2 33d32d1f 2020-12-25 op * Copyright (c) 2020 Omar Polo <op@omarpolo.com>
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.
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.
17 33d32d1f 2020-12-25 op #include <ctype.h>
18 33d32d1f 2020-12-25 op #include <string.h>
20 33d32d1f 2020-12-25 op #include "gmid.h"
23 33d32d1f 2020-12-25 op * Notes from RFC3986
25 33d32d1f 2020-12-25 op * => gemini://tanso.net/rfc/rfc3986.txt
31 33d32d1f 2020-12-25 op * pct-encoded "%" HEXDIG HEXDIG
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 * / "*" / "+" / "," / ";" / "="
38 33d32d1f 2020-12-25 op * unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"
40 33d32d1f 2020-12-25 op * URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
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
47 33d32d1f 2020-12-25 op * scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
49 33d32d1f 2020-12-25 op * authority = [ userinfo "@" ] host [ ":" port ]
51 33d32d1f 2020-12-25 op * (note that userinfo isn't used for Gemini URL)
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 )
56 33d32d1f 2020-12-25 op * port = *DIGIT
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
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
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 / ":" / "@"
75 33d32d1f 2020-12-25 op * query = *( pchar / "/" / "?" )
77 33d32d1f 2020-12-25 op * fragment = *( pchar / "/" / "?" )
83 33d32d1f 2020-12-25 op * foo://example.com:8042/over/there?name=ferret#nose
84 33d32d1f 2020-12-25 op * \_/ \______________/\_________/ \_________/ \__/
86 33d32d1f 2020-12-25 op * scheme authority path query fragment
90 33d32d1f 2020-12-25 op struct parser {
92 33d32d1f 2020-12-25 op struct uri *parsed;
93 33d32d1f 2020-12-25 op const char *err;
96 df6ca41d 2020-12-25 op #define CONT_BYTE(b) ((b & 0xC0) == 0x80)
98 33d32d1f 2020-12-25 op /* XXX: these macros will expand multiple times their argument */
100 33d32d1f 2020-12-25 op #define UNRESERVED(p) \
101 33d32d1f 2020-12-25 op (isalnum(p) \
102 33d32d1f 2020-12-25 op || p == '-' \
103 33d32d1f 2020-12-25 op || p == '.' \
104 33d32d1f 2020-12-25 op || p == '_' \
107 33d32d1f 2020-12-25 op #define SUB_DELIMITERS(p) \
109 33d32d1f 2020-12-25 op || p == '$' \
110 33d32d1f 2020-12-25 op || p == '&' \
111 33d32d1f 2020-12-25 op || p == '\'' \
112 33d32d1f 2020-12-25 op || p == '(' \
113 33d32d1f 2020-12-25 op || p == ')' \
114 33d32d1f 2020-12-25 op || p == '*' \
115 33d32d1f 2020-12-25 op || p == '+' \
116 33d32d1f 2020-12-25 op || p == ',' \
117 33d32d1f 2020-12-25 op || p == ';' \
120 f722f3c5 2020-12-26 op /* NOTE: the increments are one less what they should be, because the
121 df6ca41d 2020-12-25 op * caller will add one byte after we return. */
123 df6ca41d 2020-12-25 op valid_multibyte_utf8(struct parser *p)
131 df6ca41d 2020-12-25 op if ((s & 0xE0) == 0xC0) {
132 df6ca41d 2020-12-25 op if (!CONT_BYTE(*(p->uri+1)))
134 df6ca41d 2020-12-25 op c = ((s & 0x1F) << 6) | (*(p->uri+1) & 0x3F);
136 df6ca41d 2020-12-25 op } else if ((s & 0xF0) == 0xE0) {
137 df6ca41d 2020-12-25 op if (!CONT_BYTE(*(p->uri+1)) ||
138 df6ca41d 2020-12-25 op !CONT_BYTE(*(p->uri+2)))
140 df6ca41d 2020-12-25 op c = (s & 0x0F) << 12
141 df6ca41d 2020-12-25 op | ((*(p->uri+1) & 0x3F) << 6)
142 df6ca41d 2020-12-25 op | ((*(p->uri+2) & 0x3F));
144 df6ca41d 2020-12-25 op } else if ((s & 0xF8) == 0xF0) {
145 df6ca41d 2020-12-25 op if (!CONT_BYTE(*(p->uri+1)) ||
146 df6ca41d 2020-12-25 op !CONT_BYTE(*(p->uri+2)) ||
147 df6ca41d 2020-12-25 op !CONT_BYTE(*(p->uri+3)))
149 df6ca41d 2020-12-25 op c = (s & 0x07) << 18
150 df6ca41d 2020-12-25 op | ((*(p->uri+1) & 0x3F) << 12)
151 df6ca41d 2020-12-25 op | ((*(p->uri+2) & 0x3F) << 6)
152 df6ca41d 2020-12-25 op | ((*(p->uri+3) & 0x3F));
157 df6ca41d 2020-12-25 op return (((0x080 <= c) && (c <= 0x7FF))
158 df6ca41d 2020-12-25 op || (((0x800 <= c) && (c <= 0xFFFF)))
159 df6ca41d 2020-12-25 op || (((0x10000 <= c) && (c <= 0x10FFFF))));
163 33d32d1f 2020-12-25 op parse_pct_encoded(struct parser *p)
165 33d32d1f 2020-12-25 op if (*p->uri != '%')
168 33d32d1f 2020-12-25 op if (!isxdigit(*(p->uri+1)) || !isxdigit(*(p->uri+2))) {
169 33d32d1f 2020-12-25 op p->err = "illegal percent-encoding";
173 33d32d1f 2020-12-25 op sscanf(p->uri+1, "%2hhx", p->uri);
174 33d32d1f 2020-12-25 op memmove(p->uri+1, p->uri+3, strlen(p->uri+3)+1);
175 00781742 2020-12-25 op if (*p->uri == '\0') {
176 00781742 2020-12-25 op p->err = "illegal percent-encoding";
183 33d32d1f 2020-12-25 op /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
185 33d32d1f 2020-12-25 op parse_scheme(struct parser *p)
187 33d32d1f 2020-12-25 op p->parsed->schema = p->uri;
189 33d32d1f 2020-12-25 op if (!isalpha(*p->uri)) {
190 33d32d1f 2020-12-25 op p->err = "illegal character in scheme";
195 33d32d1f 2020-12-25 op while (isalnum(*p->uri)
196 33d32d1f 2020-12-25 op || *p->uri == '+'
197 33d32d1f 2020-12-25 op || *p->uri == '-'
198 33d32d1f 2020-12-25 op || *p->uri == '.')
201 33d32d1f 2020-12-25 op if (*p->uri != ':') {
202 33d32d1f 2020-12-25 op p->err = "illegal character in scheme";
206 33d32d1f 2020-12-25 op *p->uri = '\0';
207 33d32d1f 2020-12-25 op if (*(++p->uri) != '/' || *(++p->uri) != '/') {
208 33d32d1f 2020-12-25 op p->err = "invalid marker after scheme";
218 33d32d1f 2020-12-25 op parse_port(struct parser *p)
220 33d32d1f 2020-12-25 op uint32_t i = 0;
222 33d32d1f 2020-12-25 op p->parsed->port = p->uri;
224 33d32d1f 2020-12-25 op for (; isdigit(*p->uri); p->uri++) {
225 33d32d1f 2020-12-25 op i = i * 10 + *p->uri - '0';
226 33d32d1f 2020-12-25 op if (i > UINT16_MAX) {
227 33d32d1f 2020-12-25 op p->err = "port number too large";
232 33d32d1f 2020-12-25 op if (*p->uri != '/' && *p->uri != '\0') {
233 33d32d1f 2020-12-25 op p->err = "illegal character in port number";
237 33d32d1f 2020-12-25 op p->parsed->port_no = i;
239 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
240 33d32d1f 2020-12-25 op *p->uri = '\0';
247 33d32d1f 2020-12-25 op /* TODO: add support for ip-literal and ipv4addr ? */
248 33d32d1f 2020-12-25 op /* *( unreserved / sub-delims / pct-encoded ) */
250 33d32d1f 2020-12-25 op parse_authority(struct parser *p)
252 33d32d1f 2020-12-25 op p->parsed->host = p->uri;
254 33d32d1f 2020-12-25 op while (UNRESERVED(*p->uri)
255 33d32d1f 2020-12-25 op || SUB_DELIMITERS(*p->uri)
256 33d32d1f 2020-12-25 op || parse_pct_encoded(p))
259 00781742 2020-12-25 op if (p->err != NULL)
262 33d32d1f 2020-12-25 op if (*p->uri == ':') {
263 33d32d1f 2020-12-25 op *p->uri = '\0';
265 33d32d1f 2020-12-25 op return parse_port(p);
268 33d32d1f 2020-12-25 op if (*p->uri == '/') {
269 33d32d1f 2020-12-25 op *p->uri = '\0';
274 33d32d1f 2020-12-25 op if (*p->uri == '\0')
277 33d32d1f 2020-12-25 op p->err = "illegal character in authority section";
281 33d32d1f 2020-12-25 op /* Routine for path_clean. Elide the pointed .. with the preceding
282 33d32d1f 2020-12-25 op * element. Return 0 if it's not possible. incr is the length of
283 33d32d1f 2020-12-25 op * the increment, 3 for ../ and 2 for .. */
285 33d32d1f 2020-12-25 op path_elide_dotdot(char *path, char *i, int incr)
289 33d32d1f 2020-12-25 op if (i == path)
291 33d32d1f 2020-12-25 op for (j = i-2; j != path && *j != '/'; j--)
293 33d32d1f 2020-12-25 op if (*j == '/')
296 33d32d1f 2020-12-25 op memmove(j, i, strlen(i)+1);
301 33d32d1f 2020-12-25 op * Use an algorithm similar to the one implemented in go' path.Clean:
303 33d32d1f 2020-12-25 op * 1. Replace multiple slashes with a single slash
304 33d32d1f 2020-12-25 op * 2. Eliminate each . path name element
305 33d32d1f 2020-12-25 op * 3. Eliminate each inner .. along with the non-.. element that precedes it
306 33d32d1f 2020-12-25 op * 4. Eliminate trailing .. if possible or error (go would only discard)
308 33d32d1f 2020-12-25 op * Unlike path.Clean, this function return the empty string if the
309 33d32d1f 2020-12-25 op * original path is equivalent to "/".
312 33d32d1f 2020-12-25 op path_clean(char *path)
316 33d32d1f 2020-12-25 op /* 1. replace multiple slashes with a single one */
317 33d32d1f 2020-12-25 op for (i = path; *i; ++i) {
318 33d32d1f 2020-12-25 op if (*i == '/' && *(i+1) == '/') {
319 33d32d1f 2020-12-25 op memmove(i, i+1, strlen(i)); /* move also the \0 */
324 33d32d1f 2020-12-25 op /* 2. eliminate each . path name element */
325 33d32d1f 2020-12-25 op for (i = path; *i; ++i) {
326 33d32d1f 2020-12-25 op if ((i == path || *i == '/') && *(i+1) == '.' &&
327 33d32d1f 2020-12-25 op *(i+2) == '/') {
328 33d32d1f 2020-12-25 op /* move also the \0 */
329 33d32d1f 2020-12-25 op memmove(i, i+2, strlen(i)-1);
333 33d32d1f 2020-12-25 op if (!strcmp(path, ".") || !strcmp(path, "/.")) {
334 33d32d1f 2020-12-25 op *path = '\0';
338 33d32d1f 2020-12-25 op /* 3. eliminate each inner .. along with the preceding non-.. */
339 33d32d1f 2020-12-25 op for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
340 33d32d1f 2020-12-25 op if (!path_elide_dotdot(path, i, 3))
343 33d32d1f 2020-12-25 op /* 4. eliminate trailing ..*/
344 33d32d1f 2020-12-25 op if ((i = strstr(path, "..")) != NULL)
345 33d32d1f 2020-12-25 op if (!path_elide_dotdot(path, i, 2))
352 33d32d1f 2020-12-25 op parse_query(struct parser *p)
354 33d32d1f 2020-12-25 op p->parsed->query = p->uri;
355 33d32d1f 2020-12-25 op if (*p->uri == '\0')
358 33d32d1f 2020-12-25 op while (UNRESERVED(*p->uri)
359 33d32d1f 2020-12-25 op || SUB_DELIMITERS(*p->uri)
360 33d32d1f 2020-12-25 op || *p->uri == '/'
361 33d32d1f 2020-12-25 op || *p->uri == '?'
362 df6ca41d 2020-12-25 op || parse_pct_encoded(p)
363 df6ca41d 2020-12-25 op || valid_multibyte_utf8(p))
366 00781742 2020-12-25 op if (p->err != NULL)
369 33d32d1f 2020-12-25 op if (*p->uri != '\0' && *p->uri != '#') {
370 33d32d1f 2020-12-25 op p->err = "illegal character in query";
374 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
375 33d32d1f 2020-12-25 op *p->uri = '\0';
382 33d32d1f 2020-12-25 op /* don't even bother */
384 33d32d1f 2020-12-25 op parse_fragment(struct parser *p)
386 33d32d1f 2020-12-25 op p->parsed->fragment = p->uri;
390 33d32d1f 2020-12-25 op /* XXX: is it too broad? */
391 33d32d1f 2020-12-25 op /* *(pchar / "/") */
393 33d32d1f 2020-12-25 op parse_path(struct parser *p)
397 33d32d1f 2020-12-25 op p->parsed->path = p->uri;
398 33d32d1f 2020-12-25 op if (*p->uri == '\0') {
399 33d32d1f 2020-12-25 op p->parsed->query = p->parsed->fragment = p->uri;
403 33d32d1f 2020-12-25 op while (UNRESERVED(*p->uri)
404 33d32d1f 2020-12-25 op || SUB_DELIMITERS(*p->uri)
405 33d32d1f 2020-12-25 op || *p->uri == '/'
406 df6ca41d 2020-12-25 op || parse_pct_encoded(p)
407 df6ca41d 2020-12-25 op || valid_multibyte_utf8(p))
410 00781742 2020-12-25 op if (p->err != NULL)
413 33d32d1f 2020-12-25 op if (*p->uri != '\0' && *p->uri != '?' && *p->uri != '#') {
414 33d32d1f 2020-12-25 op p->err = "illegal character in path";
418 33d32d1f 2020-12-25 op if (*p->uri != '\0') {
420 33d32d1f 2020-12-25 op *p->uri = '\0';
423 33d32d1f 2020-12-25 op if (c == '#') {
424 33d32d1f 2020-12-25 op if (!parse_fragment(p))
427 33d32d1f 2020-12-25 op if (!parse_query(p) || !parse_fragment(p))
431 33d32d1f 2020-12-25 op if (!path_clean(p->parsed->path)) {
432 33d32d1f 2020-12-25 op p->err = "illegal path";
440 33d32d1f 2020-12-25 op parse_uri(char *uri, struct uri *ret, const char **err_ret)
443 33d32d1f 2020-12-25 op struct parser p = {uri, ret, NULL};
445 33d32d1f 2020-12-25 op bzero(ret, sizeof(*ret));
447 33d32d1f 2020-12-25 op /* initialize optional stuff to the empty string */
448 33d32d1f 2020-12-25 op end = uri + strlen(uri);
449 33d32d1f 2020-12-25 op p.parsed->port = end;
450 33d32d1f 2020-12-25 op p.parsed->path = end;
451 33d32d1f 2020-12-25 op p.parsed->query = end;
452 33d32d1f 2020-12-25 op p.parsed->fragment = end;
454 33d32d1f 2020-12-25 op if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
455 33d32d1f 2020-12-25 op *err_ret = p.err;
459 33d32d1f 2020-12-25 op *err_ret = NULL;
464 33d32d1f 2020-12-25 op trim_req_uri(char *uri)
468 33d32d1f 2020-12-25 op if ((i = strstr(uri, "\r\n")) == NULL)