Blob


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 */
17 #include <ctype.h>
18 #include <errno.h>
19 #include <stddef.h>
20 #include <stdint.h>
21 #include <stdlib.h>
22 #include <string.h>
24 #include "iri.h"
26 /* TODO: URI -> IRI. accept IRI but emit always URI */
28 static inline int
29 cpstr(const char *start, const char *till, char *buf, size_t len)
30 {
31 size_t slen = till - start;
33 if (slen + 1 >= len)
34 return (-1);
35 memcpy(buf, start, slen);
36 buf[slen] = '\0';
37 return (0);
38 }
40 static inline int
41 unreserved(int c)
42 {
43 return (isalnum((unsigned char)c) ||
44 c == '-' ||
45 c == '.' ||
46 c == '_' ||
47 c == '~');
48 }
50 static inline int
51 pctenc(const char *s)
52 {
53 const char *t = s;
55 return (t[0] == '%' &&
56 isxdigit((unsigned char)t[1]) &&
57 isxdigit((unsigned char)t[2]));
58 }
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 }
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 }
78 static inline const char *
79 advance_segment(const char *s)
80 {
81 const char *t = s;
83 while ((t = advance_pchar(s)) != NULL)
84 s = t;
85 return (s);
86 }
88 static inline const char *
89 advance_segment_nz(const char *s)
90 {
91 const char *t;
93 if ((t = advance_pchar(s)) == NULL)
94 return (NULL);
95 return (advance_segment(t));
96 }
98 static inline const char *
99 advance_segment_nz_nc(const char *s)
101 const char *t = s;
103 for (;;) {
104 if (unreserved(*t) || sub_delims(*t) || *t == '@')
105 t++;
106 else if (pctenc(t))
107 t += 3;
108 else
109 break;
112 return (t != s ? t : NULL);
115 static const char *
116 parse_scheme(const char *s, struct iri *iri)
118 const char *t = s;
120 if (!isalpha((unsigned char)*t))
121 return (NULL);
123 while (isalnum((unsigned char)*t) ||
124 *t == '+' ||
125 *t == '-' ||
126 *t == '.')
127 t++;
129 if (cpstr(s, t, iri->iri_scheme, sizeof(iri->iri_scheme)) == -1)
130 return (NULL);
132 iri->iri_flags |= IH_SCHEME;
133 return (t);
136 /* userinfo is always optional */
137 static const char *
138 parse_uinfo(const char *s, struct iri *iri)
140 const char *t = s;
142 for (;;) {
143 if (unreserved(*t) || sub_delims(*t) || *t == ':')
144 t++;
145 else if (pctenc(t))
146 t += 3;
147 else
148 break;
151 if (*t != '@')
152 return (s);
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);
160 static const char *
161 parse_host(const char *s, struct iri *iri)
163 const char *t = s;
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 */
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);
183 for (;;) {
184 if (unreserved(*t) || sub_delims(*t))
185 t++;
186 else if (pctenc(t))
187 t += 3;
188 else
189 break;
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);
198 static const char *
199 parse_port(const char *s, struct iri *iri)
201 const char *t = s;
202 const char *errstr;
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);
215 static const char *
216 parse_authority(const char *s, struct iri *iri)
218 const char *t;
220 if ((t = parse_uinfo(s, iri)) == NULL)
221 return (NULL);
223 if ((t = parse_host(t, iri)) == NULL)
224 return (NULL);
226 if (*t == ':')
227 return (parse_port(t + 1, iri));
229 return (t);
232 static const char *
233 parse_path_abempty(const char *s, struct iri *iri)
235 const char *t = s;
237 while (*t == '/')
238 t = advance_segment(t + 1);
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);
246 static const char *
247 parse_path_absolute(const char *s, struct iri *iri)
249 const char *t;
251 if (*s != '/')
252 return (NULL);
254 if ((t = advance_segment_nz(s + 1)) == NULL)
255 return (s + 1);
257 while (*t == '/')
258 t = advance_segment(t + 1);
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);
266 static const char *
267 parse_path_rootless(const char *s, struct iri *iri)
269 const char *t;
271 if ((t = advance_segment_nz(s)) == NULL)
272 return (NULL);
274 while (*t == '/')
275 t = advance_segment(t + 1);
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);
283 static const char *
284 parse_path_noscheme(const char *s, struct iri *iri)
286 const char *t;
288 if ((t = advance_segment_nz_nc(s)) == NULL)
289 return (NULL);
291 while (*t == '/')
292 t = advance_segment(t + 1);
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);
300 static const char *
301 parse_path_empty(const char *s, struct iri *iri)
303 iri->iri_path[0] = '\0';
304 iri->iri_flags |= IH_PATH;
305 return (s);
308 static const char *
309 parse_hier(const char *s, struct iri *iri)
311 const char *t;
313 if (!strncmp(s, "//", 2)) {
314 if ((t = parse_authority(s + 2, iri)) == NULL)
315 return (NULL);
316 return (parse_path_abempty(t, iri));
319 if ((t = parse_path_absolute(s, iri)) != NULL)
320 return (t);
322 if ((t = parse_path_rootless(s, iri)) != NULL)
323 return (t);
325 return (parse_path_empty(s, iri));
328 static const char *
329 parse_relative(const char *s, struct iri *iri)
331 const char *t = s;
333 if (!strncmp(s, "//", 2)) {
334 if ((t = parse_authority(s + 2, iri)) == NULL)
335 return (NULL);
336 return (parse_path_abempty(t, iri));
339 if ((t = parse_path_absolute(s, iri)) != NULL)
340 return (t);
342 if ((t = parse_path_noscheme(s, iri)) != NULL)
343 return (t);
345 return (parse_path_empty(s, iri));
348 static const char *
349 parse_query(const char *s, struct iri *iri)
351 const char *n, *t = s;
353 for (;;) {
354 if ((n = advance_pchar(t)) != NULL)
355 t = n;
356 else if (*t == '/' || *t == '?')
357 t++;
358 else
359 break;
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);
368 static const char *
369 parse_fragment(const char *s, struct iri *iri)
371 const char *n, *t = s;
373 for (;;) {
374 if ((n = advance_pchar(t)) != NULL)
375 t = n;
376 else if (*t == '/' || *t == '?')
377 t++;
378 else
379 break;
382 if (cpstr(s, t, iri->iri_fragment, sizeof(iri->iri_fragment)) == -1)
383 return (NULL);
384 iri->iri_flags |= IH_FRAGMENT;
385 return (t);
388 static int
389 parse_uri(const char *s, struct iri *iri)
391 if ((s = parse_scheme(s, iri)) == NULL)
392 return (-1);
394 if (*s != ':')
395 return (-1);
397 if ((s = parse_hier(s + 1, iri)) == NULL)
398 return (-1);
400 if (*s == '?' && (s = parse_query(s + 1, iri)) == NULL)
401 return (-1);
403 if (*s == '#' && (s = parse_fragment(s + 1, iri)) == NULL)
404 return (-1);
406 if (*s == '\0')
407 return (0);
409 return (-1);
412 static int
413 parse_relative_ref(const char *s, struct iri *iri)
415 if ((s = parse_relative(s, iri)) == NULL)
416 return (-1);
418 if (*s == '?' && (s = parse_query(s + 1, iri)) == NULL)
419 return (-1);
421 if (*s == '#' && (s = parse_fragment(s + 1, iri)) == NULL)
422 return (-1);
424 if (*s == '\0')
425 return (0);
427 return (-1);
430 static int
431 parse(const char *s, struct iri *iri)
433 iri->iri_flags = 0;
435 if (s == NULL)
436 return (0);
438 if (parse_uri(s, iri) == -1) {
439 iri->iri_flags = 0;
440 if (parse_relative_ref(s, iri) == -1)
441 return (-1);
444 return (0);
447 static inline void
448 lowerify(char *s)
450 for (; *s; ++s)
451 *s = tolower((unsigned char)*s);
454 static void
455 cpfields(struct iri *dest, const struct iri *src, int flags)
457 if (flags & IH_SCHEME) {
458 dest->iri_flags |= IH_SCHEME;
459 if (src->iri_flags & IH_SCHEME)
460 memcpy(dest->iri_scheme, src->iri_scheme,
461 sizeof(dest->iri_scheme));
462 lowerify(dest->iri_scheme);
464 if (flags & IH_UINFO) {
465 if (src->iri_flags & IH_UINFO) {
466 memcpy(dest->iri_uinfo, src->iri_uinfo,
467 sizeof(dest->iri_uinfo));
468 dest->iri_flags |= IH_UINFO;
471 if (flags & IH_HOST) {
472 dest->iri_flags |= IH_HOST;
473 if (src->iri_flags & IH_HOST)
474 memcpy(dest->iri_host, src->iri_host,
475 sizeof(dest->iri_host));
476 lowerify(dest->iri_host);
478 if (flags & IH_PORT) {
479 if (src->iri_flags & IH_PORT) {
480 dest->iri_port = src->iri_port;
481 dest->iri_flags |= IH_PORT;
484 if (flags & IH_PATH) {
485 dest->iri_flags |= IH_PATH;
486 if (src->iri_flags & IH_PATH)
487 memcpy(dest->iri_path, src->iri_path,
488 sizeof(dest->iri_path));
490 if (flags & IH_QUERY) {
491 if (src->iri_flags & IH_QUERY) {
492 dest->iri_flags |= IH_QUERY;
493 memcpy(dest->iri_query, src->iri_query,
494 sizeof(dest->iri_query));
497 if (flags & IH_FRAGMENT) {
498 if (src->iri_flags & IH_FRAGMENT) {
499 dest->iri_flags |= IH_FRAGMENT;
500 memcpy(dest->iri_fragment, src->iri_fragment,
501 sizeof(dest->iri_fragment));
506 static inline int
507 remove_dot_segments(struct iri *i)
509 char *p, *q, *buf;
510 ptrdiff_t bufsize;
512 buf = p = q = i->iri_path;
513 bufsize = sizeof(i->iri_path);
514 while (*p && (q - buf < bufsize)) {
515 if (p[0] == '/' && p[1] == '.' &&
516 (p[2] == '/' || p[2] == '\0')) {
517 p += 2;
518 if (*p != '/')
519 *q++ = '/';
520 } else if (p[0] == '/' && p[1] == '.' && p[2] == '.' &&
521 (p[3] == '/' || p[3] == '\0')) {
522 p += 3;
523 while (q > buf && *--q != '/')
524 continue;
525 if (*p != '/' && (q > buf && q[-1] != '/'))
526 *q++ = '/';
527 } else
528 *q++ = *p++;
530 if ((*p == '\0') && (q - buf < bufsize)) {
531 *q = '\0';
532 return (0);
535 errno = ENAMETOOLONG;
536 return (-1);
539 static inline int
540 mergepath(struct iri *i, struct iri *base, struct iri *r)
542 const char *bpath, *rpath, *s;
544 bpath = (base->iri_flags & IH_PATH) ? base->iri_path : "/";
545 rpath = (r->iri_flags & IH_PATH) ? r->iri_path : "/";
547 i->iri_flags |= IH_PATH;
548 i->iri_path[0] = '\0';
550 if ((base->iri_flags & IH_AUTHORITY) &&
551 (*bpath == '\0' || !strcmp(bpath, "/"))) {
552 if (*rpath == '/')
553 rpath++;
554 strlcpy(i->iri_path, "/", sizeof(i->iri_path));
555 strlcat(i->iri_path, rpath, sizeof(i->iri_path));
556 return (0);
559 if ((s = strrchr(bpath, '/')) != NULL) {
560 cpstr(bpath, s + 1, i->iri_path, sizeof(i->iri_path));
561 if (*rpath == '/')
562 rpath++;
564 if (strlcat(i->iri_path, rpath, sizeof(i->iri_path)) >=
565 sizeof(i->iri_path)) {
566 errno = ENAMETOOLONG;
567 return (-1);
570 return (0);
573 int
574 iri_parse(const char *base, const char *str, struct iri *iri)
576 static struct iri ibase, iparsed;
578 memset(iri, 0, sizeof(*iri));
580 if (base == NULL) {
581 ibase.iri_flags = 0;
582 if (parse_uri(str, &iparsed) == -1) {
583 errno = EINVAL;
584 return (-1);
586 } else {
587 if (parse_uri(base, &ibase) == -1 ||
588 parse(str, &iparsed) == -1) {
589 errno = EINVAL;
590 return (-1);
594 cpfields(iri, &iparsed, IH_FRAGMENT);
596 if (iparsed.iri_flags & IH_SCHEME) {
597 cpfields(iri, &iparsed, iparsed.iri_flags);
598 remove_dot_segments(iri);
599 return (0);
602 cpfields(iri, &ibase, IH_SCHEME);
604 if (iparsed.iri_flags & IH_HOST) {
605 cpfields(iri, &iparsed, IH_AUTHORITY|IH_PATH|IH_QUERY);
606 remove_dot_segments(iri);
607 return (0);
610 cpfields(iri, &ibase, IH_AUTHORITY);
612 if ((iparsed.iri_flags & IH_PATH) && *iparsed.iri_path == '\0') {
613 cpfields(iri, &ibase, IH_PATH);
614 if (iparsed.iri_flags & IH_QUERY)
615 cpfields(iri, &iparsed, IH_QUERY);
616 else
617 cpfields(iri, &ibase, IH_QUERY);
618 return (0);
621 cpfields(iri, &iparsed, IH_QUERY);
622 if ((iparsed.iri_flags & IH_PATH) && *iparsed.iri_path == '/')
623 cpfields(iri, &iparsed, IH_PATH);
624 else {
625 if (!(ibase.iri_flags & IH_PATH))
626 ibase.iri_path[0] = '\0';
627 if (!(iparsed.iri_flags & IH_PATH))
628 iparsed.iri_path[0] = '\0';
629 if (mergepath(iri, &ibase, &iparsed) == -1)
630 return (-1);
632 if (remove_dot_segments(iri) == -1)
633 return (-1);
634 return (0);
637 int
638 iri_unparse(const struct iri *i, char *buf, size_t buflen)
640 if (buflen == 0)
641 goto err;
643 /* TODO: should %enc octets if needed */
645 buf[0] = '\0';
647 if (i->iri_flags & IH_SCHEME) {
648 if (strlcat(buf, i->iri_scheme, buflen) >= buflen ||
649 strlcat(buf, ":", buflen) >= buflen)
650 goto err;
653 if (i->iri_flags & IH_AUTHORITY) {
654 if (strlcat(buf, "//", buflen) >= buflen)
655 goto err;
658 if (i->iri_flags & IH_UINFO) {
659 if (strlcat(buf, i->iri_uinfo, buflen) >= buflen ||
660 strlcat(buf, "@", buflen) >= buflen)
661 goto err;
663 if (i->iri_flags & IH_HOST) {
664 if (strlcat(buf, i->iri_host, buflen) >= buflen)
665 goto err;
667 if (i->iri_flags & IH_PORT) {
668 if (strlcat(buf, ":", buflen) >= buflen ||
669 strlcat(buf, i->iri_portstr, buflen) >= buflen)
670 goto err;
673 if (i->iri_flags & IH_PATH) {
674 if (i->iri_flags & IH_AUTHORITY &&
675 i->iri_path[0] != '/' &&
676 strlcat(buf, "/", buflen) >= buflen)
677 goto err;
678 if (strlcat(buf, i->iri_path, buflen) >= buflen)
679 goto err;
682 if (i->iri_flags & IH_QUERY) {
683 if (strlcat(buf, "?", buflen) >= buflen ||
684 strlcat(buf, i->iri_query, buflen) >= buflen)
685 goto err;
688 if (i->iri_flags & IH_FRAGMENT) {
689 if (strlcat(buf, "#", buflen) >= buflen ||
690 strlcat(buf, i->iri_fragment, buflen) >= buflen)
691 goto err;
694 return (0);
696 err:
697 errno = ENOBUFS;
698 return (-1);
701 int
702 iri_human(const struct iri *iri, char *buf, size_t buflen)
704 memset(buf, 0, buflen);
705 return (-1);
708 int
709 iri_setquery(struct iri *iri, const char *text)
711 return (-1);