2 3c1cf9d0 2021-01-11 op * Copyright (c) 2020 Omar Polo <op@omarpolo.com>
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.
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.
17 3c1cf9d0 2021-01-11 op #include <ctype.h>
18 3c1cf9d0 2021-01-11 op #include <string.h>
20 3c1cf9d0 2021-01-11 op #include "gmid.h"
22 3c1cf9d0 2021-01-11 op static inline int
23 3c1cf9d0 2021-01-11 op unreserved(int p)
25 3c1cf9d0 2021-01-11 op return isalnum(p)
32 3c1cf9d0 2021-01-11 op static inline int
33 3c1cf9d0 2021-01-11 op sub_delimiters(int p)
35 3c1cf9d0 2021-01-11 op return p == '!'
49 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
51 3c1cf9d0 2021-01-11 op if (*p->iri != '%')
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";
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";
69 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
71 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
73 3c1cf9d0 2021-01-11 op p->parsed->schema = p->iri;
75 3c1cf9d0 2021-01-11 op if (!isalpha(*p->iri)) {
76 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
81 de428fff 2021-01-13 op /* normalize the scheme (i.e. lowercase it)
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);
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 == '.');
94 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
95 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
99 3c1cf9d0 2021-01-11 op *p->iri = '\0';
100 f7b816dc 2021-01-15 op if (p->iri[1] != '/' || p->iri[2] != '/') {
101 3c1cf9d0 2021-01-11 op p->err = "invalid marker after scheme";
111 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
113 3c1cf9d0 2021-01-11 op uint32_t i = 0;
115 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
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";
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";
130 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
132 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
133 3c1cf9d0 2021-01-11 op *p->iri = '\0';
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 ) */
143 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
145 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
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);
156 3c1cf9d0 2021-01-11 op if (p->err != NULL)
159 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
160 3c1cf9d0 2021-01-11 op *p->iri = '\0';
162 3c1cf9d0 2021-01-11 op return parse_port(p);
164 b777bf4b 2021-01-15 op p->parsed->port_no = 1965;
166 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
167 3c1cf9d0 2021-01-11 op *p->iri = '\0';
172 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
175 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
179 3c1cf9d0 2021-01-11 op /* Routine for path_clean. Elide the pointed .. with the preceding
180 3c1cf9d0 2021-01-11 op * element. Return 0 if it's not possible. incr is the length of
181 3c1cf9d0 2021-01-11 op * the increment, 3 for ../ and 2 for .. */
183 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
187 3c1cf9d0 2021-01-11 op if (i == path)
189 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
191 3c1cf9d0 2021-01-11 op if (*j == '/')
194 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
199 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
201 3c1cf9d0 2021-01-11 op * 1. Replace multiple slashes with a single slash
202 3c1cf9d0 2021-01-11 op * 2. Eliminate each . path name element
203 3c1cf9d0 2021-01-11 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
204 3c1cf9d0 2021-01-11 op * 4. Eliminate trailing .. if possible or error (go would only discard)
206 3c1cf9d0 2021-01-11 op * Unlike path.Clean, this function return the empty string if the
207 3c1cf9d0 2021-01-11 op * original path is equivalent to "/".
210 3c1cf9d0 2021-01-11 op path_clean(char *path)
214 3c1cf9d0 2021-01-11 op /* 1. replace multiple slashes with a single one */
215 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
216 3c1cf9d0 2021-01-11 op if (*i == '/' && *(i+1) == '/') {
217 3c1cf9d0 2021-01-11 op memmove(i, i+1, strlen(i)); /* move also the \0 */
222 3c1cf9d0 2021-01-11 op /* 2. eliminate each . path name element */
223 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
224 6a9ae707 2021-01-11 op if ((i == path || *i == '/') &&
225 6a9ae707 2021-01-11 op *i != '.' && i[1] == '.' && i[2] == '/') {
226 3c1cf9d0 2021-01-11 op /* move also the \0 */
227 3c1cf9d0 2021-01-11 op memmove(i, i+2, strlen(i)-1);
231 3c1cf9d0 2021-01-11 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
232 3c1cf9d0 2021-01-11 op *path = '\0';
236 3c1cf9d0 2021-01-11 op /* 3. eliminate each inner .. along with the preceding non-.. */
237 3c1cf9d0 2021-01-11 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
238 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 3))
241 3c1cf9d0 2021-01-11 op /* 4. eliminate trailing ..*/
242 3c1cf9d0 2021-01-11 op if ((i = strstr(path, "..")) != NULL)
243 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 2))
250 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
252 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
253 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
256 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
257 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
258 3c1cf9d0 2021-01-11 op || *p->iri == '/'
259 3c1cf9d0 2021-01-11 op || *p->iri == '?'
260 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
261 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
264 3c1cf9d0 2021-01-11 op if (p->err != NULL)
267 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
268 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
272 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
273 3c1cf9d0 2021-01-11 op *p->iri = '\0';
280 3c1cf9d0 2021-01-11 op /* don't even bother */
282 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
284 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
288 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
289 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
291 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
295 3c1cf9d0 2021-01-11 op p->parsed->path = p->iri;
296 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
297 3c1cf9d0 2021-01-11 op p->parsed->query = p->parsed->fragment = p->iri;
301 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
302 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
303 3c1cf9d0 2021-01-11 op || *p->iri == '/'
304 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
305 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
308 3c1cf9d0 2021-01-11 op if (p->err != NULL)
311 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
312 3c1cf9d0 2021-01-11 op p->err = "illegal character in path";
316 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
318 3c1cf9d0 2021-01-11 op *p->iri = '\0';
321 3c1cf9d0 2021-01-11 op if (c == '#') {
322 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
325 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
329 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
330 3c1cf9d0 2021-01-11 op p->err = "illegal path";
338 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
341 3c1cf9d0 2021-01-11 op struct parser p = {iri, ret, NULL};
343 3c1cf9d0 2021-01-11 op bzero(ret, sizeof(*ret));
345 3c1cf9d0 2021-01-11 op /* initialize optional stuff to the empty string */
346 3c1cf9d0 2021-01-11 op end = iri + strlen(iri);
347 3c1cf9d0 2021-01-11 op p.parsed->port = end;
348 3c1cf9d0 2021-01-11 op p.parsed->path = end;
349 3c1cf9d0 2021-01-11 op p.parsed->query = end;
350 3c1cf9d0 2021-01-11 op p.parsed->fragment = end;
352 3c1cf9d0 2021-01-11 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
353 3c1cf9d0 2021-01-11 op *err_ret = p.err;
357 3c1cf9d0 2021-01-11 op *err_ret = NULL;
362 3c1cf9d0 2021-01-11 op trim_req_iri(char *iri)
366 3c1cf9d0 2021-01-11 op if ((i = strstr(iri, "\r\n")) == NULL)