF
29 33d32d1f 2020-12-25 op * ====
30 33d32d1f 2020-12-25 op *
31 33d32d1f 2020-12-25 op * pct-encoded "%" HEXDIG HEXDIG
32 33d32d1f 2020-12-25 op *
33 33d32d1f 2020-12-25 op * reserved = gen-delims / sub-delimis
34 33d32d1f 2020-12-25 op * gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@"
35 33d32d1f 2020-12-25 op * sub-delims = "!" / "$" / "&" / "'" / "(" / ")"
36 33d32d1f 2020-12-25 op * / "*" / "+" / "," / ";" / "="
37 33d32d1f 2020-12-25 op *
38 33d32d1f 2020-12-25 op * unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
39 33d32d1f 2020-12-25 op *
40 33d32d1f 2020-12-25 op * URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
41 33d32d1f 2020-12-25 op *
42 33d32d1f 2020-12-25 op * hier-part = "//" authority path-abempty
43 33d32d1f 2020-12-25 op * / path-absolute
44 33d32d1f 2020-12-25 op * / path-rootless
45 33d32d1f 2020-12-25 op * / path-empty
46 33d32d1f 2020-12-25 op *
47 33d32d1f 2020-12-25 op * scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
48 33d32d1f 2020-12-25 op *
49 33d32d1f 2020-12-25 op * authority = [ userinfo "@" ] host [ ":" port ]
50 33d32d1f 2020-12-25 op *
51 33d32d1f 2020-12-25 op * (note that userinfo isn't used for Gemini URL)
52 33d32d1f 2020-12-25 op *
53 33d32d1f 2020-12-25 op * host = IP-literal / IPv4address / reg-name
54 33d32d1f 2020-12-25 op * reg-name = *( unreserved / pct-encoded / sub-delims )
55 33d32d1f 2020-12-25 op *
56 33d32d1f 2020-12-25 op * port = *DIGIT
57 33d32d1f 2020-12-25 op *
58 33d32d1f 2020-12-25 op * path = path-abemty ; begins with "/" or is empty
59 33d32d1f 2020-12-25 op * / path-absolute ; begins with "/" but not "//"
60 33d32d1f 2020-12-25 op * / path-noscheme ; begins with a non-colon segment
61 33d32d1f 2020-12-25 op * / path-rootless ; begins with a segment
62 33d32d1f 2020-12-25 op * / path-empty ; zero characters
63 33d32d1f 2020-12-25 op *
64 33d32d1f 2020-12-25 op * path-abemty = *( "/" segment )
65 33d32d1f 2020-12-25 op * path-absolute = "/" [ segment-nz *( "/" segment ) ]
66 33d32d1f 2020-12-25 op * path-noscheme = ; not used
67 33d32d1f 2020-12-25 op * path-rootless = ; not used
68 33d32d1f 2020-12-25 op * path-empty = ; not used
69 33d32d1f 2020-12-25 op *
70 33d32d1f 2020-12-25 op * segment = *pchar
71 33d32d1f 2020-12-25 op * segment-nz = 1*pchar
72 33d32d1f 2020-12-25 op * segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" )
73 33d32d1f 2020-12-25 op * pchar = unreserved / pct-encoded / sub-delims / ":" / "@"
74 33d32d1f 2020-12-25 op *
75 33d32d1f 2020-12-25 op * query = *( pchar / "/" / "?" )
76 33d32d1f 2020-12-25 op *
77 33d32d1f 2020-12-25 op * fragment = *( pchar / "/" / "?" )
78 33d32d1f 2020-12-25 op *
79 33d32d1f 2020-12-25 op *
80 33d32d1f 2020-12-25 op * EXAMPLE
81 33d32d1f 2020-12-25 op * =======
82 33d32d1f 2020-12-25 op *
83 33d32d1f 2020-12-25 op * foo://example.com:8042/over/there?name=ferret#nose
84 33d32d1f 2020-12-25 op * \_/ \______________/\_________/ \_________/ \__/
85 33d32d1f 2020-12-25 op * | | | | |
86 33d32d1f 2020-12-25 op * scheme authority path query fragment
87 33d32d1f 2020-12-25 op *
88 33d32d1f 2020-12-25 op */
89 33d32d1f 2020-12-25 op
90 579e21b3 2021-01-09 op static inline int
91 579e21b3 2021-01-09 op unreserved(int p)
92 579e21b3 2021-01-09 op {
93 579e21b3 2021-01-09 op return isalnum(p)
94 579e21b3 2021-01-09 op || p == '-'
95 579e21b3 2021-01-09 op || p == '.'
96 579e21b3 2021-01-09 op || p == '_'
97 579e21b3 2021-01-09 op || p == '~';
98 579e21b3 2021-01-09 op }
99 33d32d1f 2020-12-25 op
100 579e21b3 2021-01-09 op static inline int
101 579e21b3 2021-01-09 op sub_delimiters(int p)
102 579e21b3 2021-01-09 op {
103 579e21b3 2021-01-09 op return p == '!'
104 579e21b3 2021-01-09 op || p == '$'
105 579e21b3 2021-01-09 op || p == '&'
106 579e21b3 2021-01-09 op || p == '\''
107 579e21b3 2021-01-09 op || p == '('
108 579e21b3 2021-01-09 op || p == ')'
109 579e21b3 2021-01-09 op || p == '*'
110 579e21b3 2021-01-09 op || p == '+'
111 579e21b3 2021-01-09 op || p == ','
112 579e21b3 2021-01-09 op || p == ';'
113 579e21b3 2021-01-09 op || p == '=';
114 579e21b3 2021-01-09 op }
115 33d32d1f 2020-12-25 op
116 33d32d1f 2020-12-25 op static int
117 33d32d1f 2020-12-25 op parse_pct_encoded(struct parser *p)
118 33d32d1f 2020-12-25 op {
119 33d32d1f 2020-12-25 op if (*p->uri != '%')
120 33d32d1f 2020-12-25 op return 0;
121 33d32d1f 2020-12-25 op
122 33d32d1f 2020-12-25 op if (!isxdigit(*(p->uri+1)) || !isxdigit(*(p->uri+2))) {
123 33d32d1f 2020-12-25 op p->err = "illegal percent-encoding";
124 33d32d1f 2020-12-25 op return 0;
125 33d32d1f 2020-12-25 op }
126 33d32d1f 2020-12-25 op
127 33d32d1f 2020-12-25 op sscanf(p->uri+1, "%2hhx", p->uri);
128 33d32d1f 2020-12-25 op memmove(p->uri+1, p->uri+3, strlen(p->uri+3)+1);
129 00781742 2020-12-25 op if (*p->uri == '\0') {
130 00781742 2020-12-25 op p->err = "illegal percent-encoding";
131 00781742 2020-12-25 op return 0;
132 00781742 2020-12-25 op }
133 33d32d1f 2020-12-25 op
134 33d32d1f 2020-12-25 op return 1;
135 33d32d1f 2020-12-25 op }
136 33d32d1f 2020-12-25 op
137 33d32d1f 2020-12-25 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
138 33d32d1f 2020-12-25 op static int
139 33d32d1f 2020-12-25 op parse_scheme(struct parser *p)
140 33d32d1f 2020-12-25 op {
141 33d32d1f 2020-12-25 op p->parsed->schema = p->uri;
142 33d32d1f 2020-12-25 op
143 33d32d1f 2020-12-25 op if (!isalpha(*p->uri)) {
144 33d32d1f 2020-12-25 op p->err = "illegal character in scheme";
145 33d32d1f 2020-12-25 op return 0;
146 33d32d1f 2020-12-25 op }
147 33d32d1f 2020-12-25 op
148 33d32d1f 2020-12-25 op p->uri++;
149 33d32d1f 2020-12-25 op while (isalnum(*p->uri)
150 33d32d1f 2020-12-25 op || *p->uri == '+'
151 33d32d1f 2020-12-25 op || *p->uri == '-'
152 33d32d1f 2020-12-25 op || *p->uri == '.')
153 33d32d1f 2020-12-25 op p->uri++;
154 33d32d1f 2020-12-25 op
155 33d32d1f 2020-12-25 op if (*p->uri != ':') {
156 33d32d1f 2020-12-25 op p->err = "illegal character in scheme";
157 33d32d1f 2020-12-25 op return 0;
158 33d32d1f 2020-12-25 op }
159 33d32d1f 2020-12-25 op
160 33d32d1f 2020-12-25 op *p->uri = '\0';
161 33d32d1f 2020-12-25 op if (*(++p->uri) != '/' || *(++p->uri) != '/') {
162 33d32d1f 2020-12-25 op p->err = "invalid marker after scheme";
163 33d32d1f 2020-12-25 op return 0;
164 33d32d1f 2020-12-25 op }
165 33d32d1f 2020-12-25 op
166 33d32d1f 2020-12-25 op p->uri++;
167 33d32d1f 2020-12-25 op return 1;
168 33d32d1f 2020-12-25 op }
169 33d32d1f 2020-12-25 op
170 33d32d1f 2020-12-25 op /* *DIGIT */
171 33d32d1f 2020-12-25 op static int
172 33d32d1f 2020-12-25 op parse_port(struct parser *p)
173 33d32d1f 2020-12-25 op {
174 33d32d1f 2020-12-25 op uint32_t i = 0;
175 33d32d1f 2020-12-25 op
176 33d32d1f 2020-12-25 op p->parsed->port = p->uri;
177 33d32d1f 2020-12-25 op
178 33d32d1f 2020-12-25 op for (; isdigit(*p->uri); p->uri++) {
179 33d32d1f 2020-12-25 op i = i * 10 + *p->uri - '0';
180 33d32d1f 2020-12-25 op if (i > UINT16_MAX) {
181 33d32d1f 2020-12-25 op p->err = "port number too large";
182 33d32d1f 2020-12-25 op return 0;
183 33d32d1f 2020-12-25 op }
184 33d32d1f 2020-12-25 op }
185 33d32d1f 2020-12-25 op
186 33d32d1f 2020-12-25 op if (*p->uri != '/' && *p->uri != '\0') {
187 33d32d1f 2020-12-25 op p->err = "illegal character in port number";
188 33d32d1f 2020-12-25 op return 0;
189 33d32d1f 2020-12-25 op }
190 33d32d1f 2020-12-25 op
191 33d32d1f 2020-12-25 op p->parsed->port_no = i;
192 33d32d1f 2020-12-25 op
193 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
194 33d32d1f 2020-12-25 op *p->uri = '\0';
195 33d32d1f 2020-12-25 op p->uri++;
196 33d32d1f 2020-12-25 op }
197 33d32d1f 2020-12-25 op
198 33d32d1f 2020-12-25 op return 1;
199 33d32d1f 2020-12-25 op }
200 33d32d1f 2020-12-25 op
201 33d32d1f 2020-12-25 op /* TODO: add support for ip-literal and ipv4addr ? */
202 33d32d1f 2020-12-25 op /* *( unreserved / sub-delims / pct-encoded ) */
203 33d32d1f 2020-12-25 op static int
204 33d32d1f 2020-12-25 op parse_authority(struct parser *p)
205 33d32d1f 2020-12-25 op {
206 33d32d1f 2020-12-25 op p->parsed->host = p->uri;
207 33d32d1f 2020-12-25 op
208 579e21b3 2021-01-09 op while (unreserved(*p->uri)
209 579e21b3 2021-01-09 op || sub_delimiters(*p->uri)
210 33d32d1f 2020-12-25 op || parse_pct_encoded(p))
211 33d32d1f 2020-12-25 op p->uri++;
212 33d32d1f 2020-12-25 op
213 00781742 2020-12-25 op if (p->err != NULL)
214 00781742 2020-12-25 op return 0;
215 00781742 2020-12-25 op
216 33d32d1f 2020-12-25 op if (*p->uri == ':') {
217 33d32d1f 2020-12-25 op *p->uri = '\0';
218 33d32d1f 2020-12-25 op p->uri++;
219 33d32d1f 2020-12-25 op return parse_port(p);
220 33d32d1f 2020-12-25 op }
221 33d32d1f 2020-12-25 op
222 33d32d1f 2020-12-25 op if (*p->uri == '/') {
223 33d32d1f 2020-12-25 op *p->uri = '\0';
224 33d32d1f 2020-12-25 op p->uri++;
225 33d32d1f 2020-12-25 op return 1;
226 33d32d1f 2020-12-25 op }
227 33d32d1f 2020-12-25 op
228 33d32d1f 2020-12-25 op if (*p->uri == '\0')
229 33d32d1f 2020-12-25 op return 1;
230 33d32d1f 2020-12-25 op
231 33d32d1f 2020-12-25 op p->err = "illegal character in authority section";
232 33d32d1f 2020-12-25 op return 0;
233 33d32d1f 2020-12-25 op }
234 33d32d1f 2020-12-25 op
235 33d32d1f 2020-12-25 op /* Routine for path_clean. Elide the pointed .. with the preceding
236 33d32d1f 2020-12-25 op * element. Return 0 if it's not possible. incr is the length of
237 33d32d1f 2020-12-25 op * the increment, 3 for ../ and 2 for .. */
238 33d32d1f 2020-12-25 op static int
239 33d32d1f 2020-12-25 op path_elide_dotdot(char *path, char *i, int incr)
240 33d32d1f 2020-12-25 op {
241 33d32d1f 2020-12-25 op char *j;
242 33d32d1f 2020-12-25 op
243 33d32d1f 2020-12-25 op if (i == path)
244 33d32d1f 2020-12-25 op return 0;
245 33d32d1f 2020-12-25 op for (j = i-2; j != path && *j != '/'; j--)
246 33d32d1f 2020-12-25 op /* noop */ ;
247 33d32d1f 2020-12-25 op if (*j == '/')
248 33d32d1f 2020-12-25 op j++;
249 33d32d1f 2020-12-25 op i += incr;
250 33d32d1f 2020-12-25 op memmove(j, i, strlen(i)+1);
251 33d32d1f 2020-12-25 op return 1;
252 33d32d1f 2020-12-25 op }
253 33d32d1f 2020-12-25 op
254 33d32d1f 2020-12-25 op /*
255 33d32d1f 2020-12-25 op * Use an algorithm similar to the one implemented in go' path.Clean:
256 33d32d1f 2020-12-25 op *
257 33d32d1f 2020-12-25 op * 1. Replace multiple slashes with a single slash
258 33d32d1f 2020-12-25 op * 2. Eliminate each . path name element
259 33d32d1f 2020-12-25 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
260 33d32d1f 2020-12-25 op * 4. Eliminate trailing .. if possible or error (go would only discard)
261 33d32d1f 2020-12-25 op *
262 33d32d1f 2020-12-25 op * Unlike path.Clean, this function return the empty string if the
263 33d32d1f 2020-12-25 op * original path is equivalent to "/".
264 33d32d1f 2020-12-25 op */
265 33d32d1f 2020-12-25 op static int
266 33d32d1f 2020-12-25 op path_clean(char *path)
267 33d32d1f 2020-12-25 op {
268 33d32d1f 2020-12-25 op char *i;
269 33d32d1f 2020-12-25 op
270 33d32d1f 2020-12-25 op /* 1. replace multiple slashes with a single one */
271 33d32d1f 2020-12-25 op for (i = path; *i; ++i) {
272 33d32d1f 2020-12-25 op if (*i == '/' && *(i+1) == '/') {
273 33d32d1f 2020-12-25 op memmove(i, i+1, strlen(i)); /* move also the \0 */
274 33d32d1f 2020-12-25 op i--;
275 33d32d1f 2020-12-25 op }
276 33d32d1f 2020-12-25 op }
277 33d32d1f 2020-12-25 op
278 33d32d1f 2020-12-25 op /* 2. eliminate each . path name element */
279 33d32d1f 2020-12-25 op for (i = path; *i; ++i) {
280 33d32d1f 2020-12-25 op if ((i == path || *i == '/') && *(i+1) == '.' &&
281 33d32d1f 2020-12-25 op *(i+2) == '/') {
282 33d32d1f 2020-12-25 op /* move also the \0 */
283 33d32d1f 2020-12-25 op memmove(i, i+2, strlen(i)-1);
284 33d32d1f 2020-12-25 op i--;
285 33d32d1f 2020-12-25 op }
286 33d32d1f 2020-12-25 op }
287 33d32d1f 2020-12-25 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
288 33d32d1f 2020-12-25 op *path = '\0';
289 33d32d1f 2020-12-25 op return 1;
290 33d32d1f 2020-12-25 op }
291 33d32d1f 2020-12-25 op
292 33d32d1f 2020-12-25 op /* 3. eliminate each inner .. along with the preceding non-.. */
293 33d32d1f 2020-12-25 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
294 33d32d1f 2020-12-25 op if (!path_elide_dotdot(path, i, 3))
295 33d32d1f 2020-12-25 op return 0;
296 33d32d1f 2020-12-25 op
297 33d32d1f 2020-12-25 op /* 4. eliminate trailing ..*/
298 33d32d1f 2020-12-25 op if ((i = strstr(path, "..")) != NULL)
299 33d32d1f 2020-12-25 op if (!path_elide_dotdot(path, i, 2))
300 33d32d1f 2020-12-25 op return 0;
301 33d32d1f 2020-12-25 op
302 33d32d1f 2020-12-25 op return 1;
303 33d32d1f 2020-12-25 op }
304 33d32d1f 2020-12-25 op
305 33d32d1f 2020-12-25 op static int
306 33d32d1f 2020-12-25 op parse_query(struct parser *p)
307 33d32d1f 2020-12-25 op {
308 33d32d1f 2020-12-25 op p->parsed->query = p->uri;
309 33d32d1f 2020-12-25 op if (*p->uri == '\0')
310 33d32d1f 2020-12-25 op return 1;
311 33d32d1f 2020-12-25 op
312 579e21b3 2021-01-09 op while (unreserved(*p->uri)
313 579e21b3 2021-01-09 op || sub_delimiters(*p->uri)
314 33d32d1f 2020-12-25 op || *p->uri == '/'
315 33d32d1f 2020-12-25 op || *p->uri == '?'
316 df6ca41d 2020-12-25 op || parse_pct_encoded(p)
317 df6ca41d 2020-12-25 op || valid_multibyte_utf8(p))
318 33d32d1f 2020-12-25 op p->uri++;
319 33d32d1f 2020-12-25 op
320 00781742 2020-12-25 op if (p->err != NULL)
321 00781742 2020-12-25 op return 0;
322 00781742 2020-12-25 op
323 33d32d1f 2020-12-25 op if (*p->uri != '\0' && *p->uri != '#') {
324 33d32d1f 2020-12-25 op p->err = "illegal character in query";
325 33d32d1f 2020-12-25 op return 0;
326 33d32d1f 2020-12-25 op }
327 33d32d1f 2020-12-25 op
328 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
329 33d32d1f 2020-12-25 op *p->uri = '\0';
330 33d32d1f 2020-12-25 op p->uri++;
331 33d32d1f 2020-12-25 op }
332 33d32d1f 2020-12-25 op
333 33d32d1f 2020-12-25 op return 1;
334 33d32d1f 2020-12-25 op }
335 33d32d1f 2020-12-25 op
336 33d32d1f 2020-12-25 op /* don't even bother */
337 33d32d1f 2020-12-25 op static int
338 33d32d1f 2020-12-25 op parse_fragment(struct parser *p)
339 33d32d1f 2020-12-25 op {
340 33d32d1f 2020-12-25 op p->parsed->fragment = p->uri;
341 33d32d1f 2020-12-25 op return 1;
342 33d32d1f 2020-12-25 op }
343 33d32d1f 2020-12-25 op
344 33d32d1f 2020-12-25 op /* XXX: is it too broad? */
345 33d32d1f 2020-12-25 op /* *(pchar / "/") */
346 33d32d1f 2020-12-25 op static int
347 33d32d1f 2020-12-25 op parse_path(struct parser *p)
348 33d32d1f 2020-12-25 op {
349 33d32d1f 2020-12-25 op char c;
350 33d32d1f 2020-12-25 op
351 33d32d1f 2020-12-25 op p->parsed->path = p->uri;
352 33d32d1f 2020-12-25 op if (*p->uri == '\0') {
353 33d32d1f 2020-12-25 op p->parsed->query = p->parsed->fragment = p->uri;
354 33d32d1f 2020-12-25 op return 1;
355 33d32d1f 2020-12-25 op }
356 33d32d1f 2020-12-25 op
357 579e21b3 2021-01-09 op while (unreserved(*p->uri)
358 579e21b3 2021-01-09 op || sub_delimiters(*p->uri)
359 33d32d1f 2020-12-25 op || *p->uri == '/'
360 df6ca41d 2020-12-25 op || parse_pct_encoded(p)
361 df6ca41d 2020-12-25 op || valid_multibyte_utf8(p))
362 33d32d1f 2020-12-25 op p->uri++;
363 33d32d1f 2020-12-25 op
364 00781742 2020-12-25 op if (p->err != NULL)
365 00781742 2020-12-25 op return 0;
366 00781742 2020-12-25 op
367 33d32d1f 2020-12-25 op if (*p->uri != '\0' && *p->uri != '?' && *p->uri != '#') {
368 33d32d1f 2020-12-25 op p->err = "illegal character in path";
369 33d32d1f 2020-12-25 op return 0;
370 33d32d1f 2020-12-25 op }
371 33d32d1f 2020-12-25 op
372 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
373 33d32d1f 2020-12-25 op c = *p->uri;
374 33d32d1f 2020-12-25 op *p->uri = '\0';
375 33d32d1f 2020-12-25 op p->uri++;
376 33d32d1f 2020-12-25 op
377 33d32d1f 2020-12-25 op if (c == '#') {
378 33d32d1f 2020-12-25 op if (!parse_fragment(p))
379 33d32d1f 2020-12-25 op return 0;
380 33d32d1f 2020-12-25 op } else
381 33d32d1f 2020-12-25 op if (!parse_query(p) || !parse_fragment(p))
382 33d32d1f 2020-12-25 op return 0;
383 33d32d1f 2020-12-25 op }
384 33d32d1f 2020-12-25 op
385 33d32d1f 2020-12-25 op if (!path_clean(p->parsed->path)) {
386 33d32d1f 2020-12-25 op p->err = "illegal path";
387 33d32d1f 2020-12-25 op return 0;
388 33d32d1f 2020-12-25 op }
389 33d32d1f 2020-12-25 op
390 33d32d1f 2020-12-25 op return 1;
391 33d32d1f 2020-12-25 op }
392 33d32d1f 2020-12-25 op
393 33d32d1f 2020-12-25 op int
394 33d32d1f 2020-12-25 op parse_uri(char *uri, struct uri *ret, const char **err_ret)
395 33d32d1f 2020-12-25 op {
396 33d32d1f 2020-12-25 op char *end;
397 33d32d1f 2020-12-25 op struct parser p = {uri, ret, NULL};
398 33d32d1f 2020-12-25 op
399 33d32d1f 2020-12-25 op bzero(ret, sizeof(*ret));
400 33d32d1f 2020-12-25 op
401 33d32d1f 2020-12-25 op /* initialize optional stuff to the empty string */
402 33d32d1f 2020-12-25 op end = uri + strlen(uri);
403 33d32d1f 2020-12-25 op p.parsed->port = end;
404 33d32d1f 2020-12-25 op p.parsed->path = end;
405 33d32d1f 2020-12-25 op p.parsed->query = end;
406 33d32d1f 2020-12-25 op p.parsed->fragment = end;
407 33d32d1f 2020-12-25 op
408 33d32d1f 2020-12-25 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
409 33d32d1f 2020-12-25 op *err_ret = p.err;
410 33d32d1f 2020-12-25 op return 0;
411 33d32d1f 2020-12-25 op }
412 33d32d1f 2020-12-25 op
413 33d32d1f 2020-12-25 op *err_ret = NULL;
414 33d32d1f 2020-12-25 op return 1;
415 33d32d1f 2020-12-25 op }
416 33d32d1f 2020-12-25 op
417 33d32d1f 2020-12-25 op int
418 33d32d1f 2020-12-25 op trim_req_uri(char *uri)
419 33d32d1f 2020-12-25 op {
420 33d32d1f 2020-12-25 op char *i;
421 33d32d1f 2020-12-25 op
422 33d32d1f 2020-12-25 op if ((i = strstr(uri, "\r\n")) == NULL)
423 33d32d1f 2020-12-25 op return 0;
424 33d32d1f 2020-12-25 op *i = '\0';
425 33d32d1f 2020-12-25 op return 1;
426 33d32d1f 2020-12-25 op }