1 /*
2  * Copyright (c) 2020 Omar Polo <op@omarpolo.com>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <ctype.h>
18 #include <string.h>
19 
20 #include "gmid.h"
21 
22 /*
23  * Notes from RFC3986
24  *
25  * => gemini://tanso.net/rfc/rfc3986.txt
26  *
27  *
28  * ABNF
29  * ====
30  *
31  * pct-encoded	"%" HEXDIG HEXDIG
32  *
33  * reserved	= gen-delims / sub-delimis
34  * gen-delims	= ":" / "/" / "?" / "#" / "[" / "]" / "@"
35  * sub-delims	= "!" / "$" / "&" / "'" / "(" / ")"
36  * 		/ "*" / "+" / "," / ";" / "="
37  *
38  * unreserved	= ALPHA / DIGIT / "-" / "." / "_" / "~"
39  *
40  * URI		= scheme ":" hier-part [ "?" query ] [ "#" fragment ]
41  *
42  * hier-part	= "//" authority path-abempty
43  * 		/ path-absolute
44  * 		/ path-rootless
45  * 		/ path-empty
46  *
47  * scheme	= ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
48  *
49  * authority	= [ userinfo "@" ] host [ ":" port ]
50  *
51  * (note that userinfo isn't used for Gemini URL)
52  *
53  * host		= IP-literal / IPv4address / reg-name
54  * reg-name	= *( unreserved / pct-encoded / sub-delims )
55  *
56  * port		= *DIGIT
57  *
58  * path		= path-abemty	; begins with "/" or is empty
59  * 		/ path-absolute	; begins with "/" but not "//"
60  * 		/ path-noscheme	; begins with a non-colon segment
61  * 		/ path-rootless ; begins with a segment
62  * 		/ path-empty	; zero characters
63  *
64  * path-abemty		= *( "/" segment )
65  * path-absolute	= "/" [ segment-nz *( "/" segment ) ]
66  * path-noscheme	= ; not used
67  * path-rootless	= ; not used
68  * path-empty		= ; not used
69  *
70  * segment		= *pchar
71  * segment-nz	= 1*pchar
72  * segment-nz-nc	= 1*( unreserved / pct-encoded / sub-delims / "@" )
73  * pchar		= unreserved / pct-encoded / sub-delims / ":" / "@"
74  *
75  * query		= *( pchar / "/" / "?" )
76  *
77  * fragment		= *( pchar / "/" / "?" )
78  *
79  *
80  * EXAMPLE
81  * =======
82  *
83  *    foo://example.com:8042/over/there?name=ferret#nose
84  *    \_/   \______________/\_________/ \_________/ \__/
85  *     |           |            |            |        |
86  *  scheme     authority       path        query   fragment
87  *
88  */
89 
90 static inline int
91 unreserved(int p)
92 {
93 	return isalnum(p)
94 		|| p == '-'
95 		|| p == '.'
96 		|| p == '_'
97 		|| p == '~';
98 }
99 
100 static inline int
101 sub_delimiters(int p)
102 {
103 	return p == '!'
104 		|| p == '$'
105 		|| p == '&'
106 		|| p == '\''
107 		|| p == '('
108 		|| p == ')'
109 		|| p == '*'
110 		|| p == '+'
111 		|| p == ','
112 		|| p == ';'
113 		|| p == '=';
114 }
115 
116 static int
117 parse_pct_encoded(struct parser *p)
118 {
119 	if (*p->uri != '%')
120 		return 0;
121 
122 	if (!isxdigit(*(p->uri+1)) || !isxdigit(*(p->uri+2))) {
123 		p->err = "illegal percent-encoding";
124 		return 0;
125 	}
126 
127 	sscanf(p->uri+1, "%2hhx", p->uri);
128 	memmove(p->uri+1, p->uri+3, strlen(p->uri+3)+1);
129 	if (*p->uri == '\0') {
130 		p->err = "illegal percent-encoding";
131 		return 0;
132 	}
133 
134 	return 1;
135 }
136 
137 /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
138 static int
139 parse_scheme(struct parser *p)
140 {
141 	p->parsed->schema = p->uri;
142 
143 	if (!isalpha(*p->uri)) {
144 		p->err = "illegal character in scheme";
145 		return 0;
146 	}
147 
148 	p->uri++;
149 	while (isalnum(*p->uri)
150 	    || *p->uri == '+'
151 	    || *p->uri == '-'
152 	    || *p->uri == '.')
153 		p->uri++;
154 
155 	if (*p->uri != ':') {
156 		p->err = "illegal character in scheme";
157 		return 0;
158 	}
159 
160 	*p->uri = '\0';
161 	if (*(++p->uri) != '/' || *(++p->uri) != '/') {
162 		p->err = "invalid marker after scheme";
163 		return 0;
164 	}
165 
166 	p->uri++;
167 	return 1;
168 }
169 
170 /* *DIGIT */
171 static int
172 parse_port(struct parser *p)
173 {
174 	uint32_t i = 0;
175 
176 	p->parsed->port = p->uri;
177 
178 	for (; isdigit(*p->uri); p->uri++) {
179 		i = i * 10 + *p->uri - '0';
180 		if (i > UINT16_MAX) {
181 			p->err = "port number too large";
182 			return 0;
183 		}
184 	}
185 
186 	if (*p->uri != '/' && *p->uri != '\0') {
187 		p->err = "illegal character in port number";
188 		return 0;
189 	}
190 
191 	p->parsed->port_no = i;
192 
193 	if (*p->uri != '\0') {
194 		*p->uri = '\0';
195 		p->uri++;
196 	}
197 
198 	return 1;
199 }
200 
201 /* TODO: add support for ip-literal and ipv4addr ? */
202 /* *( unreserved / sub-delims / pct-encoded ) */
203 static int
204 parse_authority(struct parser *p)
205 {
206 	p->parsed->host = p->uri;
207 
208 	while (unreserved(*p->uri)
209 	    || sub_delimiters(*p->uri)
210 	    || parse_pct_encoded(p))
211 		p->uri++;
212 
213 	if (p->err != NULL)
214 		return 0;
215 
216 	if (*p->uri == ':') {
217 		*p->uri = '\0';
218 		p->uri++;
219 		return parse_port(p);
220 	}
221 
222 	if (*p->uri == '/') {
223 		*p->uri = '\0';
224 		p->uri++;
225 		return 1;
226 	}
227 
228 	if (*p->uri == '\0')
229 		return 1;
230 
231 	p->err = "illegal character in authority section";
232 	return 0;
233 }
234 
235 /* Routine for path_clean.  Elide the pointed .. with the preceding
236  * element.  Return 0 if it's not possible.  incr is the length of
237  * the increment, 3 for ../ and 2 for .. */
238 static int
239 path_elide_dotdot(char *path, char *i, int incr)
240 {
241 	char *j;
242 
243 	if (i == path)
244 		return 0;
245 	for (j = i-2; j != path && *j != '/'; j--)
246                 /* noop */ ;
247 	if (*j == '/')
248 		j++;
249 	i += incr;
250 	memmove(j, i, strlen(i)+1);
251 	return 1;
252 }
253 
254 /*
255  * Use an algorithm similar to the one implemented in go' path.Clean:
256  *
257  * 1. Replace multiple slashes with a single slash
258  * 2. Eliminate each . path name element
259  * 3. Eliminate each inner .. along with the non-.. element that precedes it
260  * 4. Eliminate trailing .. if possible or error (go would only discard)
261  *
262  * Unlike path.Clean, this function return the empty string if the
263  * original path is equivalent to "/".
264  */
265 static int
266 path_clean(char *path)
267 {
268 	char *i;
269 
270 	/* 1. replace multiple slashes with a single one */
271 	for (i = path; *i; ++i) {
272 		if (*i == '/' && *(i+1) == '/') {
273 			memmove(i, i+1, strlen(i)); /* move also the \0 */
274 			i--;
275 		}
276 	}
277 
278 	/* 2. eliminate each . path name element */
279 	for (i = path; *i; ++i) {
280 		if ((i == path || *i == '/') && *(i+1) == '.' &&
281 		    *(i+2) == '/') {
282 			/* move also the \0 */
283 			memmove(i, i+2, strlen(i)-1);
284 			i--;
285 		}
286 	}
287 	if (!strcmp(path, ".") || !strcmp(path, "/.")) {
288 		*path = '\0';
289 		return 1;
290 	}
291 
292 	/* 3. eliminate each inner .. along with the preceding non-.. */
293 	for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
294 		if (!path_elide_dotdot(path, i, 3))
295 			return 0;
296 
297 	/* 4. eliminate trailing ..*/
298 	if ((i = strstr(path, "..")) != NULL)
299 		if (!path_elide_dotdot(path, i, 2))
300 			return 0;
301 
302 	return 1;
303 }
304 
305 static int
306 parse_query(struct parser *p)
307 {
308 	p->parsed->query = p->uri;
309 	if (*p->uri == '\0')
310 		return 1;
311 
312 	while (unreserved(*p->uri)
313 	    || sub_delimiters(*p->uri)
314 	    || *p->uri == '/'
315 	    || *p->uri == '?'
316 	    || parse_pct_encoded(p)
317 	    || valid_multibyte_utf8(p))
318 		p->uri++;
319 
320 	if (p->err != NULL)
321 		return 0;
322 
323 	if (*p->uri != '\0' && *p->uri != '#') {
324 		p->err = "illegal character in query";
325 		return 0;
326 	}
327 
328 	if (*p->uri != '\0') {
329 		*p->uri = '\0';
330 		p->uri++;
331 	}
332 
333 	return 1;
334 }
335 
336 /* don't even bother */
337 static int
338 parse_fragment(struct parser *p)
339 {
340 	p->parsed->fragment = p->uri;
341 	return 1;
342 }
343 
344 /* XXX: is it too broad? */
345 /* *(pchar / "/") */
346 static int
347 parse_path(struct parser *p)
348 {
349 	char c;
350 
351 	p->parsed->path = p->uri;
352 	if (*p->uri == '\0') {
353 		p->parsed->query = p->parsed->fragment = p->uri;
354 		return 1;
355 	}
356 
357 	while (unreserved(*p->uri)
358 	    || sub_delimiters(*p->uri)
359 	    || *p->uri == '/'
360 	    || parse_pct_encoded(p)
361 	    || valid_multibyte_utf8(p))
362 		p->uri++;
363 
364 	if (p->err != NULL)
365 		return 0;
366 
367 	if (*p->uri != '\0' && *p->uri != '?' && *p->uri != '#') {
368 		p->err = "illegal character in path";
369 		return 0;
370 	}
371 
372 	if (*p->uri != '\0') {
373 		c = *p->uri;
374 		*p->uri = '\0';
375 		p->uri++;
376 
377 		if (c == '#') {
378 			if (!parse_fragment(p))
379 				return 0;
380 		} else
381 			if (!parse_query(p) || !parse_fragment(p))
382 				return 0;
383 	}
384 
385 	if (!path_clean(p->parsed->path)) {
386 		p->err = "illegal path";
387 		return 0;
388 	}
389 
390 	return 1;
391 }
392 
393 int
394 parse_uri(char *uri, struct uri *ret, const char **err_ret)
395 {
396 	char *end;
397 	struct parser p = {uri, ret, NULL};
398 
399 	bzero(ret, sizeof(*ret));
400 
401 	/* initialize optional stuff to the empty string */
402 	end = uri + strlen(uri);
403 	p.parsed->port = end;
404 	p.parsed->path = end;
405 	p.parsed->query = end;
406 	p.parsed->fragment = end;
407 
408 	if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
409 		*err_ret = p.err;
410 		return 0;
411 	}
412 
413 	*err_ret = NULL;
414 	return 1;
415 }
416 
417 int
418 trim_req_uri(char *uri)
419 {
420 	char *i;
421 
422 	if ((i = strstr(uri, "\r\n")) == NULL)
423 		return 0;
424 	*i = '\0';
425 	return 1;
426 }