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 struct parser {
91 	char		*uri;
92 	struct uri	*parsed;
93 	const char	*err;
94 };
95 
96 /* XXX: these macros will expand multiple times their argument */
97 
98 #define UNRESERVED(p)				\
99 	(isalnum(p)				\
100 	    || p == '-'				\
101 	    || p == '.'				\
102 	    || p == '_'				\
103 	    || p == '~')
104 
105 #define SUB_DELIMITERS(p)			\
106 	(p == '!'				\
107 	    || p == '$'				\
108 	    || p == '&'				\
109 	    || p == '\''			\
110 	    || p == '('				\
111 	    || p == ')'				\
112 	    || p == '*'				\
113 	    || p == '+'				\
114 	    || p == ','				\
115 	    || p == ';'				\
116 	    || p == '=')
117 
118 static int
119 parse_pct_encoded(struct parser *p)
120 {
121 	if (*p->uri != '%')
122 		return 0;
123 
124 	if (!isxdigit(*(p->uri+1)) || !isxdigit(*(p->uri+2))) {
125 		p->err = "illegal percent-encoding";
126 		return 0;
127 	}
128 
129 	sscanf(p->uri+1, "%2hhx", p->uri);
130 	memmove(p->uri+1, p->uri+3, strlen(p->uri+3)+1);
131 
132 	return 1;
133 }
134 
135 /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
136 static int
137 parse_scheme(struct parser *p)
138 {
139 	p->parsed->schema = p->uri;
140 
141 	if (!isalpha(*p->uri)) {
142 		p->err = "illegal character in scheme";
143 		return 0;
144 	}
145 
146 	p->uri++;
147 	while (isalnum(*p->uri)
148 	    || *p->uri == '+'
149 	    || *p->uri == '-'
150 	    || *p->uri == '.')
151 		p->uri++;
152 
153 	if (*p->uri != ':') {
154 		p->err = "illegal character in scheme";
155 		return 0;
156 	}
157 
158 	*p->uri = '\0';
159 	if (*(++p->uri) != '/' || *(++p->uri) != '/') {
160 		p->err = "invalid marker after scheme";
161 		return 0;
162 	}
163 
164 	p->uri++;
165 	return 1;
166 }
167 
168 /* *DIGIT */
169 static int
170 parse_port(struct parser *p)
171 {
172 	uint32_t i = 0;
173 
174 	p->parsed->port = p->uri;
175 
176 	for (; isdigit(*p->uri); p->uri++) {
177 		i = i * 10 + *p->uri - '0';
178 		if (i > UINT16_MAX) {
179 			p->err = "port number too large";
180 			return 0;
181 		}
182 	}
183 
184 	if (*p->uri != '/' && *p->uri != '\0') {
185 		p->err = "illegal character in port number";
186 		return 0;
187 	}
188 
189 	p->parsed->port_no = i;
190 
191 	if (*p->uri != '\0') {
192 		*p->uri = '\0';
193 		p->uri++;
194 	}
195 
196 	return 1;
197 }
198 
199 /* TODO: add support for ip-literal and ipv4addr ? */
200 /* *( unreserved / sub-delims / pct-encoded ) */
201 static int
202 parse_authority(struct parser *p)
203 {
204 	p->parsed->host = p->uri;
205 
206 	while (UNRESERVED(*p->uri)
207 	    || SUB_DELIMITERS(*p->uri)
208 	    || parse_pct_encoded(p))
209 		p->uri++;
210 
211 	if (*p->uri == ':') {
212 		*p->uri = '\0';
213 		p->uri++;
214 		return parse_port(p);
215 	}
216 
217 	if (*p->uri == '/') {
218 		*p->uri = '\0';
219 		p->uri++;
220 		return 1;
221 	}
222 
223 	if (*p->uri == '\0')
224 		return 1;
225 
226 	p->err = "illegal character in authority section";
227 	return 0;
228 }
229 
230 /* Routine for path_clean.  Elide the pointed .. with the preceding
231  * element.  Return 0 if it's not possible.  incr is the length of
232  * the increment, 3 for ../ and 2 for .. */
233 static int
234 path_elide_dotdot(char *path, char *i, int incr)
235 {
236 	char *j;
237 
238 	if (i == path)
239 		return 0;
240 	for (j = i-2; j != path && *j != '/'; j--)
241                 /* noop */ ;
242 	if (*j == '/')
243 		j++;
244 	i += incr;
245 	memmove(j, i, strlen(i)+1);
246 	return 1;
247 }
248 
249 /*
250  * Use an algorithm similar to the one implemented in go' path.Clean:
251  *
252  * 1. Replace multiple slashes with a single slash
253  * 2. Eliminate each . path name element
254  * 3. Eliminate each inner .. along with the non-.. element that precedes it
255  * 4. Eliminate trailing .. if possible or error (go would only discard)
256  *
257  * Unlike path.Clean, this function return the empty string if the
258  * original path is equivalent to "/".
259  */
260 static int
261 path_clean(char *path)
262 {
263 	char *i;
264 
265 	/* 1. replace multiple slashes with a single one */
266 	for (i = path; *i; ++i) {
267 		if (*i == '/' && *(i+1) == '/') {
268 			memmove(i, i+1, strlen(i)); /* move also the \0 */
269 			i--;
270 		}
271 	}
272 
273 	/* 2. eliminate each . path name element */
274 	for (i = path; *i; ++i) {
275 		if ((i == path || *i == '/') && *(i+1) == '.' &&
276 		    *(i+2) == '/') {
277 			/* move also the \0 */
278 			memmove(i, i+2, strlen(i)-1);
279 			i--;
280 		}
281 	}
282 	if (!strcmp(path, ".") || !strcmp(path, "/.")) {
283 		*path = '\0';
284 		return 1;
285 	}
286 
287 	/* 3. eliminate each inner .. along with the preceding non-.. */
288 	for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
289 		if (!path_elide_dotdot(path, i, 3))
290 			return 0;
291 
292 	/* 4. eliminate trailing ..*/
293 	if ((i = strstr(path, "..")) != NULL)
294 		if (!path_elide_dotdot(path, i, 2))
295 			return 0;
296 
297 	return 1;
298 }
299 
300 static int
301 parse_query(struct parser *p)
302 {
303 	p->parsed->query = p->uri;
304 	if (*p->uri == '\0')
305 		return 1;
306 
307 	while (UNRESERVED(*p->uri)
308 	    || SUB_DELIMITERS(*p->uri)
309 	    || *p->uri == '/'
310 	    || *p->uri == '?'
311 	    || parse_pct_encoded(p))
312 		p->uri++;
313 
314 	if (*p->uri != '\0' && *p->uri != '#') {
315 		p->err = "illegal character in query";
316 		return 0;
317 	}
318 
319 	if (*p->uri != '\0') {
320 		*p->uri = '\0';
321 		p->uri++;
322 	}
323 
324 	return 1;
325 }
326 
327 /* don't even bother */
328 static int
329 parse_fragment(struct parser *p)
330 {
331 	p->parsed->fragment = p->uri;
332 	return 1;
333 }
334 
335 /* XXX: is it too broad? */
336 /* *(pchar / "/") */
337 static int
338 parse_path(struct parser *p)
339 {
340 	char c;
341 
342 	p->parsed->path = p->uri;
343 	if (*p->uri == '\0') {
344 		p->parsed->query = p->parsed->fragment = p->uri;
345 		return 1;
346 	}
347 
348 	while (UNRESERVED(*p->uri)
349 	    || SUB_DELIMITERS(*p->uri)
350 	    || *p->uri == '/'
351 	    || parse_pct_encoded(p))
352 		p->uri++;
353 
354 	if (*p->uri != '\0' && *p->uri != '?' && *p->uri != '#') {
355 		p->err = "illegal character in path";
356 		return 0;
357 	}
358 
359 	if (*p->uri != '\0') {
360 		c = *p->uri;
361 		*p->uri = '\0';
362 		p->uri++;
363 
364 		if (c == '#') {
365 			if (!parse_fragment(p))
366 				return 0;
367 		} else
368 			if (!parse_query(p) || !parse_fragment(p))
369 				return 0;
370 	}
371 
372 	if (!path_clean(p->parsed->path)) {
373 		p->err = "illegal path";
374 		return 0;
375 	}
376 
377 	return 1;
378 }
379 
380 int
381 parse_uri(char *uri, struct uri *ret, const char **err_ret)
382 {
383 	char *end;
384 	struct parser p = {uri, ret, NULL};
385 
386 	bzero(ret, sizeof(*ret));
387 
388 	/* initialize optional stuff to the empty string */
389 	end = uri + strlen(uri);
390 	p.parsed->port = end;
391 	p.parsed->path = end;
392 	p.parsed->query = end;
393 	p.parsed->fragment = end;
394 
395 	if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
396 		*err_ret = p.err;
397 		return 0;
398 	}
399 
400 	*err_ret = NULL;
401 	return 1;
402 }
403 
404 int
405 trim_req_uri(char *uri)
406 {
407 	char *i;
408 
409 	if ((i = strstr(uri, "\r\n")) == NULL)
410 		return 0;
411 	*i = '\0';
412 	return 1;
413 }