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 52418c8d 2021-02-12 op #include "gmid.h"
19 3c1cf9d0 2021-01-11 op #include <ctype.h>
20 3c1cf9d0 2021-01-11 op #include <string.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 9f006a21 2021-02-07 op pct_decode(char *s)
81 9f006a21 2021-02-07 op sscanf(s+1, "%2hhx", s);
82 9f006a21 2021-02-07 op memmove(s+1, s+3, strlen(s+3)+1);
86 3c1cf9d0 2021-01-11 op parse_pct_encoded(struct parser *p)
88 9a672b37 2021-01-28 op if (p->iri[0] != '%')
91 8404ec30 2021-02-05 op if (!valid_pct_enc_string(p->iri)) {
92 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
96 9f006a21 2021-02-07 op pct_decode(p->iri);
97 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
98 3c1cf9d0 2021-01-11 op p->err = "illegal percent-encoding";
105 3c1cf9d0 2021-01-11 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
107 3c1cf9d0 2021-01-11 op parse_scheme(struct parser *p)
109 3c1cf9d0 2021-01-11 op p->parsed->schema = p->iri;
111 3c1cf9d0 2021-01-11 op if (!isalpha(*p->iri)) {
112 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
117 de428fff 2021-01-13 op /* normalize the scheme (i.e. lowercase it)
119 de428fff 2021-01-13 op * XXX: since we cannot have good things, tolower
120 881dc835 2021-01-16 op * behaviour depends on the LC_CTYPE locale. The good
121 881dc835 2021-01-16 op * news is that we're sure p->iri points to something
122 881dc835 2021-01-16 op * that's in the ASCII range, so tolower can't
123 881dc835 2021-01-16 op * mis-behave on some systems due to the locale. */
124 de428fff 2021-01-13 op *p->iri = tolower(*p->iri);
126 de428fff 2021-01-13 op } while (isalnum(*p->iri)
127 3c1cf9d0 2021-01-11 op || *p->iri == '+'
128 3c1cf9d0 2021-01-11 op || *p->iri == '-'
129 de428fff 2021-01-13 op || *p->iri == '.');
131 3c1cf9d0 2021-01-11 op if (*p->iri != ':') {
132 3c1cf9d0 2021-01-11 op p->err = "illegal character in scheme";
136 3c1cf9d0 2021-01-11 op *p->iri = '\0';
137 f7b816dc 2021-01-15 op if (p->iri[1] != '/' || p->iri[2] != '/') {
138 3c1cf9d0 2021-01-11 op p->err = "invalid marker after scheme";
148 3c1cf9d0 2021-01-11 op parse_port(struct parser *p)
150 3c1cf9d0 2021-01-11 op uint32_t i = 0;
152 3c1cf9d0 2021-01-11 op p->parsed->port = p->iri;
154 3c1cf9d0 2021-01-11 op for (; isdigit(*p->iri); p->iri++) {
155 3c1cf9d0 2021-01-11 op i = i * 10 + *p->iri - '0';
156 3c1cf9d0 2021-01-11 op if (i > UINT16_MAX) {
157 3c1cf9d0 2021-01-11 op p->err = "port number too large";
162 3c1cf9d0 2021-01-11 op if (*p->iri != '/' && *p->iri != '\0') {
163 3c1cf9d0 2021-01-11 op p->err = "illegal character in port number";
167 3c1cf9d0 2021-01-11 op p->parsed->port_no = i;
169 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
170 3c1cf9d0 2021-01-11 op *p->iri = '\0';
177 3c1cf9d0 2021-01-11 op /* TODO: add support for ip-literal and ipv4addr ? */
178 3c1cf9d0 2021-01-11 op /* *( unreserved / sub-delims / pct-encoded ) */
180 3c1cf9d0 2021-01-11 op parse_authority(struct parser *p)
182 3c1cf9d0 2021-01-11 op p->parsed->host = p->iri;
184 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
185 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
186 117ac52c 2021-01-29 op || parse_pct_encoded(p)
187 117ac52c 2021-01-29 op || valid_multibyte_utf8(p)) {
188 e4d82bec 2021-01-15 op /* normalize the host name. */
189 e4d82bec 2021-01-15 op if (*p->iri < 0x7F)
190 e4d82bec 2021-01-15 op *p->iri = tolower(*p->iri);
194 3c1cf9d0 2021-01-11 op if (p->err != NULL)
197 3c1cf9d0 2021-01-11 op if (*p->iri == ':') {
198 3c1cf9d0 2021-01-11 op *p->iri = '\0';
200 3c1cf9d0 2021-01-11 op return parse_port(p);
202 b777bf4b 2021-01-15 op p->parsed->port_no = 1965;
204 3c1cf9d0 2021-01-11 op if (*p->iri == '/') {
205 3c1cf9d0 2021-01-11 op *p->iri = '\0';
210 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
213 3c1cf9d0 2021-01-11 op p->err = "illegal character in authority section";
217 3c1cf9d0 2021-01-11 op /* Routine for path_clean. Elide the pointed .. with the preceding
218 3c1cf9d0 2021-01-11 op * element. Return 0 if it's not possible. incr is the length of
219 3c1cf9d0 2021-01-11 op * the increment, 3 for ../ and 2 for .. */
221 3c1cf9d0 2021-01-11 op path_elide_dotdot(char *path, char *i, int incr)
225 3c1cf9d0 2021-01-11 op if (i == path)
227 3c1cf9d0 2021-01-11 op for (j = i-2; j != path && *j != '/'; j--)
229 3c1cf9d0 2021-01-11 op if (*j == '/')
232 3c1cf9d0 2021-01-11 op memmove(j, i, strlen(i)+1);
237 3c1cf9d0 2021-01-11 op * Use an algorithm similar to the one implemented in go' path.Clean:
239 3c1cf9d0 2021-01-11 op * 1. Replace multiple slashes with a single slash
240 3c1cf9d0 2021-01-11 op * 2. Eliminate each . path name element
241 3c1cf9d0 2021-01-11 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
242 3c1cf9d0 2021-01-11 op * 4. Eliminate trailing .. if possible or error (go would only discard)
244 3c1cf9d0 2021-01-11 op * Unlike path.Clean, this function return the empty string if the
245 3c1cf9d0 2021-01-11 op * original path is equivalent to "/".
248 3c1cf9d0 2021-01-11 op path_clean(char *path)
252 3c1cf9d0 2021-01-11 op /* 1. replace multiple slashes with a single one */
253 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
254 3c1cf9d0 2021-01-11 op if (*i == '/' && *(i+1) == '/') {
255 3c1cf9d0 2021-01-11 op memmove(i, i+1, strlen(i)); /* move also the \0 */
260 3c1cf9d0 2021-01-11 op /* 2. eliminate each . path name element */
261 3c1cf9d0 2021-01-11 op for (i = path; *i; ++i) {
262 6a9ae707 2021-01-11 op if ((i == path || *i == '/') &&
263 6a9ae707 2021-01-11 op *i != '.' && i[1] == '.' && i[2] == '/') {
264 3c1cf9d0 2021-01-11 op /* move also the \0 */
265 3c1cf9d0 2021-01-11 op memmove(i, i+2, strlen(i)-1);
269 3c1cf9d0 2021-01-11 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
270 3c1cf9d0 2021-01-11 op *path = '\0';
274 3c1cf9d0 2021-01-11 op /* 3. eliminate each inner .. along with the preceding non-.. */
275 3c1cf9d0 2021-01-11 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
276 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 3))
279 3c1cf9d0 2021-01-11 op /* 4. eliminate trailing ..*/
280 3c1cf9d0 2021-01-11 op if ((i = strstr(path, "..")) != NULL)
281 3c1cf9d0 2021-01-11 op if (!path_elide_dotdot(path, i, 2))
288 3c1cf9d0 2021-01-11 op parse_query(struct parser *p)
290 3c1cf9d0 2021-01-11 op p->parsed->query = p->iri;
291 3c1cf9d0 2021-01-11 op if (*p->iri == '\0')
294 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
295 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
296 3c1cf9d0 2021-01-11 op || *p->iri == '/'
297 3c1cf9d0 2021-01-11 op || *p->iri == '?'
298 19e7bd00 2021-02-06 op || *p->iri == ':'
299 19e7bd00 2021-02-06 op || *p->iri == '@'
300 8404ec30 2021-02-05 op || valid_pct_encoded(p)
301 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
304 3c1cf9d0 2021-01-11 op if (p->err != NULL)
307 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '#') {
308 3c1cf9d0 2021-01-11 op p->err = "illegal character in query";
312 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
313 3c1cf9d0 2021-01-11 op *p->iri = '\0';
320 3c1cf9d0 2021-01-11 op /* don't even bother */
322 3c1cf9d0 2021-01-11 op parse_fragment(struct parser *p)
324 3c1cf9d0 2021-01-11 op p->parsed->fragment = p->iri;
328 3c1cf9d0 2021-01-11 op /* XXX: is it too broad? */
329 3c1cf9d0 2021-01-11 op /* *(pchar / "/") */
331 3c1cf9d0 2021-01-11 op parse_path(struct parser *p)
335 42bbdc79 2021-01-21 op /* trim initial slashes */
336 42bbdc79 2021-01-21 op while (*p->iri == '/')
339 3c1cf9d0 2021-01-11 op p->parsed->path = p->iri;
340 3c1cf9d0 2021-01-11 op if (*p->iri == '\0') {
341 3c1cf9d0 2021-01-11 op p->parsed->query = p->parsed->fragment = p->iri;
345 3c1cf9d0 2021-01-11 op while (unreserved(*p->iri)
346 3c1cf9d0 2021-01-11 op || sub_delimiters(*p->iri)
347 3c1cf9d0 2021-01-11 op || *p->iri == '/'
348 3c1cf9d0 2021-01-11 op || parse_pct_encoded(p)
349 3c1cf9d0 2021-01-11 op || valid_multibyte_utf8(p))
352 3c1cf9d0 2021-01-11 op if (p->err != NULL)
355 3c1cf9d0 2021-01-11 op if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
356 3c1cf9d0 2021-01-11 op p->err = "illegal character in path";
360 3c1cf9d0 2021-01-11 op if (*p->iri != '\0') {
362 3c1cf9d0 2021-01-11 op *p->iri = '\0';
365 3c1cf9d0 2021-01-11 op if (c == '#') {
366 3c1cf9d0 2021-01-11 op if (!parse_fragment(p))
369 3c1cf9d0 2021-01-11 op if (!parse_query(p) || !parse_fragment(p))
373 3c1cf9d0 2021-01-11 op if (!path_clean(p->parsed->path)) {
374 3c1cf9d0 2021-01-11 op p->err = "illegal path";
382 3c1cf9d0 2021-01-11 op parse_iri(char *iri, struct iri *ret, const char **err_ret)
385 3c1cf9d0 2021-01-11 op struct parser p = {iri, ret, NULL};
387 3c1cf9d0 2021-01-11 op bzero(ret, sizeof(*ret));
389 3c1cf9d0 2021-01-11 op /* initialize optional stuff to the empty string */
390 3c1cf9d0 2021-01-11 op end = iri + strlen(iri);
391 57d0d0ad 2021-01-31 op p.parsed->host = end;
392 3c1cf9d0 2021-01-11 op p.parsed->port = end;
393 3c1cf9d0 2021-01-11 op p.parsed->path = end;
394 3c1cf9d0 2021-01-11 op p.parsed->query = end;
395 3c1cf9d0 2021-01-11 op p.parsed->fragment = end;
397 3c1cf9d0 2021-01-11 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
398 3c1cf9d0 2021-01-11 op *err_ret = p.err;
402 3c1cf9d0 2021-01-11 op *err_ret = NULL;
407 c4f682f8 2021-01-27 op trim_req_iri(char *iri, const char **err)
411 c4f682f8 2021-01-27 op if ((i = strstr(iri, "\r\n")) == NULL) {
412 c4f682f8 2021-01-27 op *err = "missing CRLF";
421 2fafa2d2 2021-02-01 op serialize_iri(struct iri *i, char *buf, size_t len)
425 2fafa2d2 2021-02-01 op /* in ex.c we receive empty "" strings as NULL */
426 2fafa2d2 2021-02-01 op if (i->schema == NULL || i->host == NULL) {
427 2fafa2d2 2021-02-01 op memset(buf, 0, len);
431 2fafa2d2 2021-02-01 op strlcpy(buf, i->schema, len);
432 2fafa2d2 2021-02-01 op strlcat(buf, "://", len);
433 2fafa2d2 2021-02-01 op strlcat(buf, i->host, len);
434 2fafa2d2 2021-02-01 op strlcat(buf, "/", len);
436 2fafa2d2 2021-02-01 op if (i->path != NULL)
437 2fafa2d2 2021-02-01 op l = strlcat(buf, i->path, len);
439 2fafa2d2 2021-02-01 op if (i->query != NULL && *i->query != '\0') {
440 2fafa2d2 2021-02-01 op strlcat(buf, "?", len);
441 2fafa2d2 2021-02-01 op l = strlcat(buf, i->query, len);
444 2fafa2d2 2021-02-01 op return l < len;
448 9f006a21 2021-02-07 op pct_decode_str(char *s)
452 9f006a21 2021-02-07 op for (t = s; *t; ++t) {
453 9f006a21 2021-02-07 op if (*t == '%' && valid_pct_enc_string(t))
454 9f006a21 2021-02-07 op pct_decode(t);