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 8404ec30 2021-02-05 op valid_pct_enc_string(char *s)
51 8404ec30 2021-02-05 op if (*s != '%')
54 8404ec30 2021-02-05 op if (!isxdigit(s[1]) || !isxdigit(s[2]))
57 8404ec30 2021-02-05 op if (s[1] == '0' && s[2] == '0')
64 8404ec30 2021-02-05 op valid_pct_encoded(struct parser *p)
66 8404ec30 2021-02-05 op if (p->iri[0] != '%')
69 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
70 8404ec30 2021-02-05 op p->err = "illegal percent-encoding";
79 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
81 9a672b37 2021-01-28 op if (p->iri[0] != '%')
84 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
85 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
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";
99 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
101 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
103 3c1cf9d0 2021-01-11 op p->parsed->schema = p->iri;
105 3c1cf9d0 2021-01-11 op if (!isalpha(*p->iri)) {
106 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
111 de428fff 2021-01-13 op /* normalize the scheme (i.e. lowercase it)
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);
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 == '.');
125 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
126 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
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";
142 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
144 3c1cf9d0 2021-01-11 op uint32_t i = 0;
146 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
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";
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";
161 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
163 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
164 3c1cf9d0 2021-01-11 op *p->iri = '\0';
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 ) */
174 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
176 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
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);
188 3c1cf9d0 2021-01-11 op if (p->err != NULL)
191 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
192 3c1cf9d0 2021-01-11 op *p->iri = '\0';
194 3c1cf9d0 2021-01-11 op return parse_port(p);
196 b777bf4b 2021-01-15 op p->parsed->port_no = 1965;
198 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
199 3c1cf9d0 2021-01-11 op *p->iri = '\0';
204 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
207 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
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 .. */
215 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
219 3c1cf9d0 2021-01-11 op if (i == path)
221 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
223 3c1cf9d0 2021-01-11 op if (*j == '/')
226 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
231 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
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)
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 "/".
242 3c1cf9d0 2021-01-11 op path_clean(char *path)
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 */
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);
263 3c1cf9d0 2021-01-11 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
264 3c1cf9d0 2021-01-11 op *path = '\0';
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))
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))
282 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
284 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
285 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
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))
298 3c1cf9d0 2021-01-11 op if (p->err != NULL)
301 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
302 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
306 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
307 3c1cf9d0 2021-01-11 op *p->iri = '\0';
314 3c1cf9d0 2021-01-11 op /* don't even bother */
316 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
318 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
322 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
323 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
325 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
329 42bbdc79 2021-01-21 op /* trim initial slashes */
330 42bbdc79 2021-01-21 op while (*p->iri == '/')
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;
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))
346 3c1cf9d0 2021-01-11 op if (p->err != NULL)
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";
354 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
356 3c1cf9d0 2021-01-11 op *p->iri = '\0';
359 3c1cf9d0 2021-01-11 op if (c == '#') {
360 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
363 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
367 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
368 3c1cf9d0 2021-01-11 op p->err = "illegal path";
376 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
379 3c1cf9d0 2021-01-11 op struct parser p = {iri, ret, NULL};
381 3c1cf9d0 2021-01-11 op bzero(ret, sizeof(*ret));
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;
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;
396 3c1cf9d0 2021-01-11 op *err_ret = NULL;
401 c4f682f8 2021-01-27 op trim_req_iri(char *iri, const char **err)
405 c4f682f8 2021-01-27 op if ((i = strstr(iri, "\r\n")) == NULL) {
406 c4f682f8 2021-01-27 op *err = "missing CRLF";
415 2fafa2d2 2021-02-01 op serialize_iri(struct iri *i, char *buf, size_t len)
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);
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);
430 2fafa2d2 2021-02-01 op if (i->path != NULL)
431 2fafa2d2 2021-02-01 op l = strlcat(buf, i->path, len);
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);
438 2fafa2d2 2021-02-01 op return l < len;