op public repos

Blob

Date:: Sat Dec 24 18:16:47 2022 UTC
Message:: iri: first try at remove_dot_segments and mergepath
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2022 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 <errno.h>
19 #include <stddef.h>
20 #include <stdint.h>
21 #include <stdlib.h>
22 #include <string.h>
23 
24 #include "iri.h"
25 
26 /* TODO: URI -> IRI.  accept IRI but emit always URI */
27 
28 static inline int
29 cpstr(const char *start, const char *till, char *buf, size_t len)
30 {
31 	size_t		slen = till - start;
32 
33 	if (slen + 1 >= len)
34 		return (-1);
35 	memcpy(buf, start, slen);
36 	buf[slen] = '\0';
37 	return (0);
38 }
39 
40 static inline int
41 unreserved(int c)
42 {
43 	return (isalnum((unsigned char)c) ||
44 	    c == '-' ||
45 	    c == '.' ||
46 	    c == '_' ||
47 	    c == '~');
48 }
49 
50 static inline int
51 pctenc(const char *s)
52 {
53 	const char	*t = s;
54 
55 	return (t[0] == '%' &&
56 	    isxdigit((unsigned char)t[1]) &&
57 	    isxdigit((unsigned char)t[2]));
58 }
59 
60 static inline int
61 sub_delims(int c)
62 {
63 	return (c == '!' || c == '$' || c == '&' || c == '\'' ||
64 	    c == '(' || c == ')' || c == '*' || c == '+' || c == ',' ||
65 	    c == ';' || c == '=');
66 }
67 
68 static inline const char *
69 advance_pchar(const char *s)
70 {
71 	if (unreserved(*s) || sub_delims(*s) || *s == ':' || *s == '@')
72 		return (s + 1);
73 	if (pctenc(s))
74 		return (s + 3);
75 	return (NULL);
76 }
77 
78 static inline const char *
79 advance_segment(const char *s)
80 {
81 	const char	*t = s;
82 
83 	while ((t = advance_pchar(s)) != NULL)
84 		s = t;
85 	return (s);
86 }
87 
88 static inline const char *
89 advance_segment_nz(const char *s)
90 {
91 	const char	*t;
92 
93 	if ((t = advance_pchar(s)) == NULL)
94 		return (NULL);
95 	return (advance_segment(t));
96 }
97 
98 static inline const char *
99 advance_segment_nz_nc(const char *s)
100 {
101 	const char	*t = s;
102 
103 	for (;;) {
104 		if (unreserved(*t) || sub_delims(*t) || *t == '@')
105 			t++;
106 		else if (pctenc(t))
107 			t += 3;
108 		else
109 			break;
110 	}
111 
112 	return (t != s ? t : NULL);
113 }
114 
115 static const char *
116 parse_scheme(const char *s, struct iri *iri)
117 {
118 	const char	*t = s;
119 
120 	if (!isalpha((unsigned char)*t))
121 		return (NULL);
122 
123 	while (isalnum((unsigned char)*t) ||
124 	    *t == '+' ||
125 	    *t == '-' ||
126 	    *t == '.')
127 		t++;
128 
129 	if (cpstr(s, t, iri->iri_scheme, sizeof(iri->iri_scheme)) == -1)
130 		return (NULL);
131 
132 	iri->iri_flags |= IH_SCHEME;
133 	return (t);
134 }
135 
136 /* userinfo is always optional */
137 static const char *
138 parse_uinfo(const char *s, struct iri *iri)
139 {
140 	const char	*t = s;
141 
142 	for (;;) {
143 		if (unreserved(*t) || sub_delims(*t) || *t == ':')
144 			t++;
145 		else if (pctenc(t))
146 			t += 3;
147 		else
148 			break;
149 	}
150 
151 	if (*t != '@')
152 		return (s);
153 
154 	if (cpstr(s, t, iri->iri_uinfo, sizeof(iri->iri_uinfo)) == -1)
155 		return (NULL);
156 	iri->iri_flags |= IH_UINFO;
157 	return (t + 1);
158 }
159 
160 static const char *
161 parse_host(const char *s, struct iri *iri)
162 {
163 	const char	*t = s;
164 
165 	/*
166 	 * cheating a bit by relaxing and merging the rule for
167 	 * IPv6address and IPvFuture and by merging IPv4address and
168 	 * reg-name.
169 	 */
170 
171 	if (*t == '[') {
172 		while (*t && *t != ']')
173 			++t;
174 		if (*t == '\0')
175 			return (NULL);
176 		t++;
177 		if (cpstr(s, t, iri->iri_host, sizeof(iri->iri_host)) == -1)
178 			return (NULL);
179 		iri->iri_flags |= IH_HOST;
180 		return (t);
181 	}
182 
183 	for (;;) {
184 		if (unreserved(*t) || sub_delims(*t))
185 			t++;
186 		else if (pctenc(t))
187 			t += 3;
188 		else
189 			break;
190 	}
191 
192 	if (cpstr(s, t, iri->iri_host, sizeof(iri->iri_host)) == -1)
193 		return (NULL);
194 	iri->iri_flags |= IH_HOST;
195 	return (t);
196 }
197 
198 static const char *
199 parse_port(const char *s, struct iri *iri)
200 {
201 	const char	*t = s;
202 	const char	*errstr;
203 
204 	while (isdigit((unsigned char)*t))
205 		t++;
206 	if (cpstr(s, t, iri->iri_portstr, sizeof(iri->iri_portstr)) == -1)
207 		return (NULL);
208 	iri->iri_port = strtonum(iri->iri_portstr, 1, UINT16_MAX, &errstr);
209 	if (errstr)
210 		return (NULL);
211 	iri->iri_flags |= IH_PORT;
212 	return (t);
213 }
214 
215 static const char *
216 parse_authority(const char *s, struct iri *iri)
217 {
218 	const char	*t;
219 
220 	if ((t = parse_uinfo(s, iri)) == NULL)
221 		return (NULL);
222 
223 	if ((t = parse_host(t, iri)) == NULL)
224 		return (NULL);
225 
226 	if (*t == ':')
227 		return (parse_port(t, iri));
228 
229 	return (t);
230 }
231 
232 static const char *
233 parse_path_abempty(const char *s, struct iri *iri)
234 {
235 	const char	*t = s;
236 
237 	while (*t == '/')
238 		t = advance_segment(t + 1);
239 
240 	if (cpstr(s, t, iri->iri_path, sizeof(iri->iri_path)) == -1)
241 		return (NULL);
242 	iri->iri_flags |= IH_PATH;
243 	return (t);
244 }
245 
246 static const char *
247 parse_path_absolute(const char *s, struct iri *iri)
248 {
249 	const char	*t;
250 
251 	if (*s != '/')
252 		return (NULL);
253 
254 	if ((t = advance_segment_nz(s + 1)) == NULL)
255 		return (s + 1);
256 
257 	while (*t == '/')
258 		t = advance_segment(t + 1);
259 
260 	if (cpstr(s, t, iri->iri_path, sizeof(iri->iri_path)) == -1)
261 		return (NULL);
262 	iri->iri_flags |= IH_PATH;
263 	return (t);
264 }
265 
266 static const char *
267 parse_path_rootless(const char *s, struct iri *iri)
268 {
269 	const char	*t;
270 
271 	if ((t = advance_segment_nz(s)) == NULL)
272 		return (NULL);
273 
274 	while (*t == '/')
275 		t = advance_segment(t + 1);
276 
277 	if (cpstr(s, t, iri->iri_path, sizeof(iri->iri_path)) == -1)
278 		return (NULL);
279 	iri->iri_flags |= IH_PATH;
280 	return (t);
281 }
282 
283 static const char *
284 parse_path_noscheme(const char *s, struct iri *iri)
285 {
286 	const char	*t;
287 
288 	if ((t = advance_segment_nz_nc(s)) == NULL)
289 		return (NULL);
290 
291 	while (*t == '/')
292 		t = advance_segment(t + 1);
293 
294 	if (cpstr(s, t, iri->iri_path, sizeof(iri->iri_path)) == -1)
295 		return (NULL);
296 	iri->iri_flags |= IH_PATH;
297 	return (t);
298 }
299 
300 static const char *
301 parse_path_empty(const char *s, struct iri *iri)
302 {
303 	iri->iri_path[0] = '\0';
304 	iri->iri_flags |= IH_PATH;
305 	return (s);
306 }
307 
308 static const char *
309 parse_hier(const char *s, struct iri *iri)
310 {
311 	const char	*t;
312 
313 	if (!strncmp(s, "//", 2)) {
314 		if ((t = parse_authority(s + 2, iri)) == NULL)
315 			return (NULL);
316 		return (parse_path_abempty(t, iri));
317 	}
318 
319 	if ((t = parse_path_absolute(s, iri)) != NULL)
320 		return (t);
321 
322 	if ((t = parse_path_rootless(s, iri)) != NULL)
323 		return (t);
324 
325 	return (parse_path_empty(s, iri));
326 }
327 
328 static const char *
329 parse_relative(const char *s, struct iri *iri)
330 {
331 	const char	*t = s;
332 
333 	if (!strncmp(s, "//", 2)) {
334 		if ((t = parse_authority(s + 2, iri)) == NULL)
335 			return (NULL);
336 		return (parse_path_abempty(t, iri));
337 	}
338 
339 	if ((t = parse_path_absolute(s, iri)) != NULL)
340 		return (t);
341 
342 	if ((t = parse_path_noscheme(s, iri)) != NULL)
343 		return (t);
344 
345 	return (parse_path_empty(s, iri));
346 }
347 
348 static const char *
349 parse_query(const char *s, struct iri *iri)
350 {
351 	const char	*n, *t = s;
352 
353 	for (;;) {
354 		if ((n = advance_pchar(t)) != NULL)
355 			t = n;
356 		else if (*t == '/' || *t == '?')
357 			t++;
358 		else
359 			break;
360 	}
361 
362 	if (cpstr(s, t, iri->iri_query, sizeof(iri->iri_query)) == -1)
363 		return (NULL);
364 	iri->iri_flags |= IH_QUERY;
365 	return (t);
366 }
367 
368 static int
369 parse_uri(const char *s, struct iri *iri)
370 {
371 	if ((s = parse_scheme(s, iri)) == NULL)
372 		return (-1);
373 
374 	if (*s != ':')
375 		return (-1);
376 
377 	if ((s = parse_hier(s + 1, iri)) == NULL)
378 		return (-1);
379 
380 	if (*s == '?' && (s = parse_query(s + 1, iri)) == NULL)
381 		return (-1);
382 
383 	/* skip fragments */
384 	if (*s == '#' || *s == '\0')
385 		return (0);
386 
387 	return (-1);
388 }
389 
390 static int
391 parse_relative_ref(const char *s, struct iri *iri)
392 {
393 	if ((s = parse_relative(s, iri)) == NULL)
394 		return (-1);
395 
396 	if (*s == '?' && (s = parse_query(s + 1, iri)) == NULL)
397 		return (-1);
398 
399 	/* skip fragments */
400 	if (*s == '#' || *s == '\0')
401 		return (0);
402 
403 	return (-1);
404 }
405 
406 static int
407 parse(const char *s, struct iri *iri)
408 {
409 	iri->iri_flags = 0;
410 
411 	if (s == NULL)
412 		return (0);
413 
414 	if (parse_uri(s, iri) == -1) {
415 		iri->iri_flags = 0;
416 		if (parse_relative_ref(s, iri) == -1)
417 			return (-1);
418 	}
419 
420 	return (0);
421 }
422 
423 static inline void
424 lowerify(char *s)
425 {
426 	for (; *s; ++s)
427 		*s = tolower((unsigned char)*s);
428 }
429 
430 static void
431 cpfields(struct iri *dest, const struct iri *src, int flags)
432 {
433 	if (flags & IH_SCHEME) {
434 		dest->iri_flags |= IH_SCHEME;
435 		if (src->iri_flags & IH_SCHEME)
436 			memcpy(dest->iri_scheme, src->iri_scheme,
437 			    sizeof(dest->iri_scheme));
438 		lowerify(dest->iri_scheme);
439 	}
440 	if (flags & IH_UINFO) {
441 		dest->iri_flags |= IH_UINFO;
442 		if (src->iri_flags & IH_UINFO)
443 			memcpy(dest->iri_uinfo, src->iri_uinfo,
444 			    sizeof(dest->iri_uinfo));
445 	}
446 	if (flags & IH_HOST) {
447 		dest->iri_flags |= IH_HOST;
448 		if (src->iri_flags & IH_HOST)
449 			memcpy(dest->iri_host, src->iri_host,
450 			    sizeof(dest->iri_host));
451 		lowerify(dest->iri_host);
452 	}
453 	if (flags & IH_PORT) {
454 		dest->iri_flags |= IH_PORT;
455 		if (src->iri_flags & IH_PORT)
456 			dest->iri_port = src->iri_port;
457 	}
458 	if (flags & IH_PATH) {
459 		dest->iri_flags |= IH_PATH;
460 		if (src->iri_flags & IH_PATH)
461 			memcpy(dest->iri_path, src->iri_path,
462 			    sizeof(dest->iri_path));
463 	}
464 	if (flags & IH_QUERY) {
465 		dest->iri_flags |= IH_QUERY;
466 		if (src->iri_flags & IH_QUERY)
467 			memcpy(dest->iri_query, src->iri_query,
468 			    sizeof(dest->iri_query));
469 	}
470 }
471 
472 static inline int
473 remove_dot_segments(struct iri *iri)
474 {
475 	char		*p, *q, *buf;
476 	ptrdiff_t	 bufsize;
477 
478 	buf = p = q = iri->iri_path;
479 	bufsize = sizeof(iri->iri_path);
480 
481 	while (*p && (q - buf < bufsize)) {
482 		/* A */
483 		if (!strncmp(p, "../", 3)) {
484 			p += 3;
485 			continue;
486 		}
487 		if (!strncmp(p, "./", 3)) {
488 			p += 2;
489 			continue;
490 		}
491 		/* B */
492 		if (!strncmp(p, "/./", 3)) {
493 			*q++ = '/';
494 			p += 3;
495 			continue;
496 		}
497 		if (!strcmp(p, "/.")) {
498 			p += 2;
499 			break;
500 		}
501 		/* C */
502 		if (p[0] == '/' && p[1] == '.' && p[2] == '.' &&
503 		    (p[3] == '/' || p[3] == '\0')) {
504 			p += 3;
505 			while (q != buf && *--q != '/')
506 				continue;
507 			continue;
508 		}
509 		/* D */
510 		if (!strcmp(p, ".")) {
511 			p++;
512 			break;
513 		}
514 		if (!strcmp(p, "..")) {
515 			p += 2;
516 			break;
517 		}
518 		/* E */
519 		while (*p && *p != '/' && (q - buf < bufsize))
520 			*q++ = *p++;
521 	}
522 
523 	if (*p == '\0' && (q - buf < bufsize)) {
524 		*q = '\0';
525 		return (0);
526 	}
527 
528 	errno = ENAMETOOLONG;
529 	return (-1);
530 }
531 
532 static inline int
533 mergepath(struct iri *i, struct iri *base, struct iri *r)
534 {
535 	const char	*bpath, *rpath, *s;
536 
537 	bpath = (base->iri_flags & IH_PATH) ? base->iri_path : "/";
538 	rpath = (r->iri_flags & IH_PATH) ? r->iri_path : "/";
539 
540 	i->iri_flags |= IH_PATH;
541 	i->iri_path[0] = '\0';
542 
543 	if ((base->iri_flags & IH_AUTHORITY) &&
544 	    (*bpath == '\0' || !strcmp(bpath, "/"))) {
545 		if (*rpath == '/')
546 			rpath++;
547 		strlcpy(i->iri_path, "/", sizeof(i->iri_path));
548 		strlcat(i->iri_path, rpath, sizeof(i->iri_path));
549 		return (0);
550 	}
551 
552 	if ((s = strrchr(bpath, '/')) != NULL) {
553 		cpstr(bpath, s + 1, i->iri_path, sizeof(i->iri_path));
554 		if (*rpath == '/')
555 			rpath++;
556 	}
557 	if (strlcat(i->iri_path, rpath, sizeof(i->iri_path)) >=
558 	    sizeof(i->iri_path)) {
559 		errno = ENAMETOOLONG;
560 		return (-1);
561 	}
562 
563 	return (0);
564 }
565 
566 int
567 iri_parse(const char *base, const char *str, struct iri *iri)
568 {
569 	static struct iri	ibase, iparsed;
570 
571 	memset(iri, 0, sizeof(*iri));
572 
573 	if (base == NULL) {
574 		ibase.iri_flags = 0;
575 		if (parse_uri(str, &iparsed) == -1)
576 			return (-1);
577 	} else {
578 		if (parse_uri(base, &ibase) == -1)
579 			return (-1);
580 		if (parse(str, &iparsed) == -1)
581 			return (-1);
582 	}
583 
584 	if (iparsed.iri_flags & IH_SCHEME) {
585 		cpfields(iri, &iparsed, iparsed.iri_flags);
586 		remove_dot_segments(iri);
587 		return (0);
588 	}
589 
590 	/* if fragments are supported, copy iparsed fragment to iri */
591 
592 	cpfields(iri, &ibase, IH_SCHEME);
593 
594 	if (iparsed.iri_flags & IH_HOST) {
595 		cpfields(iri, &iparsed, IH_AUTHORITY|IH_PATH|IH_QUERY);
596 		remove_dot_segments(iri);
597 		return (0);
598 	}
599 
600 	cpfields(iri, &ibase, IH_AUTHORITY);
601 
602 	if ((iparsed.iri_flags & IH_PATH) && *iparsed.iri_path == '\0') {
603 		cpfields(iri, &ibase, IH_PATH);
604 		if (iparsed.iri_flags & IH_QUERY)
605 			cpfields(iri, &iparsed, IH_QUERY);
606 		else
607 			cpfields(iri, &ibase, IH_QUERY);
608 		return (0);
609 	}
610 
611 	cpfields(iri, &iparsed, IH_QUERY);
612 	if ((iparsed.iri_flags & IH_PATH) && !strcmp(iparsed.iri_path, "/"))
613 		cpfields(iri, &iparsed, IH_PATH);
614 	else {
615 		if (!(ibase.iri_flags & IH_PATH))
616 			ibase.iri_path[0] = '\0';
617 		if (!(iparsed.iri_flags & IH_PATH))
618 			iparsed.iri_path[0] = '\0';
619 		mergepath(iri->iri_path, sizeof(iri->iri_path),
620 		    ibase.iri_path, iparsed.iri_path);
621 	}
622 	remove_dot_segments(iri);
623 	cpfields(iri, &ibase, IH_QUERY);
624 	return (0);
625 }
626 
627 int
628 iri_unparse(const struct iri *iri, char *buf, size_t buflen)
629 {
630 	memset(buf, 0, buflen);
631 	return (-1);
632 }
633 
634 int
635 iri_human(const struct iri *iri, char *buf, size_t buflen)
636 {
637 	memset(buf, 0, buflen);
638 	return (-1);
639 }
640 
641 int
642 iri_setquery(struct iri *iri, const char *text)
643 {
644 	return (-1);
645 }