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 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 3c1cf9d0 2021-01-11 op p->iri++;
81 3c1cf9d0 2021-01-11 op while (isalnum(*p->iri)
82 3c1cf9d0 2021-01-11 op || *p->iri == '+'
83 3c1cf9d0 2021-01-11 op || *p->iri == '-'
84 3c1cf9d0 2021-01-11 op || *p->iri == '.')
85 3c1cf9d0 2021-01-11 op p->iri++;
86 3c1cf9d0 2021-01-11 op
87 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
88 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
89 3c1cf9d0 2021-01-11 op return 0;
90 3c1cf9d0 2021-01-11 op }
91 3c1cf9d0 2021-01-11 op
92 3c1cf9d0 2021-01-11 op *p->iri = '\0';
93 3c1cf9d0 2021-01-11 op if (*(++p->iri) != '/' || *(++p->iri) != '/') {
94 3c1cf9d0 2021-01-11 op p->err = "invalid marker after scheme";
95 3c1cf9d0 2021-01-11 op return 0;
96 3c1cf9d0 2021-01-11 op }
97 3c1cf9d0 2021-01-11 op
98 3c1cf9d0 2021-01-11 op p->iri++;
99 3c1cf9d0 2021-01-11 op return 1;
100 3c1cf9d0 2021-01-11 op }
101 3c1cf9d0 2021-01-11 op
102 3c1cf9d0 2021-01-11 op /* *DIGIT */
103 3c1cf9d0 2021-01-11 op static int
104 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
105 3c1cf9d0 2021-01-11 op {
106 3c1cf9d0 2021-01-11 op uint32_t i = 0;
107 3c1cf9d0 2021-01-11 op
108 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
109 3c1cf9d0 2021-01-11 op
110 3c1cf9d0 2021-01-11 op for (; isdigit(*p->iri); p->iri++) {
111 3c1cf9d0 2021-01-11 op i = i * 10 + *p->iri - '0';
112 3c1cf9d0 2021-01-11 op if (i > UINT16_MAX) {
113 3c1cf9d0 2021-01-11 op p->err = "port number too large";
114 3c1cf9d0 2021-01-11 op return 0;
115 3c1cf9d0 2021-01-11 op }
116 3c1cf9d0 2021-01-11 op }
117 3c1cf9d0 2021-01-11 op
118 3c1cf9d0 2021-01-11 op if (*p->iri != '/' && *p->iri != '\0') {
119 3c1cf9d0 2021-01-11 op p->err = "illegal character in port number";
120 3c1cf9d0 2021-01-11 op return 0;
121 3c1cf9d0 2021-01-11 op }
122 3c1cf9d0 2021-01-11 op
123 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
124 3c1cf9d0 2021-01-11 op
125 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
126 3c1cf9d0 2021-01-11 op *p->iri = '\0';
127 3c1cf9d0 2021-01-11 op p->iri++;
128 3c1cf9d0 2021-01-11 op }
129 3c1cf9d0 2021-01-11 op
130 3c1cf9d0 2021-01-11 op return 1;
131 3c1cf9d0 2021-01-11 op }
132 3c1cf9d0 2021-01-11 op
133 3c1cf9d0 2021-01-11 op /* TODO: add support for ip-literal and ipv4addr ? */
134 3c1cf9d0 2021-01-11 op /* *( unreserved / sub-delims / pct-encoded ) */
135 3c1cf9d0 2021-01-11 op static int
136 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
137 3c1cf9d0 2021-01-11 op {
138 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
139 3c1cf9d0 2021-01-11 op
140 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
141 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
142 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p))
143 3c1cf9d0 2021-01-11 op p->iri++;
144 3c1cf9d0 2021-01-11 op
145 3c1cf9d0 2021-01-11 op if (p->err != NULL)
146 3c1cf9d0 2021-01-11 op return 0;
147 3c1cf9d0 2021-01-11 op
148 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
149 3c1cf9d0 2021-01-11 op *p->iri = '\0';
150 3c1cf9d0 2021-01-11 op p->iri++;
151 3c1cf9d0 2021-01-11 op return parse_port(p);
152 3c1cf9d0 2021-01-11 op }
153 3c1cf9d0 2021-01-11 op
154 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
155 3c1cf9d0 2021-01-11 op *p->iri = '\0';
156 3c1cf9d0 2021-01-11 op p->iri++;
157 3c1cf9d0 2021-01-11 op return 1;
158 3c1cf9d0 2021-01-11 op }
159 3c1cf9d0 2021-01-11 op
160 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
161 3c1cf9d0 2021-01-11 op return 1;
162 3c1cf9d0 2021-01-11 op
163 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
164 3c1cf9d0 2021-01-11 op return 0;
165 3c1cf9d0 2021-01-11 op }
166 3c1cf9d0 2021-01-11 op
167 3c1cf9d0 2021-01-11 op /* Routine for path_clean. Elide the pointed .. with the preceding
168 3c1cf9d0 2021-01-11 op * element. Return 0 if it's not possible. incr is the length of
169 3c1cf9d0 2021-01-11 op * the increment, 3 for ../ and 2 for .. */
170 3c1cf9d0 2021-01-11 op static int
171 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
172 3c1cf9d0 2021-01-11 op {
173 3c1cf9d0 2021-01-11 op char *j;
174 3c1cf9d0 2021-01-11 op
175 3c1cf9d0 2021-01-11 op if (i == path)
176 3c1cf9d0 2021-01-11 op return 0;
177 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
178 3c1cf9d0 2021-01-11 op /* noop */ ;
179 3c1cf9d0 2021-01-11 op if (*j == '/')
180 3c1cf9d0 2021-01-11 op j++;
181 3c1cf9d0 2021-01-11 op i += incr;
182 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
183 3c1cf9d0 2021-01-11 op return 1;
184 3c1cf9d0 2021-01-11 op }
185 3c1cf9d0 2021-01-11 op
186 3c1cf9d0 2021-01-11 op /*
187 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
188 3c1cf9d0 2021-01-11 op *
189 3c1cf9d0 2021-01-11 op * 1. Replace multiple slashes with a single slash
190 3c1cf9d0 2021-01-11 op * 2. Eliminate each . path name element
191 3c1cf9d0 2021-01-11 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
192 3c1cf9d0 2021-01-11 op * 4. Eliminate trailing .. if possible or error (go would only discard)
193 3c1cf9d0 2021-01-11 op *
194 3c1cf9d0 2021-01-11 op * Unlike path.Clean, this function return the empty string if the
195 3c1cf9d0 2021-01-11 op * original path is equivalent to "/".
196 3c1cf9d0 2021-01-11 op */
197 3c1cf9d0 2021-01-11 op static int
198 3c1cf9d0 2021-01-11 op path_clean(char *path)
199 3c1cf9d0 2021-01-11 op {
200 3c1cf9d0 2021-01-11 op char *i;
201 3c1cf9d0 2021-01-11 op
202 3c1cf9d0 2021-01-11 op /* 1. replace multiple slashes with a single one */
203 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
204 3c1cf9d0 2021-01-11 op if (*i == '/' && *(i+1) == '/') {
205 3c1cf9d0 2021-01-11 op memmove(i, i+1, strlen(i)); /* move also the \0 */
206 3c1cf9d0 2021-01-11 op i--;
207 3c1cf9d0 2021-01-11 op }
208 3c1cf9d0 2021-01-11 op }
209 3c1cf9d0 2021-01-11 op
210 3c1cf9d0 2021-01-11 op /* 2. eliminate each . path name element */
211 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
212 3c1cf9d0 2021-01-11 op if ((i == path || *i == '/') && *(i+1) == '.' &&
213 3c1cf9d0 2021-01-11 op *(i+2) == '/') {
214 3c1cf9d0 2021-01-11 op /* move also the \0 */
215 3c1cf9d0 2021-01-11 op memmove(i, i+2, strlen(i)-1);
216 3c1cf9d0 2021-01-11 op i--;
217 3c1cf9d0 2021-01-11 op }
218 3c1cf9d0 2021-01-11 op }
219 3c1cf9d0 2021-01-11 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
220 3c1cf9d0 2021-01-11 op *path = '\0';
221 3c1cf9d0 2021-01-11 op return 1;
222 3c1cf9d0 2021-01-11 op }
223 3c1cf9d0 2021-01-11 op
224 3c1cf9d0 2021-01-11 op /* 3. eliminate each inner .. along with the preceding non-.. */
225 3c1cf9d0 2021-01-11 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
226 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 3))
227 3c1cf9d0 2021-01-11 op return 0;
228 3c1cf9d0 2021-01-11 op
229 3c1cf9d0 2021-01-11 op /* 4. eliminate trailing ..*/
230 3c1cf9d0 2021-01-11 op if ((i = strstr(path, "..")) != NULL)
231 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 2))
232 3c1cf9d0 2021-01-11 op return 0;
233 3c1cf9d0 2021-01-11 op
234 3c1cf9d0 2021-01-11 op return 1;
235 3c1cf9d0 2021-01-11 op }
236 3c1cf9d0 2021-01-11 op
237 3c1cf9d0 2021-01-11 op static int
238 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
239 3c1cf9d0 2021-01-11 op {
240 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
241 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
242 3c1cf9d0 2021-01-11 op return 1;
243 3c1cf9d0 2021-01-11 op
244 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
245 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
246 3c1cf9d0 2021-01-11 op || *p->iri == '/'
247 3c1cf9d0 2021-01-11 op || *p->iri == '?'
248 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
249 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
250 3c1cf9d0 2021-01-11 op p->iri++;
251 3c1cf9d0 2021-01-11 op
252 3c1cf9d0 2021-01-11 op if (p->err != NULL)
253 3c1cf9d0 2021-01-11 op return 0;
254 3c1cf9d0 2021-01-11 op
255 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
256 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
257 3c1cf9d0 2021-01-11 op return 0;
258 3c1cf9d0 2021-01-11 op }
259 3c1cf9d0 2021-01-11 op
260 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
261 3c1cf9d0 2021-01-11 op *p->iri = '\0';
262 3c1cf9d0 2021-01-11 op p->iri++;
263 3c1cf9d0 2021-01-11 op }
264 3c1cf9d0 2021-01-11 op
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 /* don't even bother */
269 3c1cf9d0 2021-01-11 op static int
270 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
271 3c1cf9d0 2021-01-11 op {
272 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
273 3c1cf9d0 2021-01-11 op return 1;
274 3c1cf9d0 2021-01-11 op }
275 3c1cf9d0 2021-01-11 op
276 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
277 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
278 3c1cf9d0 2021-01-11 op static int
279 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
280 3c1cf9d0 2021-01-11 op {
281 3c1cf9d0 2021-01-11 op char c;
282 3c1cf9d0 2021-01-11 op
283 3c1cf9d0 2021-01-11 op p->parsed->path = p->iri;
284 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
285 3c1cf9d0 2021-01-11 op p->parsed->query = p->parsed->fragment = p->iri;
286 3c1cf9d0 2021-01-11 op return 1;
287 3c1cf9d0 2021-01-11 op }
288 3c1cf9d0 2021-01-11 op
289 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
290 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
291 3c1cf9d0 2021-01-11 op || *p->iri == '/'
292 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
293 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
294 3c1cf9d0 2021-01-11 op p->iri++;
295 3c1cf9d0 2021-01-11 op
296 3c1cf9d0 2021-01-11 op if (p->err != NULL)
297 3c1cf9d0 2021-01-11 op return 0;
298 3c1cf9d0 2021-01-11 op
299 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
300 3c1cf9d0 2021-01-11 op p->err = "illegal character in path";
301 3c1cf9d0 2021-01-11 op return 0;
302 3c1cf9d0 2021-01-11 op }
303 3c1cf9d0 2021-01-11 op
304 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
305 3c1cf9d0 2021-01-11 op c = *p->iri;
306 3c1cf9d0 2021-01-11 op *p->iri = '\0';
307 3c1cf9d0 2021-01-11 op p->iri++;
308 3c1cf9d0 2021-01-11 op
309 3c1cf9d0 2021-01-11 op if (c == '#') {
310 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
311 3c1cf9d0 2021-01-11 op return 0;
312 3c1cf9d0 2021-01-11 op } else
313 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
314 3c1cf9d0 2021-01-11 op return 0;
315 3c1cf9d0 2021-01-11 op }
316 3c1cf9d0 2021-01-11 op
317 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
318 3c1cf9d0 2021-01-11 op p->err = "illegal path";
319 3c1cf9d0 2021-01-11 op return 0;
320 3c1cf9d0 2021-01-11 op }
321 3c1cf9d0 2021-01-11 op
322 3c1cf9d0 2021-01-11 op return 1;
323 3c1cf9d0 2021-01-11 op }
324 3c1cf9d0 2021-01-11 op
325 3c1cf9d0 2021-01-11 op int
326 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
327 3c1cf9d0 2021-01-11 op {
328 3c1cf9d0 2021-01-11 op char *end;
329 3c1cf9d0 2021-01-11 op struct parser p = {iri, ret, NULL};
330 3c1cf9d0 2021-01-11 op
331 3c1cf9d0 2021-01-11 op bzero(ret, sizeof(*ret));
332 3c1cf9d0 2021-01-11 op
333 3c1cf9d0 2021-01-11 op /* initialize optional stuff to the empty string */
334 3c1cf9d0 2021-01-11 op end = iri + strlen(iri);
335 3c1cf9d0 2021-01-11 op p.parsed->port = end;
336 3c1cf9d0 2021-01-11 op p.parsed->path = end;
337 3c1cf9d0 2021-01-11 op p.parsed->query = end;
338 3c1cf9d0 2021-01-11 op p.parsed->fragment = end;
339 3c1cf9d0 2021-01-11 op
340 3c1cf9d0 2021-01-11 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
341 3c1cf9d0 2021-01-11 op *err_ret = p.err;
342 3c1cf9d0 2021-01-11 op return 0;
343 3c1cf9d0 2021-01-11 op }
344 3c1cf9d0 2021-01-11 op
345 3c1cf9d0 2021-01-11 op *err_ret = NULL;
346 3c1cf9d0 2021-01-11 op return 1;
347 3c1cf9d0 2021-01-11 op }
348 3c1cf9d0 2021-01-11 op
349 3c1cf9d0 2021-01-11 op int
350 3c1cf9d0 2021-01-11 op trim_req_iri(char *iri)
351 3c1cf9d0 2021-01-11 op {
352 3c1cf9d0 2021-01-11 op char *i;
353 3c1cf9d0 2021-01-11 op
354 3c1cf9d0 2021-01-11 op if ((i = strstr(iri, "\r\n")) == NULL)
355 3c1cf9d0 2021-01-11 op return 0;
356 3c1cf9d0 2021-01-11 op *i = '\0';
357 3c1cf9d0 2021-01-11 op return 1;
358 3c1cf9d0 2021-01-11 op }