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 static inline int
23 unreserved(int p)
24 {
25 	return isalnum(p)
26 		|| p == '-'
27 		|| p == '.'
28 		|| p == '_'
29 		|| p == '~';
30 }
31 
32 static inline int
33 sub_delimiters(int p)
34 {
35 	return p == '!'
36 		|| p == '$'
37 		|| p == '&'
38 		|| p == '\''
39 		|| p == '('
40 		|| p == ')'
41 		|| p == '*'
42 		|| p == '+'
43 		|| p == ','
44 		|| p == ';'
45 		|| p == '=';
46 }
47 
48 static int
49 valid_pct_enc_string(char *s)
50 {
51 	if (*s != '%')
52 		return 1;
53 
54 	if (!isxdigit(s[1]) || !isxdigit(s[2]))
55 		return 0;
56 
57 	if (s[1] == '0' && s[2] == '0')
58 		return 0;
59 
60 	return 1;
61 }
62 
63 static int
64 valid_pct_encoded(struct parser *p)
65 {
66 	if (p->iri[0] != '%')
67 		return 0;
68 
69 	if (!valid_pct_enc_string(p->iri)) {
70 		p->err = "illegal percent-encoding";
71 		return 0;
72 	}
73 
74 	p->iri += 2;
75 	return 1;
76 }
77 
78 static int
79 parse_pct_encoded(struct parser *p)
80 {
81 	if (p->iri[0] != '%')
82 		return 0;
83 
84 	if (!valid_pct_enc_string(p->iri)) {
85 		p->err = "illegal percent-encoding";
86 		return 0;
87 	}
88 
89 	sscanf(p->iri+1, "%2hhx", p->iri);
90 	memmove(p->iri+1, p->iri+3, strlen(p->iri+3)+1);
91 	if (*p->iri == '\0') {
92 		p->err = "illegal percent-encoding";
93 		return 0;
94 	}
95 
96 	return 1;
97 }
98 
99 /* ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) "://" */
100 static int
101 parse_scheme(struct parser *p)
102 {
103 	p->parsed->schema = p->iri;
104 
105 	if (!isalpha(*p->iri)) {
106 		p->err = "illegal character in scheme";
107 		return 0;
108 	}
109 
110 	do {
111 		/* normalize the scheme (i.e. lowercase it)
112 		 *
113 		 * XXX: since we cannot have good things, tolower
114 		 * behaviour depends on the LC_CTYPE locale.  The good
115 		 * news is that we're sure p->iri points to something
116 		 * that's in the ASCII range, so tolower can't
117 		 * mis-behave on some systems due to the locale. */
118 		*p->iri = tolower(*p->iri);
119 		p->iri++;
120 	} while (isalnum(*p->iri)
121 	    || *p->iri == '+'
122 	    || *p->iri == '-'
123 	    || *p->iri == '.');
124 
125 	if (*p->iri != ':') {
126 		p->err = "illegal character in scheme";
127 		return 0;
128 	}
129 
130 	*p->iri = '\0';
131 	if (p->iri[1] != '/' || p->iri[2] != '/') {
132 		p->err = "invalid marker after scheme";
133 		return 0;
134 	}
135 
136 	p->iri += 3;
137 	return 1;
138 }
139 
140 /* *DIGIT */
141 static int
142 parse_port(struct parser *p)
143 {
144 	uint32_t i = 0;
145 
146 	p->parsed->port = p->iri;
147 
148 	for (; isdigit(*p->iri); p->iri++) {
149 		i = i * 10 + *p->iri - '0';
150 		if (i > UINT16_MAX) {
151 			p->err = "port number too large";
152 			return 0;
153 		}
154 	}
155 
156 	if (*p->iri != '/' && *p->iri != '\0') {
157 		p->err = "illegal character in port number";
158 		return 0;
159 	}
160 
161 	p->parsed->port_no = i;
162 
163 	if (*p->iri != '\0') {
164 		*p->iri = '\0';
165 		p->iri++;
166 	}
167 
168 	return 1;
169 }
170 
171 /* TODO: add support for ip-literal and ipv4addr ? */
172 /* *( unreserved / sub-delims / pct-encoded ) */
173 static int
174 parse_authority(struct parser *p)
175 {
176 	p->parsed->host = p->iri;
177 
178 	while (unreserved(*p->iri)
179 	    || sub_delimiters(*p->iri)
180 	    || parse_pct_encoded(p)
181 	    || valid_multibyte_utf8(p)) {
182 		/* normalize the host name. */
183 		if (*p->iri < 0x7F)
184 			*p->iri = tolower(*p->iri);
185 		p->iri++;
186 	}
187 
188 	if (p->err != NULL)
189 		return 0;
190 
191 	if (*p->iri == ':') {
192 		*p->iri = '\0';
193 		p->iri++;
194 		return parse_port(p);
195 	} else
196 		p->parsed->port_no = 1965;
197 
198 	if (*p->iri == '/') {
199 		*p->iri = '\0';
200 		p->iri++;
201 		return 1;
202 	}
203 
204 	if (*p->iri == '\0')
205 		return 1;
206 
207 	p->err = "illegal character in authority section";
208 	return 0;
209 }
210 
211 /* Routine for path_clean.  Elide the pointed .. with the preceding
212  * element.  Return 0 if it's not possible.  incr is the length of
213  * the increment, 3 for ../ and 2 for .. */
214 static int
215 path_elide_dotdot(char *path, char *i, int incr)
216 {
217 	char *j;
218 
219 	if (i == path)
220 		return 0;
221 	for (j = i-2; j != path && *j != '/'; j--)
222                 /* noop */ ;
223 	if (*j == '/')
224 		j++;
225 	i += incr;
226 	memmove(j, i, strlen(i)+1);
227 	return 1;
228 }
229 
230 /*
231  * Use an algorithm similar to the one implemented in go' path.Clean:
232  *
233  * 1. Replace multiple slashes with a single slash
234  * 2. Eliminate each . path name element
235  * 3. Eliminate each inner .. along with the non-.. element that precedes it
236  * 4. Eliminate trailing .. if possible or error (go would only discard)
237  *
238  * Unlike path.Clean, this function return the empty string if the
239  * original path is equivalent to "/".
240  */
241 static int
242 path_clean(char *path)
243 {
244 	char *i;
245 
246 	/* 1. replace multiple slashes with a single one */
247 	for (i = path; *i; ++i) {
248 		if (*i == '/' && *(i+1) == '/') {
249 			memmove(i, i+1, strlen(i)); /* move also the \0 */
250 			i--;
251 		}
252 	}
253 
254 	/* 2. eliminate each . path name element */
255 	for (i = path; *i; ++i) {
256 		if ((i == path || *i == '/') &&
257 		    *i != '.' && i[1] == '.' && i[2] == '/') {
258 			/* move also the \0 */
259 			memmove(i, i+2, strlen(i)-1);
260 			i--;
261 		}
262 	}
263 	if (!strcmp(path, ".") || !strcmp(path, "/.")) {
264 		*path = '\0';
265 		return 1;
266 	}
267 
268 	/* 3. eliminate each inner .. along with the preceding non-.. */
269 	for (i = strstr(path, "../"); i != NULL; i = strstr(path, ".."))
270 		if (!path_elide_dotdot(path, i, 3))
271 			return 0;
272 
273 	/* 4. eliminate trailing ..*/
274 	if ((i = strstr(path, "..")) != NULL)
275 		if (!path_elide_dotdot(path, i, 2))
276 			return 0;
277 
278 	return 1;
279 }
280 
281 static int
282 parse_query(struct parser *p)
283 {
284 	p->parsed->query = p->iri;
285 	if (*p->iri == '\0')
286 		return 1;
287 
288 	while (unreserved(*p->iri)
289 	    || sub_delimiters(*p->iri)
290 	    || *p->iri == '/'
291 	    || *p->iri == '?'
292 	    || *p->iri == ':'
293 	    || *p->iri == '@'
294 	    || valid_pct_encoded(p)
295 	    || valid_multibyte_utf8(p))
296 		p->iri++;
297 
298 	if (p->err != NULL)
299 		return 0;
300 
301 	if (*p->iri != '\0' && *p->iri != '#') {
302 		p->err = "illegal character in query";
303 		return 0;
304 	}
305 
306 	if (*p->iri != '\0') {
307 		*p->iri = '\0';
308 		p->iri++;
309 	}
310 
311 	return 1;
312 }
313 
314 /* don't even bother */
315 static int
316 parse_fragment(struct parser *p)
317 {
318 	p->parsed->fragment = p->iri;
319 	return 1;
320 }
321 
322 /* XXX: is it too broad? */
323 /* *(pchar / "/") */
324 static int
325 parse_path(struct parser *p)
326 {
327 	char c;
328 
329 	/* trim initial slashes */
330 	while (*p->iri == '/')
331 		p->iri++;
332 
333 	p->parsed->path = p->iri;
334 	if (*p->iri == '\0') {
335 		p->parsed->query = p->parsed->fragment = p->iri;
336 		return 1;
337 	}
338 
339 	while (unreserved(*p->iri)
340 	    || sub_delimiters(*p->iri)
341 	    || *p->iri == '/'
342 	    || parse_pct_encoded(p)
343 	    || valid_multibyte_utf8(p))
344 		p->iri++;
345 
346 	if (p->err != NULL)
347 		return 0;
348 
349 	if (*p->iri != '\0' && *p->iri != '?' && *p->iri != '#') {
350 		p->err = "illegal character in path";
351 		return 0;
352 	}
353 
354 	if (*p->iri != '\0') {
355 		c = *p->iri;
356 		*p->iri = '\0';
357 		p->iri++;
358 
359 		if (c == '#') {
360 			if (!parse_fragment(p))
361 				return 0;
362 		} else
363 			if (!parse_query(p) || !parse_fragment(p))
364 				return 0;
365 	}
366 
367 	if (!path_clean(p->parsed->path)) {
368 		p->err = "illegal path";
369 		return 0;
370 	}
371 
372 	return 1;
373 }
374 
375 int
376 parse_iri(char *iri, struct iri *ret, const char **err_ret)
377 {
378 	char *end;
379 	struct parser p = {iri, ret, NULL};
380 
381 	bzero(ret, sizeof(*ret));
382 
383 	/* initialize optional stuff to the empty string */
384 	end = iri + strlen(iri);
385 	p.parsed->host = end;
386 	p.parsed->port = end;
387 	p.parsed->path = end;
388 	p.parsed->query = end;
389 	p.parsed->fragment = end;
390 
391 	if (!parse_scheme(&p) || !parse_authority(&p) || !parse_path(&p)) {
392 		*err_ret = p.err;
393 		return 0;
394 	}
395 
396 	*err_ret = NULL;
397 	return 1;
398 }
399 
400 int
401 trim_req_iri(char *iri, const char **err)
402 {
403 	char *i;
404 
405 	if ((i = strstr(iri, "\r\n")) == NULL) {
406 		*err = "missing CRLF";
407 		return 0;
408 	}
409 	*i = '\0';
410 	return 1;
411 }
412 
413 
414 int
415 serialize_iri(struct iri *i, char *buf, size_t len)
416 {
417 	size_t l;
418 
419 	/* in ex.c we receive empty "" strings as NULL */
420 	if (i->schema == NULL || i->host == NULL) {
421 		memset(buf, 0, len);
422 		return 0;
423 	}
424 
425 	strlcpy(buf, i->schema, len);
426 	strlcat(buf, "://", len);
427         strlcat(buf, i->host, len);
428 	strlcat(buf, "/", len);
429 
430 	if (i->path != NULL)
431 		l = strlcat(buf, i->path, len);
432 
433 	if (i->query != NULL && *i->query != '\0') {
434 		strlcat(buf, "?", len);
435 		l = strlcat(buf, i->query, len);
436 	}
437 
438 	return l < len;
439 }